Chapter 2

Mycosis fungoides: disease evolution and

prognosis of 309 Dutch patients

Arch Dermatol. 2000 Apr; 136(4): 504-10

 STUDY

Mycosis Fungoides
Disease Evolution and Prognosis of 309 Dutch Patients
Remco van Doorn, MD; Christian W. Van Haselen, MD; Pieter C. van Voorst Vader, MD;
Marie-Louise Geerts, MD; Freerk Heule, MD; Menno de Rie, MD; Peter M. Steijlen, MD;
Sybren K. Dekker, MD; Willem A. van Vloten, MD; Rein Willemze, MD

Objectives: To determine the disease course of Dutch
patients with mycosis fungoides and to define factors re-
lated to disease progression and survival.
Design: A multicenter, 13-year, retrospective cohort
analysis.
Setting: Eight dermatology departments collaborating
in the Dutch Cutaneous Lymphoma Group.
Patients: Three hundred nine patients with mycosis fun-
goides registered between October 1985 and May 1997, in-
cluding 89 patients with limited patches or plaques (stage
Ia), 135 with generalized patches or plaques (stage Ib), 46
with skin tumors (stage Ic), 18 with enlarged but uninvolved
lymph nodes (stage II), 18 with lymph node involvement
(stage III), and 3 with visceral involvement (stage IV).
Main Outcome Measures: Response to initial treat-
ment, sustained complete remission, actuarial disease pro-
gression, and overall and disease-specific survival per clini-
cal stage.
Results: The median follow-up was 62 months (range,
1-113 months). For the entire group, the actuarial over-
all and disease-specific survival was 80% and 89% at 5
years, and 57% and 75% at 10 years, respectively. The
actuarial 5-year disease-specific survival of patients
with stage Ia, Ib, and Ic disease was 100%, 96%, and
80%, respectively, and only 40% for patients with stage
III disease. Using multivariate analysis, the presence of
extracutaneous disease, the type and extent of skin in-
volvement, the response to initial treatment, and the
presence of follicular mucinosis were independently as-
sociated with higher disease progression and mortality
rates. The calculated risks of disease progression at 5
and 10 years gradually increased from 4% to 10% for
those with stage Ia disease, from 21% to 39% for those
with stage Ib disease, and from 32% to 60% for those
with stage Ic disease; for those with stage III disease, the
risk remained at 70% at 5 and 10 years. The overall risk
of disease progression at 5 and 10 years was 24% and
38%, respectively, for the total study group.
Conclusion: At least within the first 10 years after di-
agnosis, disease progression and mycosis fungoides–
related mortality occur in only a subset of patients gen-
erally presenting with advanced disease.
Arch Dermatol. 2000;136:504-510

From the Departments of
Dermatology, Free University
Hospital (Drs van Doorn and
Willemze) and Academic
Medical Center (Dr de Rie),
Amsterdam, University Medical
Center, Utrecht (Drs Van
Haselen and van Vloten),
University Hospital Groningen,
Groningen (Dr van Voorst
Vader), University Hospital
Rotterdam, Rotterdam
(Dr Heule), University of
Nijmegen, Nijmegen
(Dr Steijlen), and Leiden
University Medical Center,
Leiden (Drs Dekker and
Willemze), the Netherlands; and
University Hospital Gent, Gent,
Belgium (Dr Geerts).
Dr Willemze is now with the
Department of Dermatology,
Leiden University Medical
Center, Leiden, the Netherlands.
M
YCOSIS fungoides (MF)
is the most common
type of cutaneous T-
cell lymphoma (CTCL),
with an estimated in-
cidence of 0.5 per 100 000 per year in the
1
western world. According to the major
textbooks, MF is an indolent type of CTCL
that slowly evolves through patch, plaque,
and tumor stages before lymph nodes and
visceral organs become involved, and ul-
timately a rapidly progressive and fatal dis-
ease develops. Long-term follow-up stud-
ies2-14 on large groups of patients with MF
are rare, and most come from a few US-
based centers. Comparison of published
series is often difficult, because of differ-
ent inclusion criteria (eg, inclusion or ex-
clusion of patients with large-plaque para-
psoriasis) and an inconsistent use of the
terms MF and CTCL. Several studies, in-
cluding one of the largest European stud-
ies of 92 Dutch patients published by Ham-
minga et al4 in 1982, included not only
patients with classic MF but also patients
with Sézary syndrome and other types of
CTCL defined more recently. For in-
stance, patients with CTCL presenting with
tumors with the histological appearance
of a diffuse, large, T-cell lymphoma and
without prior or concurrent patches or
plaques were designated previously as hav-
ing “MF d’emblée,” whereas such pa-
tients are now classified as having either
CD30+ or CD30− large-cell CTCL, which
are considered distinct disease entities sepa-
rate from MF.15,16
It is well recognized that the evolu-
tion from skin-limited to widespread dis-
seminated disease in patients with MF may
take years or even decades. However, clini-
cal experience also suggests that only a pro-
portion of patients with MF presenting with
only skin lesions will develop extracuta-

ARCH DERMATOL / VOL 136, APR 2000 WWW.ARCHDERMATOL.COM

504

33
©2000 American Medical Association. All rights reserved.
 PATIENTS AND METHODS
Between October 1985 and May 1997, 345 patients with MF
were included in the registry of the Dutch Cutaneous Lym-
phoma Group, and follow-up data had been collected yearly.
For the present study, only patients with clinical and histo-
logical features consistent with MF, and who underwent a
follow-up period of at least 12 months after histological con-
firmation of the diagnosis, unless death due to MF oc-
curred earlier, were selected. Thirty-six patients were ex-
cluded because of insufficient clinical information or follow-
up. The final study group comprised 309 patients. In each
patient, the diagnosis had been made by an expert panel of
dermatologists and pathologists at 1 of the quarterly meet-
ings of the Dutch Cutaneous Lymphoma Group. The his-
tological criteria for the diagnosis of early MF are essen-
tially the same as those described by Nickoloff,17 and require
the presence of hyperchromatic, slightly to markedly atypi-
cal lymphoid cells in the epidermis, either as single, often
haloed, cells, or in a linear configuration at the dermal-
epidermal junction. Patients with patches or plaques clini-
cally suspected, but histologically not diagnostic of MF, are
included as a separate category in the Dutch registry and were
not included in the present study, unless at a later point a
definite diagnosis of MF was made. The time of the first di-
agnostic biopsy was taken as the time of diagnosis. The study
group did not include patients with Sézary syndrome, pag-
etoid reticulosis, or other CTCL, recognized as distinct en-
tities in the European Organization for Research and Treat-
ment of Cancer classification for primary cutaneous
lymphomas.16 However, to allow comparison with other pub-
lished series, patients with MF-associated follicular muci-
nosis, included as a distinct variant of MF in the European
Organization for Research and Treatment of Cancer classi-
fication, were included. Association with follicular mucino-
sis was included as one of the variables in univariate and mul-
tivariate analyses of survival and disease progression.
The stage of the disease was determined based on the
type and extent of skin involvement and the presence of
lymph node, visceral, or blood involvement according to
a modification of the Fuks classification scheme,4,18 which
can easily be translated into the TNM system19 (Table 1).
Staging evaluation consisted of obtaining a complete medi-
cal history and a complete blood cell count and perform-
ing a physical examination, serum chemistry studies, and
a skin biopsy. In the presence of lymphadenopathy, a lymph
node biopsyand thoracic and abdominal computed tomo-
graphic scans were performed, and a chest x-ray film was
obtained. Lymph node involvement was assessed through-
out the study (1985-1997) with the same criteria, de-
neous and ultimately fatal disease. Whereas previous stud-
ies mainly focused on prognostic variables and survival
data in the different stages of MF, data regarding the fre-
quency of disease progression have been published only
recently.10-13 Obviously, clinical information to patients with
MF should not only include the message that disease pro-
gression may occur but also how often and after which pe-
riod such a development can be expected.
In the present study, clinical and follow-up data of all
patients with MF included between October 1985 and May
1997 in the registry of the Dutch Cutaneous Lymphoma
ARCH DERMATOL / VOL 136, APR 2000
505
©2000 American Medical Association. All rights reserved.
34
scribed previously.20 When indicated clinically, addi-
tional staging studies to determine visceral involvement were
performed. The following variables were recorded: age; sex;
clinical stage at the time of diagnosis; duration of skin le-
sions before diagnosis; type of initial therapy; whether there
was complete remission after initial therapy; disease course
after initial therapy; the date of disease progression, if ap-
plicable; and the date of last contact and cause of death, if
applicable. In addition, the presence of follicular mucino-
sis in the first diagnostic biopsy specimen and the pres-
ence of lymphomatoid papulosis or B-cell neoplasms prior
to, concurrent with, or following the development of MF
were recorded. Complete remission on initial treatment was
defined as complete disappearance of all skin lesions. In
most cases, histological confirmation of complete remis-
sion was not obtained. No distinction was made between
partial or no responses on initial therapy. For clinical course,
distinction was made between sustained complete remis-
sion, defined as the total disappearance of all (extra) cuta-
neous lesions after initial therapy without subsequent
relapse (without maintenance treatment); continued dis-
ease, disease without progression to a higher clinical stage;
and disease progression, the development of skin tumors in
patients with a previous patch or plaque (stage Ia-Ib), the
development of histologically documented nodal involve-
ment (stage III) in patients with previous skin-limited dis-
ease, the development of visceral involvement in patients
with prior skin and/or lymph node involvement, and death
due to MF.
As indicators of survival, disease-specific survival, in-
cluding only death related to MF as the event, and overall
survival, including death due to any cause as the event, were
investigated. Actuarial survival and disease progression curves
were calculated from the date of diagnosis to the date of death
or last contact and the date of disease progression, respec-
tively, using the Kaplan-Meier technique.21 Patients lost to
follow-up were considered to be censored at the time of last
contact. Differences between survival and disease progres-
sion rates were analyzed using the log-rank test. Compara-
tive analysis of groups of numerical variables was per-
formed using 2-tailed t tests. P.05 was considered
significant. Relative risks and 95% confidence intervals were
determined using standard methods. Univariate analysis of
possible prognostic factors was performed using the log-
rank test and Cox proportional hazards regression analysis.
Multivariate analysis was performed by entering significant
univariate variables for survival and disease progression in
Cox proportional hazards regression analysis22 to establish
their independence as prognostic factors. All analyses were
performed using Statistical Product and Services Solutions
software (SPSS Inc, Chicago, Ill).
Group were evaluated. This study determines disease-
specific and overall survival and the risk of disease pro-
gression for patients with different stages of MF and de-
fines variables predictive of survival and disease progression.
RESULTS
CLINICAL CHARACTERISTICS AT PRESENTATION
Of the 309 patients included in this study, 72.5% had ei-
ther stage Ia or Ib MF at the time of diagnosis, whereas
WWW.ARCHDERMATOL.COM
 Table 1. Clinical Stage of 309 Patients With MF at the Time of Diagnosis*

Patches and Plaques

Covering the Skin Surface
Skin
Stage 10% (a) 10% (b) Tumor (c) Erythroderma (d) Total
MF confined to the skin (I) 89 135 46 0 270
MF with dermatopathic lymphadenopathy (II) 0 9 7 2 18
MF with lymph node involvement (III) 1 7 8 2 18
MF with visceral involvement (IV) 0 0 3 0 3
Total 90 151 64 4 309

*The clinical stage is given according to the modified Fuks classification.4 Translation into the TNM classification18 is as follows; Ia, T1 N0 M0; Ib, T2 N0 M0;
Ic, T3 N0 M0; Id, T4 N0 M0; IIa through IId, T1 through T4 N1 M0; IIIa through IIId, T1 through T4 N3 M0; and IVa through IVd, T1 through T4 N0 through N3
M1. MF indicates mycosis fungoides.

Table 2. Patient Characteristics and Disease Outcome per Clinical Stage*

Stage

Variable Ia Ib Ic II III IV All

No. of patients† 89 (28.8) 135 (43.7) 46 (14.9) 18 (5.8) 18 (5.8) 3 (1.0) 309 (100)
Age at diagnosis, 58.0 (19-90) 61.0 (14-92) 67.5 (35-88) 62.5 (37-82) 57.5 (29-84) 67.0 (53-69) 61.0 (14-92)
median (range), y
Male-female ratio 59:30 83:52 29:17 10:8 13:5 2:1 196:113
Duration of skin 60 (1-372) 48 (1-600) 48 (2-600) 48 (10-648) 24 (5-300) 24 (10-72) 48 (1-648)
lesions before diagnosis,
median (range), mo
Complete remission 42 (47) 41 (30) 12 (26) 2 (11) 1 (6) 0 (0) 98 (32)
on initial therapy
Duration of follow-up, 71 (12-203) 70 (12-313) 48 (6-249) 45 (14-184) 49 (8-239) 2 (1-9) 62 (1-313)
median (range), mo
Disease course
Sustained CR 16 (18) 10 (7) 5 (11) 1 (6) 1 (6) 0 (0) 33 (11)
Continued disease 69 (78) 92 (68) 23 (50) 8 (44) 6 (33) 1 (33) 199 (64)
Progression 4 (4) 33 (24) 18 (39) 9 (50) 11 (61) 2 (67) 77 (25)
Risk of disease progression, %
At 5 y 4 21 32 65 70 100 24
At 10 y 10 39 60 65 70 ... 38
Current status
Alive without disease 39 (44) 29 (21) 5 (11) 1 (6) 1 (6) 0 (0) 75 (24)
Alive with disease 41 (46) 71 (53) 14 (30) 10 (56) 5 (28) 0 (0) 141 (46)
Died of other cause 7 (8) 22 (16) 12 (26) 4 (22) 1 (6) 1 (33) 45 (15)
Died of MF 2 (2) 13 (10) 15 (33) 3 (17) 11 (61) 2 (67) 47 (15)
Disease-specific survival, %
At 5 y 100 96 80 68 40 0 89
At 10 y 97 83 42 68 20 0 75
Overall survival, %
At 5 y 99 86 65 49 40 0 80
At 10 y 84 61 27 49 20 0 57

*Data are given as number (percentage) of patients within each column unless otherwise indicated. Percentages may not total 100 because of rounding.
The clinical stages are explained in Table 1. CR indicates complete remission; MF, mycosis fungoides; and ellipses, data not applicable.
†The row total was used to obtain the percentages.

14.9% presented with 1 or more skin tumors in addi-
tion to patches and plaques, but no evidence of extracu-
taneous disease. Of the 309 patients, 270 (87.4%) had
only skin lesions at the time of diagnosis; 18 (5.8%), en-
larged but histologically uninvolved lymph nodes; and
21 (6.8%), nodal and/or visceral involvement (Table 2).
Considering the entire group of patients, the age at
diagnosis varied between 14 and 92 years (median, 61
years). Only 2 patients (0.6%) were younger than 20 years
at the time of diagnosis. Patients presenting with stage
Ic MF were significantly older than patients with stage
Ia MF (67.5 vs 58.0 years; P.001). There was a male
predominance, with a male-female ratio of 196:113, which
is consistent with that of other large series.2,3,13 The du-
ration of skin lesions before a definite diagnosis could
be made varied between 1 month and more than 50 years
(median, 48 months) and was significantly shorter in the
21 patients presenting with extracutaneous disease (me-
dian, 24 months) compared with patients with only skin
lesions at presentation (median, 48 months) (P = .006).
Of the 309 patients with MF, 32 (10.4%) had asso-
ciated follicular mucinosis at the time of diagnosis. This

ARCH DERMATOL / VOL 136, APR 2000 WWW.ARCHDERMATOL.COM

506

©2000 American Medical Association. All rights reserved.

35
 Table 3. Initial Treatment per Clinical Stage*

Stage
All
Initial Treatment Ia (n = 89) Ib (n = 135) Ic (n = 46) II (n = 18) III (n = 18) IV (n = 3) (N = 309)
Topical corticosteroids 15 (17) 11 (8) 0 0 0 0 26 (8)
PUVA therapy 51 (57) 89 (66) 9 (20) 7 (39) 0 0 156 (51)
UV-B therapy 14 (16) 15 (11) 0 0 0 0 29 (9)
Topical mechlorethamine hydrochloride 6 (7) 10 (7) 4 (9) 0 1 (6) 0 21 (7)
Total skin electron beam irradiation 0 7 (5) 6 (13) 1 (6) 4 (22) 0 18 (6)
Radiotherapy with or without skin-directed therapy† 3 (3) 2 (2) 21 (46) 5 (28) 2 (11) 1 (33) 34 (11)
Polychemotherapy with or without skin-directed therapy 0 0 4 (9) 2 (11) 10 (56) 2 (67) 18 (6)
Other‡ 0 1 (1) 2 (4) 3 (17) 1 (6) 0 7 (2)
Complete remission on initial treatment, % 47 30 26 11 6 0 32

*Data are given as number (percentage) of patients unless otherwise indicated. Column percentages may not total 100 because of rounding. The clinical stages
are explained in the footnote to Table 1. PUVA indicates psoralen–UV-A.
†Skin-directed therapies include topical corticosteroids, PUVA therapy, UV-B therapy, topical mechlorethamine hydrochloride, radiotherapy, or total skin
electron beam irradiation.
‡Other therapies include monochemotherapy, retinoids, interferons, or combinations of these with skin-directed therapy.

group included 28 patients with stage I, 3 with stage II,
and 1 with stage IV MF. The combination of MF—stage
Ia in 2 and Ib in 6 patients—and lymphomatoid papu-
losis was noticed in 8 (2.6%) of the 309 patients. Asso-
ciated B-cell lymphoproliferations were documented in
5 patients with stage Ia or Ib MF, including 4 with B-cell
chronic lymphocytic leukemia and 1 with nodal follicu-
lar lymphoma.
TREATMENT AND FOLLOW-UP
The initial therapies at the different stages of MF are listed
in Table 3, and reflect the approach used in the treat-
ment of MF in the Netherlands. The treatment modality
most commonly used for stage Ia or Ib disease was pso-
ralen–UV-A therapy (140 [62.5%] of 224 cases); less
frequently used modalities included topical corticoste-
roids, UV-B therapy, topical mechlorethamine hydro-
chloride, and, in case of extensive skin lesions, total skin
electron beam irradiation. Patients with stage Ic disease
were treated similarly, often with additional local radio-
therapy for persistent tumors (Table 3). Systemic poly-
chemotherapy consisting of cyclophosphamide, vincris-
tine sulfate, doxorubicine, and prednisone was mainly
given to patients presenting with nodal (stage III) or vis-
ceral (stage IV) involvement, often in combination with
or followed by skin-directed therapies.
Initial treatment resulted in clinical complete re-
missions in 98 (31.7%) of 309 patients. However, in most
patients, these complete remissions were short-lived. Sus-
tained complete remissions on initial treatment were ob-
served in only 33 (10.7%) of the 309 patients, among
whom 26 had stage Ia or Ib disease. The disease-free sur-
vival in these 33 patients varied between 10 and 163
months (median, 68 months). In 199 (64.4%) of the 309
patients, there was continued disease without progres-
sion, typically having a fluctuating course, while in the
remaining 77 (24.9%), disease progression, including
death due to MF, occurred.
The median follow-up was 62 months (range, 1-313
months). During that period, 92 of the 309 patients died,
including 47 of MF. The disease-specific survival at 5 and
10 years for the whole group of 309 patients was 89%
and 75%, respectively; the 5- and 10-year overall sur-
vival was 80% and 57%, respectively. The survival rates
according to clinical stage are presented in Table 2,
Figure 1, and Figure 2. Consistent with prior re-
ports,23,24 patients with MF and lymphomatoid papulo-
sis had an excellent prognosis. None of these patients
showed disease progression after a median follow-up of
158 months (range, 23-244 months).
PROGNOSTIC VARIABLES
Univariate analysis of variables possibly influencing dis-
ease-specific survival in the entire group of 309 patients
showed that the following factors were statistically sig-
nificant: stage at diagnosis (P.001), including the pres-
ence of extracutaneous disease (P.001) and the type and
extent of skin involvement (P.001); no complete re-
mission on initial treatment (P.001); associated fol-
licular mucinosis (P = .005); and older age (P = .01). Sex
(P = .69) and duration of skin lesions before diagnosis
(P = .34) were not significantly related to survival, when
the total group was considered.
Univariate analysis of the prognostic variables per
clinical stage showed that only complete remission on
initial treatment within the group of patients with stage
Ib MF was significantly related to survival (P = .04)
(Figure 3). Multivariate analysis revealed that—in or-
der of predictive value—presence of extracutaneous dis-
ease, type and extent of skin involvement, no complete
response to initial treatment, and presence of follicular
mucinosis were independently associated with MF-
related mortality. The relative risks for MF-related mor-
tality are presented in Table 4.
Regarding clinical stage, patients with stage Ia and
stage Ib MF had a significantly better survival than pa-
tients with stage Ic MF (P.001). However, no signifi-
cant difference in survival was found between patients
with stage Ia and stage Ib disease (P = .11). Notably, not
only patients with histologically documented lymph node
involvement (stage III) but also patients with enlarged,
but histologically uninvolved, lymph nodes (stage II) had

ARCH DERMATOL / VOL 136, APR 2000 WWW.ARCHDERMATOL.COM

507

©2000 American Medical Association. All rights reserved.

36
 100 100

80 80

60 Patients With Stage I MF (n = 270) 60

Survival, %

Survival, %
Patients With Stage II MF (n = 18)
Patients With Stage III MF (n = 18)
Patients With Stage IV MF (n = 3)
40 40

Patients With Complete

20 20 Remission (n = 41)
Patients Without Complete
Remission (n = 94)

0 60 120 180 0 60 120 180 240

Follow-up Duration, mo Follow-up Duration, mo

Figure 1. Actuarial disease-specific survival of 309 patients with mycosis Figure 3. Actuarial disease-specific survival of patients with stage Ib
fungoides (MF). The differences between the patients with each stage of MF mycosis fungoides with or without complete remission after initial treatment.
are as follows: stage I vs stage II, P = .02; stage I vs stage III, P.001; stage The difference between those with complete remission vs those without
II vs stage III, P = .13; and stage III vs stage IV, P.001. complete remission was significant ( P = .05).

100
Table 4. Factors Independently Influencing
Disease-Specific Survival*
80
Relative 95% Confidence
Factor Risk Interval P
60
Survival, %

Extracutaneous involvement
I (absent)† 1.0 ... ...
40
II (LN enlargement) 3.3 1.1-9.5 .02
III (LN involvement) 7.3 3.6-14.8 .001
IV (visceral involvement) 227.0 31.3-1646.3 .001
20 Patients With Stage Ia MF (n = 89) Type and extent of skin lesions
Patients With Stage Ib MF (n = 135) Ia (limited plaques)† 1.0 ... ...
Patients With Stage Ic MF (n = 46)
Ib (generalized plaques) 3.2 0.7-14.4 .11
Ic (skin tumors) 20.9 4.7-92.0 .001
0 60 120 180 Complete remission after
Follow-up Duration, mo initial treatment
Yes† 1.0 ... ...
Figure 2. Actuarial disease-specific survival of 270 patients with stage I No 7.0 2.2-22.5 .001
mycosis fungoides (MF). The differences between the patients with each
Follicular mucinosis
variation of stage I MF are as follows: stage Ia vs stage Ib, P = .11; stage Ia
vs stage Ic, P.001; and stage Ib vs stage Ic, P.001. Absent† 1.0 ... ...
Present 2.3 1.1-5.0 .001

*Factors are presented in order of predictive value. The relative risk of

a significantly lower survival compared with patients with
stage I MF (P.001 and P = .02, respectively).
DISEASE PROGRESSION
One of the goals of this study was to assess the risk of
disease progression, including death due to MF, for pa-
tients with different stages of MF. The calculated risks
of disease progression at 5 and 10 years gradually in-
creased from 4% to 10% for those with stage Ia disease,
from 21% to 39% for those with stage Ib disease, and from
32% to 60% for those with stage Ic disease; for those with
stage III disease, the risk remained at 70% at 5 and 10
years (Table 2). Table 5 shows the actual frequency of
disease progression in the group of 309 patients after a
median follow-up of 62 months. It also shows that the
higher the clinical stage at diagnosis, the shorter the du-
ration until disease progression.
Multivariate analysis revealed that—as for disease-
specific survival—the presence of extracutaneous dis-
ease, the type and extent of skin involvement, the re-
different types and extents of skin lesions has been calculated for stage I
only. LN indicates lymph node; ellipses, data not applicable.
†Referent value.
sponse to initial treatment, and the presence of follicular
mucinosis were independently associated with disease
progression.
COMMENT
In the present study, the main clinical characteristics, dis-
ease evolution, and survival of 309 Dutch patients with
MF, included in the Dutch registry for cutaneous lym-
phomas between October 1985 and May 1997, were evalu-
ated. In addition, prognostic variables and the risk of dis-
ease progression for those with different stages of MF were
analyzed. The median age at diagnosis of 61 years and
the male-female ratio of 196:113 were similar to those
given in the few other large series studied.2,3,13 At the time
of first presentation, 93.2% of the patients with MF had
only skin lesions, including 5.8% with concurrent en-

ARCH DERMATOL / VOL 136, APR 2000 WWW.ARCHDERMATOL.COM

508

©2000 American Medical Association. All rights reserved.

37
 Table 5. Actual Disease Progression in 309 Patients With MF After a Median Follow-up of 62 Months*
Lymph Node
Stage at Diagnosis Skin Tumors Involvement
Ia (n = 89) 3 (3) 2 (2)
Ib (n = 135) 28 (21) 19 (14)
Ic (n = 46) 0 11 (24)
II (n = 18) 3 (17) 5 (28)
III (n = 18) 2 (11) 0 3 (17)
IV (n = 3) 0 0
Total (N = 309) 36 (12) 37 (12)
*Data are given as number (percentage) of patients unless otherwise indicated. The clinical stages are explained in the footnote to Table 1.
MF indicates mycosis fungoides.
†Total number (percentage) of patients showing disease progression.
‡Data are given as the median (range).
larged but histologically uninvolved lymph nodes, whereas
only 6.8% presented with concurrent nodal or visceral
involvement.
In the Netherlands, patients with MF are treated tra-
ditionally with skin-directed therapies, including topi-
cal corticosteroids, psoralen–UV-A therapy, UV-B therapy,
or topical mechlorethamine hydrochloride, and addi-
tional radiotherapy in case of concurrent skin tumors,
whereas multiagent chemotherapy is generally only used
in patients with extracutaneous localizations. Initial treat-
ment according to this classic approach resulted in a com-
plete remission in almost one third of the patients (98
of the 309 patients). As expected, in most patients, this
complete remission was short-lived. However, in 33 (34%)
of the 98 patients achieving complete remission on ini-
tial treatment, and not receiving any type of mainte-
nance treatment, there was no subsequent relapse. Eigh-
teen of these 33 patients, including 9 with stage Ia, 5 with
stage Ib, 3 with stage Ic, and 1 with stage IIb MF, have
been in complete remission for more than 5 years, and,
since most relapses occur within 5 years after achieving
complete remission,10,13 may be considered as potential
cures. The present retrospective study does not allow a
representative comparison of the effects of the different
treatment modalities, since treatment selection may have
been affected by disease severity. It is, therefore, not sur-
prising that we did not find a relation between the re-
sults of initial treatment and the type of treatment given
(data not shown). The complete response of only 10
(55.6%) of the 18 patients to total skin electron beam ir-
radiation at initial therapy may be explained by the fact
that 4 of 18 patients had stage III disease, and 8 of the
14 remaining patients had MF-associated follicular mu-
cinosis, a combination known to be rather refractory to
total skin electron beam irradiation.25
The disease-related and overall survival for the
whole group of 309 patients was 89% and 80% at 5
years, and 75% and 57% at 10 years, respectively. For
the survival rates for the different stages of MF, the over-
all and disease-specific survival rates at 5 and 10 years
for patients with stage Ia and Ib MF (Table 2) were simi-
lar to those reported in previous studies.4,5,10-14 In this
study, we did find a difference in survival between those
with stage Ia and those with stage Ib MF, but this did
not reach statistical significance. The fact that disease
ARCH DERMATOL / VOL 136, APR 2000
©2000 American Medical Association. All rights reserved.
Visceral Death Due Duration Until
Involvement to MF Progression† Disease Progression, mo‡
1 (1) 2 (2) 4 (4) 73 (49-96)
5 (4) 13 (10) 33 (24) 45 (8-109)
7 (15) 15 (33) 18 (39) 37 (1-242)
2 (11) 4 (22) 9 (50) 24 (14-51)
11 (61) 11 (61) 32 (8-56)
0 2 (67) 2 (67) 6 (2-9)
18 (6) 47 (15) 77 (25) 40 (1-242)
progression was much more frequent in patients with
stage Ib MF than in those with stage Ia MF suggests that
with longer follow-up, the difference in survival might
become statistically significant.
Our study did not include patients with large-
plaque parapsoriasis, characterized by the presence of
patches or slightly infiltrated plaques but with histologi-
cal changes not diagnostic of MF. While we consider such
lesions as a potential precursor of MF, other groups26 have
expressed the view that large-plaque parapsoriasis, and
even small-plaque parapsoriasis, should be considered
as MF, and not as precursors of MF. However, in view
of the excellent prognosis of patients with stage Ia MF—
similar to that of a race-, age-, and sex-matched control
population10—and the low tendency to progress, it is ques-
tionable which patient with small-plaque parapsoriasis
will benefit from being marked as a patient with CTCL.
Because of this ongoing controversy and the inconsis-
tent use of the terms large-plaque parapsoriasis and small-
plaque parapsoriasis, it is perhaps better to abandon these
terms. The only question that really matters is the fol-
lowing: is it MF or is it not MF?
Tumor stage MF is often associated with a poor prog-
nosis. However, this and other studies11,12,14 clearly indi-
cate that patients with skin tumors without concurrent
extracutaneous disease (stage Ic) still have a disease-
related 5-year survival of approximately 70% to 80%.
Whether patients with enlarged, but histologically unin-
volved, lymph nodes (stage II) have a more unfavorable
prognosis compared with patients without clinically en-
larged lymph nodes is a matter of controversy.4,11,13 In the
present study, patients with stage II disease had a signifi-
cantly lower survival rate than patients with stage I MF
(P.001). Previous studies27 suggested that T-cell recep-
tor gene rearrangement analysis is a more accurate and
objective method of diagnosing early lymph node in-
volvement in patients with MF than routine histological
features. However, in prior studies,28 which included
lymph nodes of 7 of the 18 patients with stage II MF in-
cluded in this study, no clonal T-cell populations were
found in any of the dermatopathic, but noninvolved,
lymph nodes. There is, therefore, no evidence that the
lymph nodes of these patients with stage II MF were ac-
tually involved, which leaves the more unfavorable prog-
nosis of this group unexplained.
WWW.ARCHDERMATOL.COM
509
38
 In the literature, the following prognostic variables
have been described: stage of disease,3-5,11-14 age,2,10,12,13
race,2 response to initial treatment,3,10,13 and prior ma-
lignant neoplasm.2 In the present study, stage at diagno-
sis (ie, the presence of extracutaneous disease and the
type of skin involvement), complete remission after ini-
tial treatment, and the presence of follicular mucinosis
proved independently predictive of disease-specific sur-
vival and disease progression.
In the group of 32 patients with MF-associated fol-
licular mucinosis, disease progression occurred more
often and disease-specific survival rates were signifi-
cantly lower than in the 277 patients without follicular
mucinosis. In the 32 patients with follicular mucinosis,
disease progression was estimated to occur in 89%
within 10 years after diagnosis vs 32% in the 277 pa-
tients without follicular mucinosis. The disease-related
survival at 5 and 10 years was 81% and 36%, and the
overall survival 75% and 21%, respectively. A more de-
tailed clinical and histological analysis of a group of 40
patients with MF-associated follicular mucinosis will be
published separately.
Older patients had significantly lower disease-
specific survival rates and higher disease progression rates.
However, older age was associated with more advanced
stages of MF, and it appeared not to be an independent
prognostic factor.
Mycosis fungoides is generally depicted as a malig-
nant disease that slowly evolves through patch, plaque,
and tumor stages, and ultimately may develop into an
extracutaneous and generally fatal disease. Because of the
increased accessibility of medical literature through In-
ternet services, we are confronted more frequently with
patients with newly diagnosed MF who have come to be-
lieve that this sequence of events invariably takes place.
On the other hand, clinical experience suggests that many
patients with MF, in particular those with stage Ia and
perhaps also many with stage Ib disease, may have stable
disease for decades, and that only a proportion of pa-
tients with MF progresses and will develop extracutane-
ous disease. Consistently, the results of this and other
recent studies10,12,13 indicate that the risk of disease pro-
gression within the first 10 years after diagnosis is about
5% to 10% for patients with stage Ia and between 17%
and 39% for patients with stage Ib disease. In patients
with more advanced stages of MF, the risk of disease pro-
gression was higher and the duration until progression
shorter (Tables 2 and 5). These results confirm the clini-
cal impression that, at least within the first 10 years af-
ter diagnosis, disease progression occurs in only a few
patients. Further studies are warranted to elucidate the
risk factors associated with disease progression within
these early stages of MF.
Accepted for publication October 26, 1999.
We thank P. D. Bezemer, PhD, P. J. Kostense, PhD, and
J. J. Oudejans, MD, for their statistical advice.
Reprints: Rein Willemze, MD, Department of Derma-
tology, B1-Q-93, Leiden University Medical Center, PO
Box 9600, 2300 RC Leiden, the Netherlands (e-mail:
willemze.dermatology@lumc.nl).
ARCH DERMATOL / VOL 136, APR 2000
510
©2000 American Medical Association. All rights reserved.
39
REFERENCES
1. Weinstock M, Horm J. Mycosis fungoides in the United States: increasing inci-
dence and descriptive epidemiology. JAMA. 1988;260:42-46.
2. Weinstock M, Horm J. Population-based estimate of survival and determinants
of prognosis in patients with mycosis fungoides. Cancer. 1988;62:1658-1661.
3. Lamberg S, Green S, Byar D, et al. Status report of 376 mycosis fungoides pa-
tients at 4 years: Mycosis Fungoides Cooperative Group. Cancer Treat Rep. 1979;
63:701-707.
4. Hamminga L, Hermans J, Noordijk E, Meijer C, Scheffer E, van Vloten W. Cuta-
neous T-cell lymphoma: clinicopathological relationships, therapy and survival
in ninety-two patients. Br J Dermatol. 1982;107:145-155.
5. Sausville E, Eddy J, Makuch R, et al. Histopathological staging at initial diagnosis
of mycosis fungoides and the Sézary syndrome: definition of three distinctive prog-
nostic groups. Ann Intern Med. 1988;109:372-382.
6. Hoppe R, Wood G, Abel E. Mycosis fungoides and the Sézary syndrome: pathol-
ogy, staging and treatment. Curr Probl Cancer. 1990;14:293-371.
7. Hermann J, Roenigk H, Hurria A, et al. Treatment of mycosis fungoides with
photochemotherapy (PUVA): long-term follow-up. J Am Acad Dermatol. 1995;
33:234-242.
8. Zackheim H. Topical carmustine (BCNU) for patch/plaque mycosis fungoides.
Semin Dermatol. 1994;13:202-206.
9. Quiros P, Kacinski B, Wilson L. Extent of skin involvement as a prognostic indi-
cator of disease free survival and overall survival of patients with T3 cutaneous
T-cell lymphoma treated with total skin electron beam radiation therapy. Can-
cer. 1996;77:1912-1917.
10. Kim Y, Jensen R, Watanabe G, Varghese A, Hoppe R. Clinical stage IA (limited
patch and plaque) mycosis fungoides. Arch Dermatol. 1996;132:1309-1313.
11. Toro J, Stoll H, Stomper P, Oseroff A. Prognostic factors and evaluation of mycosis
fungoides and Sézary syndrome. J Am Acad Dermatol. 1997;37:58-67.
12. Vonderheid E, Ekbote S, Kerrigan K, et al. The prognostic significance of delayed
hypersensitivity to dinitrochlorobenzene and mechlorethamine hydrochloride in
cutaneous T cell lymphoma. J Invest Dermatol. 1998;110:946-950.
13. Kim Y, Chow S, Varghese A, Hoppe R. Clinical characteristics and long-term
outcome of patients with generalized patch and/or plaque (T2) mycosis fungoi-
des. Arch Dermatol. 1999;135:26-32.
14. Zackheim H, Amin S, Kashani-Sabet M, McMillan A. Prognosis in cutaneous T-
cell lymphoma by skin stage: long-term survival in 489 patients. J Am Acad Der-
matol. 1999;40:418-425.
15. Beljaards R, Meijer C, Scheffer E, et al. Prognostic significance of CD30 (Ki-1/
Ber-H2) expression in primary cutaneous large-cell lymphomas of T-cell origin:
a clinico-pathological and histochemical study in 20 patients. Am J Pathol. 1989;
135:1169-1178.
16. Willemze R, Kerl H, Sterry W, et al. EORTC classification for primary cutaneous
lymphomas: a proposal from the Cutaneous Lymphoma Study Group of the Eu-
ropean Organization for Research and Treatment of Cancer. Blood. 1997;90:
354-371.
17. Nickoloff BJ. Light-microscopic assessment of 100 patients with patch/plaque
stage mycosis fungoides. Am J Dermatopathol. 1988;10:469-477.
18. Fuks Z, Bagshaw M, Farber E. Prognostic signs and the management of the my-
cosis fungoides. Cancer. 1973;32:1385-1395.
19. Bunn P, Lamberg S. Report of the Committee on Staging and Classification of
Cutaneous T-Cell Lymphomas. Cancer Treat Rep. 1979;63:725-728.
20. Scheffer E, Meijer C, Van Vloten W. Dermatopathic lymphadenopathy and lymph
node involvement in mycosis fungoides. Cancer. 1980;45:137-148.
21. Kaplan E, Meier P. Nonparametric estimation from incomplete observations.
J Am Stat Assoc. 1958;53:475-480.
22. Cox D, Snell E. Analysis of Binary Data. New York, NY: Chapman & Hall; 1989.
23. Beljaards R, Willemze R. The prognosis of patients with lymphomatoid papulosis
associated with other types of malignancies. Br J Dermatol. 1992;126:596-602.
24. Basarab T, Fraser-Andrews E, Orchard G, et al. Lymphomatoid papulosis in as-
sociation with mycosis fungoides: a study of 15 cases. Br J Dermatol. 1998;139:
630-638.
25. Wilson L, Cooper D, Goodrich A, et al. Impact of non-CTL dermatologic diagno-
sis on cutaneous T-cell lymphoma patients treated with total skin electron beam
radiation therapy. Int J Radiat Oncol Biol Phys. 1994;28:829-837.
26. King-Ismael D, Ackerman AB. Guttate parapsoriasis/digitate dermatosis (small plaque
parapsoriasis) is mycosis fungoides. Am J Dermatopathol. 1992;14:518-530.
27. Weiss LM, Hu E, Wood GS, et al. Clonal rearrangements of T-cell receptor genes
in mycosis fungoides and dermatopathic lymphadenopathy. N Engl J Med. 1985;
313:539-544.
28. Bakels V, van Oostveen JW, Geerts ML, et al. Diagnostic and prognostic signifi-
cance of clonal T-cell receptor beta gene rearrangements in lymph nodes of pa-
tients with mycosis fungoides. J Pathol. 1993;170:249-255.
WWW.ARCHDERMATOL.COM
40
 Chapter 3

Follicular mycosis fungoides, a distinct disease entity

with or without associated follicular mucinosis:
a clinicopathologic and follow-up study of 51 patients

Arch Dermatol. 2002 Feb: 138(2):191-8

41
 STUDY

Follicular Mycosis Fungoides, a Distinct Disease

Entity With or Without Associated Follicular Mucinosis
A Clinicopathologic and Follow-up Study of 51 Patients
Remco van Doorn, MD; Erik Scheffer, MD; Rein Willemze, MD; for the Dutch Cutaneous Lymphoma Group

Objective: To determine the clinicopathologic fea-
tures and the disease course of patients with follicular
mycosis fungoides (MF).
Design: A multicenter, 14-year, retrospective cohort
analysis.
Setting: Dutch Cutaneous Lymphoma Group.
Patients: Fifty-one patients with the clinicopathologic
features of follicular MF with (n=49) or without (n=2)
associated follicular mucinosis. Follow-up data were com-
pared with those of 158 patients with the classic epider-
motropic type of MF, including 122 patients with gen-
eralized plaque-stage MF (T2 N0 M0) and 36 patients
with tumor-stage MF (T3 N0 M0).
Observations: Characteristic clinical features not or
rarely observed in classic MF were the preferential lo-
calization of the skin lesions in the head and neck re-
gion (45 of 51 patients), the presence of follicular pap-
ules, alopecia, acneiform lesions, mucinorrhoea, and often
severe pruritus. Characteristic histologic findings were
the presence of perifollicular neoplastic infiltrates with
a variable degree of folliculotropism, but generally no epi-
dermotropism, follicular mucinosis (49 of 51 cases), and
often a considerable admixture of eosinophils and plasma
cells. Response on initial treatment, risk of disease pro-
gression (development of extracutaneous disease and/or
death from lymphoma), and disease-specific and over-
all survival of patients with follicular MF were worse than
in classic MF patients. The actuarial disease-specific sur-
vival was 68% at 5 years and 26% at 10 years.
Conclusions: Follicular MF shows distinctive clinico-
pathologic features, is more refractory to treatment, and
has a worse prognosis than the classic type of MF; it should
be considered a distinct type of cutaneous T-cell lym-
phoma. Based on these results and those of other stud-
ies, we suggest the term follicular MF for cases with or
without associated follicular mucinosis.
Arch Dermatol. 2002;138:191-198

From the Departments of
Dermatology (Dr van Doorn)
and Pathology (Dr Scheffer),
Vrije Universiteit Medical
Center, Amsterdam, and
Department of Dermatology
(Dr Willemze), Leiden
University Medical Center,
the Netherlands. A complete list
of the participants in the Dutch
Cutaneous Lymphoma Group is
available from the authors.
M
YCOSIS fungoides (MF)
is the most common
type of cutaneous T-cell
lymphoma, character-
ized clinically by an
indolent clinical course with the subse-
quent evolution of patches, plaques, and
tumors, and histologically by the infiltra-
tion of the epidermis by medium-sized to
large atypical T cells with cerebriform
nuclei.1
See also pages 182 and 244
In the first 10 years after diagnosis, dis-
ease progression, including development
of extracutaneous disease and disease-
related deaths, occurs in only a minority of
patients.2 Apart from this so-called classic
Alibert-Bazin type of MF, many clinical and
histologic subtypes have been reported,
including hypopigmented, vesicular,
pustular, granulomatous, and other types
of MF. Since these variant types of MF
have the same clinical course and clinical
behavior and require the same therapeu-
tic approach as the classic type of MF, they
are generally not considered separate
entities. In both the EORTC (European
Organization for Research on the Treat-
ment of Cancer)1 classification for pri-
mary cutaneous lymphomas and in the
WHO (World Health Organization) clas-
sification,3 only pagetoid reticulosis and
MF-associated follicular mucinosis have
been categorized as separate entities.
Hereinafter, the latter condition will be
referred to as follicular MF, a term also
used in the WHO classification for cases
with or without associated follicular
mucinosis.
Follicular MF has been classified as a
separate entity because it has distinctive
clinical and histologic features, is more re-
fractory to standard treatment, and has a
worse prognosis than classic MF. How-
ever, this observation is particularly based
on clinical experience of members of the
classification committee of the EORTC Cu-
taneous Lymphoma Group and is not sub-

(REPRINTED) ARCH DERMATOL / VOL 138, FEB 2002 WWW.ARCHDERMATOL.COM

191

©2002 American Medical Association. All rights reserved.

43
 PATIENTS AND METHODS
Between October 1985 and December 1998, 57 patients with
follicular MF were included in the registry of the Dutch Cu-
taneous Lymphoma Group. Three of the 57 had to be ex-
cluded from the present study, 1 because of incomplete fol-
low-up data and 2 because of lack of representative skin
biopsy specimens. Another 3 patients were excluded be-
cause they had a history of classic epidermotropic MF for 4
to 7 years before they developed skin tumors on the face with
the histologic features of MF-associated follicular mucino-
sis. The final study group consisted of 51 patients. In each
patient the diagnosis had been made by an expert panel of
dermatologists and pathologists at one of the quarterly meet-
ings of the Dutch Cutaneous Lymphoma Group. The main
criteria for diagnosis and for inclusion in the study were the
presence of perifollicular-to-diffuse dermal infiltrates with
variable numbers of atypical T cells with cerebriform nu-
clei infiltrating the follicular epithelium and the presence of
mucinous degeneration of the follicular epithelium, as con-
firmed by Alcian blue staining of the first diagnostic biopsy
specimens (diagnostic specimens).6,7 Forty-nine patients met
both criteria. Two cases with the same cytoarchitectural fea-
tures in the diagnostic specimen but without associated fol-
licular mucinosis were included as well. The time of evalu-
ation of the first diagnostic specimen was considered the time
of diagnosis.
In all cases diagnostic evaluation at the time of diag-
nosis consisted of a thorough physical examination, com-
plete blood cell count, serum chemistry studies, and skin
specimen evaluation. Lymph node biopsies and thoracic
and abdominal computed tomographic scans were per-
formed only in patients with enlarged peripheral lymph
nodes. Lymph node involvement was assessed using crite-
ria described previously.10 When indicated clinically,
stantiated sufficiently in the literature. In fact, pub-
lished reports on follicular MF are relatively scarce,
generally concern case reports or small series of pa-
tients, and have mainly been focused on differentiation
between MF-associated follicular mucinosis and alope-
cia mucinosa, the benign idiopathic form of follicular
mucinosis.4-9
In a recent study by members of our group2 on 309
patients with MF, the 32 patients with follicular MF
seemed to have significantly higher disease progression
and mortality rates than the 277 patients without fol-
licular mucinosis. These observations prompted us to re-
view all cases of follicular MF registered at the Dutch Cu-
taneous Lymphoma Group between 1985 and 1998.
RESULTS
CLINICAL CHARACTERISTICS
The main clinical features and relevant follow-up data
have been summarized in Table 1. Fifty-one patients,
including 42 male and 9 female, were included in this
study. Four (8%) of 51 patients were younger than 40
years at the time of diagnosis. At the time of diagnosis,
(REPRINTED) ARCH DERMATOL / VOL 138, FEB 2002
192
©2002 American Medical Association. All rights reserved.
44
additional studies to determine visceral involvement were
performed.
In all cases clinical records and follow-up data, which
had been collected yearly, were evaluated. The following vari-
ables were recorded: age; sex; duration of skin lesions be-
fore diagnosis; type of initial therapy; whether there was a
complete remission after initial therapy; the date of disease
progression if applicable; and the date of last contact (or death
if applicable). Because of difficulties in defining skin stage
in patients with follicular MF, disease progression was de-
fined by the development of histologically documented nodal
involvement in patients with previously skin-limited dis-
ease, the development of visceral involvement in patients with
prior skin and/or lymph node involvement, and death due
to lymphoma. In all cases, one or multiple skin biopsy speci-
mens obtained at the time of diagnosis, and in most cases
also obtained during follow-up, were reviewed.
To evaluate differences in clinical behavior between
follicular MF and classic MF, 49 patients with follicular MF
presenting with disease confined to the skin were com-
pared with 158 patients with classic MF included in an ear-
lier study.2 This latter group included 122 patients with gen-
eralized plaque-stage MF (T2 N0 M0) and 36 patients with
tumor-stage MF (T3 N0 M0) without evidence of extracu-
taneous disease and without associated follicular mucino-
sis at the time of diagnosis.
Actuarial survival curves were calculated from the date
of diagnosis to the date of last contact (or the date of death)
using the Kaplan-Meier technique. Differences between sur-
vival and disease progression curves were analyzed using
the log-rank test. Univariate analysis of possible prognos-
tic factors was performed using the log-rank test and Cox
proportional hazards regression analysis. Patients lost to
follow-up were considered censored at the time of last con-
tact. Analyses were performed using the SPSS statistical soft-
ware (SPSS Inc, Chicago, Ill).
49 patients (96%) had disease confined to the skin, in-
cluding 4 patients with enlarged but histologically un-
involved lymph nodes, whereas 1 patient had concur-
rent lymph node involvement and another, concurrent
visceral involvement.
At the time of diagnosis, 34 of 51 patients had only
patches, plaques, or (grouped) follicular papules, often
associated with alopecia; 14 had (concurrent) nodules
or tumors; and 3 had erythroderma (Figure 1). Acne-
iform lesions, including comedolike lesions and epider-
mal cysts, were a prominent feature in 4 patients.
Mucinorrhea (ie, discharge of mucinous substance from
the follicular orifices) was noted in 3 patients. During
the course of their disease, 2 patients developed a leo-
nine face (Figure 2). In 45 of 51 patients, the skin le-
sions were preferentially localized in the head and neck
region at the time of diagnosis. A characteristic finding
in 25 of these 45 patients was the presence of plaques or
tumors on the head or neck, whereas the trunk and ex-
tremities showed only patches or slightly infiltrated
plaques and/or grouped follicular papules (Figure 1A-B).
Infiltrated plaques in the eyebrows with concurrent alo-
pecia were a common finding. Most patients had mod-
erate to severe pruritus.
WWW.ARCHDERMATOL.COM
 HISTOLOGIC FEATURES
Table 1. Clinical Characteristics and Follow-up Data
A total of 74 representative skin biopsy specimens from these of 51 Patients With Follicular Mycosis Fungoides*
51 patients were reviewed. These included the 51 diagnos-
Characteristic Finding
tic specimens, 8 prediagnostic specimens obtained 4 to 30
months (median, 12 months) prior to diagnosis, and 15 Age at diagnosis, median (range), y 57 (15-84)
Male-female ratio 4.7 (42:9)
specimens obtained during follow-up at the time of re-
Duration of skin lesions before diagnosis, 48 (4-156)
lapse or disease progression. Characteristically, the diag- median (range), mo
nostic specimens showed perifollicular and perivascular- Type of skin lesions at diagnosis
to-diffuse dermal infiltrates with variable infiltration of the Patches/plaques/papules 44 (86)
follicular epithelium by medium-sized to large atypical T Nodules/tumors 14 (27)
cells with cerebriform nuclei (Figure 3). Pautrier micro- Cysts/comedones 4 (8)
Erythroderma 3 (6)
abscesses appeared in only a minority of specimens. Infil-
Clinical stage at diagnosis
tration of the interfollicular epidermis by (atypical) T cells, Only skin lesions 45 (88)
as in classic MF, was rare. Only 5 of 51 specimens showed Enlarged, but uninvolved nodes (DL) 4 (8)
infiltration of both the epidermis (epidermotropism) and Lymph node involvement 1 (2)
the follicular epithelium (folliculotropism). Prominent in- Visceral involvement 1 (2)
filtration of the eccrine sweat glands was observed in 3 speci- Folicular mucinosis
Present 49 (96)
mens. In all but 2 cases, the skin specimens showed mu-
Absent 2 (4)
cinous degeneration of the hair follicles, varying from focal Initial treatment
spots of mucin deposition (which had to be searched for PUVA 22 (43)
in serial sections) to lakes of mucin (Figure 3B). TSEBI 11 (22)
The number of atypical T cells infiltrating the follicu- Radiotherapy + other 7 (14)
lar epithelium was generally low and did not correlate with Other 11 (22)
the amount of mucin deposition. The perifollicular infil- Complete remission on initial therapy 8 (16)
Calculated risk of disease progression†, %
trates consisted of variable numbers of medium-sized to At 5 y 37
large atypical T cells with cerebriform nuclei and blast cells At 10 y 66
and admixed small lymphocytes, histiocytes, eosinophils Current status
(which were numerous in 13 of 51 specimens), and plasma Alive without disease 5 (10)
cells, in particular in patients with secondary bacterial in- Alive with disease 20 (39)
fection (Figure 3C). Concurrent patches on the trunk Died of other cause 6 (12)
Died of FMF 20 (39)
showed essentially the same histologic features, though the
Disease-specific survival, %
number of atypical T cells was often less than in the more At 5 y 68
infiltrated lesions in the head and neck, which made a defi- At 10 y 26
nite diagnosis in these specimens more difficult or even Overall survival, %
impossible. At 5 y 64
Immunohistochemical analysis demonstrated a At 10 y 14
CD3+-CD4+-CD8− phenotype of the neoplastic T cells in
*Unless otherwise indicated, data are number (percentage) of patients.
all cases studied. Small numbers of scattered CD30+ blast DL indicates histologic features of dermatopathic lymphadenopathy;
cells were regularly observed, as were small clusters of ad- PUVA, psoralen plus UV-A therapy; TSEBI, total skin electron beam
mixed B cells. In the initial diagnostic specimens of 7 of irradiation; and FMF, follicular mycosis fungoides.
the 51 patients, we found a considerable number of blast †Disease progression means development of extracutaneous disease
and/or death from lymphoma.
cells (�15%; generally a mixture of CD30+ and CD30− blast
cells). Six of these 7 patients died of lymphoma 11 to 100
months (median, 40 months) after diagnosis.
In follow-up specimens taken during disease pro- tion (TSEBI) in 11 patients (Table 1). Seven patients were
gression, the dermal infiltrates tended to become more initially treated with local radiotherapy in combination with
diffuse, sometimes showed complete effacement of the other therapies, including PUVA with or without reti-
follicular structures, and invariably showed increasing noids, UV-B, topical mechlorethamine, or topical steroids.
numbers of CD30− and/or CD30+ blast cells. In the 8 pre- For the remaining patients treatments included topical ste-
diagnostic specimens, the dermal infiltrates were mainly roids (3 patients), topical mechlorethamine (1 patient), UV-B
confined to the perifollicular areas. Although some of these (1 patient), PUVA in combination with retinoids (1 pa-
specimens already contained small numbers of atypical tient), prednisone (1 patient), azathioprine in combina-
T cells in the follicular epithelium and in the perifollicu- tion with topical mechlorethamine (1 patient), or poly-
lar infiltrates, the size and morphologic characteristics chemotherapy (2 patients). One patient refused treatment.
of the infiltrating T cells did not warrant a definite diag- Only 8 (16%) of 51 patients, each with disease con-
nosis of follicular MF. fined to the skin, achieved complete remission on initial
treatment. Six of them had been treated with TSEBI, 1 with
THERAPY AND FOLLOW-UP PUVA, and 1 with a combination of PUVA, retinoids, and
local radiotherapy. Two of these 7 patients were still in com-
Initial therapy consisted of psoralen plus UV-A (PUVA) treat- plete remission after a follow-up of 38 and 192 months,
ment in 22 patients and total skin electron beam irradia- respectively, and may be considered cured.

(REPRINTED) ARCH DERMATOL / VOL 138, FEB 2002 WWW.ARCHDERMATOL.COM

193

©2002 American Medical Association. All rights reserved.

45
 A B

C D

E F

Figure 1. Clinical appearances of follicular mycosis fungoides. A, Boggy infiltrates on the cheek and neck; B, concurrent grouped follicular papules on the trunk;
C, Infiltrated plaque with alopecia and cystic lesions above the left eye; D, the same patient had numerous cysts and comedolike lesions on the trunk;
E, Diffuse erythema and alopecia of the eyebrows (and eyelashes, not shown); F, Infiltrated plaque on the forehead with alopecia of the left eyebrow.

The follow-up period varied between 8 and 239
months (median, 48 months; mean, 58 months). Devel-
opment of lymph node or visceral involvement was docu-
mented in 14 and 7 patients, respectively. Disease pro-
gression defined as the development of extracutaneous
disease or death from lymphoma occurred in 20 (39%)
of 51 patients, and occurred 11 to 168 months (median,
45 months) after diagnosis. The calculated risk of dis-
ease progression during the first 5 years after diagnosis
was 37%; at 10 years, 66%. At the conclusion of the study,
26 patients had died, 20 from lymphoma. The 5- and 10-
year disease-specific survival rate was 68% and 26%, re-
spectively. The respective overall survival rates at 5 and
10 years were 64% and 14%. Univariate analysis dem-
onstrated no association between disease-specific sur-
vival and age at diagnosis, sex, duration of skin lesions
before diagnosis, or response to initial treatment.
FOLLICULAR MF VS CLASSIC MF
To evaluate differences in clinical behavior and progno-
sis between follicular MF and classic MF, we compared

(REPRINTED) ARCH DERMATOL / VOL 138, FEB 2002 WWW.ARCHDERMATOL.COM

194

©2002 American Medical Association. All rights reserved.

46
 A B

C D

Figure 2. Sequential photographs of a 39-year-old man with follicular mycosis fungoides. A, At the time of diagnosis; B, 26 months after diagnosis with a leonine
face; C, 29 months after diagnosis; and D, 35 months after diagnosis following total skin electron beam irradiation, which had not resulted in complete remission.
The patient died of lymphoma 42 months after diagnosis.

relevant clinical features of the 49 patients with follicular
MF who had disease confined to the skin with the fea-
tures of 122 patients with generalized plaque-stage MF and
36 patients with tumor-stage MF without evidence of ex-
tracutaneous disease and without associated follicular mu-
cinosis (Table 2). Patients with follicular MF showed a
significantly higher male-female ratio and less frequently
achieved complete remission on initial treatment. The cal-
culated risk of disease progression, as defined in this study,
within the first 5 years after diagnosis was 36% for fol-
licular MF vs 12% for classic plaque-stage MF and 24%
for tumor-stage MF. Disease-specific and overall survival
in patients with follicular MF were significantly lower than
in patients with generalized plaque-stage MF, and were
roughly similar to patients with tumor-stage MF without
associated follicular mucinosis (Figure 4).
COMMENT
The results of the present study clearly demonstrate that
follicular MF has distinctive clinicopathologic features and
should be considered a distinct disease entity. Character-
istic histologic features include the primary perifollicular
localization of the dermal infiltrates, with variable infiltra-
tion of the follicular epithelium by medium-sized to large
atypical T cells with cerebriform nuclei. In most cases the
epidermis is spared (folliculotropism instead of epidermo-
tropism). Mucinous degeneration of the follicular epithe-
lium occurs in most cases, and a considerable admixture
with eosinophils and plasma cells is frequently present.
Clinical characteristics include the preferential lo-
calization of the skin lesions in the head and neck area (45
of 51 patients), the presence of papules (often grouped),
alopecia, frequent secondary bacterial infection, and, less
commonly, the presence of acneiform lesions and muci-
norrhea. Unlike in classic MF, pruritus is often severe and
may represent a good parameter of disease activity: in sev-
eral patients, a relapse after initial therapy was preceded
by the reappearance of pruritus. In addition, patients with
follicular MF proved generally more refractory to stan-
dard classic MF therapies, showed more frequent disease
progression, and had a less favorable prognosis (Table 2).
This more unfavorable prognosis suggests a true bio-
logical difference in clinical behavior between patients with
follicular MF and patients with the classic epidermotropic
type of MF, which is consistent with the conclusion of a
recent study.11 The similar duration of skin lesions before
diagnosis in patients with follicular MF and patients with
classic-type MF indicates that the difference in survival
does not simply result from a selection of patients with

(REPRINTED) ARCH DERMATOL / VOL 138, FEB 2002 WWW.ARCHDERMATOL.COM

195

©2002 American Medical Association. All rights reserved.

47
 A B
C perifollicular infiltrate of part A showing atypical
more advanced disease in the present study (Table 2). Com-
parison of the disease-specific and overall survival data in-
dicate that patients with follicular MF have a similar (at
5 years) or worse (at 10 years) survival than patients with
tumor-stage MF (Table 2). Nevertheless, under the clas-
sic MF classification systems,12,13 most of our patients with
follicular MF would have been classified as stage IA (T1
N0 M0) or IB (T2 N0 M0), and only 14 of them had nod-
ules or tumors at the time of diagnosis. This supports our
contention that these clinical staging systems for MF are
not very useful in patients with follicular MF. For instance,
patients presenting with a solitary patch or plaque on the
face do not have stage IA or T1 N0 M0–stage disease. Be-
cause of the perifollicular localization of the dermal in-
filtrates, such patients should always be considered to have
tumor-stage disease, regardless of the clinical appearance
of the skin lesion, and should be treated accordingly.
MF-ASSOCIATED FOLLICULAR MUCINOSIS
VS FOLLICULAR MF
In recent years the term follicular MF or cutaneous T-cell
lymphoma (also folliculocentric MF or pilotropic MF) has been
(REPRINTED) ARCH DERMATOL / VOL 138, FEB 2002
196
©2002 American Medical Association. All rights reserved.
48
Figure 3. Characteristic histologic features of
follicular mycosis fungoides. Perifollicular infiltrates
with marked folliculotropism and associated follicular
mucinosis. A, Note the absence of epidermal
involvement. B, Alcian blue staining of a concurrent
lesion showing mucin deposits. C, Detail of
hyperchromatic T cells, blast cells, and admixed
histiocytes and eosinophils.
introduced for a rare clinical variant of MF characterized
by follicular papules, follicular keratoses, comedolike le-
sions and epidermal cysts. Histologically, perifollicular in-
filtrates are present showing marked folliculotropism, but
there is generally no epidermotropism or follicular muci-
nosis.14-22 Evaluation of published reports of “follicular MF”
demonstrates considerable clinical heterogeneity and sug-
gests that this term has been used for the diagnoses of at
least 3 different groups of patients.
The largest group comprises patients with clinically
and histologically classic MF prior to or, less often, con-
current with the development of the follicular le-
sions.17,21,22 In the present study, 3 patients with 4- to 7-year
histories of classic epidermotropic MF developing skin tu-
mors with the histologic features of follicular MF were ex-
cluded because such cases show the clinical behavior of clas-
sic MF developing tumor-stage disease.
The second group includes patients presenting with
acneiform lesions as the predominant or only manifesta-
tion of the disease.17,19,21,22 However, a similar clinical pre-
sentation may also occur in patients with associated fol-
licular mucinosis23,24 and was a predominant feature in 4
of 51 patients in the present study. One of our patients was
WWW.ARCHDERMATOL.COM
 Table 2. Characteristics of Patients With Folllicular, Generalized Plaque-Stage, and Tumor-Stage Mycosis Fungoides*

P Value (FMF
vs Classic MF)
FMF† Plaque-Stage MF† Tumor-Stage MF†
Characteristic (n = 49) (n = 122) (n = 36) Plaque Tumor
Age at diagnosis, median (range), y 57 (15-84) 61 (14-92) 47 (35-88) .61 .01
Male-female ratio 4.4 (40:9) 1.39 (71:51) 1.25 (20:16) .004 .02
Duration of skin lesions before diagnosis (range), mo 48 (4-156) 48 (1-840) 48 (2-600) .09 .12
Complete remission on initial therapy, No. (%) 8 (16) 38 (31) 10 (28) .05 .20
Risk of disease progression, %
At 5 y 36 12 24 �.001 .49
At 10 y 65 18 47
Disease-specific survival, %
At 5 y 69 95 79 �.001 .25
At 10 y 26 84 61
Overall survival, %
At 5 y 65 89 61 �.001 .80
At 10 y 14 68 34
Follow-up duration, median (range), mo 48 (11-239) 72 (14-313) 47 (6-249)

*FMF indicates follicular mycosis fungiodes; MF, mycosis fungoides.

†Columns “FMF,” “Plaque-Stage MF,” and “Tumor-Stage MF” denote patients without extracutaneous involvement at time of diagnosis with FMF, MF with
patches and plaques covering 10% or more of the skin surface (T2 N0 M0), and MF with skin tumors (T3 N0 M0), respectively.

a 16-year-old boy with a history of severely pruritic acne- 100

iform lesions and alopecia for more than 5 years before the 90 Plaque-Stage MF (n = 122)
diagnosis MF-associated follicular mucinosis was made. De-
Disease-Specific Survival, %

80
spite radiotherapy and multiagent chemotherapy, he died 70
5 years after diagnosis of systemic lymphoma. Interest- 60
Tumor-Stage MF (n = 36)
ingly, while several of his acneiform lesions showed mu- 50
40
cin deposits, others did not, even after further sectioning.
30
Finally, some of the reported cases demonstrate all
20 FMF (n = 49)
of the characteristic clinical and histologic features of the
10
cases reported herein except for the presence of follicular
mucinosis.18,20 The present study includes 2 such cases in 0 60 120 180
patients who otherwise did not differ clinically or histo- Follow-up Duration, mo

logically from the 49 patients with associated follicular
mucinosis. Based on these observations, we do not be-
lieve that it is useful to differentiate between follicular MF
with and without associated follicular mucinosis in these
2 latter groups. From a biological point of view, the most
relevant feature in follicular MF with or without follicu-
lar mucinosis is the deep, perifollicular localization of the
neoplastic infiltrates, which makes them less accessible to
skin-targeted therapies. On the basis of our own observa-
tions and the available literature,11 we are inclined to be-
lieve that follicular MF with or without follicular muci-
nosis should not be considered separately, and that cases
with a preferential (peri)follicular distribution of the neo-
plastic infiltrates, regardless of the presence of mucinous
degeneration, should be termed follicular MF.
DIFFERENTIAL DIAGNOSIS
Although distinctive clinical and histologic features should
facilitate an early and correct diagnosis, it is our experi-
ence over the last 15 years that the diagnosis of follicular
MF is often overlooked. Because of the preferential in-
volvement of the head and neck area, the absence of patches
and plaques on the trunk or buttocks, and the absence of
epidermotropic atypical T cells, the diagnosis of MF or cu-
taneous T-cell lymphoma is often not considered. The fol-
lowing incorrect diagnoses have been made more than once
Figure 4. Actuarial disease-related survival of 49 patients with follicular
mycosis fungoides (FMF), 122 with generalized plaque-stage mycosis
fungoides (MF) (T2 N0 M0), and 36 with tumor-stage MF (T3 N0 M0). For
FMF vs plaque-stage MF, P�.001; for FMF vs tumor-stage MF, P =.25.
prior to referral: seborrheic dermatitis in patients present-
ing with erythematous lesions on the scalp and eye-
brows; atopic dermatitis, because of the severe pruritus;
and facial granuloma with eosinophilia in patients pre-
senting with a solitary plaque on the face and an eosinophil-
rich infiltrate. Finally, it should be noted that even when
follicular MF is suspected, it may require several biopsies
to make a definite diagnosis. It is important that biopsy
specimens be taken from the most infiltrated skin le-
sions, generally in the face or neck, and not only from
patches with or without follicular papules on the trunk.
THERAPY
The results of the present study confirm our clinical im-
pression and the scattered data in literature that patients
with follicular MF are generally less responsive to stan-
dard therapies used in patients with classic MF.11,20,25 Our
retrospective study does not allow a meaningful compari-
son of the effects of the different treatment methods be-
cause patients were treated at different institutions, and treat-
ment selection may have been affected by disease severity.

(REPRINTED) ARCH DERMATOL / VOL 138, FEB 2002 WWW.ARCHDERMATOL.COM

197

©2002 American Medical Association. All rights reserved.

49
 Because of the perifollicular localization of the der-
mal infiltrates, patients with follicular MF should always
be considered to have tumor-stage disease, regardless of
the clinical appearance of the skin lesions. Therefore, in
patients presenting with a single plaque or tumor or a few
clustered skin lesions, but without patches or follicular pap-
ules at other sites, radiotherapy is the first choice for treat-
ment. In selected cases with superficial lesions present-
ing at multiple sites, PUVA treatment might be attempted
first. However, in most cases this approach will not result
in complete remission. In the present series, complete re-
mission was achieved with PUVA therapy in only 1 of 22
patients. In patients with more infiltrated skin lesions, in
particular those who do not respond to PUVA therapy
alone, TSEBI is the preferred method of treatment. How-
ever, only 6 of 11 patients treated with TSEBI reached com-
plete remission, and 3 of the 6 complete responders had a
relapse within 6 months.
The relative unresponsiveness of follicular MF to
TSEBI has been reported.18 In some patients relapsing skin
disease may be controlled effectively by a maintenance
treatment with topical nitrogen mustard. If TSEBI is not
available, PUVA in combination with interferon alfa or
retinoids and local radiotherapy for thick tumors may be
an alternative. The same approach can be used for re-
lapsing disease following TSEBI. In our experience, mul-
tiagent chemotherapy does not generally result in com-
plete remission in patients with skin-limited disease and
should therefore be reserved for patients developing
extracutaneous disease.
Accepted for publication July 2, 2001.
Corresponding author and reprints: Rein Willemze, MD,
Department of Dermatology, B1-Q-93, Leiden University
Medical Center, Albinusdreef 2, 2333 ZA Leiden, the Neth-
erlands (e-mail: willemze.dermatology@lumc.nl).
REFERENCES
1. Willemze R, Kerl H, Sterry W, et al. EORTC classification for primary cutaneous
lymphomas: a proposal from the Cutaneous Lymphoma Study Group of the Eu-
ropean Organization for Research and Treatment of Cancer. Blood. 1997;90:
354-371.
2. Van Doorn R, Van Haselen CW, van Voorst Vader PC, et al. Mycosis fungoides:
disease evolution and prognosis of 309 Dutch patients. Arch Dermatol. 2000;
136:504-510.
3. Harris NL, Jaffe ES, Diebold J, et al. World Health Organization classification of
neoplastic diseases of the hematopoietic and lymphoid tissues: report of the Clini-
cal Advisory Committee meeting, Airlie House, Virginia, November 1997. J Clin
Oncol. 1999;17:3835-3849.
4. Pinkus H. The relationship of alopecia mucinosa to malignant lymphoma. Der-
matologica. 1964;129:266-270.
5. Emmerson RW. Follicular mucinosis: a study of 47 patients. Br J Dermatol. 1969;
81:395-413.
6. Nickoloff BJ, Wood C. Benign idiopathic versus mycosis fungoides–associated
follicular mucinosis. Pediatr Dermatol. 1985;2:201-206.
7. Sentis HJ, Willemze R, Scheffer E. Alopecia mucinosa progressing into mycosis
fungoides: a long-term follow-up study of two patients. Am J Dermatopathol.
1988;10:478-486.
8. Gibson LE, Muller SA, Leiferman KM, Peters MS. Follicular mucinosis: clinical
and histopathologic study. J Am Acad Dermatol. 1989;20:441-446.
9. Mehregan DA, Gibson LE, Muller SA. Follicular mucinosis: histopathologic re-
view of 33 cases. Mayo Clin Proc. 1991;66:387-390.
10. Scheffer E, Meijer CJ, Van Vloten WA. Dermatopathic lymphadenopathy and lymph
node involvement in mycosis fungoides. Cancer. 1980;45:137-148.
11. Bonta MD, Tannous ZS, Demierre MF, Gonzalez E, Harris NL, Duncan LM. Rap-
idly progressing mycosis fungoides presenting as follicular mucinosis. J Am Acad
Dermatol. 2000;43:635-640.
12. Bunn P, Lamberg S. Report of the Committee on Staging and Classification of
Cutaneous T-cell Lymphomas. Cancer Treat Rep. 1979;63:725-728.
13. Fuks Z, Bagshaw M, Farber E. Prognostic signs and the management of the my-
cosis fungoides. Cancer. 1973;32:1385-1395.
14. Kim SY. Follicular mycosis fungoides. Am J Dermatopathol. 1985;7:300-301.
15. Lacour JP, Castanet J, Perrin C, Ortonne JP. Follicular mycosis fungoides—a
clinical and histologic variant of cutaneous T-cell lymphoma: report of two cases.
J Am Acad Dermatol. 1993;29:330-334.
16. Goldenhersh MA, Zlotogorski A, Rosenmann E. Follicular mycosis fungoides. Am
J Dermatopathol. 1994;16:52-55.
17. Vergier B, Beylot-Barry M, Beylot C, et al. Pilotropic cutaneous T-cell lymphoma
without mucinosis: a variant of mycosis fungoides? Arch Dermatol. 1996;132:
683-687.
18. Pereyo NG, Requena L, Galloway J, Sangueza OP. Follicular mycosis fungoides:
a clinicohistopathologic study. J Am Acad Dermatol. 1997;36:563-568.
19. Fraser-Andrews E, Ashton R, Russell-Jones R. Pilotropic mycosis fungoides pre-
senting with multiple cysts, comedones and alopecia. Br J Dermatol. 1999;
140:141-144.
20. Klemke CD, Dippel E, Assaf C, et al. Follicular mycosis fungoides. Br J Dermatol.
1999;141:137-140.
21. Hodak E, Feinmesser M, Segal T, et al. Follicular cutaneous T-cell lymphoma: a
clinicopathological study of nine cases. Br J Dermatol. 1999;141:315-322.
22. Grau C, Pont V, Matarredona J, Fortea JM, Aliaga A. Follicular mycosis fungoi-
des: presentation of a case and review of the literature. J Eur Acad Dermatol
Venereol. 1999;13:131-136.
23. Wilkinson JD, Black MM, Chu A. Follicular mucinosis associated with mycosis
fungoides presenting with gross cystic changes on the face. Clin Exp Dermatol.
1982;7:333-339.
24. Wittenberg GP, Gibson LE, Pittelkow MR, el-Azhary RA. Follicular mucinosis pre-
senting as an acneiform eruption: report of four cases. J Am Acad Dermatol. 1998;
38:849-851.
25. Gilliam AC, Lessin SR, Wilson DM, Salhany KE. Folliculotropic mycosis fungoi-
des with large-cell transformation presenting as dissecting cellulitis of the scalp.
J Cutan Pathol. 1997;24:169-175.

News and Notes

News and Notes

T he Regional Conference on Dermatological Laser and Facial Cos-

metic Surgery 2002 will be held from September 13 through Septem-
ber 15, 2002, at the new wing of the Hong Kong Convention and Ex-
hibition Center. The conference is jointly organized by the University of Hong
Kong, the Hong Kong Society of Dermatology and Venereology, and the Hong
Kong Society of Plastic & Reconstructive Surgeons.
Renowned authorities to speak at the conference include Dr Yung-lung
Lai (Chang Gung Memorial Hospital, Taiwan), Dr Dieter Manstein (Harvard
Medical School, United States), Prof Rolf Nordström (Nordström Hospital for
Plastic and Reconstructive Surgery, Finland), Dr Niwat Polnikorn (Ramathi-
bodi Hospital, Thailand), and Dr Woffles Wu (Woffles Wu Aesthetic Surgery
and Laser Center, Singapore).
For more information on the conference, please contact the secretariat at
phone (852) 25278898; fax: (852) 28667530, or e-mail: cosfmshk@netvigator
.com.

(REPRINTED) ARCH DERMATOL / VOL 138, FEB 2002 WWW.ARCHDERMATOL.COM

198

©2002 American Medical Association. All rights reserved.

50
 Chapter 4

CD8+ T cells in cutaneous T-cell lymphoma:

expression of cytotoxic proteins, Fas ligand,
and killing inhibitory receptors and
their relationship with clinical behaviour

J. Clin. Oncol. 2001 Dec 1;19(23):4322-9

 CD8� T Cells in Cutaneous T-Cell Lymphoma: Expression
of Cytotoxic Proteins, Fas Ligand, and Killing Inhibitory
Receptors and Their Relationship With Clinical Behavior
By Maarten H. Vermeer, Remco van Doorn, Danny Dukers, Marcel W. Bekkenk, Chris J.L.M. Meijer, and Rein Willemze
Purpose: We investigated the number, phenotype,
and prognostic significance of CD8� T cells in patients
with mycosis fungoides (MF) and CD30� primary cuta-
neous large T-cell lymphoma (PCLTCL).
Patients and Methods: Immunohistochemical stain-
ings for CD8, granzyme B (GrB), T cell–restricted intra-
cellular antigen (TIA-1), Fas ligand (FasL), and killer-cell
inhibitory receptors (KIRs; CD95, CD158a, and CD158b)
were performed on 83 first-diagnostic biopsy samples
obtained from patients with plaque-stage MF (n � 42),
tumor-stage MF (n � 20), and CD30� PCLTCL (n � 21).
Results: Serial sections and double-staining experi-
ments showed that the large majority of CD8� T cells in
MF and CD30� PCLTCL expressed TIA-1 and FasL,
whereas only a minority expressed GrB, which sug-
gested that these CD8� T cells were partly activated
cytotoxic T lymphocytes (CTLs). These CD8� CTLs never
or rarely expressed KIRs. This phenotype was a con-
stant feature of CD8� CTLs and did not alter with
C UTANEOUS T-CELL lymphomas (CTCLs) are a
group of non-Hodgkin’s lymphomas mainly with a
CD3�CD4� memory T-cell phenotype that preferentially
home to the skin.1 Mycosis fungoides (MF), the most
common type of CTCL, is characterized by erythematous,
scaly patches, and plaques in the early stages of disease and
generally has an indolent clinical course.2 However, in
some patients, progression to tumor-stage MF is observed,
which is associated with more aggressive clinical behavior.
In contrast, patients with CD30� primary cutaneous large
T-cell lymphoma (PCLTCL) generally present with multi-
ple tumors that rapidly spread to extracutaneous sites, and
these patients have a poor prognosis, with a 5-year survival
rate of 15%.1
The pathophysiologic mechanisms underlying these dif-
ferences in biologic behavior are largely unknown. How-
ever, the protracted clinical course in most MF patients, the
beneficial effect of immunomodulating therapies, and the
From the Departments of Dermatology and Pathology, Free Uni-
versity Hospital, Amsterdam, the Netherlands.
Submitted January 22, 2001; accepted July 23, 2001.
Address reprint requests to M.H. Vermeer, MD, Department of
Dermatology, Leiden University Medical Center, Albinusdreef 2 2300
RC Leiden, the Netherlands; email: m.h.vermeer@lumc.nl.
© 2001 by American Society of Clinical Oncology.
0732-183X/01/1923-4322/$20.00
4322 Journal of Clinical Oncology, Vol 19, No 23 (December 1), 2001: pp 4322-4329
53
disease progression. In contrast, the median percent-
age of CD8� CTLs in plaque-stage MF (22%), tumor-
stage MF (7%), and CD30� PCLTCL (3%) differed signif-
icantly (P < .0001) and was associated with a
significant decrease in 5-year survival. Also within the
group of tumor-stage MF, a significant relation be-
tween CD8� CTLs and survival was found. Multivariate
analysis in the total group of MF demonstrated that
both skin stage and percentage of CD8� CTLs were
independent parameters of survival.
Conclusion: Our results demonstrated that partly
activated CD8� CTLs were present in CTCL and that
high proportions of these cells correlated with a bet-
ter prognosis. This suggested that these CD8� CTLs
could play an important role in the antitumor re-
sponse in these conditions.
J Clin Oncol 19:4322-4329. © 2001 by American
Society of Clinical Oncology.
influx of CD8� cytotoxic T lymphocytes (CTLs) in regress-
ing MF tumors on intralesional administration of interleu-
kin-12 suggest that a tumor-specific immune response may
play an important role.3 In vitro studies have consistently
shown that malignant cells in MF display tumor-specific
antigens that can be recognized by autologous CD8�
CTLs.4-7 Taken together, these observations suggest that
CD8� CTLs are a critical component in the antitumor
immune response in CTCL. However, studies that evaluate
the number and immunophenotype of CD8� CTLs in
different stages of CTCL are scarce, and correlation with
clinical behavior leads to contradicting results.8,9
CD8� CTL–mediated tumor-cell lysis is achieved by at
least two distinct pathways: (1) by exocytosis of intracyto-
plasmic granules that contain perforin, granzymes, and T
cell–restricted intracellular antigen (TIA-1), and (2) by the
Fas-mediated pathway in which membrane-bound Fas li-
gand (FasL) expressed on cytotoxic T cells interacts with
Fas (CD95/Apo-1) on target cells.10,11 Both pathways acti-
vate a cascade of subcellular events that ultimately lead to
the activation of caspases and target-cell death. The avail-
ability of monoclonal antibodies (MoAbs) against perforin,
the serine proteases granzyme A and granzyme B (GrB),
TIA-1, FasL, and Fas has enabled the study of these
components involved in T cell–mediated cytotoxicity in
tissue sections.12-15 Recently, killing inhibitory receptors
(KIRs) were discovered as a new class of molecules that are
CD8� T CELLS CORRELATE WITH SURVIVAL IN CTCL 4323

able to modulate cytotoxic function of natural killer (NK) Table 1. Clinical Characteristics of Patients Included in This Study
and T cells. At present, two groups of KIRs are known: the Plaque-Stage MF Tumor-Stage MF CD30-Negative
Diagnosis (T2N0M0) (T3N0M0) PCLTCL
immunoglobulin (Ig) superfamily-like receptors (CD158a,
CD158b)16 and the lectin-like receptors (CD94).17 These No. of patients 42 20 21
receptor molecules are expressed by subpopulations of NK Age, years
Median 65 67 71
cells and CD8� CTLs. On ligation with MHC class I Range 33-93 47-92 31-88
receptors on target cells, KIRs deliver inhibitory signals that Male/female 2.0 1.0 1.7
impair cytolytic function.18 Thus, expression of KIRs by Follow-up, months
CTLs might impair their capacity to effectively kill virus- Median 64 24 12
infected or tumor cells. Recent in vitro studies demonstrated Range 12-176 3-117 1-56
5-year survival rate, % 85 55 0
the expression of functional KIRs on a CD8�, tumor- Current status, n
specific, cytotoxic, T-cell clone isolated from the peripheral Alive without disease 0 0 1
blood of a patient with MF.19 However, studies of the Alive with disease 35 8 2
expression of KIRs by cytotoxic cells in skin infiltrates of Died of unrelated disease 3 2 1
CTCL lesions have not yet been published, and the impor- Died of lymphoma 6 12 17

tance of the induction of KIRs as a mechanism to evade the

immune response in CTCL is presently unknown.
To further characterize the role of CD8� CTLs in the
antitumor immune response in CTCL, we investigated the
proportions of CD8� T cells and the expression of TIA-1,
GrB, FasL, and KIRs by CD8� CTLs in 83 initial biopsy
samples from patients with plaque-stage MF (n � 42),
tumor-stage MF (n � 20), and CD30� PCLTCL (n � 21)
obtained at the time of diagnosis and 27 follow-up biopsy
samples and correlated the results with clinical behavior.
PATIENTS AND METHODS
Patients
Eighty-three paraffin-embedded skin biopsy samples obtained at the
time of diagnosis from 83 untreated CTCL patients, who included 62
MF patients and 21 CD30� PCLTCL patients, were studied. The
diagnosis of MF and CD30� PCLTCL was based on a combination of
clinical, histologic, and immunophenotypical data as described previ-
ously.1 The duration of skin lesions before a definite diagnosis could be
made varied between 2 months and more than 10 years (median, 103
months) in MF and between 3 and 24 months (median, 12 months) in
patients with CD30� PCLTCL.
Only cases in which the neoplastic T cells had either a CD3�/
CD4�/CD8� (n � 70) or a CD3�/CD4�/CD8� phenotype (n � 13)
were included. Rare cases of MF and CD30� PCLTCL with a
CD3�/CD4�/CD8� phenotype were not selected for this study
because it was impossible to make a reliable estimate of the number of
reactive CD8� T cells in these cases. Follow-up data, including
response to initial therapy and survival, were recorded in each case
(Table 1).
Of the 62 MF biopsy samples taken at the time of diagnosis, 42 were
obtained from plaques from patients with generalized (T2N0M0)
plaque-stage MF, and 20 were obtained from skin tumors from patients
with tumor-stage MF without concurrent lymph node involvement
(T3N0M0). For 19 of these 62 patients, 27 follow-up biopsy samples
from plaques (n � 9) or skin tumors (n � 18) obtained at the time of
relapse or disease progression 1 to 90 months (median, 24 months) after
the first diagnostic biopsy were studied as well. The results of these
biopsies are not included in the statistical analyses and will be
discussed separately.
Frozen sections obtained from the same excision biopsy sample or
another biopsy sample from a similar concurrent lesion were used for
KIR staining in 18 cases, including seven MF plaques, nine MF tumors,
and two CD30� PCLTCL biopsy samples.
Immunohistochemistry
Paraffin sections. Immunostaining on formalin-fixed, paraffin-em-
bedded skin sections with MoAbs against CD8 (DAKO, Glostrup,
Denmark), GrB,20,21 TIA-1 (Coulter Immunology, Hialeah, FL),4 FasL
(clone 33; Transduction Laboratories, Lexington, United Kingdom),
and CD4 (Novocastra, Newcastle on Tyne, United Kingdom), Ber-H2/
CD30 (DAKO) and polyclonal antibodies against CD3 (DAKO) was
performed using a standard three-step streptavidin-biotin-peroxidase–
based technique after antigen retrieval with microwave heating as
described previously.22,23 In 10 cases, which included four patients
with MF plaques, four with MF tumors, and two with CD30�
PCLTCL, double staining for CD8 with GrB and CD8 with TIA-1 was
performed as described previously.24
Frozen sections. Snap-frozen specimens were stored in liquid
nitrogen until use. NK cells were identified by their staining for CD56
(IgG1 subtype; Beckton and Dickinson, Poole, United Kingdom).
Expression of KIR was studied using MoAbs against CD94 (IgG2
subtype), CD158a, and CD158b (Coulter Immunology), as described
previously.25 Double staining for CD94 with CD56 or CD8 was
performed in eight cases, which included three patients with MF
plaques, three with MF tumors, and two with CD30� PCLTCL, by
simultaneous incubation with primary antibodies followed by incuba-
tion with horseradish peroxidase– conjugated goat antimouse IgG1 and
biotinylated goat antimouse IgG2a for the detection of CD56 or CD8
and CD94, respectively. The horseradish peroxidase was visualized by
fluorescein-conjugated tyramine,26,27 whereas the biotin label was
detected with cy3-conjugated streptavidin.
Interpretation of Immunohistochemical Staining
The percentages of CD8� T cells were expressed as a percentage of
the total number of skin-infiltrating cells (both reactive and neoplastic).
Percentages of CD8� T cells were independently estimated by two
observers (M.H.V. and R.W.) to the nearest 5% for the entire tissue
section in a blinded fashion. In the few cases in which there was
disagreement, sections were read jointly by the two investigators, and

54
4324 VERMEER ET AL

Table 2. Percentages of CD8� T Cells in MF and CD30-Negative PCLTCL
CD8 T Cells (%)*
Diagnosis No. of Patients Median Range
MF plaque 42 22† 2-50
MF tumor 20 10† 2-20
CD30� PCTLCL 21 3† 1-25
*The percentage of CD8� T cells is taken as a percentage of the total number
of skin-infiltrating cells (reactive cells and tumor cells).
†Difference between MF plaques versus MF tumors and MF tumors versus
CD30� PCTLCL is significant (Student’s t test, P � .0001).
consensus was reached. The percentages of CD8� T cells positive for
TIA-1, GrB, FasL, and KIR were studied on serial paraffin sections
stained with antibodies against CD3, CD4, CD8, GrB, TIA-1, and FasL
and serial frozen sections stained with antibodies against CD56, CD8,
CD94, CD158a, and CD158b. Stainings were scored as follows:
negative, no or occasional (� 10%) CD8� T cells stained; positive,
10% to 50%; and double positive, more than 50%. To determine
whether and to what extent TIA-1, GrB, FasL, and KIRs are expressed
by inflammatory cells other than CD8� cells, we performed double-
staining experiments as described above.
Statistical Analysis
Comparison of proportions of CD8� T cells and the proportions of
CD8� T cells that express TIA-1, GrB, and FasL between patients with
plaque-stage MF, tumor-stage MF, and CD30� PCLTCL was per-
formed using the two-tailed Student’s t test.
Univariate analysis of possible prognostic factors was performed
using the log-rank test for categorical variables (stage) and Cox
proportional regression analysis for continuous variables (percentage of
CD8� T cells, age). To assess independence of prognostic value,
multivariate analysis was performed by entering the variables of
interest in Cox proportional hazards regression analysis. Actuarial
survival curves in MF were calculated from the time of the initial
diagnostic biopsy until death from disease or end of follow-up using the
Kaplan-Meier method. P values below .05 were considered statistically
significant. All analyses were performed using Statistical Products and
Services Solutions Software (SPSS Inc, Chicago, IL).
RESULTS
Percentage of CD8� CTLs in CTCL
The proportions of CD8� CTLs in the MF and CD30-
negative PCLTCL biopsy samples taken at the time of
diagnosis are presented in Table 2. Major differences were
found between MF plaques and tumors. Percentages of 20%
or more CD8� T cells were observed in 21 (50%) of 42 MF
plaques but not in any of the 20 MF tumors. The median
number of CD8� T cells in MF was 17% (range, 2% to
50%). CD30� PCLTCL showed low percentages of CD8�
T cells, with a median value of 3% and less than 5% of
CD8� T cells in 12 (55%) of 21 cases. The differences in
the proportions of CD8� T cells between plaque-stage and
tumor-stage MF and tumor-stage MF and CD30� PCLTCL
were statistically significant (Student’s t test, P � .0001 and
P � .018, respectively).
The results of the proportions of CD8� T cells in 27 MF
biopsy samples obtained during follow-up are summarized
in Table 3. Taken together, the percentages of CD8� T cells
in nine MF plaques (median, 22%; range, 15% to 27%) and
18 MF tumors (mean, 10%; range, 2% to 17%) were similar
to the percentages observed in plaques and tumors present at
the time of diagnosis. Interestingly, examination of plaques
in three MF patients who had concurrent tumors demon-
strated high percentages of CD8� T cells in the plaques
(median, 25%) compared with the concurrent tumors (me-
dian, 10%; range, 7% to 15%).
Expression of Cytotoxic Proteins GrB, TIA-1, and FasL
by CD8� T Cells
Expression of GrB, TIA-1, and FasL was detected as a
clear, granular, cytoplasmic staining in all cases. Because a
reliable estimation of the percentages of CD8� T cells that
express cytotoxic proteins might be hampered by the fact
that these proteins are expressed not only by CD8� CTLs
but also by the neoplastic T cells of some types of
CTCL28,29 and perhaps by rare reactive CD4� T cells,30
double-staining experiments for CD8 with GrB or CD8 with
TIA-1 were first performed in 10 cases, including four
patients with MF plaques, four with MF tumors, and two
with CD30� PCLTCL. These experiments demonstrated
that TIA-1 and GrB are not or are rarely expressed by
inflammatory cells other than CD8� T cells (Fig 1).

Table 3. Percentages of CD8� T Cells in 27 MF Biopsy Samples Obtained During Follow-Up

Initial Biopsy Samples Follow-Up Biopsy Samples

CD8� T Cells (%) CD8� T Cells (%) Time Interval (months)

Type of Skin Lesion No. of Patients Median Range Type of Skin Lesion No. of Patients Median Range Median Range

MF plaque 5* 19 15-50 MF plaque 6 20 15-25 12 3-36

MF plaque 9† 20 12-40 MF tumor 11 7 2-17 36 18-90
MF plaque 3‡ 25 25-27
MF tumor 5 7 2-15 MF tumor 7 7 2-10 7 1-16

*Patients with plaque-stage MF who developed only new plaques during follow-up.
†Patients with plaque-stage MF who showed disease progression to tumor-stage MF during follow-up.
‡MF plaques in patients with concurrent skin tumors.

55
CD8� T CELLS CORRELATE WITH SURVIVAL IN CTCL
Fig 1. Double staining demonstrates that expression of TIA-1 is limited to
CD8� T cells. Expression of TIA-1 (black dot) was observed on the majority
of CD8� T cells, indicated by the arrows, but not on CD8� cells. Streptavi-
din-biotin-peroxidase technique; hematoxylin counterstain; magnification,
�400.
Moreover, because of the superior morphology in paraffin
sections, the differentiation between CD8� T cells and
neoplastic T cells was generally not difficult. Examination
of serial sections stained for CD8, GrB, TIA-1, and FasL
showed a similar topographic distribution for CD8� T cells
and TIA-1� and/or GrB� cells. In plaque-stage MF,
tumor-stage MF, and CD30� PCLTCL, most (60% to
90%) CD8� T cells expressed TIA-1, whereas only a
minority (generally less than 25%) of CD8� T cells
expressed GrB (Table 4; Fig 2). Staining for FasL
showed granular intracytoplasmic staining in more then
50% of the CD8� T cells in all but a few cases. Taken
together, no significant differences in the relative propor-
tions of CD8� T cells that express TIA-1, GrB, or FasL
Table 4. Expression of TIA-1, GrB, and FasL by CD8� T cells in MF and CD30� PCLTCL
TIA-1
Diagnosis No. of Patients � �
MF plaque 40 0 5
MF tumor 17 0 1
CD30� PCLTCL 16 0 0
NOTE. The expression of cytotoxic proteins by reactive cells is presented as a percentage of the number of CD8� T cells.
Abbreviations: �, � 10% positive CD8� T cells; �, 10%-50% positive CD8� T cells; ��, � 50% positive CD8� T cells.
4325
were detected between MF plaques, MF tumors, and
CD30� PCLTCL.
Expression of KIR by CD8� T Cells
To investigate whether the cytotoxic function of CD8� T
cells is possibly modulated by KIR, the expression of CD94,
CD158a, and CD158b was studied. All antibodies demon-
strated a membranous staining. CD94�, CD158a�, or
CD158b� cells comprised up to 5% of the total number of
reactive cells. In all cases, the number of CD94� cells was
higher than the number of CD158b� cells, which in turn
outnumbered the CD158a� cells. No differences were
observed in the number of reactive cells that expressed
CD94, CD158a, and CD158b between MF plaques, MF
tumors, and CD30� PCLTCL. Double-staining experi-
ments for CD56 or CD8 with CD94 demonstrated coexpres-
sion of CD56 with CD94 in more than 95% of CD94� cells,
whereas colocalization of CD8 with CD94 was limited to a
few scattered cells (Fig 3).
Correlation of CD8� T Cells With Clinical
Characteristics
Cox proportional hazards regression analysis in the total
group of 62 MF patients demonstrated that survival in MF
patients declined with lower numbers of CD8� T cells (P �
.001). Also, within tumor-stage MF, high numbers of
CD8� T cells correlated with an increased survival (P �
.003). In plaque-stage MF, the relation between the percent-
ages of CD8� T cells and survival did not reach signifi-
cance, although a trend was observed toward a better
survival in patients with higher numbers of CD8� T cells.
Univariate analysis demonstrated that stage of disease
(P � .0004) and the percentage of CD8� T cells (P � .001)
but not age (P � .57) correlated with survival. Multivariate
analysis showed that both stage of disease and the number
of CD8� T cells were independent prognostic parameters.
The actuarial survival curves for all MF patients divided the
patients into two groups using the median number of CD8�
T cells (17%) as the cutoff point (Fig 4).
GrB FasL
�� � � �� � � ��
35 17 20 3 0 2 38
16 4 11 2 0 0 17
16 1 11 4 0 0 16
56
4326
Fig 2. Expression of cytotoxic proteins by CD8� T cells in an MF plaque. Serial sections demonstrate (A) localization of infiltrating CD8� T cells and (B) TIA-1
expression by these cells. Inset: the granular staining of TIA-1–positive cells. Streptavidin-biotin-peroxidase technique; hematoxylin counterstain; magnification,
�400; inset, �1,000.
DISCUSSION
In the present study, the expression of cytotoxic
proteins by CD8� T cells and the relation between the
percentages of CD8� T cells and clinical behavior was
studied in MF and CD30� PCLTCL. We demonstrated
that, in all types of CTCL, the large majority of CD8� T
cells expressed TIA-1 and FasL, whereas a minority was
GrB positive.
These results are consistent with the results of a previous
study that demonstrated that the majority of CD8� T cells
expressed TIA-1 but not granzymes.31 Similarly, the neo-
plastic cells in CD8� epidermotropic cytotoxic T-cell
lymphomas, which may be considered the neoplastic equiv-
alents of these reactive CD8� cytotoxic T cells, also
showed constant expression of TIA-1 but rarely expressed
GrB or perforin.32
With respect to the cytotoxic phenotype of these CD8� T
cells, in vitro studies demonstrated that, although TIA-1 is
expressed on both resting and activated CTLs,33 perforins
and granzymes are only expressed after activation34 and
might therefore be a more reliable marker for activated
57
VERMEER ET AL
CTLs. Because TIA-1 was expressed by most CD8� T cells
and GrB was expressed by only a minority of these cells
(5% to 10%) in all groups studied, one may wonder whether
these cells are functionally active CTLs. One possible
explanation for the low GrB expression is that most CD8�
T cells had already secreted their GrB as has been demon-
strated in skin biopsy samples of lichen planus.35 Interest-
ingly, recent studies of our group demonstrated that, in skin
biopsy samples of lichen planus, lupus erythematosus, and
graft-versus-host disease, the majority of CD8� T cells
expressed TIA-1, whereas GrB was expressed by only a
small number of these cells (unpublished data). Epidermal
injury and apoptotic keratinocytes were a constant finding in
these biopsy samples, illustrating the presence of function-
ally active CTLs. Thus, these findings illustrated that the
low expression of GrB compared with TIA-1 by the CD8�
T cells in these CTCLs does not exclude the possibility that
they are functionally active CTLs directed at the neoplastic
T cells.
Tumor cells use various escape mechanisms to evade an
effective antitumor response. Recent studies of our group
CD8� T CELLS CORRELATE WITH SURVIVAL IN CTCL
Fig 3. (A) Double stainings for CD56 (red) and CD94 (green) demonstrate
a low proportion of CD56/CD94 double-positive cells stained yellow. (B) On
double staining for CD8 (red) and CD94 (green), no double-stained cells are
observed. Immunofluorescence technique; hematoxylin counterstain; mag-
nification, �400.
demonstrated that the loss of Fas expression by the neoplastic
cells in aggressive types of CTCL may be one of these
mechanisms.36 Another mechanism to evade the immune
response may be the inhibition of cytolytic function through
KIRs. Expression of KIRs was demonstrated on tumor-specific
CD8� CTLs in melanoma patients, and functional studies
showed that tumor-specific lysis was inhibited by these
KIRs.37 Recent in vitro studies demonstrated the expression of
functional KIRs on a CD8� tumor-specific cytotoxic T-cell
clone isolated from the peripheral blood of an MF patient.19
Whether the in vitro expansion of T cells may lead to the
induction of KIRs is presently unknown. In the present study,
KIRs were expressed only by a few scattered NK cells but
never or rarely by CD8� T cells, which argues against a role
for these KIRs in tumor progression in CTCL.
Correlation of the percentages of CD8� CTLs and clinical
behavior in MF and CD30� PCLTCL showed that the
proportions of CD8� CTLs were significantly higher in MF
plaques than in MF tumors, which in turn outnumbered the
58
4327
Fig 4. Actuarial survival of patients with MF, plaque and tumor stage,
stratified according to the median percentage of CD8� T cells. Large
numbers of CD8� T cells correlate with a favorable prognosis in MF.
percentages of CD8� T cells in CD30� PCLTCL. The
decline in the percentage of CD8� CTLs was associated with
a less favorable prognosis. Also, within tumor-stage MF, a
relation between higher percentages of CD8� CTLs and better
survival was found. Multivariate analysis demonstrated that
both skin stage and the proportion of CD8� T cells were
independent parameters of survival. Examination of 27 addi-
tional follow-up skin biopsy samples of MF patients showed
similar percentages of CD8� CTLs in plaques and tumors
when compared with plaques and tumors biopsied at the time
of diagnosis. Interestingly, MF plaques in patients who had
concurrent skin tumors contained considerable proportions of
CD8� CTLs (median, 25% v 10% in the concurrent tumors),
as seen in MF patients with plaque lesions only. Taken
together, these observations suggest that, in patients who have
already progressed to tumor-stage MF, an effective antitumor
response is still functioning in concurrent MF plaques that
prevents tumor progression at those sites.
Previous studies that address the relation between the
proportions of CD8� CTLs and clinical behavior in CTCL
are few, have mainly been focused on MF, and have reached
conflicting conclusions.8,9 Consistent with our results,
Hoppe et al9 also described a relation between high propor-
tions of CD8� CTLs and a better survival in MF. However,
no statistically significant difference in the percentage of
CD8� CTLs between MF plaques and tumors was found. In
contrast, in a study by Vonderheid et al,8 the percentages of
CD8� CTLs decreased from 11.9% to 6.4% to 3.8% in MF
patches, plaques, and tumors, respectively, but no relation
between the percentages of CD8� CTLs and survival was
found in this study.
In the group of CD30� PCLTCL, characterized by an
extremely poor prognosis, low percentages of CD8� CTLs
(median, 3%) were found. Taken together, high proportions
4328
of CD8� CTLs were found in the patient group with a
favorable prognosis (plaque-stage MF) and low proportions
of CD8� CTLs in groups with a poor prognosis (tumor-
stage MF, CD30� PCLTCL), which suggests that CD8�
CTLs play an important role in the antitumor immune
response in CTCL. This notion is in line with earlier studies
that demonstrated that malignant cells in MF express
tumor-specific antigens that can be recognized by autolo-
gous CD8� CTLs in vitro.4-6
In conclusion, we demonstrated that reactive CD8� T
cells in CTCL had the phenotype of partly activated
REFERENCES
1. Willemze R, Kerl H, Sterry W, et al: EORTC classification for
primary cutaneous lymphomas: A proposal from the Cutaneous Lym-
phoma Study Group of the European Organization for Research and
Treatment of Cancer. Blood 90:354-371, 1997
2. van Doorn R, Van Haselen CW, van Voorst-Vader PC, et al:
Mycosis fungoides disease evolution and prognosis of 309 Dutch
patients. Arch Dermatol 136:504-510, 2000
3. Rook AH, Yoo EK, Grossman DJ, et al: Use of biological
response modifiers in the treatment of cutaneous T-cell lymphoma.
Curr Opin Oncol 10:170-174, 1998
4. Berger CL, Wang N, Christensen I, et al: The immune response
to class I-associated tumor-specific cutaneous T-cell lymphoma anti-
gens. J Invest Dermatol 107:392-397, 1996
5. Berger CL, Longley BJ, Imaeda S, et al: Tumor-specific peptides
in cutaneous T-cell lymphoma: Association with class I major histo-
compatibility complex and possible derivation from the clonotypic
T-cell receptor. Int J Cancer 76:304-311, 1998
6. Bagot M, Echchakir H, Mami-Chouaib F, et al: Isolation of
tumor-specific cytotoxic CD4� and CD4� CD8dim� T-cell clones
infiltrating a cutaneous T-cell lymphoma. Blood 11:4331-4341,
1998
7. Echchakir H, Bagot M, Dorothee G, et al: Cutaneous T-cell
lymphoma reactive CD4� cytotoxic T lymphocyte clones display a
Th1 cytokine profile and use a fas-independent pathway for specific
tumor cell lysis. J Invest Dermatol 115:74-80, 2000
8. Vonderheid EC, Tan E, Sobel EL, et al: Clinical implications of
immunologic phenotyping in cutaneous T-cell lymphoma. J Am Acad
Dermatol 17:40-52, 1987
9. Hoppe RT, Medeiros LJ, Warnke R, et al: CD8-positive
tumor-infiltrating lymphocytes influence the long-term survival of
patients with mycosis fungoides. J Am Acad Dermatol 32:448-453,
1995
10. Kagi D, Vignaux F, Ledermann B, et al: Fas and perforin
pathways as major mechanisms of T-cell-mediated cytotoxicity. Sci-
ence 265:528-530, 1994
11. Peter ME, Krammer PH: Mechanisms of CD95 (APO-1/Fas)-
mediated apoptosis. Curr Opin Immunol 10:545-551, 1998
12. Anderson P, Nagler-Anderson C, O’Brien C, et al: A monoclo-
nal antibody reactive with a 15-kDa cytoplasmic granule-associated
protein defines a subpopulation of CD8� T lymphocytes. J Immunol
144:574-582, 1990
13. Anderson P: TIA-1 structural and functional studies on a new
class of cytolytic effector molecule. Curr Top Microbiol Immunol
198:131-143, 1995
59
VERMEER ET AL
cytotoxic T cells (TIA-1�, GrB�/�, FasL�) but did not
express KIRs. Moreover, we demonstrated that high per-
centages of infiltrating CD8� CTLs were associated with a
better prognosis. Taken together, these observations suggest
that CD8� CTLs could play an important role in the
antitumor response in CTCL.
ACKNOWLEDGMENT
We thank Els de Vries and Wim Vos for their excellent technical
assistance.
14. Griffiths GM, Mueller C: Expression of perforin and granzymes
in vivo: Potential diagnostic markers for activated cytotoxic cells.
Immunol Today 12:415-419, 1991
15. Krahenbuhl O, Rey C, Jenne D, et al: Characterization of
granzymes A and B isolated from granules of cloned human cytotoxic
T lymphocytes. J Immunol 141:3471-3477, 1988
16. Moretta A, Sivori S, Vitale M, et al: Existence of both inhibitory
(p58) and activatory (p50) receptors for HLA-C molecules in human
natural killer cells. J Exp Med 182:875-884, 1995
17. Moretta A, Vitale M, Sivori S, et al: Human natural killer cell
receptor for HLA-class I molecules: Evidence that the Kp43 (CD94)
molecule functions as a receptor for HLA-B allele. J Exp Med
180:545-555, 1994
18. Mingari MC, Schiavetti F, Ponte M, et al: Human CD8� T
lymphocyte subsets that express HLA I-specific inhibitory receptors
represent oligoclonally or monoclonally expanded cell populations.
Proc Nat Acad Science 93:12433-12438, 1996
19. Bagot M, Martinvallet H, Echchakir H, et al: Functional
inhibitory receptors expressed by a cutaneous T cell lymphoma-
specific cytolytic clonal T cell population. J Invest Dermatal
115:994-999, 2000
20. Kummer JA, Kamp AM, van Katwijk M, et al: Production and
characterization of monoclonal antibodies raised against recombinant
human granzymes A and B showing cross reactions with the natural
proteins. J Immunol Methods 163:77-83, 1993
21. Kummer JA, Kamp AM, Tadema TM, et al: Localization and
identification of granzymes A- and B-expressing cells in normal human
lymphoid tissue and peripheral blood. Clin Exp Immunol 100:164-172,
1995
22. Oudejans JJ, Kummer JA, Jiwa NM, et al: Granzyme B
expression in Reed-Sternberg cells of Hodgkin’s disease. Am J Pathol
148:233-240, 1996
23. Zoi-Toli O, Meijer CJLM, Oudejans JJ, et al: Prognostic
significance of Fas and Fas-ligand expression in cutaneous B-cell
lymphomas. J Pathol 189:533-538, 1999
24. Oudejans JJ, Jiwa NM, Kummer JA, et al: Activated cytotoxic
T-cells as prognostic marker in Hodgkin’s disease. Blood 89:1376-
1382, 1997
25. Dukers DF, Vermeer MH, Jaspars LH, et al: Expression of killer
cell inhibitory receptors is restricted to true NK cell lymphomas and a
subset of intestinal enteropathy-type T cell lymphomas with a cytotoxic
phenotype. J Clin Pathol 54:224-228, 2001
26. Raap AK, van der Corput MPC, Vervenne RAW, et al: Ultra-
sensitive FISH using peroxidase-mediated deposition of biotin or
fluorochrome tyramides. Hum Mol Genet 4:529-534, 1995
CD8� T CELLS CORRELATE WITH SURVIVAL IN CTCL
27. Dukers DF, ten Berge RL, Oudejans JJ, et al: A cytotoxic
phenotype does not predict clinical outcome in anaplastic large-cell
lymphomas. J Clin Pathol 52:129-136, 1999
28. Vermeer MH, Geelen FAMJ, Kummer JA, et al: Expression of
cytotoxic proteins by neoplastic T cells in mycosis fungoides increases
with progression from plaque stage to tumor stage disease. Am J Pathol
154:1203-1210, 1999
29. Kummer JA, Vermeer MH, Dukers DF, et al: Most primary
cutaneous CD30 positive lymphoproliferative disorders have a CD4-
positive cytotoxic T-cell phenotype. J Invest Dermatol 109:636-640, 1997
30. Hahn S, Gehri R, Erb P: Mechanism and biological significance
of CD4-mediated cytotoxicity. Immunol Rev 146:57-79, 1995
31. Wood GS, Edinger A, Hoppe RT, et al: Mycosis fungoides skin
lesions contain CD8� tumor-infiltrating lymphocytes expressing an
activated: MHC-restricted cytotoxic T-lymphocyte phenotype. J Cutan
Pathol 21:151-156, 1994
32. Berti E, Tomasini D, Vermeer MH, et al: Primary cutaneous
CD8-positive epidermotropic cytotoxic T-cell lymphomas: A distinc-
60
4329
tive clinicopathological entity with an aggressive behavior. Am J
Pathol 155:483-492, 1999
33. Podack ER: Execution and suicide: Cytotoxic T cells enforce
Draconian laws through separate molecular pathways. Curr Opin
Immunol 7:11-16, 1995
34. Griffiths GM: The cell biology of CTL killing. Curr Opin
Immunol 7:343-348, 1995
35. Shimizu M, Higaki T, Kawashima M: The role of granzyme
B-expressing CD8-positive T cells in apoptosis of keratinocytes in
lichen planus. Arch Dermatol Res 289:527-532, 1997
36. Zoi-Toli O, Vermeer MH, de Vries E, et al: Expression of Fas
and Fas-ligand in primary cutaneous T-cell lymphoma: Association
between lack of Fas expression and aggressive types of CTCL. Br J
Dermatol 143:313-319, 2000
37. Speiser DE, Pittet M, Valmori D, et al: In vivo expression of
natural killer cell inhibitory receptors by human melanoma-specific
cytolytic T lymphocytes. J Exp Med 190:775-782, 1999
 Chapter 5

A novel splice variant of the Fas gene in patients

with cutaneous T-cell lymphoma

Cancer Res. 2002 Oct. 1;62(19):5389-92

[CANCER RESEARCH 62, 5389 –5392, October 1, 2002]
Advances in Brief
A Novel Splice Variant of the Fas Gene in Patients with
Cutaneous T-Cell Lymphoma
Remco van Doorn,1,2 Remco Dijkman,1 Maarten H. Vermeer, Theo M. Starink, Rein Willemze, and
Cornelis P. Tensen3
Department of Dermatology, Vrije Universiteit Medical Center, 1081 HV Amsterdam, the Netherlands, and Department of Dermatology, Leiden University Medical Center, 2333
AL Leiden, the Netherlands
Abstract
Defective apoptosis signaling has been implicated in the pathogenesis of
primary cutaneous T-cell lymphomas (CTCLs), a group of malignancies
derived from skin-homing T cells. An important mediator of apoptosis in
T cells is the Fas receptor. We identified a novel splice variant of the Fas
gene that displays retention of intron 5 and encodes a dysfunctional Fas
protein in 13 of 22 patients (59%) in both early and advanced CTCL.
Impairment of Fas-induced apoptosis resulting from aberrant splicing
potentially contributes to the development and progression of CTCL by
allowing continued clonal expansion of activated T cells and by reducing
susceptibility to antitumor immune responses.
Introduction
CTCLs4 are a group of clinically heterogeneous malignancies of
mature skin-homing T cells (1). The low mitosis index in the early
stages of MF, the most common type of CTCL, and resistance to
treatment with genotoxic agents suggest that defective apoptosis sig-
naling plays a role in the pathogenesis of CTCL (2). One of the key
regulators of apoptosis in mature T cells is Fas, a homotrimer cell
surface receptor that mediates apoptosis upon cross-linking to Fas
ligand. The Fas protein contains an intracellular region essential for
transduction of the apoptotic signal termed the death domain; muta-
tions in this domain dominantly interfere with Fas function (3, 4). Fas
is a critical mediator of activation-induced cell death, a propriocidal
mechanism involved in homeostasis of activated T cells (5). In addi-
tion, Fas signaling renders neoplastic cells susceptible to antitumor
immune responses executed by cytotoxic T cells through cross-linking
with Fas ligand. We previously reported loss of Fas protein expression
in aggressive types of CTCL (6). In non-Hodgkin’s lymphoma, del-
eterious mutations of the Fas gene have been identified in 11% of
patients (7). In the present study, we examined the occurrence of
splice variants and mutations of the Fas gene as well as Fas protein
expression in lesional skin biopsy specimens from patients with
CTCL.
Materials and Methods
Lesional skin biopsy specimens were obtained from 15 patients with
advanced CTCL, including 7 patients with tumor-stage MF (T3N0M0) and
Received 7/23/02; accepted 8/19/02.
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance with
18 U.S.C. Section 1734 solely to indicate this fact.
1
R. v. D. and R. D. contributed equally to this study.
2
R. v. D. is supported by a grant from the Dutch foundation “De Drie Lichten.”
3
To whom requests for reprints should be addressed, at Department of Dermatology,
Leiden University Medical Center, Wassenaarseweg 72, 2333 AL Leiden, the Nether-
lands, Phone: 31-71-5271903; Fax: 31-71-5271910; E-mail:c.p.tensen@lumc.nl.
4
The abbreviations used are: CTCL, cutaneous T-cell lymphoma; MF, mycosis
fungoides; PCLTCL, primary cutaneous large T-cell lymphoma.
5389
63
8 patients with CD30� PCLTCL. For the detection of splice variants in
early CTCL, skin biopsy specimens from 7 patients with plaque-stage MF
(T2N0M0; Ref. 8) were obtained. Cultured keratinocytes, phytohemag-
glutinin-activated CD4� and CD8� T cells, and skin biopsy specimens
from patients with benign cutaneous lymphocytic infiltrates (atopic derma-
titis, graft-versus-host-disease, and chronic discoid lupus erythematosus)
were used as controls. Total cellular RNA was extracted from homogenized
samples before treatment with RQ1 DNase (Promega, Madison, WI).
cDNA synthesis was performed using Expand Reverse Transcriptase
(Roche, Mannheim, Germany) after priming with an oligo(dT)12–18 primer
(Invitrogen, Breda, the Netherlands). PCR amplification of cDNA was
performed with five overlapping primer pairs (FAS I-V) covering the entire
coding region of the Fas gene used previously (9) and an additional primer
pair encompassing intron 5 (FAS intron: sense, 5� TCAAGGAATGCA-
CACTCACC 3�; antisense, 5� CCAAACAATTAGTGGAATTG 3�). PCR
was performed under the following conditions: denaturing for 30 s at 94°C;
annealing for 60 s at 58°C for primer pairs I and II and at 50°C for primer
pairs III, IV, and V, and intron 5; and extension for 60 s at 72°C; for 35
cycles. PCR products were separated by electrophoresis on a 2% agarose
gel containing ethidium bromide and on a 12.5% acrylamide gel stained
using PlusOne DNA Silver Staining Kit (Amersham Pharmacia, Uppsala,
Sweden). PCR products were extracted from agarose gel and purified using
the QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany) for reampli-
fication under identical conditions and direct sequence analysis using the
dideoxy chain termination method. The presence of transmembrane helices
was predicted using TMHMM version 2.0 software (CBS, Denmark).
Snap-frozen skin biopsy specimens were stained with a monoclonal anti-
body to Fas (Fas6 antibody; kindly provided by Dr. L. A. Aarden) as
described previously (6).
Results and Discussion
In three of seven (43%) patients with tumor-stage MF and four
of eight (50%) patients with CD30� PCLTCL, an aberrantly
spliced transcript of the Fas gene was identified that has not been
described previously. Sequence analysis revealed that this splice
variant displays selective retention of intron 5, a 152-bp sequence,
leading to frameshift and formation of a truncated protein (Figs. 1
and 2). Aberrant splicing of Fas pre-mRNA was not due to splice
site mutations because none were detected in intron 5 or in its
boundaries. The presence of this particular splice variant was
confirmed by using a different primer set designed to amplify the
exonic sequences flanking intron 5 and cannot be due to contam-
ination with genomic DNA because this would have generated a
larger PCR product containing not only intron 5 but also intron 4.
Subsequent examination of an additional group of seven patients
with early patch/plaque-stage MF for this splice variant revealed
that it was also present in six of seven (85%) patients. It could not
be demonstrated in benign cutaneous lymphocytic infiltrates, ke-
ratinocytes (Fig. 1A), or phytohemagglutinin-activated CD4� and
CD8� T cells (data not shown). This indicates that the aberrant
 Fas SPLICE VARIANT IN CTCL

Fig. 1. Detection and analysis of Fas transcripts.

A, reverse transcription-PCR analysis of human Fas
mRNA using primer pair FAS III. In addition to the
expected PCR product of 240 bp, a second PCR
product of 392 bp (indicated by the arrow) was
detected. Input cDNA template was synthesized
from RNA isolated from the following sources. A1:
Lane 1, H2O (no cDNA) control; Lane 2, cultured
primary keratinocytes; Lanes 3, 5, 6, 7, 11, and 12,
MF tumor-stage biopsies; Lanes 4, 8, 9, and 13–17,
PCLTCL biopsies; Lane 10, benign lymphocytic
skin infiltrate biopsy. A2: Lane 1, H2O (no cDNA)
control; Lanes 2– 6, benign lymphocytic skin infil-
trate biopsies; Lanes 7–13, patch/plaque-stage bi-
opsies. Lane M, molecular size marker (Generuler;
MBI Fermentas, St. Leon-Rot, Germany). B, se-
quence analysis of reverse transcription-PCR prod-
ucts of 240 and 392 bp demonstrating the wild-type
transcript and the alternative transcript that exhibits
retention of a 152-bp sequence corresponding to
intron 5.

Fig. 2. Top, schematic representation of the FAS-encoding exons, the FAS protein, and amplified regions of FAS cDNA in this study. Bottom, location of retained intron found in
cDNA synthesized from RNA isolated from CTCL biopsies and schematic representation of altered protein encoded by mRNA with retained intron 5. CRD, cysteine-rich domains;
TM, transmembrane domain. Nucleotides are numbered starting with the ATG encoding the initiating methionine as 1 (taken from GenBank accession number M67454).
5390

64
 Fas SPLICE VARIANT IN CTCL

Table 1 Clinical features, Fas mutations and alternative transcripts, and Fas protein expression of 15 patients with advanced CTCL
Fas protein
Patient no. Diagnosis Age, sex Disease coursea expression Nucleotide change Type of mutation Protein alteration
1 MF tumor stage 46 F D� 12 mo � 836 C3T Silent
2 MF tumor stage 83 F D� 18 mo � None None
3 MF tumor stage 88 M D� 10 mo � Insertion of 152 bp at nucleotide 699 Frameshift Termination at codon 201
416 A3G Silent
4 MF tumor stage 68 M D� 15 mo � 836 C3T Silent
5 MF tumor stage 57 M A� 36 mo � (50%) Insertion of 152 bp at nucleotide 699 Frameshift Termination at codon 201
6 MF tumor stage 54 M D� 9 mo � None None
7 MF tumor stage 68 F D� 89 mo � Insertion of 152 bp at nucleotide 699 Frameshift Termination at codon 201
836 C3T Silent
8 CD30� PCLTCL 87 F D� 5 mo � Insertion of 152 bp at nucleotide 699 Frameshift Termination at codon 201
406 G3C Missense R71T
417 G3T Missense V75F
9 CD30� PCLTCL 74 M D� 18 mo � None None
10 CD30� PCLTCL 63 F D� 14 mo � None None
11 CD30� PCLTCL 82 F D� 27 mo � None None
12 CD30� PCLTCL 73 M D� 31 mo ndb Insertion of 152 bp at nucleotide 699 Frameshift Termination at codon 201
13 CD30� PCLTCL 81 M D� 10 mo � None None
14 CD30� PCLTCL 67 M D� 9 mo � Insertion of 152 bp at nucleotide 699 Frameshift Termination at codon 201
892 C3T Missense I233T
15a CD30� PCLTCL 87 M D� 24 mo � (75%) Insertion of 152 bp at nucleotide 699 Frameshift Termination at codon 201
15b CD30� PCLTCL 87 M D� 17 mo � (75%) Insertion of 152 bp at nucleotide 699 Frameshift Termination at codon 201
a
A�, alive with disease; D�, died of lymphoma; mo, months after taken biopsy; D�, died of other cause.
b
nd, not done.

splicing of Fas generating this particular variant is specific for
CTCL and is an early event in lymphomagenesis. Detection of this
transcript might therefore be useful in the early diagnosis of this
group of lymphomas. From one patient with CD30� PCLTCL, we
analyzed the presence of this splice variant in a skin biopsy
specimen obtained at the time of diagnosis and in a separate skin
biopsy specimen obtained 7 months later (Table 1, patient 15a and
15b). Analysis revealed the identical splice variant (Fig. 1A, Lanes
13 and 14), suggesting that alternative splicing of Fas is a persis-
tent event.
Translation of the aberrantly spliced variant results in a protein that
contains 32 novel amino acid residues and terminates prematurely at
codon 201. The predicted translation product encodes a form of Fas
that does not contain a functional transmembrane anchor domain and
lacks the death domain (Fig. 2). Therefore, if the aberrantly spliced
variant is translated exclusively in a cell, the resulting aberrant Fas
receptor would not be expressed on the plasma membrane. If both the
wild-type and alternative transcript are translated simultaneously,
assembly and expression of dysfunctional mixed receptor complexes
are predicted to occur (10, 11). In skin biopsy specimens from the 15
patients with advanced CTCL, Fas protein could not be detected on
neoplastic T cells in 9 cases (56%). In five of the seven skin biopsy
specimens in which the splice variant was identified, Fas protein
expression was not detectable (Table 1).
In addition, three missense point mutations and two silent point
mutations were detected in the 15 patients with advanced CTCL.
Two missense point mutations detected in one patient were located
in the extracellular domain (G to C at position 406 causing sub-
stitution of Thr for Arg at codon 71 and G to T at position 417
causing substitution of Phe for Val at codon 75). One point
mutation (C to T at position 892 causing substitution of Thr for Ile
at codon 233) was located in the cytoplasmic domain and is
predicted to interfere dominantly with the function of Fas. The
observed low frequency of deleterious point mutations in the
Fas gene in CTCL corresponds with the recent findings of Dereure
et al. (12).
Although alternative splicing is used extensively by genes involved
in apoptosis, a role for this mechanism in the pathogenesis of lym-
phoma has scarcely been described thus far (13, 14). The function of
the Fas gene is regulated at the level of pre-mRNA splicing in human
5391
thymocytes, where five splice variants have been demonstrated (11).
Another three different Fas splice variants have been detected in
leukocytes from patients with silicosis (9). One of the regulators of
alternative splicing of Fas is TIA-1, an apoptosis-promoting protein
that has been shown to be expressed by neoplastic T cells in a subset
of patients with MF (15, 16).
In conclusion, we found a splice variant of the Fas tumor suppres-
sor gene, associated with formation of a dysfunctional protein that
dominantly interferes with Fas-induced apoptosis, frequently and spe-
cifically occurring in patients with CTCL. This could be implicated in
the pathogenesis of CTCL by allowing continued clonal expansion of
activated T cells and in tumor progression by reducing the suscepti-
bility to antitumor immune responses.
Acknowledgments
We appreciate the help of Dr. Zoi-Toli, Ing. P. van Beek, and Ing. A. Mulder
in preparing sections and staining biopsy material.
References
1. Willemze, R., Kerl, H., Sterry, W., Berti, E., Cerroni, L., Chimenti, S., Diaz-Perez,
J. L., Geerts, M. L., Goos, M., Knobler, R., Ralfkiaer, E., Santucci, M., Smith, N.,
Wechsler, J., van Vloten, W. A., and Meijer, C. J. EORTC classification for primary
cutaneous lymphomas: a proposal from the Cutaneous Lymphoma Study Group of the
European Organization for Research and Treatment of Cancer. Blood, 90: 354 –371,
1997.
2. Kikuchi, A., and Nishikawa, T. Apoptotic and proliferating cells in cutaneous lym-
phoproliferative diseases. Arch. Dermatol., 133: 829 – 833, 1997.
3. Cascino, I., Papoff, G., De Maria, R., Testi, R., and Ruberti, G. Fas/Apo-1 (CD95)
receptor lacking the intracytoplasmic signaling domain protects tumor cells from
Fas-mediated apoptosis. J. Immunol., 156: 13–17, 1996.
4. Vaishnaw, A. K., Orlinick, J. R., Chu, J. L., Krammer, P. H., Chao, M. V., and Elkon,
K. B. The molecular basis for apoptotic defects in patients with CD95 (Fas/Apo-1)
mutations. J. Clin. Investig., 103: 355–363, 1999.
5. Chan, K. F., Siegel, M. R., and Lenardo, J. M. Signaling by the TNF receptor
superfamily and T cell homeostasis. Immunity, 13: 419 – 422, 2000.
6. Zoi-Toli, O., Vermeer, M. H., De Vries, E., Van Beek, P., Meijer, C. J., and
Willemze, R. Expression of Fas and Fas-ligand in primary cutaneous T-cell lym-
phoma (CTCL): association between lack of Fas expression and aggressive types of
CTCL. Br. J. Dermatol., 143: 313–319, 2000.
7. Gronbaek, K., Straten, P. T., Ralfkiaer, E., Ahrenkiel, V., Andersen, M. K., Hansen,
N. E., Zeuthen, J., Hou-Jensen, K., and Guldberg, P. Somatic Fas mutations in
non-Hodgkin’s lymphoma: association with extranodal disease and autoimmunity.
Blood, 92: 3018 –3024, 1998.
8. Bunn, P., and Lamberg, S. Report of the Committee on Staging and Classification of
Cutaneous T-Cell Lymphomas. Cancer Treat. Rep., 63: 725–728, 1979.

65
 Fas SPLICE VARIANT IN CTCL
9. Otsuki, T., Sakaguchi, H., Tomokuni, A., Aikoh, T., Matsuki, T., Isozaki, Y.,
Hyodoh, F., Kawakami, Y., Kusaka, M., Kita, S., and Ueki, A. Detection of alterna-
tively spliced variant messages of Fas gene and mutational screening of Fas and Fas
ligand coding regions in peripheral blood mononuclear cells derived from silicosis
patients. Immunol. Lett., 72: 137–143, 2000.
10. Siegel, R. M., Frederiksen, J. K., Zacharias, D. A., Chan, F. K., Johnson, M., Lynch,
D., Tsien, R. Y., and Lenardo, M. J. Fas preassociation required for apoptosis
signaling and dominant inhibition by pathogenic mutations. Science (Wash. DC),
288: 2354 –2357, 2000.
11. Jenkins, M., Keir, M., and McCune, J. M. A membrane-bound Fas decoy receptor
expressed by human thymocytes. J. Biol. Chem., 275: 7988 –7993, 2000.
12. Dereure, O., Levi, E., Vonderheid, E. C., and Kadin, M. E. Infrequent Fas mutations
but no Bax or p53 mutations in early mycosis fungoides: a possible mechanism for the
5392
66
accumulation of malignant T lymphocytes in the skin. J. Investig. Dermatol., 118:
949 –956, 2002.
13. Jiang, Z., and Wu, J. Alternative splicing and programmed cell death. Proc. Soc. Exp.
Biol. Med., 220: 64 –72, 1999.
14. Xerri, L., Hassoun, J., Devilard, E., Birnbaum, D., and Birg, F. BCL-X and the
apoptotic machinery of lymphoma cells. Leuk. Lymphoma, 28: 451– 458, 1998.
15. Forch, P., Puig, O., Kedersha, N., Martinez, C., Granneman, S., Seraphin, B.,
Anderson, P., and Valcarcel, J. The apoptosis-promoting factor TIA-1 is a regulator
of alternative pre-mRNA splicing. Mol. Cell, 6: 1089 –1098, 2000.
16. Vermeer, M. H., Geelen, F. A., Kummer, J. A., Meijer, C. J., and Willemze, R.
Expression of cytotoxic proteins by neoplastic T cells in mycosis fungoides increases
with progression from plaque stage to tumor stage disease. Am. J. Pathol., 154:
1203–1210, 1999.
 Chapter 6

Aberrant expression of the tyrosine kinase receptor

EphA4 and the transcription factor Twist in Sezary
syndrome identified by gene expression analysis

Cancer Res. 2004 Aug. 15;64(16)5578-86

[CANCER RESEARCH 64, 5578 –5586, August 15, 2004]
Aberrant Expression of the Tyrosine Kinase Receptor EphA4 and the Transcription
Factor Twist in Sézary Syndrome Identified by Gene Expression Analysis
Remco van Doorn, Remco Dijkman, Maarten H. Vermeer, Jacoba J. Out-Luiting,
Elisabeth M. H. van der Raaij-Helmer, Rein Willemze, and Cornelis P. Tensen
Department of Dermatology, Leiden University Medical Center, Leiden, the Netherlands
ABSTRACT
Sézary syndrome (Sz) is a malignancy of CD4� memory skin-homing T
cells and presents with erythroderma, lymphadenopathy, and peripheral
blood involvement. To gain more insight into the molecular features of Sz,
oligonucleotide array analysis was performed comparing gene expression
patterns of CD4� T cells from peripheral blood of patients with Sz with
those of patients with erythroderma secondary to dermatitis and healthy
controls. Using unsupervised hierarchical clustering gene, expression pat-
terns of T cells from patients with Sz were classified separately from those
of benign T cells. One hundred twenty-three genes were identified as
significantly differentially expressed and had an average fold change
exceeding 2. T cells from patients with Sz demonstrated decreased expres-
sion of the following hematopoietic malignancy-linked tumor suppressor
genes: TGF-� receptor II, Mxi1, Riz1, CREB-binding protein, BCL11a,
STAT4, and Forkhead Box O1A. Moreover, the tyrosine kinase receptor
EphA4 and the potentially oncogenic transcription factor Twist were
highly and selectively expressed in T cells of patients with Sz. High
expression of EphA4 and Twist was also observed in lesional skin biopsy
specimens of a subset of patients with cutaneous T cell lymphomas related
to Sz, whereas their expression was nearly undetectable in benign T cells
or in skin lesions of patients with inflammatory dermatoses. Detection of
EphA4 and Twist may be used in the molecular diagnosis of Sz and related
cutaneous T-cell lymphomas. Furthermore, the membrane-bound EphA4
receptor may serve as a target for directed therapeutic intervention.
INTRODUCTION
Sézary syndrome (Sz) is a leukemic variant of cutaneous T cell
lymphoma (CTCL) characterized by erythroderma, generalized
lymphadenopathy, and the presence of neoplastic CD4� skin-homing
memory T cells (Sézary cells) in the skin, lymph nodes, and peripheral
blood (1, 2). In the early phases of the disease, differentiation between
Sz and benign forms of erythroderma secondary to atopic dermatitis,
chronic dermatitis, or adverse drug reactions may be very difficult.
Because of the similarity of their histopathological and immunophe-
notypical features, Sz is generally considered to be a leukemic variant
of the more common CTCL mycosis fungoides (3). Severe pruritus,
ectropion, alopecia, palmoplantar keratoderma, and generalized im-
munosuppression are common associated features. The results of
various treatments for Sz are generally disappointing. Subjects with
Sz have an unfavorable prognosis with an estimated 5-year survival of
15% (4).
Previous studies on the pathogenesis of Sz have pointed to aberra-
tions of signaling by the STAT family of transcription factors (5– 8),
overexpression of JunB (9), diminished expression of the tumor
suppressor genes TGF-� receptor II (10), p15, p16 (11), and Fas (12).
Other studies have pointed to the presence of chromosomal alter-
ations, indicating genomic instability (13, 14). Nonetheless, compre-
Received 4/13/04; revised 6/14/04; accepted 6/22/04.
The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance with
18 U.S.C. Section 1734 solely to indicate this fact.
Note: R. van Doorn and R. Dijkman contributed equally to this study.
Requests for reprints: Cornelis P. Tensen, Department of Dermatology, Sylvius
Building Room 3042, Wassenaarseweg 72, 2333 AL Leiden, the Netherlands. Phone:
31-71-5271900; Fax: 31-71-5271910; E-mail address: c.p.tensen@lumc.nl.
5578
69
hension of the pathogenic mechanisms implicated in the development
and progression of this T-cell malignancy is still limited.
To expand the understanding of the pathogenesis and to facilitate
the molecular diagnosis of Sz, we performed oligonucleotide array
analysis on T cells isolated from the peripheral blood of patients with
Sz. Gene expression patterns were compared with those of CD4� T
cells isolated from the blood of patients with erythroderma secondary
to atopic or chronic dermatitis and of healthy volunteers.
The gene expression patterns of malignant T cells from patients
with Sz demonstrated consistent differences with those of benign T
cells. The transcriptional program distinctive for the malignant phe-
notype revealed evidence for the dysregulation of multiple growth-
regulatory signal transduction pathways, such as the TGF-� receptor
and c-myc pathway. We identified high and selective expression in Sz
and related CTCLs of the tyrosine kinase receptor EphA4 and the
transcription factor Twist. These genes, which may also be implicated
in the pathogenesis of Sz, could serve as molecular markers in the
diagnosis or as therapeutic targets in the treatment of this malignancy.
MATERIALS AND METHODS
Selection of Patients. Cryopreserved blood samples from 10 patients with
Sz (seven males, three females; median age 62 years) were available for
inclusion in this study. Sz was defined by the criteria of the European
Organization for Research and Treatment of Cancer classification (4). All
patients showed highly elevated CD4/CD8 ratios and clonal T cells in the
peripheral blood, as described previously (15). Follow-up data revealed that all
patients had died of Sz; the median survival time was 26 months.
For comparative analysis, blood samples were obtained from a group of
eight control patients that included five patients with a benign form of eryth-
roderma (BE; three males, two females; median age, 57 years) and three
healthy volunteers. The BE group included three patients with atopic dermatitis
and two with idiopathic chronic dermatitis. Examination of the peripheral
blood in these five BE patients showed an absence of atypical T cells, normal
CD4/CD8 ratios, and no evidence of a T cell clone. Follow-up (median
duration 29 months) was unremarkable. From all Sz patients and controls,
peripheral blood samples were collected at the time of diagnosis before
systemic or phototherapeutic treatment had been given. In addition, for real-
time quantitative PCR experiments, lesional skin biopsy specimens from
patients with plaque-stage (T2N0M0) and tumor-stage (T3N0M0) mycosis fun-
goides as well as primary cutaneous CD30-negative large T cell lymphoma
were obtained. As controls for these experiments phytohemagglutinin-acti-
vated and interleukin 2-expanded peripheral blood T cells as well as biopsies
of benign cutaneous lymphocytic infiltrates from patients with inflammatory
skin diseases (M. Jessner, chronic discoid lupus erythematosus, graft-versus-
host disease, benign erythroderma secondary to dermatitis, lichen planus, and
cutaneous vasculitis) were used.
T-Cell Isolation and RNA Isolation. Peripheral blood mononuclear cells
were obtained by Ficoll density centrifugation. The percentage of CD4� Sz cells
in the peripheral blood mononuclear cells samples of Sz patients ranged from 90
to 97%, as verified by fluorescence-activated cell sorter analysis. Mononuclear
cells of the patients with BE and from healthy volunteers were subjected to further
purification by negative selection using magnetic beads (CD4� T-cell isolation
kit, Miltenyi Biotec, Bergisch Gladbach, Germany). The purification of control
samples yielded �95% CD4� T cells, similar to the percentage of CD4� T cells
in peripheral blood mononuclear cell samples from Sz patients. RNA was ex-
tracted from T-cell samples and homogenized cryopreserved skin biopsy samples
using the RNeasy kit (Qiagen, Hilden, Germany).
 GENE EXPRESSION PROFILING OF SÉZARY SYNDROME
Table 1 Sequences of primers used for amplification of selected transcripts by qPCR
Name Accession Forward primer (5�-3�)
EphA4 L36645 ggaaggcgtggtcactaaat
Twist BC036704 cactgaaaggaaaggcatca
STAT4 L78440 cagaaaggggtgacaaaggt
TNFSF7 L08096 atcacacaggacctcagcag
LCK AF228313 ctacgggacattcaccatca
CREBBP AJ251843 cacaagtccatttggacagc
TGF-� R2 M85079 atctcgctgtaatgcagtgg
SATB1 NM_002971 gacttttagcccagcagtcc
CCR4 NM_005508 actgctgccttaatcccatc
KIR3DL2 AF263617 gggacctcagtggtcatctt
NOTE. Also indicated are sizes of the PCR products and accession numbers of the respective genes.
Abbreviation: bp, base pair(s).
Fig. 1. Unsupervised hierarchical clustering of gene expres-
sion profiles generated from CD4� T cells from peripheral
blood of 10 patients with Sz, CD4� T cells from three healthy
volunteers, and five patients with BE. The dendrograms are
generated using a hierarchical clustering algorithm on 12,625
transcripts based on the average-linkage method.
Preparation and Hybridization of Fluorescent-Labeled antisense RNA.
Samples and microarrays were processed according to the manufacturer’s protocol
(available from Affymetrix, Santa Clara, CA). In brief, using the MessageAmp
antisense RNA kit (Ambion, Huntingdon, United Kingdom), total RNA was
reverse transcribed using an oligodeoxythymidylic acid-T7 promoter primer to
prime first-strand synthesis. After second-strand synthesis, the purified cDNA
product was in vitro transcribed using T7 RNA polymerase, biotin-UTP, and
biotin-CTP to generate fragmented biotinylated antisense RNA.
Fragmented antisense RNA (5 �g) was hybridized to a Human Genome
U95Av2 Array (Affymetrix), interrogating 12,625 human transcripts, for 16 h
at 45°C with constant rotation at 60 rpm. After hybridization, the microarray
was washed, stained on an Affymetrix fluidics station, and scanned with an
argon-ion confocal laser with 488-nm excitation and 570-nm detection wave-
lengths.
Microarray Experimental Design and Data Analysis. The array images
were quantified using MicroArray Suite v5.0 software (Affymetrix, Santa
Clara, CA). The average fluorescence intensity was determined for each
microarray, and then the output of each experiment was globally scaled to 200.
Normalization was performed using variant stability and normalization (16)
part of the R statistical software package.3 Significance analysis of microarrays
(SAM; ref. 17) was applied to compare gene expression patterns of T cells
from 10 Sz patients with those of the eight controls. A false discovery rate of
�1 was chosen to select genes significantly up- or down-regulated. Gene
expression patterns were further analyzed, and output was visualized using
Spotfire DecisionSite (Spotfire, Göteborg, Sweden) and MicroArray Suite v5.0
software.
Real-Time Quantitative PCR. cDNA synthesis was performed on 1 �g of
total RNA after treatment with RQ1 DNase I (Promega, Madison, WI) using
Superscript III reverse transcriptase (Invitrogen, Breda, the Netherlands) and
an oligo(dT)12–18 primer (Invitrogen, Breda, the Netherlands) in a final volume
of 20 �l. Real-time quantitative PCR (qPCR) was performed using SYBR
Green PCR Master Mix in an ABI-Prism 7700 sequence detection system
(Applied Biosystems, Nieuwerkerk aan den IJssel, the Netherlands). The
primer sequences (Invitrogen) of selected transcripts are given in Table 1. The
cycle parameters for these transcripts and for the housekeeping genes U1A and
RPS11 used for normalization were as follows: denaturing for 15 s at 95°C;
annealing and extension for 60 s at 60°C, for 40 cycles. Specificity of the PCR
product was confirmed by agarose gel electrophoresis and subsequent DNA
3
www.bioconductor.org.
5579
70
Reverse primer (5�-3�) Product size (bp)
tctgccatcatttttcctga 96
ggccagtttgatcccagtat 107
atgggaagaaggtctgatgg 102
atacgtagctgccccttgtc 105
gttggtcatccctgggtaag 99
gttgaccatgctctgtttgc 99
cacgttgtccttcatgcttt 99
gtgttggtcgaaacctgttg 102
ccacagtattggcagagcac 111
gctcttggtccattacagca 99
sequence analysis of test samples and melting curve analysis in the case of
patient material. The crossover point (Ct) values, defined as the cycle number
at which each curve intersects the threshold detection value, were used to
calculate the gene-specific input mRNA amount for each tissue according to
the calibration curve method. Data were evaluated using the SDS software
version 1.9.1 (Applied Biosystems) and the second derivative maximum algo-
rithm.
RESULTS
Distribution of Gene Expression. Initial analysis of the hybrid-
ization signal of the 12,625 transcripts represented on the oligonu-
cleotide array revealed that malignant T cells of Sz patients on
average expressed 6,100 of the interrogated genes; in CD4� T cells
from peripheral blood of BE patients and healthy volunteers 5,771
respectively 5,975 of the genes were expressed.
Analysis of the entire set of expressed genes using an unsupervised
hierarchical clustering analysis algorithm, grouping the samples on
the basis of similarity of their expression profiles, showed that the Sz
samples display a relatively homogeneous gene expression pattern
that is classified separately from that of benign CD4� T cells (see
Fig. 1 for dendrogram representing cluster analysis of the entire set of
expressed genes). The transcript profiles of T cells from patients with
BE were markedly different from those of Sz patients and were more
related to transcript profiles of T cells from healthy volunteers.
Sézary Syndrome-Specific Gene Expression Pattern. Compara-
tive analysis of the 10 Sz samples with the 8 control samples imple-
menting the SAM algorithm demonstrated that 176 genes were sta-
tistically significantly differentially expressed at P � 0.01. Of these
genes significant in the separation of Sz CD4� T cells and benign
CD4� T cells, 69 were relatively overexpressed, with fold changes
ranging from 1.7 to 19.8. One hundred and seven genes were down-
regulated in the malignant T cells, with fold changes ranging from 1.4
to 13. Of the 176 significantly differentially expressed genes, 123
genes were up- or down-regulated with fold changes exceeding 2.
Comparative gene expression profiles of these 59 overexpressed and
64 underexpressed genes are shown in Fig. 2.
 GENE EXPRESSION PROFILING OF SÉZARY SYNDROME

A Sézary Syndrome Healthy Benign

Controls Erythroderma

Fold Change
12,59 tumor necrosis factor (ligand) superfamily, member 11 (RANKL)
6,91 protein tyrosine phosphatase, receptor type, N polypeptide 2
4,97 EphA4
4,83 protein kinase, cAMP-dependent, regulatory, type K, alpha
3,86 interleukin 6 receptor
3,41 tumor necrosis factor (ligand) superfamily, member 7 (DC70)
3,37 SHP2 interacting transmembrane adaptor
Signal 3,22 SH2 domain protein 1A
transduction 3,11 protein kinase (cAMP-depenent, catalytic) inhibitor alpha
3,06 calcium/calmodulin-dependent serine protein kinase
2,46 integrin, beta 1
2,32 mitogen-activated protein kinase 1
2,3 serine/threonine kinase 38
2,19 PI3-kinase, regulatory subunit, polypeptide 3 (p55, gamma)
2,16 RAS p21 protein activator (GTPase activating protein) 1
2 ADP-ribosylation factor related protein 1
19,82 transcription factor AP-2 alpha
10,11 Twist
4,48 RecQ protein-like (DNA helicase Q1-like)
Transcriotion 3,09 sin3-associated polypeptide, 18kDa
2,59 high mobility group nucleosomal binding domain 4
regulation 2,57 zinc finger protein 146
2,38 SMARC D2
2,36 c-myc binding protein
2,17 transcription factor 12
2,11 nuclear factor (erythroid-derived 2)-like 1
RNA processing 2,53 NS1-binding protein
2,35 nuclear cap binding protein subunit 2, 20kDa
Translation 3,78 phenylalanine-tRNA synthetase-like
Apoptosis 3,06 testis enhanced gene transcript (BAX inhibitor 1)
4,41 acyl-Coenzyme A dehydrogenase, C-4 to C-12 straight chain
3,83 cytochrome b5 outer mitochondrial membrane precursos
3,35 acyl-Coenzyme A oxidase 3, pristanoyl
2,59 aconitase 2, mitochondrial
2,55 phosphoribosyl pyrophhosphate synthetase-associated protein 1
Metabolism 2,41 15kDa selenoprotein
2,41 uncoupling protein 2
2,34 phosphoglucomutase 1
2,3 AAS dehydrogenase-phosphopantetheinyl transferase
2,13 ATP synthase f chain, mitochondrial
Transport 2,45 transmembrane 9 superfamilyy member 1
2,21 tetracycline transporter-like protein
3,73 Vacuolar ATP synthase subunit C
Degradation 2,18 acid phosphatase 2, lysosomal
5,28 unc-84 homolog A
3,46 ARP2 actin-related protein 2 homolog
Cytoskeleton 3,07 dynein, cytoplasmic, intermediate polypeptide 2
2,87 actin related protein 2/3 complex, subunit 5, 16kDa
2,1 actin related protein 2/3 complex, subunit 4, 20kDa
7,39 sialyltransferase 8A
6,03 parathyroid hormone-like hormone
4,44 ceroid-lipofuscinosis, neuronal 5
3,89 presenilin 1
3,01 coproporphyrinogen oxidase
Miscellaneous 2,97 MHC class I region ORF
2,86 tumor differentially expressed 1
2,47 neutral sphingomyelinase activation associated factor
2,4 BAI1-associated protein 3
2,37 telomeric repeat binding factor 1 (TRF1)

z-score
2 0 -2
Fig. 2. Supervised analysis performed on 10 tumor samples of CD4� T cells from peripheral blood of patients with Sz versus eight control samples, consisting of CD4� T cells
from three healthy volunteers and five patients with BE. Represented genes are selected by the SAM algorithm and have an average fold change exceeding 2. Values are visualized
according to the scale bar that represents the difference in the z-score (expression difference/SD) relative to the mean. A, the analysis identified 69 genes that are up-regulated in CD4�
T cells from peripheral blood of patients with Sz compared with the controls. Fold change denotes the ratio of average normalized expression level in the Sz and the control group
of samples. B, 107 genes are down-regulated in CD4� T cells from peripheral blood of patients with Sz compared with the controls. Fold change denotes the ratio of average normalized
expression level in the control and the Sz group of samples.

Table 2 displays the most discriminating genes, according to fold
change between average expression levels in the malignant T cells
compared with the benign T cells. The most differentially expressed
genes, with fold changes exceeding 10, were the transcription factor
activator protein 2-�; dual specificity phosphatase 8; the tumor ne-
crosis factor (TNF) receptor family ligand, receptor activator of nu-
clear factor-�B ligand (RANKL; TNFSF11); and the transcription
factor Twist. Among the genes highly overexpressed in T cells of Sz
patients, especially up-regulation of the genes Twist and EphA4 was
consistently seen. Transcripts of Twist and EphA4 were present in 9
respectively 8 of 10 Sz samples.
Expression of the following genes was undetectable in any of the
control samples but present in Sz T cells: EphA4 (expressed in 8 of 10 Sz
samples), phenylalanine-tRNA synthetase-like (expressed in 7 of 10 Sz
samples), RANKL (expressed in 7 of 10 Sz samples), transcription factor
activator protein 2-� (expressed in 3 of 10 Sz samples), and peroxisomal
5580
acyl-CoA oxidase 3 (expressed in 3 of 10 Sz samples). Although the level
of RANKL transcript level was undetectable in benign T cells in this
array analysis study, it is normally expressed by activated T cells (18).
Interestingly, phenylalanine-tRNA synthetase-like transcript has been
demonstrated previously to be expressed in a human acute-phase chronic
myeloid leukemia cell line but not in its non-tumorigenic variant, sug-
gesting that selective expression of this member of the tRNA synthetase
family is more common in hematopoietic neoplasms (19).
Real-Time Quantitative PCR of Selected Genes, Including
EphA4 and Twist. To validate the results of microarray analysis
data, qPCR was applied on a panel of 10-selected genes. As Fig. 3A
shows, qPCR results were all in agreement with gene-profiling data.
In general, differences in transcript levels between malignant and
benign T cells appeared to be more pronounced when measured by
qPCR compared with oligonucleotide microarray analysis.
Next we evaluated the expression of EphA4 and Twist, the two

71
 GENE EXPRESSION PROFILING OF SÉZARY SYNDROME

B Sézary Syndrome Healthy Benign

Controls Erythroderma
Fold Change
13,02 dual specificity phosphatase 8
5,39 iller cell lectin-like receptor subfamily B, member 1
4,75 transforming growth factor, beta receptor II (70/80kDa)
3,56 guanine nucleotide binding protein (G protein), alpha 15 (Gq class)
Signal 3,44 C-type lectin, superfamily member 2 (activation-induced)
transduction 2,83 son of sevenless homolog 2
2,73 serine/threonine kinase 17b (apoptosis-inducing)
2,69 activated p21cdc42Hs kinase
2,17 A kinase (PRKA) anchor protein 8
2,09 CD6 antigen
2,06 transforming growth factor, beta 1
5,61 signal transducer and activator of transcription 4
4,62 special AT-rich sequence binding protein 1
4,47 PR domain containing 2, with ZNF domain
3,33 ring finger protein 10
3,16 basic transcription element binding protein 1
3,14 CBF1 interacting corepressor
3,00 heat shock transcription factor 2
2,84 MXI1 gene product, mRNA sequence
2,71 mediator of RNA polymerase II transcription, subunit 6 homolog
Transcriotion 2,69 MAX binding protein
regulation 2,68 CCAAT-box-binding transcription factor
2,65 H1 histone family, member X
2,57 basic helix-loop-helix domain containing, class B, 2
2,47 POU domain, class 2, transcription factor 1
2,44 SON DNA binding protein
2,41 chromodomain helicase DNA binding protein 3
2,39 putative DNA/chromatin binding motif
2,37 CREB binding protein
2,15 forkhead box O1A
2,12 BcI-2-associated transcription factor
2,11 PHD finger protein 1
3,09 splicing factor, arginine/serine-rich 12
3,01 RNA, U17D small nucleolar
2,68 small nuclear ribonucleoprotein polypeptide A’
RNA processing 2,49 serine/arginine repetitive matrox 2
2,37 U2-associated SR140 protein
2,11 PRP8 pre-mRNA processing factor 8 homolog
2,10 heterogeneous nuclear ribonucleoprotein A1
Translation 2,32 eukaryotic translation initiation factor 5
Apoptosis 4,67 BCL2-like 11
2,45 B-cell CLL/lymphoma 2
Cell ccle 4,30 BTG family, member 3
regulation 2,75 M-phase phosphoprotein 9
5,22 short-chain dehydrogenase/reductase 1
4,31 proprotein convertase subtilisin/kexin type 5
2,42 non-metastatic cells 4, protein expressed in
Metabolism 2,24 dihydrolipoamide S-succinyltransferase
2,21 histidine decarboxylase
2,06 Rab geranylgeranyltransferase, beta subunit
2,80 Sec23-interacting protein p125
2,25 nucleoporin 98kDa
Transport 2,16 nucleoporin 88kDa
2,15 potassium voltage-gated channel, KQT-like subfamily, member 1
7,29 glutathione S-transferase M1
2,49 F-box and leucine-riche repeat protein 11
Degradation 2,41 ubiquitin-conjugating enzyme E2, J1
2,26 glutathione S-transferase M4
Cytoskeleton 3,25 diaphanous homolog 2
7,40 amyloid beta (A4) precursor protein-binding, family A, member 2
4,02 monocyte to macrophage differentiation-associated
Miscellaneous 3,28 ectodermal-nneural cortex (with BTB-like domain)
2,40 rearranged L-myc fusion sequence
2,14 cryptochrome 2 (photolyase-like)

z-score
2 0 -2
Fig. 2 Continued

most selectively and consistently up-regulated genes in the malignant
T cells of Sz patients, using qPCR in a larger set of CTCL and control
samples. To assess whether EphA4 and Twist were also expressed by
malignant T cells of patients with mycosis fungoides and CD30-
negative primary cutaneous large T-cell lymphoma, transcript levels
were analyzed in lesional skin biopsy samples of patients with these
Sz-related CTCLs. Included as additional controls were T cells
in vitro activated with phytohemagglutinin and expanded with
interleukin-2 as well as lesional skin biopsy samples from patients
with T cell-rich inflammatory dermatoses. EphA4 was highly ex-
pressed in lesional skin of three of nine patients with mycosis fun-
goides, whereas its expression was nearly undetectable in any of the

Table 2 List of 10 transcripts including fold change that are most significantly up- or down-regulated in CD4� T cells from patients with Sz when comparing average expression
levels to those in CD4� T cells from controls
Up-regulated in Sézary syndrome Down-regulated in Sézary syndrome

Gene description Fold change Gene description Fold change

Transcription factor AP2-� 19.82 Dual specificity phosphatase 8 13.02
Twist 12.59 Amyloid beta (A4) precursor protein-binding, family A, member 2 7.4
TNFSF11 (RANKL) 10.11 Glutathione S-transferase M1 7.29
Protein tyrosine phosphatase, receptor type, N polypeptide 2 7.39 STAT4 5.61
Parathyroid hormone-like hormone 6.91 Killer cell lectin-like receptor subfamily B, member 1 5.39
Sialyltransferase 8A 6.03 Short-chain dehydrogenase/reductase 1 5.22
Unc-84 homolog A 5.28 TGF-� receptor II 4.75
EphA4 4.97 BCL2-like 11 (Bcl-11a) 4.67
Protein kinase, cAMP-dependent, regulatory, type I, � 4.83 Special AT-rich sequence binding protein 1 (SATB1) 4.62
RecQ protein-like (DNA helicase Q1-like) 4.48 PR domain containing 2 (PRDM2, Riz1) 4.47

5581

72
 GENE EXPRESSION PROFILING OF SÉZARY SYNDROME

Fig. 3. Results of real-time quantitative PCR analysis. A, histogram showing expression levels of 10-selected genes as measured by oligonucleotide array analysis (f) and qPCR
(�). Fold change of oligonucleotide and qPCR experiments denote average expression level in patients with Sz relative to average expression level in controls. For qPCR experiments,
expression levels of the gene of interest were normalized to those of two housekeeping genes. B, histogram showing expression level of the EphA4 gene as measured by qPCR in
peripheral blood T cells of Sz patients, BE patients, healthy volunteers, in vitro-stimulated peripheral blood T cells, as well as in lesional skin biopsy samples from patients with
cutaneous T-cell lymphoma and various inflammatory skin diseases. C, histogram showing expression level of the Twist gene as measured by qPCR in these samples. Fold change
denotes the expression of EphA4 or Twist relative to expression of the housekeeping gene U1A. MF, mycosis fungoides; CD30-PCLTCL, CD30-negative primary cutaneous large T-cell
lymphoma; PHA, phytohemagglutinin; PBL-T, peripheral blood T cells; CDLE, chronic discoid lupus erythematosus.

isolated or in vitro-activated T cell samples (Fig. 3B). In the benign
lesional skin biopsy samples, occasionally weak expression of EphA4
was observed, which might be attributable to the presence of endo-
thelial cells in skin that have been reported to be capable of expressing
the EphA4 receptor. The Twist gene was overexpressed in four of nine
mycosis fungoides skin biopsy samples including one patient with
patch-stage disease, whereas its expression in control samples was
only weak or absent (Fig. 3C).
Expression Patterns of Genes Reported to Be Differentially
Expressed in Sézary Syndrome. Subsequently we evaluated the
transcript levels of selected genes reported in the literature to be
specifically or differentially expressed by Sz T cells. Presented in
Fig. 4 are Sz-associated genes reported in previous publications ac-
companied by a heat map indicating the relative expression levels of
each gene in Sz and control samples resulting from our microarray
analysis. It should be mentioned that for some of the selected genes,
different expression has been described on the protein level but not on
5582
the mRNA level. As Fig. 3 shows, consistent with reports from the
literature, we found high expression of JunB (9), versican (20),
TRAIL (20), T-plastin (22), Kir3DL2 (23), integrin �1 (26), as well as
low expression of STAT4 (6), TGF-� receptor II (10), Fas (12) and
CD26 (33) in Sz T cells. In addition to the reported selective expres-
sion of the killer cell immunoglobulin-like receptor KIR3DL2, we
found expression of KIR2DL4 transcript by Sz cells. We did not
observe aberrant expression of MAGE1 (34), P-glycoprotein (35), or
SOCS3 (36). Contradictory to previous reports we found increased
rather than decreased levels of TIA1 (30) and SHP1 (31) transcripts.
DISCUSSION
The purpose of this study was to apply gene expression analysis
using oligonucleotide microarrays to gain more insight into the patho-
genesis of Sz and to identify tumor-associated markers for potential
use in the diagnosis and therapy of this malignancy. We compared

73
 GENE EXPRESSION PROFILING OF SÉZARY SYNDROME
Fig. 3
gene expression patterns of malignant T cells from peripheral blood of
patients with Sz with expression of CD4� T cells from healthy
volunteers and from patients with benign forms of erythroderma.
Microarray results were validated by real-time quantitative PCR on a
subset of genes.
Transcriptional profiles of Sz T cells demonstrated to be relatively
homogeneous and unsupervised hierarchical clustering revealed that
malignant T-cell profiles were classified separately from those of
benign T cells used as controls. The expression patterns of T cells
from patients with BE were more related to those of T cells from
healthy volunteers than to those of T cells from Sz patients.
By implementing the SAM algorithm, 176 genes were identified
that were significantly differentially expressed in the 10 Sz samples
compared with the group of eight control samples. Fifty-nine of these
differentially expressed genes were up-regulated, and 64 genes were
down-regulated �2-fold (Fig. 2). Approximately half of the genes
encode proteins that function in cell signaling and transcription reg-
ulation.
First, we examined these differentially expressed genes to identify
genes specifically expressed in Sz that might serve as tumor-associ-
ated markers. Second, we attempted to discern oncogenic pathways in
this transcriptional pattern that separates malignant T cells of Sz
patients from benign T cells. Among the most highly overexpressed
genes in Sz, two genes (Twist and EphA4) were very consistently
up-regulated, whereas transcripts were nearly undetectable in any of
the control samples.
The potential tumor-associated gene Twist was expressed in 9 of 10
Sz patient T-cell samples and only weakly in one of the control
samples; the average fold change was 12.6. The Twist gene encodes
a transcription factor that functions as a regulator of mesodermal
differentiation and is normally not expressed in lymphoid cells (37,
38). Twist belongs to the basic helix-loop-helix family of transcription
factors, several members of which are known to be T-cell oncopro-
teins (39). Twist has been shown to have oncogenic properties be-
cause it can prevent c-myc-induced apoptosis by antagonizing the p53
pathway (40). The antiapoptotic properties of Twist have additionally
5583
Continued
been suggested by its interaction with components of the nuclear
factor-�B pathway regulating susceptibility to TNF-�-induced apo-
ptosis (41). Interestingly the nuclear factor-�B pathway, which both
regulates and is regulated by Twist activity in mammalian cells, has
been reported to be constitutively activated in CTCL cells (42). In
human rhabdomyosarcoma and experimental avian nephroblastoma,
increased expression of the Twist gene was noted (40, 43). It was
shown recently that overexpression of Twist in cancer cell lines is
associated with acquisition of resistance to the anticancer drugs taxol
and vincristine (44). However, it remains to be established whether
increased expression of Twist also has an important role in the
pathogenesis of Sz.
EphA4, another potential tumor-associated marker that was highly
expressed in 8 of 10 Sz T-cell samples and none of the control
samples, belongs to the Eph-receptor subfamily of transmembrane
protein-tyrosine kinases (45– 47). The expression of Eph receptors has
been found most consistently in brain where they are implicated in
regulation of neuronal migration and angiogenesis (48, 49). Recently,
EphA4 and active signaling by this receptor has been demonstrated in
human T cells (50). However, its function in T cells is not clear; nor
is it clear whether expression of EphA4 is confined to a specific subset
of T cells. Expression of Eph receptors has been linked to malignant
transformation. The EphA2 receptor, closely related to EphA4, is
highly expressed in several human cancers such as breast, colon, and
prostate carcinoma where it has been demonstrated to function as an
oncoprotein (51, 52). A possible functional role of EphA4 tyrosine
kinase activity in the pathogenesis of Sz may be suspected because
components of its downstream signal transduction pathway such as
Fyn, Grb2, and Abl were also found to be up-regulated in Sz T cells.
Recently activation of EphA4 was described to activate Jak/STAT
signaling and induce phosphorylation of STAT3, a transcriptional
activator reported to be constitutively phosphorylated in malignant T
cells of Sz patients (5, 7, 53). If EphA4 expression would also be
functionally significant in Sz T cells, this membrane-bound receptor
would constitute an attractive target for therapeutic intervention using
74
 GENE EXPRESSION PROFILING OF SÉZARY SYNDROME
Fig. 4. Expression patterns of selected genes reported in the
literature to be differentially expressed in T cells of patients with
Sz. Values are visualized according to the scale bar that repre-
sents the difference in the z-score (expression difference/SD)
relative to the mean, similar to Fig. 2. A, relative normalized
expression levels of genes reported to be overexpressed in Sz. B,
relative normalized expression levels of genes reported to be
underexpressed in Sz.
monoclonal antibodies or small molecular inhibitors of its kinase
domain.
Both EphA4 and Twist appeared to be expressed in lesional skin
biopsy samples of a subset of patients with mycosis fungoides, a more
common CTCL related to Sz, and CD30-negative primary cutaneous
T cell lymphoma, but not or only weakly in skin lesions of patients
with inflammatory dermatoses. This observation indicates that aber-
rant expression of these two genes is not limited to malignant T cells
of Sz patients and might be a feature of CTCL other than Sz.
The transcriptional profile of Sz T cells demonstrated dysregulation
of several other potentially oncogenic signal transduction pathways.
We observed a general increased expression of growth-promoting
tyrosine kinases including several mitogen-activated protein kinases
and decreased expression of inactivating phosphatases such as dual
specificity phosphatase 8. In T cells from Sz patients, the expression
of a number of tumor suppressor genes that are known to be impli-
cated in the pathogenesis of hematopoietic neoplasms was diminished,
including the histone methyltransferase PRDM2 (Riz1; 54), the pro-
apoptotic protein Bcl-11a (55), CREB-binding protein (56), TGF-�
receptor II (10), Mxi1 (57), and Forkhead box O1A (FOXO1A,
FKHR; 58).
A recurrent feature of Sz cells in our study was the decreased
expression of the tumor suppressor gene TGF-� receptor II and of its
ligand TGF-�1. Also its downstream signaling components SMAD3
and SMAD7 were significantly down-regulated in T cells from Sz
patients, with fold changes of 1.8 and 1.7, respectively. The resulting
disruption of the TGF-� receptor signaling pathway is associated with
Sézary Syndrome Healthy Benign
Controls Erythroderma
A Versican (chondroitin sulfate proteoglycan 2) 20
TRAIL (TNFSF10) 20
Ras homolog gene family, member B (ARHB) 20
T-plastin (plastin 3, T-isoform) 21
L-selectin (CD62L) 22
KIR3DL2 (CD158k, NKAT4) 23
JunB 9
Interferon inhibiting cytokine (IK cytokine) 24
Interferon-γ receptor accessory factor-1 (AF-1) 25
Integrin β1 26
IL-7 receptor 27
IL-11 receptor (α chain) 20
GATA-3 20
DUSP1 20
CD15 (fucosyltransferase 4) 28
ILT2 receptor (CD85j, LILRB1) 29
B TIA1 30
TGF-β receptor II 10
STAT4 6
SHP1 (PTPN6, PTP1c) 31
p53 32
LFA-1 (β subunit) 22
LFA-1 (α subunit) 22
interleukin 2 receptor (β subunit) 20
interleukin 1 receptor type II 20
interleukin 2 receptor type 1 20
Fas (CD95, TNFRSF6) 12
integrin α4 (α4β1, CD49d, CD29) 28
CD26 (dipeptidyl peptidase IV) 33
CD40 (TNFRSF5) 20
z-score
2 0 -2
loss of growth inhibition by TGF-�, an antiproliferative and proapop-
totic cytokine for lymphoid cells (59). Diminished cell surface ex-
pression of the TGF-� receptor II protein on CD4� T cells from Sz
patients and loss of sensitivity to TGF-� in the progression of CTCL
have been described previously (60, 61). The consistent down-regu-
lation of multiple components of the TGF-� receptor pathway in
patients with Sz supports the notion that abrogation of this signaling
pathway is a critical event in the pathogenesis of this malignancy.
The tumor suppressor genes Mxi1 and Mnt, which both antagonize
the activity of the potentially oncogenic transcription factor c-myc
(62), were consistently down-regulated in the malignant T cells. In
addition, malignant T cells in Sz demonstrate high expression of
MycBP (Amy1) that encodes a protein that stimulates transcriptional
activity of c-myc (63). In mice, targeted deletion of the Mxi1 gene
results in a phenotype that shows concordance with that of c-myc-
overexpressing mice, including predisposition to the development of
lymphomas, implying that Mxi1 functions as a tumor suppressor gene
in lymphoid cells (57). Interestingly, the Mxi1 gene is located at
chromosome 10q24 –26 and the Mnt gene at 17p13.3, chromosomal
regions that are frequently lost in Sz cells as shown by cytogenetic
studies (14). Although the c-myc gene itself was not consistently
overexpressed in the malignant T cells, potentially oncogenic activa-
tion of the c-myc pathway could result from altered expression of
three of its regulatory proteins, Mxi1, Mnt, and MycBP (62).
Another tumor suppressor gene, which may be relevant in the
pathogenesis of Sz and which was down-regulated in T-cell samples
from each Sz patient with an average fold change of 2.2, is the
5584
75
 GENE EXPRESSION PROFILING OF SÉZARY SYNDROME
transcription factor FOXO1A. FOXO1A is a downstream target of the
phosphatidylinositol 3�-kinase-PTEN-AKT signal transduction path-
way (64). Dysregulation of this pathway is a critical event in several
hematopoietic neoplasms such as T-cell chronic lymphocytic leuke-
mia (65). In addition, T cell-specific deletion of PTEN in mice has
been shown to result in the development of CD4� T-cell lymphomas
(66). An essential role of FOXO1A in the lymphomagenic properties
of this signaling pathway is suggested by the recent observation that
in PTEN-deficient cells tumorigenicity is reversed by restoration of
FOXO1A activity (67). It therefore seems likely that the loss of
FOXO1A expression has pro-oncogenic consequences, comparable
with those conferred by the loss of PTEN.
Another noticeable feature of the expression pattern particular to Sz
T cells is increased expression of the TNF receptor ligands TNFSF7
(CD70) and TNFSF11 (RANKL) in 7 respectively 8 of 10 Sz patients.
Deregulated expression of several other members of the TNF receptor
family CD40, CD40L and TRAIL has been reported previously in
related CTCLs (20, 68, 69). In Sz T cells, overexpression of the
TNFSF7 gene encoding the transmembrane protein CD70 is paral-
leled by high expression of its receptor CD27. CD70 is normally only
transiently expressed on antigen-activated T cells and interaction with
its receptor CD27 promotes T-cell proliferation and effector T-cell
functions (70, 71). In transgenic mice, constitutive expression of
CD70 leads to a state of chronic immune activation with excessive
formation of effector T cells and progressive depletion of naive T cells
resulting in lethal T-cell immunodeficiency (72). Also RANKL can
induce inappropriate immune activation through enhancing the co-
stimulatory properties of dendritic cells (18, 73). Sz patients can also
be severely immunocompromised and susceptible to opportunistic
infection, one of the main causes of mortality in these patients. It has
been proposed that the immunodeficiency observed in advanced
CTCL is caused primarily by T-cell depletion (74). Indicative of
T-cell depletion, also observed in CD70-transgenic mice, peripheral
blood of patients with Sz displays a marked decrease of absolute
normal CD4� T cells as well as a reduction of the complexity of the
T-cell receptor repertoire (75). The coordinate expression of CD70
and CD27 can thus be expected to stimulate activation and prolifer-
ation of the malignant T-cell pool in Sz patients. An additional
consequence of high expression of CD70 as well as RANKL in Sz
patients may be detrimental persistent immune activation and deple-
tion of the naive T-cell pool, resulting in immunodeficiency.
As shown in Fig. 4, many of our findings on gene expression in
malignant T cells from Sz patients are consistent with previous
findings reported in the literature. We identified the proposed tumor-
associated markers T-plastin (21), Kir3DL2 (23), and JunB (9) as
up-regulated in Sz T cells, but in the included Sz patient samples their
expression patterns were not significantly discriminating according to
the SAM algorithm. Recently Kari et al. (20) performed cDNA
microarray analysis on T cells from patients with erythrodermic
CTCL. Although the results of their study show similarities with the
present study such as decreased expression of STAT 4 and high
expression of transcription factor activator protein 2-� in malignant T
cells, differences predominate. The discrepancies between the results
are likely to be attributable to the characteristics of the patient group
they studied that included erythrodermic CTCL other than Sz, their
choice of cultured peripheral blood mononuclear cells skewed to the
Th2 phenotype as control, the use of cDNA microarrays that analyze
a different and smaller set of genes, as well as the data analysis
methods applied.
Our studies indicate that malignant T cells of Sz patients display
a gene expression pattern that separates them from benign T cells
of patients with BE and healthy controls. Among the most highly
and consistently expressed genes in T cells of Sz patients are the
5585
76
membrane-bound tyrosine kinase receptor EphA4 and Twist, a poten-
tially oncogenic transcription factor. Additional studies are necessary
to evaluate whether these tumor-associated proteins are involved in
the development and progression of Sz and if they can be used as
targets for therapeutic intervention.
ACKNOWLEDGMENTS
The authors thank Eveline Mank (Dept. of Human and Clinical Genetics,
LUMC) and Aat Mulder for their excellent technical assistance and Dr. Pieter
van der Velden for initial help with analysis of the micro-array data.
REFERENCES
1. Sézary A, Bouvrain Y. Erythrodermie avec presence de cellules monstrueuses dans le
derme et dans le sang circulant. Bull Soc Fr Dermatol Syphilogr 1938;45:254 – 60.
2. Vonderheid EC, Bernengo MG, Burg G, et al. Update on erythrodermic cutaneous
T-cell lymphoma: report of the International Society for Cutaneous Lymphomas.
J Am Acad Dermatol 2002;46:95–106.
3. Kamarashev J, Burg G, Kempf W, Hess Schmid M, Dummer R. Comparative analysis
of histological and immunohistological features in mycosis fungoides and Sezary
syndrome. J Cutan Pathol 1998;25:407–12.
4. Willemze R, Kerl H, Sterry W, et al. EORTC classification for primary cutaneous
lymphomas: a proposal from the Cutaneous Lymphoma Study Group of the European
Organization for Research and Treatment of Cancer. Blood 1997;90:354 –71.
5. Zhang Q, Nowak I, Vonderheid EC, et al. Activation of Jak/STAT proteins involved
in signal transduction pathway mediated by receptor for interleukin 2 in malignant T
lymphocytes derived from cutaneous anaplastic large T-cell lymphoma and Sezary
syndrome. Proc Natl Acad Sci USA 1996;93:9148 –53.
6. Showe LC, Fox FE, Williams D, Au K, Niu Z, Rook AH. Depressed IL-12-mediated
signal transduction in T cells from patients with Sezary syndrome is associated with
the absence of IL-12 receptor beta 2 mRNA and highly reduced levels of STAT4.
J Immunol 1999;163:4073–9.
7. Eriksen KW, Kaltoft K, Mikkelsen G, et al. Constitutive STAT3-activation in Sezary
syndrome: tyrphostin AG490 inhibits STAT3-activation, interleukin-2 receptor ex-
pression and growth of leukemic Sezary cells. Leukemia 2001;15:787–93.
8. Mitchell TJ, Whittaker SJ, John S. Dysregulated expression of COOH-terminally
truncated Stat5 and loss of IL2-inducible Stat5-dependent gene expression in Sezary
Syndrome. Cancer Res 2003;63:9048 –54.
9. Mao X, Orchard G, Lillington DM, Russell-Jones R, Young BD, Whittaker SJ. Am-
plification and overexpression of JUNB is associated with primary cutaneous T-cell
lymphomas. Blood 2003;101:1513–9.
10. Kadin ME, Cavaille-Coll MW, Gertz R, Massague J, Cheifetz S, George D. Loss of
receptors for transforming growth factor beta in human T-cell malignancies. Proc Natl
Acad Sci USA 1994;91:6002– 6.
11. Scarisbrick JJ, Woolford AJ, Calonje E, et al. Frequent abnormalities of the p15 and
p16 genes in mycosis fungoides and Sezary syndrome. J Investig Dermatol 2002;
118:493–9.
12. Dereure O, Portales P, Clot J, Guilhou JJ. Decreased expression of Fas (APO-1/
CD95) on peripheral blood CD4� T lymphocytes in cutaneous T-cell lymphomas.
Br J Dermatol 2000;143:1205–10.
13. Karenko L, Hyytinen E, Sarna S, Ranki A. Chromosomal abnormalities in cutaneous
T-cell lymphoma and in its premalignant conditions as detected by G-banding and
interphase cytogenetic methods. J Investig Dermatol 1997;108:22–9.
14. Mao X, Lillington DM, Czepulkowski B, Russell-Jones R, Young BD, Whittaker S.
Molecular cytogenetic characterization of Sezary syndrome. Genes Chromosomes
Cancer 2003;36:250 – 60.
15. Bakels V, van Oostveen JW, Preesman AH, Meijer CJ, Willemze R. Differentiation
between actinic reticuloid and cutaneous T cell lymphoma by T cell receptor gamma
gene rearrangement analysis and immunophenotyping. J Clin Pathol 1998;51:154 – 8.
16. Huber W, Von Heydebreck A, Sultmann H, Poustka A, Vingron M. Variance
stabilization applied to microarray data calibration and to the quantification of
differential expression. Bioinformatics 2002;18(Suppl 1):S96 –S104.
17. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the
ionizing radiation response. Proc Natl Acad Sci USA 2001;98:5116 –21.
18. Wong BR, Josien R, Lee SY, et al. TRANCE (tumor necrosis factor [TNF]-related
activation-induced cytokine), a new TNF family member predominantly expressed in
T cells, is a dendritic cell-specific survival factor. J Exp Med 1997;186:2075– 80.
19. Sen S, Zhou H, Ripmaster T, Hittelman WN, Schimmel P, White RA. Expression of
a gene encoding a tRNA synthetase-like protein is enhanced in tumorigenic human
myeloid leukemia cells and is cell cycle stage- and differentiation-dependent. Proc
Natl Acad Sci USA 1997;94:6164 –9.
20. Kari L, Loboda A, Nebozhyn M, et al. Classification and prediction of survival in
patients with the leukemic phase of cutaneous T cell lymphoma. J Exp Med 2003;
197:1477– 88.
21. Su MW, Dorocicz I, Dragowska WH, et al. Aberrant expression of T-plastin in Sezary
cells. Cancer Res 2003;63:7122–7.
22. Hwang ST, Fitzhugh DJ. Aberrant expression of adhesion molecules by Sezary cells:
functional consequences under physiologic shear stress conditions. J Investig Der-
matol 2001;116:466 –70.
23. Bagot M, Moretta A, Sivori S, et al. CD4(�) cutaneous T-cell lymphoma cells
express the p140-killer cell immunoglobulin-like receptor. Blood 2001;97:1388 –91.
 GENE EXPRESSION PROFILING OF SÉZARY SYNDROME
24. Willers J, Haffner A, Zepter K, et al. The interferon inhibiting cytokine IK is
overexpressed in cutaneous T cell lymphoma derived tumor cells that fail to upregu-
late major histocompatibility complex class II upon interferon-gamma stimulation.
J Investig Dermatol 2001;116:874 –9.
25. Dummer R, Heald PW, Nestle FO, et al. Sezary syndrome T-cell clones display
T-helper 2 cytokines and express the accessory factor-1 (interferon-gamma receptor
beta-chain). Blood 1996;88:1383–9.
26. Paulin Y, Boukhelifa M, Derappe C, Giner M, Font J, Aubery M. Activity of proximal
promoter of the human beta(1)-integrin gene was increased in Sezary syndrome. Leuk
Res 2001;25:487–92.
27. Bagot M, Charue D, Boulland ML, et al. Interleukin-7 receptor expression in
cutaneous T-cell lymphomas. Br J Dermatol 1996;135:572–5.
28. Rappl G, Muche JM, Abken H, et al. CD4(�)CD7(�) T cells compose the dominant
T-cell clone in the peripheral blood of patients with Sezary syndrome. J Am Acad
Dermatol 2001;44:456 – 61.
29. Nikolova M, Musette P, Bagot M, Boumsell L, Bensussan A. Engagement of
ILT2/CD85j in Sezary syndrome cells inhibits their CD3/TCR signaling. Blood
2002;100:1019 –25.
30. Matutes E, Coelho E, Aguado MJ, et al. Expression of TIA-1 and TIA-2 in T cell
malignancies and T cell lymphocytosis. J Clin Pathol 1996;49:154 – 8.
31. Leon F, Cespon C, Franco A, et al. SHP-1 expression in peripheral T cells from
patients with Sezary syndrome and in the T cell line HUT-78: implications in
JAK3-mediated signaling. Leukemia 2002;16:1470 –7.
32. Brito-Babapulle V, Hamoudi R, Matutes E, et al. p53 allele deletion and protein
accumulation occurs in the absence of p53 gene mutation in T-prolymphocytic
leukaemia and Sezary syndrome. Br J Haematol 2000;110:180 –7.
33. Bernengo MG, Novelli M, Quaglino P, et al. The relevance of the CD4� CD26-
subset in the identification of circulating Sezary cells. Br J Dermatol 2001;144:
125–35.
34. Haffner AC, Tassis A, Zepter K, et al. Expression of cancer/testis antigens in
cutaneous T cell lymphomas. Int J Cancer 2002;97:668 –70.
35. Jillella AP, Murren JR, Hamid KK, Longley BJ, Edelson RL, Cooper DL. P-
glycoprotein expression and multidrug resistance in cutaneous T-cell lymphoma.
Cancer Investig 2000;18:609 –13.
36. Brender C, Nielsen M, Kaltoft K, et al. STAT3-mediated constitutive expression of
SOCS-3 in cutaneous T-cell lymphoma. Blood 2001;97:1056 – 62.
37. Spicer DB, Rhee J, Cheung WL, Lassar AB. Inhibition of myogenic bHLH and MEF2
transcription factors by the bHLH protein Twist. Science (Lond) 1996;272:1476 – 80.
38. Wang SM, Coljee VW, Pignolo RJ, Rotenberg MO, Cristofalo VJ, Sierra F. Cloning
of the human twist gene: its expression is retained in adult mesodermally-derived
tissues. Gene 1997;187:83–92.
39. Baer R. TAL1, TAL2 and LYL1: a family of basic helix-loop-helix proteins impli-
cated in T cell acute leukaemia. Semin Cancer Biol 1993;4:341–7.
40. Maestro R, Dei Tos AP, Hamamori Y, et al. Twist is a potential oncogene that inhibits
apoptosis. Genes Dev 1999;13:2207–17.
41. Sosic D, Richardson JA, Yu K, Ornitz DM, Olson EN. Twist regulates cytokine gene
expression through a negative feedback loop that represses NF-kappaB activity. Cell
2003;112:169 – 80.
42. Izban KF, Ergin M, Qin JZ, et al. Constitutive expression of NF-kappa B is a
characteristic feature of mycosis fungoides: implications for apoptosis resistance and
pathogenesis. Hum Pathol 2000;31:1482–90.
43. Pajer P, Pecenka V, Karafiat V, Kralova J, Horejsi Z, Dvorak M. The twist gene is
a common target of retroviral integration and transcriptional deregulation in experi-
mental nephroblastoma. Oncogene 2003;22:665–73.
44. Wang X, Ling MT, Guan XY, et al. Identification of a novel function of TWIST, a
bHLH protein, in the development of acquired taxol resistance in human cancer cells.
Oncogene 2004;23:474 – 82.
45. Kullander K, Klein R. Mechanisms and functions of Eph and ephrin signalling. Nat
Rev Mol Cell Biol 2002;3:475– 86.
46. Lindberg RA, Hunter T. cDNA cloning and characterization of eck, an epithelial cell
receptor protein-tyrosine kinase in the eph/elk family of protein kinases. Mol Cell
Biol 1990;10:6316 –24.
47. Wilkinson DG. Multiple roles of EPH receptors and ephrins in neural development.
Nat Rev Neurosci 2001;2:155– 64.
48. Ramaswamy S, Tamayo P, Rifkin R, et al. Multiclass cancer diagnosis using tumor
gene expression signatures. Proc Natl Acad Sci USA 2001;98:15149 –54.
5586
77
49. Hafner C, Schmitz G, Meyer S, et al. Differential gene expression of eph receptors
and ephrins in benign human tissues and cancers. Clin Chem 2004;50:490 –9.
50. Sharfe N, Freywald A, Toro A, Roifman CM. Ephrin-A1 induces c-Cbl phosphoryl-
ation and EphA receptor down-regulation in T cells. J Immunol 2003;170:6024 –32.
51. Walker-Daniels J, Coffman K, Azimi M, et al. Overexpression of the EphA2 tyrosine
kinase in prostate cancer. Prostate 1999;41:275– 80.
52. Zelinski DP, Zantek ND, Stewart JC, Irizarry AR, Kinch MS. EphA2 overexpression
causes tumorigenesis of mammary epithelial cells. Cancer Res 2001;61:2301– 6.
53. Lai KO, Chen Y, Po HM, Lok KC, Gong K, Ip NY. Identification of the Jak/Stat
Proteins as Novel Downstream Targets of EphA4 Signaling in Muscle: implications
in the regulation of acetylcholinesterase expression. J Biol Chem 2004;279:13383–92.
54. Sasaki O, Meguro K, Tohmiya Y, Funato T, Shibahara S, Sasaki T. Altered expres-
sion of retinoblastoma protein-interacting zinc finger gene, RIZ, in human leukaemia.
Br J Haematol 2002;119:940 – 8.
55. Satterwhite E, Sonoki T, Willis TG, et al. The BCL11 gene family: involvement of
BCL11A in lymphoid malignancies. Blood 2001;98:3413–20.
56. Kang-Decker N, Tong C, Boussouar F, et al. Loss of CBP causes T cell lym-
phomagenesis in synergy with p27Kip1 insufficiency. Cancer Cell 2004;5:177– 89.
57. Schreiber-Agus N, Meng Y, Hoang T, et al. Role of Mxi1 in ageing organ systems
and the regulation of normal and neoplastic growth. Nature (Lond) 1998;393:483–7.
58. Hideshima T, Nakamura N, Chauhan D, Anderson KC. Biologic sequelae of inter-
leukin-6 induced PI3-K/Akt signaling in multiple myeloma. Oncogene 2001;20:
5991– 6000.
59. Totth A, Sebestyen A, Barna G, et al. TGF beta 1 induces caspase-dependent but
death-receptor independent apoptosis in lymphoid cells. Anticancer Res 2001;21:
1207–12.
60. Capocasale RJ, Lamb RJ, Vonderheid EC, et al. Reduced surface expression of
transforming growth factor beta receptor type II in mitogen-activated T cells from
Sezary patients. Proc Natl Acad Sci USA 1995;92:5501–5.
61. Kadin ME, Levi E, Kempf W. Progression of lymphomatoid papulosis to systemic
lymphoma is associated with escape from growth inhibition by transforming growth
factor-beta and CD30 ligand. Ann N Y Acad Sci 2001;941:59 – 68.
62. Pelengaris S, Khan M, Evan G. c-MYC: more than just a matter of life and death. Nat
Rev Cancer 2002;2:764 –76.
63. Taira T, Maeda J, Onishi T, et al. AMY-1, a novel C-MYC binding protein that
stimulates transcription activity of C-MYC. Genes Cells 1998;3:549 – 65.
64. Burgering BM, Kops GJ. Cell cycle and death control: long live Forkheads. Trends
Biochem Sci 2002;27:352– 60.
65. Pekarsky Y, Hallas C, Croce CM. Targeting mature T cell leukemia: new under-
standing of molecular pathways. Am J Pharmacogenomics 2003;3:31– 6.
66. Suzuki A, Yamaguchi MT, Ohteki T, et al. T cell-specific loss of Pten leads to defects
in central and peripheral tolerance. Immunity 2001;14:523–34.
67. Nakamura N, Ramaswamy S, Vazquez F, Signoretti S, Loda M, Sellers WR. Fork-
head transcription factors are critical effectors of cell death and cell cycle arrest
downstream of PTEN. Mol Cell Biol 2000;20:8969 – 82.
68. Tracey L, Villuendas R, Dotor AM, et al. Mycosis fungoides shows concurrent
deregulation of multiple genes involved in the TNF signaling pathway: an expression
profile study. Blood 2003;102:1042–50.
69. Storz M, Zepter K, Kamarashev J, Dummer R, Burg G, Haffner AC. Coexpression of
CD40 and CD40 ligand in cutaneous T-cell lymphoma (mycosis fungoides). Cancer
Res 2001;61:452–54.
70. Lens SM, Baars PA, Hooibrink B, van Oers MH, van Lier RA. Antigen-presenting
cell-derived signals determine expression levels of CD70 on primed T cells. Immu-
nology 1997;90:38 – 45.
71. Tesselaar K, Xiao Y, Arens R, et al. Expression of the murine CD27 ligand CD70 in
vitro and in vivo. J Immunol 2003;170:33– 40.
72. Tesselaar K, Arens R, van Schijndel GM, et al. Lethal T cell immunodeficiency
induced by chronic costimulation via CD27-CD70 interactions. Nat Immunol 2003;
4:49 –54.
73. Josien R, Li HL, Ingulli E, et al. TRANCE, a tumor necrosis factor family member,
enhances the longevity and adjuvant properties of dendritic cells in vivo. J Exp Med
2000;191:495–502.
74. Rook AH, Heald P. The immunopathogenesis of cutaneous T-cell lymphoma. He-
matol Oncol Clin North Am 1995;9:997–1010.
75. Yawalkar N, Ferenczi K, Jones DA, et al. Profound loss of T-cell receptor repertoire
complexity in cutaneous T-cell lymphoma. Blood 2003;102:4059 – 66.
78
 Chapter 7

Epigenetic profiling of cutaneous T-cell lymphoma:

promoter hypermethylation of multiple tumor
suppressor genes including BL7a, PTPRG and p73

J. Clin. Oncol. 2005 May 16

 Published Ahead of Print on May 16, 2005 as 10.1200/JCO.2005.11.353
VOLUME 23 d
NUMBER 17 d
JUNE 10 2005

JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T

Epigenetic Profiling of Cutaneous T-Cell Lymphoma:

From the Departments of
Dermatology and Medical Statistics,
Leiden University Medical Center,
Leiden, the Netherlands; and Division
of Human Cancer Genetics,
Comprehensive Cancer Center, Ohio
State University, Columbus, OH.
Submitted January 11, 2005; accepted
April 4, 2005.
R. van Doorn was supported by a
Genomics Fellowship, awarded by
the Netherlands Organization for
Scientific Research.
Authors’ disclosures of potential
conflicts of interest are found at
the end of this article.
Address reprint requests to Cornelius P.
Tensen, MD, Leiden University Medical
Center, Dermatology, Wassenaarseweg
72, Leiden ZH 2333AL, the Netherlands;
e-mail: c.p.tensen@lumc.nl.
� 2005 by American Society of Clinical
Oncology
0732-183X/05/2317-1/$20.00
DOI: 10.1200/JCO.2005.11.353
Promoter Hypermethylation of Multiple Tumor
Suppressor Genes Including BCL7a, PTPRG, and p73
Remco van Doorn, Willem H. Zoutman, Remco Dijkman, Renee X. de Menezes,
Suzan Commandeur, Aat A. Mulder, Pieter A. van der Velden, Maarten H. Vermeer,
Rein Willemze, Pearlly S. Yan, Tim H. Huang, and Cornelis P. Tensen
A B S T R A C T
Purpose
To analyze the occurrence of promoter hypermethylation in primary cutaneous T-cell
lymphoma (CTCL) on a genome-wide scale, focusing on epigenetic alterations with pathoge-
netic significance.
Materials and Methods
DNA isolated from biopsy specimens of 28 patients with CTCL, including aggressive CTCL
entities (transformed mycosis fungoides and CD30-negative large T-cell lymphoma) and an
indolent entity (CD30-positive large T-cell lymphoma), were investigated. For genome-wide
DNA methylation screening, differential methylation hybridization using CpG island microar-
rays was applied, which allows simultaneous detection of the methylation status of 8640 CpG
islands. Bisulfite sequence analysis was applied for confirmation and detection of hyper-
methylation of eight selected tumor suppressor genes.
Results
The DNA methylation patterns of CTCLs emerging from differential methylation hybridization
analysis included 35 CpG islands hypermethylated in at least four of the 28 studied CTCL sam-
ples when compared with benign T-cell samples. Hypermethylation of the putative tumor sup-
pressor genes BCL7a (in 48% of CTCL samples), PTPRG (27%), and thrombospondin 4 (52%)
was confirmed and demonstrated to be associated with transcriptional downregulation.
BCL7a was hypermethylated at a higher frequency in aggressive (64%) than in indolent
(14%) CTCL entities. In addition, the promoters of the selected tumor suppressor genes
p73 (48%), p16 (33%), CHFR (19%), p15 (10%), and TMS1 (10%) were hypermethylated
in CTCL.
Conclusion
Malignant T cells of patients with CTCL display widespread promoter hypermethylation as-
sociated with inactivation of several tumor suppressor genes involved in DNA repair, cell cy-
cle, and apoptosis signaling pathways. In view of this, CTCL may be amenable to treatment
with demethylating agents.
J Clin Oncol 23. � 2005 by American Society of Clinical Oncology

INTRODUCTION CTCL may be very difficult as few therapies

Primary cutaneous T-cell lymphomas
(CTCLs) are extranodal non-Hodgkin’s
lymphomas derived from mature T cells pre-
senting in the skin. CTCLs encompass a
group of distinct entities with different clin-
ical behavior.1 Management of patients with
are able to influence the generally progres-
sive disease course of these malignancies.
In malignant T cells of patients with
CTCL, many oncogenic alterations have
been demonstrated, such as functional in-
activation of the Fas receptor, the trans-
forming growth factor beta receptors,

81
Copyright 2005 by American Society of Clinical Oncology
 van Doorn et al
p16, constitutive activity of STAT3, and chromosomal in-
stability.2-6 Although only few causative genetic lesions
have been identified, the epigenetic mechanism of pro-
moter hypermethylation has been found to be associated
with inactivation of several tumor suppressor genes
in CTCL.6-9
Methylation of CpG islands located in the promoter
or first exon of genes is associated with transcriptional
downregulation through alteration of the chromatin con-
formation and interference with binding of transcription
factors.10 Aberrant promoter methylation is increasingly
recognized as an important factor in the pathogenesis of
cancer as many tumor suppressor genes can be inactivated
through this epigenetic mechanism.11 Both the frequency
and the patterns of promoter hypermethylation vary
markedly between different tumor types.12,13 Promoter
hypermethylation in cancer cells is interesting from a clin-
ical point of view as its detection can be exploited in the
diagnosis and prognosis of cancer patients.14 The potential
reversibility of gene silencing associated with promoter
hypermethylation through pharmacologic manipulation
with demethylating agents, such as 5-aza-2#-deoxycyti-
dine, zebularine, and MG98, promises epigenetic ap-
proaches to cancer therapy.15
The notion that epigenetic dysregulation has an impor-
tant role in the development and progression of CTCL is
supported by the observation that these malignancies
respond favorably to treatment with histone deacetylase
inhibitors.16 Moreover, promoter hypermethylation, that
often coincides with repressive histone modifications, is
displayed by hematopoietic malignancies at a higher fre-
quency than by other tumor types.12,13,17,18 Consistently,
in the few studies performed thus far, promoter hypermeth-
ylation of the p15, p16, MLH1, and SHP1 genes has been
demonstrated in CTCL.6-9
The purpose of this study was to analyze the occur-
rence of promoter hypermethylation in CTCL on a
genome-wide scale. We analyzed DNA isolated from
skin tumor biopsy specimens of 28 patients with CTCL,
including CTCL entities with aggressive clinical behavior
(transformed mycosis fungoides and CD30-negative pri-
mary cutaneous large T-cell lymphoma) and indolent
CTCL (CD30-positive primary cutaneous large T-cell lym-
phoma). To identify novel methylation targets and to
screen for global DNA methylation patterns, differential
methylation hybridization (DMH) using CpG island mi-
croarrays was applied.19,20 Additionally, the methylation
status of eight tumor suppressor genes, selected for their
known involvement in lymphomagenesis, was examined
using bisulfite sequence analysis (BSA).
In this study, we show that malignant T cells in CTCL
exhibit widespread promoter hypermethylation suggestive
of epigenetic instability. The novel methylation targets we
identified in CTCL include the putative tumor suppressor
2 JOURNAL
Downloaded from www.jco.org at WALAEUS LIBRARY on May 20, 2005 .
Copyright © 2005 by the American Society of Clinical Oncology. All rights reserved.
82
genes BCL7a (B-cell CLL/Lymphoma 7a), PTPRG (protein
tyrosine phosphatase receptor gamma), and THBS4
(thrombospondin 4). We found six of eight selected tumor
suppressor gene promoters hypermethylated in CTCL in-
cluding MGMT, p73, p16, p15, CHFR, and TMS1.
MATERIALS AND METHODS
Patient Samples
Snap-frozen biopsy specimens were obtained from skin
tumors of 28 patients with CTCL, including 12 patients with
tumor-stage mycosis fungoides with blast cell transformation
(MF-TR) characterized by the presence of � 25% blast cells,
five patients with CD30-negative primary cutaneous large
T-cell lymphoma (CD30� LTCL) and 11 patients with CD30-
positive primary cutaneous large T-cell lymphoma (CD30�
LTCL), defined by criteria of the European Organisation for
Research and Treatment of Cancer classification for cutaneous
lymphomas.1 MF-TR and CD30� LTCL have an estimated
5-year survival of approximately 25%; CD30� LTCL has a 5-year
survival of 96%.1,21,22 Skin biopsy specimens were required to
contain at least 70% malignant T cells for inclusion in the study.
As controls, seven CD4� T-cell samples isolated from peripheral
blood of healthy volunteers and three skin biopsy specimens
from patients with inflammatory dermatoses (atopic dermatitis,
discoid lupus erythematosus) were used. DNA was extracted us-
ing the Genomic Tip kit (Qiagen, Hilden, Germany). All 28
CTCL samples were analyzed using DMH; 21 CTCL samples
were analyzed using BSA.
DMH Using CpG Island Microarrays
DMH was performed according to the detailed protocol
published by Yan et al,20 with the only modification that tumor
and control amplicons were hybridized to microarrays separately
and fluorescence signal intensities instead of ratios were used as
a measure of CpG island methylation. Briefly, DNA isolated from
tumor tissue was digested by MseI, ligated to linkers, and sequen-
tially digested with methylation-sensitive restriction enzymes
(HpaII and BstUI). The digested linker-ligated DNA was used
as template for polymerase chain reaction (PCR) amplification
(20 cycles) and fluorescence labeling using Cy5 before hybridiza-
tion to the CpG island microarray. As probes, a panel containing
8640 CpG island tags prepared from a genomic library arrayed on
glass slides were used. The identity of selected CpG islands
(CpGIs) was determined by sequence analysis of plasmids con-
tained in the CGI library.
Statistical Analysis
Data from DMH experiments were normalized using vari-
ance stabilizing normalization implemented in the R statistical
software package.23 A Wilcoxon rank-sums test was applied
separately to each CpG island in order to identify those with
consistently different intensity patterns in the 28 tumors when
compared with the 10 controls. The resulting list of P values was
corrected for multiple testing via the false discovery rate step-up
procedure of Benjamini & Hochberg.24 To identify CpG islands
methylated sporadically in only a subset of tumors, for which
the Wilcoxon test is not suited, we first estimated the CpG island–
specific intensity distribution based upon the 10 control samples
and a gamma-distribution model. These estimates were used to
compute upper-tail probabilities corresponding to the intensity
OF CLINICAL ONCOLOGY
 Epigenetic Profiling of CTCL

of each tumor. A threshold D Z 0.01 for these probabilities was
applied, below which tumor intensities were classified as po-
tentially methylated. Under the hypothesis that tumor intensities
behave in the same way as control intensities, the number of
potentially methylated intensities per CpG island follows a bino-
mial distribution, with sample size 28 and probability D Z 0.01.
This yields a list of P values, which were then corrected for multiple
testing as those from the Wilcoxon test. Considered as frequently
hypermethylated were CpG islands that exhibited values exceeding
this threshold in at least four tumor samples. The false discovery
rate for CpG islands selected using the Wilcoxon test was 30%
and for the sporadic test, 2.1%. Subsequently excluded were
CpG islands that had fluorescence intensity values that were below
a predetermined background level in all tumor samples or above
this level in any of the control samples.
Bisulfite Sequence Analysis
Two �g of patient sample–derived DNA was bisulfite-
converted using the EZ DNA methylation kit (Zymo Research,
Orange, CA). Primers were designed to anneal to bisulfite-
converted DNA as template. Primer sequences (Invitrogen,
Breda, the Netherlands) are given in Table 1. PCR reactions
were carried out in 25-�L volume using 50-ng bisulfite-
converted DNA as template. Cycle parameters for all analyzed
CpG islands were as follows: denaturing at 94�C for 5 minutes;
annealing at temperatures varying from 63�C to 56�C according
to primer set for 1 minute and extension at 72�C for 1 minute for
five cycles; followed by denaturing for 30 seconds, annealing for 1
minute and extension for 1 minute at similar temperatures for 30
cycles. Following electrophoresis, PCR products were excised
from the agarose gel and purified using the Qiaquick Gel Extrac-
tion Kit (Qiagen). Sequence analysis was performed on an ABI
Prism 3700 DNA analyzer (Applied Biosystems, Nieuwerkerk
aan den IJssel, the Netherlands) under standard conditions.
Validation experiments using mixtures (1:1 and 1:3) of un-
methylated semen and completely methylated CpGnome univer-
sal methylated DNA (Chemicon, Hampshire, United Kingdom)
for each primer combination demonstrated absence of PCR bias
and showed that methylation could accurately be detected if
a minimum of 25% of total analyzed DNA was methylated. Cy-
tosine peaks in the chromatogram with a height of 0.4 or more
relative to the thymine peak height were considered as indicative
of methylation. A CpG island was defined as aberrantly methyl-
ated if the density of methylated CpG dinucleotides contained in
the amplified sequence exceeded 15%, consistent with the thresh-
old value implemented in other studies.25,26
Cell Culture, 5-Aza-2#-Deoxycytidine Treatment
The MF cell line MyLa (courtesty of Dr K. Kaltoft) was cul-
tured in RPMI 1640 (Invitrogen) supplemented with 200 U/mL
penicillin, 200 �g/mL streptomycin, 2 �mol/mL L-glutamine,
20% fetal calf serum, and 12.5% crude T-cell extract at 37�C,
5% CO2.27 For demethylation studies, cells were cultured in
this medium to which 5-aza-2#-deoxycytidine (Sigma, St. Louis,
MO) was added to a concentration of 2 �mol/L for 4 days before
extraction of RNA using the RNeasy kit (Qiagen).
Real-Time Quantitative PCR
cDNA synthesis was performed on 1-�g total RNA after
treatment with RQ1 DNase I (Promega, Madison, WI) using
Superscript III reverse transcriptase (Invitrogen) and an
oligo(dT)12-18 primer. Quantitative PCR (qPCR) was performed
using DyNAmo HS SYBR Green qPCR kit (Finnzymes, Espoo,
Finland) in an ABI-Prism 7700 sequence detection system (Ap-
plied Biosystems). For normalization, cleavage and polyadenyla-
tion specific factor 6 (CPSF6) and TATA-box-binding protein
(TBP) were used, genes stably expressed also after treatment
with 5-aza-2#-deoxycytidine (T. van Wezel, personal communi-
cation, September 2004). Primer sequences are given in Table 2.
Cycle parameters were as follows: denaturing for 15 seconds at
95�C, and annealing and extension for 60 seconds at 60�C for
40 cycles. Specificity of PCR products was confirmed by agarose

Table 1. Primers for Bisulfite Sequence Analysis

CpGs in Position of Amplicon Relative Amplicon
Gene Primer Sequence (5# - 3#) Amplicon to Transcription Start Site Size (bp)
BCL7a S GAGAGTTTTTGGTGGTTTGTTTGGTA 26 �192, �546 355
AS GGTATTTGTAGTTTTCGAGGAAGGGTTAG
PTPRg S GGAGTTGTTTGTTTGAAGTTCGGAGA 60 �464, �30 494
AS CCACCAACACGATTCCAATAACCT
THBS4 S TTAGGATAGGGGATTTCGTTAAGGG 56 �86, �532 446
AS CAAATCCCTCCACTCACTCAACA
p15 S GGATAGGGGGCGGAGTTTAAGG 32 �109, �266 375
AS CTCTTCCCTTCTTTCCCACGCTACTC
p16 S AGTATTAGGAGGAAGAAAGAGGAG 32 �204, �215 419
AS TCCAATTCCCCTACAAACTCC
p73 S GGTTGGGGTTGGGAGAGTAGTTT 21 �740, -448 292
AS CCAAACGACCCATCTTTCCTAACA
MGMT S GGAGCGGTATTAGGAGGGGAGAGA 21 �568, -339 229
AS CCTTCCCAACTTCCGCCTAAAACT
MLH1 S GAGTGAAGGAGGTTACGGGTAAGT 33 �697, -354 343
AS CTCAAACTCCTCCTCTCCCCTTA
RB1 S GGAGGGGGTGGTTTTGGGTAG 22 �250, -8 242
AS CCCCCGCCCGACAACTAAAC
SOCS1 S GTAGGGGAGGAGAGGATAGGGTTT 25 �446, -108 338
AS CCCCCAACTCCACTTTTAATTTCTC
TMS1 S GAGGGGATTAAGGGTGTAGTAAGGAAG 44 �85, �458 439
AS CTACAACTACCCGACCATCTCCTACA
CHFR S GTTTTGTGTTTTAGATTTCGGATTTGTG 15 �400, -130 270
AS CTCGACCATCTTTAATCCTAACCAAAC

www.jco.org 3

83
 van Doorn et al
Table 2. Primers for Real-Time Quantitative Polymerase Chain Reaction
cDNA Primer Sequence (5# - 3#)
BCL7a S TGGTGACACATCCCTACGAA
AS CACTTCTCGTCCTTGCCTTT
PTPRG S TTGGGATCATAACGCACAGA
AS CTCGACTTGGCCAGTACACA
CPSF6 S AAGATTGCCTTCATGGAATTGAG
AS TCGTGATCTACTATGGTCCCTCTCT
TBP S CACGAACCACGGCACTGATT
AS TTTTCTTGCTGCCAGTCTGGAC
gel electrophoresis, melting curve, and sequence analysis of test
samples. Experiments were performed in duplicate. Cross-over
point values were used to calculate the gene-specific input
cDNA amount. Data were evaluated using Sequence Detection
System software version 1.9.1 (Applied Biosystems) and the sec-
ond derivative maximum algorithm.
RESULTS
Examination of DNA Methylation Patterns
Using CpG Island Microarrays
To screen for hypermethylated CpG islands in the 28
CTCL tumor specimens, we applied DMH using CpG is-
land microarrays. DMH allows screening of methylation
of 8640 CpG islands, most of which are located in gene
promoters, and has been established in previous studies
to accurately identify DNA methylation patterns.20,28,29
This method is based on the use of methylation-sensitive
endonucleases that digest unmethylated DNA sequences,
whereas methylated sequences are protected from diges-
tion. Following methylation-sensitive digestion of DNA
extracted from tumor and control samples, methylated
DNA sequences can therefore be selectively amplified by
PCR, labeled, and hybridized to a high-density microarray
of CpG islands.19 Differences in methylation of a particular
CpG island between tumor and control samples are
reflected in differences in fluorescence intensities of CpG
island tags on the microarray corresponding to the
amplified DNA sequences. Since T cells have been found
to display heterogeneous DNA methylation patterns, we
compared methylation characteristics of 28 CTCL samples
with those of multiple controls including seven T-cell sam-
ples and three inflammatory skin disease biopsy speci-
mens.30 A statistical algorithm suited for the analysis of
methylation profiling data was used, which detects CpG
islands methylated consistently as well as CpG islands
methylated sporadically in four or more of the tumor sam-
ples. Thus 35 CpG islands were identified as differentially
methylated in CTCL. The methylation patterns, identity,
and location of these 35 CpG islands are presented in Fig-
ure 1. Fourteen of these CpG islands were identified as
consistently methylated using the Wilcoxon test, 15 as spo-
radically methylated, and six CpG islands were identified
by both methods. The CpG island associated with the
4 JOURNAL
Downloaded from www.jco.org at WALAEUS LIBRARY on May 20, 2005 .
Copyright © 2005 by the American Society of Clinical Oncology. All rights reserved.
84
Accession No. Exonic Location Amplicon Size (bp)
NM_020993 2,3 99
NM_002841 26,27 86
NM_007007 9,10 137
NM_003194 5,6 89
BCL7a gene was among the most discriminating between
CTCL and control samples. Methylation in CTCL was
evenly distributed over the various chromosomes, in
contrast to acute myeloid leukemia, where preferential
methylation of CpG islands located on one particular
chromosome was observed.31 Among the CpG islands hy-
permethylated in CTCL were the promoters of collagen
alpha II type I and THBS4, genes previously shown to
be methylated in colon cancer cells using different methy-
lation detection methods.18,32
Confirmation of BCL7a, PTPRG, and THBS4 as
Novel Methylation Targets in CTCL
Among the 35 CpG islands identified as hypermethy-
lated in CTCL by DMH, we selected for confirmation studies
CpG islands of BCL7a, PTPRG, and THBS4, as their epige-
netic inactivation may be relevant in the pathogenesis of
CTCL. These putative tumor suppressor genes were de-
tected as consistently hypermethylated (THBS4), sporadi-
cally hypermethylated (PTPRG), or identified by both
methods (BCL7a). Direct BSA, applied to confirm the re-
sults of DMH experiments, provides a map of average meth-
ylation densities for each of the cytosines contained in the
DNA sequences amplified by PCR following bisulfite con-
version.25,26,33 This method makes use of the fact that
sodium bisulfite deaminates cytosine to uracil, which is re-
placed in subsequent PCR amplification by thymine, while
leaving methylated cytosine unaltered. Each cytosine base
present in the DNA sequence following bisulfite conversion
and PCR amplification is indicative of methylated cytosine
in the original DNA sequence. Figure 2 illustrates the re-
sults of BSA of the CpG island located in the first exon of
BCL7a. As presented in Table 3, the CpG islands of
BCL7a, PTPRG, and THBS4 were hypermethylated in
48%, 24%, and 52% of CTCL samples respectively, con-
firming DMH results. These three novel methylation targets
were methylated among all three CTCL entities, as well as in
the CTCL cell line MyLa, but not in any of the six control
samples analyzed.
Demethylation and Reactivation of BCL7a and
PTPRG Expression
Next we analyzed whether methylation of the CpG
island located in the first exon of BCL7a and in the
OF CLINICAL ONCOLOGY
 Epigenetic Profiling of CTCL

MF-TR CD30 - LTCL CD30 + LTCL Controls

Gene name Location Locus Acc. no.
predicted gene NT_005079.5 promoter 2q14.3 AC067962
SIN3A promoter 15q24.2 AC105137
GRIK2 exon 1 6q16.3-q21 BC037954
angiomotin-like 1 intron 1 11q21 AF453742
collagen, type I, alpha 1 intron 1 17q21.3-q22.1 BC036531
Sideroflexin 3 promoter 10q24.32 BC000124
clone RP11-603F24 / 2p21-p22 AF106953
clone 10PTEL013 / 10q26.1 Z96142
Ells1 intron 1, exon 2 7p15.1 AY054121
Paired Box Protein PAX-6 intron 5 11p13 BC011953
SIN3A promoter 15q23 AL832463
Serpine 1 exon 1 7q21.3-q22 BC010860
aldehyde dehydrogenase 9 family A1 promoter 1q23.1 BC070030
PDE4DIP (myomegalin) intron 1 1q21.1 AB007923
NIMA-related kinase 11 promoter, exon 1 3q22.1 BC028587
deleted in lymphocytic leukemia 1 promoter 13q14.3 Y15227
protein tyrosine phosphatase receptor gamma promoter 3p21-p14 L09247
C6orf102 promoter 6p21.2 AL832634
C9orf64 promoter 9q21.32 AK090882
DCOHM intron 2 5q31.1 AL136721
protein tyrosine phosphatase receptor N2 intron 2 7q36 U81561
clone IMAGE:4793694 promoter, exon 1 19q13.43 BC036412
clone IMAGE:4793694 promoter, exon 1 19q13.43 BC036412
collagen, type I, alpha 2 exon 1 7q22.1 J03464
C22orf4 promoter 22q13.3 BC029897
B-cell CLL/Lymphoma 7A (BCL7A) exon 1 12q24.13 X89984
Gene BC030533 promoter 7q34 BC030533
PACE4 exon 16 15q26 AB001909
latent TGF-beta binding protein 2 intron 2 14q24.3 Z37976
cadherin-like 26 exon 10 20q13.2-q13.33 AF169690
thrombospondin 4 promoter 5q13 NM_003248
SSX2IP intron 1 1p22.3 AB023140
NYD-SP29 promoter, exon 1 1p22.3 AY049724
C6orf117 promoter, exon 1 6q14.3 AK090775
clone IMAGE:4793694 promoter, exon 1 19q13.43 BC036412
Syntaxin 17 promoter 9q31.1 AK000658

Fig 1. Methylation patterns of 35 CpG islands identified as hypermethylated when comparing 28 cutaneous T-cell lymphoma (CTCL) samples (mycosis fungoides
with blast cell transformation [MF-TR], CD30-negative large T-cell lymphoma [LTCL], and CD30-positive LTCL) with 10 control samples using differential
methylation hybridization (DMH) on CpG microarrays. The colors reflect normalized fluorescence values that correspond to likelihood of methylation of the
particular CpG island (red Z high, white Z low). The gene name, location of the CpG island, and locus and accession number (Acc. no.) of the gene are indicated.
Patient samples were ordered according to CTCL entity; the 35 CpG islands were ordered using a hierarchical clustering algorithm based on the Euclidean distance
and average-linkage method.

promoter of PTPRG is associated with gene silencing.
Promoter methylation of the THBS4 gene had previously
been shown to be associated with gene silencing in cancer
cells.18 The MyLa CTCL cell line with demonstrated
methylation of BCL7a and PTPRG was exposed to
5-aza-2#-deoxycytidine and effects on expression of these
genes before and after treatment with this demethylating
agent was assessed by real-time quantitative PCR. As illus-
trated in Figure 3, expression of these genes, nearly unde-
tectable before exposure to 5-aza-2#-deoxycytidine, was
dramatically increased upon demethylation. Expression
of BCL7a increased with a fold change, denoting the aver-
age expression of the gene of interest relative to two stably
expressed genes, of approximately 489. Expression of
PTPRG increased more than seven-fold. These results sug-
gest that expression of BCL7a and PTPRG is epigenetically
regulated and that promoter methylation as occurs in
CTCL is associated with transcriptional downregulation.
Promoter Hypermethylation of Eight Selected
Tumor Suppressor Genes in CTCL
Subsequently the methylation characteristics of pro-
moters of eight selected tumor suppressor genes were
examined in 21 CTCL patient samples using BSA. Hyper-
methylation of these genes was recently shown in MF (p15,
p16, and MLH1), in nodal lymphomas (p73, MGMT, and
SOCS1), or in solid tumors (TMS1 and CHFR).34-38 In ad-
dition, promoter hypermethylation of these genes is known
to be associated with gene silencing in cancer cells. As pre-
sented in Table 4, aberrant promoter methylation of p73
(48%), p16 (33%), CHFR (19%), p15 (10%), and TMS1
(10%) could be detected in CTCL, but not in control
samples. The CpG island located in the promoter of the
MGMT gene was found to be methylated in all tested
CTCL samples, but unexpectedly was also methylated in
three of the six CD4� T-cell samples included as controls.
An overview of clinical features, disease course data of the
CTCL patients, and the methylation status of the 11 genes
investigated using BSA is presented in Table 5.
Promoter Hypermethylation in CTCL Entities
With Different Clinical Behavior
To evaluate possible correlations between DNA meth-
ylation and CTCL entities with different clinical behav-
ior, we compared the methylation characteristics of
CTCL entities with known aggressive behavior (MF-TR,
CD30� LTCL) with those of an indolent CTCL entity
(CD30� LTCL). As noticeable in Figure 1, the CpG is-
land methylation patterns of aggressive and indolent
CTCLs as identified using DMH show great similarity.
Hierarchical clustering of patient samples using the entire
set of 8640 CpG islands or the selection of 35 CpG

www.jco.org 5

Downloaded from www.jco.org at WALAEUS LIBRARY on May 20, 2005 .

Copyright © 2005 by the American Society of Clinical Oncology. All rights reserved.

85
 van Doorn et al

A BCL7A CpG Island

252 918

5' Exon 1 3'

1 1063
ATG 972
192 546
Amplicon

gagagtctctggtggtctgcctggcaccaggcaccttcctacaaccct
1 2 3
agttttccaaaaggacaaagcctggggcaggcgacgtcctagctcgca
4 5 6 7
tttgaacagggccgcgggccagcagagatgcgcgatgcccaactcttt
8 9 10 11 12 Fig 2. Hypermethylation of the CpG
ccaagagcacctcgcgtcccgaaccggtgccttcaactcggagaagtc
13 14 15 16 17 18
island located in the first exon of the
aagagacccgcaagaaacttgcacgactgcacccgccgccgcgctctg BCL7a gene. (A) Schematic represen-
19 20 21 tation of the first exon of BCL7a
ggggctgggcaggggcagctgggctggctcccggggaacgcgaccccc
22 23 24 25 26 (accession number AC069503) on
ccgcgccccgcagaccggctgtctcccatggacccctcggcacctgca 12q24.31, the location of the CpG island
gcctccgaggaagggtcag
and of the region amplified for bisulfite
sequence analysis. (B) Exemplary re-
sults of bisulfite sequence analysis of
B 8 9 10 11 the BCL7a CpG island in a tumor
sample of a cutaneous T-cell lymphoma
T T T TG T G T T T T G A A T T G G T G (CTCL) patient with an unmethylated
(upper) and a CTCL patient with a hyper-
methylated (lower) BCL7a CpG island.
Unmethylated and methylated bisulfite-
Nonmethylated converted sequences are indicated
CTCL Patient above the chromatograms.
Sample

8 9 10 11
T T T CG C G T T T C G A A T C G G T G

Hypermethylated
CTCL Patient
Sample

islands hypermethylated in CTCL revealed that the two
groups of CTCL entities with different clinical behavior
could not be separated on the basis of their methyla-
tion pattern.
In the included CTCL patient samples, the frequency of
promoter methylation did not differ significantly between
patients with an aggressive CTCL entity (3.2 of 11 pro-
moters) and an indolent CTCL entity (3.1 of 11 promoters).
This contrasts with acute lymphoblastic leukemia, where
prognostic significance of promoter hypermethylation
frequency was recently reported.39 Of the 11 investigated
tumor suppressor genes, only BCL7a was methylated signif-
icantly more frequently in the aggressive CTCLs MF-TR and
CD30� LTCL than in CD30� LTCL (64% v 14%; �2-test,
P � .05). Comparison of survival rates using the log-rank
test showed no significant association between hypermeth-
ylation of any of these 11 genes and survival, possibly related
to the limited number of 21 patients and short follow-up
duration in this study group.
DISCUSSION
In this study we evaluated the occurrence of aberrant DNA
methylation in skin biopsy specimens obtained from pa-
tients with CTCL using a combination of a screening ge-
nomic approach, DMH on CpG island microarrays, and
a candidate-gene approach, BSA. Using DMH analysis, 35
CpG islands were identified as hypermethylated in at least
four of 28 studied CTCL patient samples. Hypermethyla-
tion of the putative tumor suppressor genes BCL7a,
PTPRG, and THBS4 was confirmed. Subsequently, we
showed that promoter hypermethylation of BCL7a and
PTPRG is associated with gene silencing, as expression

6 JOURNAL OF CLINICAL ONCOLOGY

Downloaded from www.jco.org at WALAEUS LIBRARY on May 20, 2005 .

Copyright © 2005 by the American Society of Clinical Oncology. All rights reserved.

86
 Epigenetic Profiling of CTCL

C CpG-Dinucleotides
Patient Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
MF-TR #1
MF-TR #2
MF-TR #3
MF-TR #4
MF-TR #5
MF-TR #6
MF-TR #7
MF-TR #8
MF-TR #9
MF-TR #10
MF-TR #11 Fig 2. (continued) (C) Methylation density map
of the BCL7a CpG island amplicon, containing
CD30 - LTCL #1 26 CpG dinucleotides, in 21 CTCL samples and
CD30 - LTCL #2 nine controls. Unmethylated cytosine bases
CD30 - LTCL #3 are indicated as (h), methylated cytosines as
(-). Marked in gray are the patient samples
CD30 + LTCL #1
considered as having a hypermethylated
CD30 + LTCL #2
CD30 + LTCL #3 BCL7a CpG island. MF-TR, mycosis fungoides
CD30 + LTCL #4 with blast cell transformation; LTCL, large
CD30 + LTCL #5 T-cell lymphoma.
CD30 + LTCL #6
CD30 + LTCL #7

Control #1
Control #2
Control #3
Control #4
Control #5
Control #6
Control #7
Control #8
Control #9

of these genes was restored by chemical demethylation in
the MyLa CTCL cell line that exhibits hypermethylation of
these genes. The THBS4 gene was previously shown to be
silenced through promoter methylation in cancer cells.18
Additionally, we demonstrated hypermethylation of
six of eight selected tumor suppressor gene promoters
in CTCL using BSA. Promoters of the MGMT, p73, p16,
p15, CHFR, and TMS1 genes were found to be hyper-
methylated in CTCL at varying frequencies.
Concerning the results of DMH analysis, the DNA
methylation pattern exhibited by CTCL is tumor type–
specific, as it was different from that of colon, ovarian,
and breast cancer analyzed previously with similar meth-
ods.20,28,29 Hierarchical clustering revealed that MF-TR
and CD30� LTCL could not be distinguished from
Table 3. Hypermethylation of Methylation Targets Identified by
Differential Methylation Hybridization Analysis in Different
Cutaneous T-Cell Lymphoma Entities, As Analyzed
Using Bisulfite Sequence Analysis
BCL7a PTPRG
(%) (%)
CTCL (n Z 21) 48 24
MF-TR (n Z 11) 64 27
CD30� LTCL (n Z 3) 67 0 100
CD30� LTCL (n Z 7) 14 29
Controls (n Z 6) 0 0
Abbreviations: CTCL, cutaneous T-cell lymphoma; MF-TR, mycosis
fungoides with blast cell transformation; LTCL, large T-cell lymphoma.
CD30� LTCL, which has a more favorable prognosis,
on the basis of their methylation patterns. This may not
be considered surprising as the three entities are very
closely related, all originating from CD4�, CD45RO�,
CLA� skin-homing memory T cells. Akin to loss of hetero-
zygosity, recurrent promoter methylation of genes in can-
cer cells may indicate that these genes encode proteins with
tumor suppressive functions, loss of which confers growth
advantage. In line with this assumption, the CTCL-specific
methylation pattern included several putative tumor sup-
pressor genes.
The CpG island of BCL7a, located in its first exon, was
methylated in 48% of CTCL samples. Epigenetic inactiva-
tion of BCL7a may be related to clinical behavior of CTCL,
as its CpG island was hypermethylated at a higher fre-
quency in aggressive than in indolent CTCL entities.
The BCL7a locus is on chromosome 12q24.31 at the site
of a recurrent breaking point in B-cell lymphomas. The
BCL7a gene was cloned as part of a complex chromosomal
translocation in a Burkitt’s lymphoma cell line and was
THBS4 subsequently found to be rearranged in another cell line
(%)
derived from mediastinal large B-cell lymphoma.40 The re-
52
45 sultant disruption of this gene is considered to be impli-
cated in the pathogenesis of these lymphomas. However,
43
0 the exact cellular function of BCL7a, which is expressed
at low levels in a wide variety of normal tissues, and the
consequences of its functional inactivation are currently
unknown. Notably, gene expression profiling has shown

www.jco.org 7

Downloaded from www.jco.org at WALAEUS LIBRARY on May 20, 2005 .

Copyright © 2005 by the American Society of Clinical Oncology. All rights reserved.

87
 van Doorn et al

A
CpG-Dinucleotides
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

BCL7a

CpG-Dinucleotides
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

PTPRG

B
0.020
BCL7a
0.018

0.016

0.014 Fig 3. Results of demethylation using

5-aza-2#-deoxycytidine on expression of
0.012 the BCL7a and PTPRG genes as mea-
2-�Ct

sured using real-time quantitative poly-

0.010 merase chain reaction analysis.
Expression levels of the genes of in-
0.008 terest are represented as the average
of 2-DCt(gene of interest-CPSF6) and 2-DCt(gene
0.006 of interest-TBP)
. Error bars indicate stan-
dard deviation from duplicate experi-
0.004 ments. (A) Methylation density of the
CpG island located in the first exon of
0.002 BCL7a and the CpG island in the
promoter of PTPRG in the MyLa cell
0 line before demethylation. (B) Histo-
Untreated After 5-aza-2'-deoxycytidine gram showing BCL7a gene expression
in the MyLa cell line before and after
exposure to 5-aza-2#deoxycytidine. (C)
C Histogram showing PTPRG gene ex-
pression in the MyLa cell line before
0.12 and after exposure to 5-aza-2#-deoxy-
PTPRG cytidine.

0.10

0.08
2-�Ct

0.06

0.04

0.02

Untreated After 5-aza-2'-deoxycytidine

that expression of BCL7a is diminished in MF skin lesions:
BCL7a was among the seven downregulated genes, signif-
icant in separating MF from benign inflammatory skin
diseases.41 In addition, BCL7a was recently found to be ex-
pressed at significantly lower levels in peripheral T-cell
lymphomas when compared with lymphoblastic lympho-
mas.42 These observations combined with the finding that
diminished expression of BCL7a is an unfavorable prog-
nostic sign in patients with B-cell lymphoma strongly sug-
gests that this gene functions as a tumor suppressor in

8 JOURNAL OF CLINICAL ONCOLOGY

Downloaded from www.jco.org at WALAEUS LIBRARY on May 20, 2005 .

Copyright © 2005 by the American Society of Clinical Oncology. All rights reserved.

88
 Epigenetic Profiling of CTCL

event in tumorigenesis. The PTPRG tumor suppressor

Table 4. Hypermethylation of Selected Tumor Suppressor Genes in
Different CTCL Entities, As Analyzed Using gene frequently contains missense mutations in colon
Bisulfite Sequence Analysis carcinomas and is often deleted in lung and renal
p15 p16 p73 MGMT MLH1 SOCS1 TMS1 CHFR carcinoma.45,46 Therefore, it has been concluded that
(%) (%) (%) (%) (%) (%) (%) (%)
PTPRG, which is normally expressed in lymphocytes, is
CTCL (n Z 21) 10 33 48 100 0 0 10 19
MF-TR (n Z 11) 18 28 40 100 0 0 9 18
a tumor suppressor gene. However, the functional sig-
CD30� LTCL 0 0 67 100 0 0 0 0 nificance of epigenetic silencing of PTPRG in lymphoid
(n Z 3)
CD30� LTCL 0 57 57 100 0 0 14 29
cells has not yet been established and demands fur-
(n Z 7) ther investigation.
Controls (n Z 6) 0 0 0 50 0 0 0 0
THBS4, hypermethylated in 11 (52%) of 21 CTCL
Abbreviations: CTCL, cutaneous T-cell lymphoma; MF-TR, mycosis samples, encodes an extracellular calcium-binding protein
fungoides with blast cell transformation; LTCL, large T-cell lymphoma.
involved in proliferation, adhesion, and migration. Pro-
moter hypermethylation of THBS4, as well of THBS1,
has been reported in colon carcinoma, but as yet not
lymphoid cells.43 Promoter hypermethylation is an impor- in lymphoma.18,47
tant mechanism of its inactivation in T-cell lymphomas. The promoter of MGMT was methylated in all CTCL
PTPRG, another novel target of epigenetic inactiva- samples, but also in three of six control samples, preclud-
tion in CTCL, was hypermethylated in five (24%) of 21 ing its use as a marker for malignancy in CTCL. MGMT
CTCL patients as confirmed by BSA. Because we addition- encodes a DNA repair enzyme that acts through removal
ally observed frequent methylation of PTPRN2 using of alkylating lesions at the guanine base protecting against
DMH in CTCL, and because methylation of PTPRO has mutagenesis and malignant transformation, loss of which
been described in hepatocellular carcinomas, members increases sensitivity to treatment with alkylating agents.
of the tyrosine phosphatase gene superfamily may be a The region amplified for BSA was located in the promoter
common target for epigenetic inactivation.44 Protein ty- of the MGMT gene, stretching 279 to 508 nucleotides
rosine phosphatases play a role in setting the levels of ty- upstream of the transcription start site and 365 nucleotides
rosine phosphorylation in cells by balancing the activity of upstream of the region reported to be methylated in
tyrosine kinases, thereby regulating signaling pathways B-cell lymphomas and gliomas (accession number
that control cellular growth. Whereas constitutive activity X61657).35,48,49 The observation of MGMT promoter
of tyrosine kinases has been long recognized as an im- methylation in T cells of healthy subjects is analogous to
portant oncogenic alteration, inactivation of tyrosine the finding of promoter methylation of the tumor sup-
phosphatases is increasingly considered as an important pressor gene DAPK in benign as well as malignant B cells.50

Table 5. Clinical Features, Disease Course Data, and Methylation Status of Tumor Suppressor Genes in 21 CTCL Patients
Follow-up
Patients and Age Duration
Diagnosis (years) Sex (months) Current Status BCL7A PTPRG THBS4 p15 p16 p73 MGMT MLH1 SOCS1 TMS1 CHFR
MF-TR #1 62 M 25 A� - -
MF-TR #2 88 F 38 A� --- -
MF-TR #3 58 M 15 D� - - - - - -
MF-TR #4 77 M 134 D� - -
MF-TR #5 70 M 48 A� ---- -
MF-TR #6 63 M 102 A� - - -
MF-TR #7 48 F 77 A� - - - -
MF-TR #8 79 M 29 A� - - -
MF-TR #9 77 M 34 D� - -
MF-TR #10 73 M 73 D� - - - - -
MF-TR #11 64 M 58 D� -
CD30� LTCL #1 74 M 14 D� - -
CD30� LTCL #2 81 F 27 D� - - - -
CD30� LTCL #3 71 M 8 A� - - -
CD30� LTCL #1 65 M 102 A� - - -
CD30� LTCL #2 72 M 71 A� - -
CD30� LTCL #3 62 M 63 A� -- -
CD30� LTCL #4 40 F 113 A� - -
CD30� LTCL #5 69 M 107 A� - - -- - -
CD30� LTCL #6 58 M 24 A� - - - --
CD30� LTCL #7 62 F 60 D� -
NOTE. A� denotes alive with disease; A� denotes alive with disease symptoms; D� denotes died due to lymphoma.
Abbreviations: CTCL, cutaneous T-cell lymphoma; MF-TR, mycosis fungoides with blast cell transformation; LTCL, large T-cell lymphoma.

www.jco.org 9

Downloaded from www.jco.org at WALAEUS LIBRARY on May 20, 2005 .

Copyright © 2005 by the American Society of Clinical Oncology. All rights reserved.

89
 van Doorn et al

The p73 gene promoter was hypermethylated in 11
(48%) of 21 CTCL patient samples. Epigenetic inactiva-
tion of the p73 gene appears to be particularly relevant
in the pathogenesis of CTCL, since activation-induced
cell death in T cells is dependent on the activity of
p73.51 Activation-induced cell death is a mechanism es-
sential in preventing unrestricted clonal expansion of
activated T cells, such as occurs in CTCL. Although
hypermethylation of p73 has been described in nodal
B-cell lymphomas and natural-killer cell lymphomas, its
methylation had not been previously reported in lympho-
mas of T-cell origin.35,52,53
The promoter of the CHFR gene, which encodes a
protein regulating the mitotic checkpoint pathway govern-
ing the transition to metaphase, was hypermethylated in
a minority of CTCL patients (19%). Its epigenetic inacti-
vation, previously shown in colon and gastric cancer,
has been analyzed only in advanced tumor stage MF with
blast cell transformation. In previous studies, promoter
hypermethylation of the p16 and MLH1 gene has also
been detected in the patch/plaque stage of MF, suggesting
that these epigenetic alterations may arise early in the
development of this T-cell malignancy.6,7,9 The demethy-
lating agents 5-aza-2#-deoxycytidine, zebularine, and
MG98, which can reverse silencing due to promoter hy-
permethylation, are currently tested in clinical trials for
treatment of various malignancies and the first agent
has already proven to be efficacious in chronic myeloid
leukemia.54-56 The finding of promoter hypermethylation
in CTCL not only gives insight into the molecular patho-
genesis of these malignancies, but in addition provides
a rationale for treatment of CTCL patients with demeth-
ylating agents.
- - -

may contribute to chromosomal instability.38 Consistent

Acknowledgment
with results of other research groups, we showed hyper-
We thank Joseph Liu (Division of Human Cancer
methylation of the p15 and p16 promoter in MF patient
Genetics, Comprehensive Cancer Center, Ohio State
samples in 18% and 28%6,7 of MF patients. Promoter
University) for technical assistance related to production
hypermethylation of p16 was additionally found in
of CpG island microarrays, Eddy Wierenga (Department
CD30� LTCL. Methylation of MLH1, previously found
of Cell Biology and Histology, AMC) for providing DNA
in a subset of MF patients with demonstrated microsatel-
from T cells of healthy subjects, Judith Boer (Department
lite instability, could not be detected in patients included
of Human and Clinical Genetics, LUMC) for advice
in this study.9
concerning microarray data analysis, and Jan Willem
Our results provide evidence for epigenetic instability
Dierssen and Tom van Wezel (Department of Pathology,
and widespread promoter methylation in CTCL associated
LUMC) for advice on measurement of gene expression in
with inactivation of several tumor suppressor genes, that
cells treated with 5-aza-2#-deoxycytidine.
may lead to cell cycle dysregulation (p15, p16, p73), defec-
tive DNA repair (MGMT), disruption of apoptosis signal- Authors’ Disclosures of Potential
ing (TMS1, p73), and chromosomal instability (CHFR). In Conflicts of Interest
this study, the occurrence of promoter hypermethylation The authors indicated no potential conflicts of interest.

REFERENCES
1. Willemze R, Kerl H, Sterry W, et al: EORTC
classification for primary cutaneous lymphomas:
A proposal from the Cutaneous Lymphoma
Study Group of the European Organization for
Research and Treatment of Cancer. Blood 90:
354-371, 1997
2. van Doorn R, Dijkman R, Vermeer MH,
et al: A novel splice variant of the Fas gene in
patients with cutaneous T-cell lymphoma. Can-
cer Res 62:5389-5392, 2002
3. Capocasale RJ, Lamb RJ, Vonderheid EC,
et al: Reduced surface expression of trans-
forming growth factor beta receptor type II in
mitogen-activated T cells from Sezary patients.
Proc Natl Acad Sci U S A 92:5501-5505, 1995
4. Sommer VH, Clemmensen OJ, Nielsen O,
et al: In vivo activation of STAT3 in cutaneous
T-cell lymphoma. Evidence for an antiapoptotic
function of STAT3. Leukemia 18:1288-1295,
2004
5. Mao X, Lillington DM, Czepulkowski B,
et al: Molecular cytogenetic characterization of
Sezary syndrome. Genes Chromosomes Cancer
36:250-260, 2003
6. Navas IC, Ortiz-Romero PL, Villuendas R,
et al: p16(INK4a) gene alterations are frequent
in lesions of mycosis fungoides. Am J Pathol
156:1565-1572, 2000
7. Scarisbrick JJ, Woolford AJ, Calonje E,
et al: Frequent abnormalities of the p15 and p16
genes in mycosis fungoides and Sezary syn-
drome. J Invest Dermatol 118:493-499, 2002
8. Zhang Q, Raghunath PN, Vonderheid E,
et al: Lack of phosphotyrosine phosphatase
SHP-1 expression in malignant T-cell lymphoma
cells results from methylation of the SHP-1
promoter. Am J Pathol 157:1137-1146, 2000
9. Scarisbrick JJ, Mitchell TJ, Calonje E, et al:
Microsatellite instability is associated with hy-
permethylation of the hMLH1 gene and reduced
gene expression in mycosis fungoides. J Invest
Dermatol 121:894-901, 2003
10. Deng G, Song GA, Pong E, et al: Promoter
methylation inhibits APC gene expression by
causing changes in chromatin conformation and
interfering with the binding of transcription
factor CCAAT-binding factor. Cancer Res 64:
2692-2698, 2004
11. Herman JG, Baylin SB: Gene silencing in
cancer in association with promoter hyper-
methylation. N Engl J Med 349:2042-2054, 2003
12. Costello JF, Fruhwald MC, Smiraglia DJ,
et al: Aberrant CpG-island methylation has non-
random and tumour-type-specific patterns. Nat
Genet 24:132-138, 2000
13. Esteller M, Corn PG, Baylin SB, et al: A
gene hypermethylation profile of human cancer.
Cancer Res 61:3225-3229, 2001
14. Laird PW: The power and the promise of
DNA methylation markers. Nat Rev Cancer 3:
253-266, 2003
15. Egger G, Liang G, Aparicio A, et al:
Epigenetics in human disease and prospects
for epigenetic therapy. Nature 429:457-463,
2004
16. Piekarz RL, Robey R, Sandor V, et al:
Inhibitor of histone deacetylation, depsipeptide
(FR901228), in the treatment of peripheral and
cutaneous T cell lymphoma: A case report. Blood
98:2865-2868, 2001

10 JOURNAL OF CLINICAL ONCOLOGY

Downloaded from www.jco.org at WALAEUS LIBRARY on May 20, 2005 .

Copyright © 2005 by the American Society of Clinical Oncology. All rights reserved.

90

17. Esteller M: Profiling aberrant DNA methyl-
ation in hematologic neoplasms: A view from the
tip of the iceberg. Clin Immunol 109:80-88, 2003
18. Kondo Y, Shen L, Yan PS, et al: Chromatin
immunoprecipitation microarrays for identifica-
tion of genes silenced by histone H3 lysine 9
methylation. Proc Natl Acad Sci U S A 101:7398-
7403, 2004
19. Yan PS, Chen CM, Shi H, et al: Ap-
plications of CpG island microarrays for high-
throughput analysis of DNA methylation. J Nutr
132:2430S-2434S, 2002 (suppl)
20. Yan PS, Efferth T, Chen HL, et al: Use of
CpG island microarrays to identify colorectal
tumors with a high degree of concurrent
methylation. Methods 27:162-169, 2002
21. Vergier B, de Muret A, Beylot-Barry M,
et al: Transformation of mycosis fungoides:
clinicopathological and prognostic features of
45 cases. French Study Group of Cutaneious
Lymphomas. Blood 95:2212-2218, 2000
22. Bekkenk MW, Geelen FA, van Voorst
Vader PC, et al: Primary and secondary cutane-
ous CD30(�) lymphoproliferative disorders: A
report from the Dutch Cutaneous Lymphoma
Group on the long-term follow-up data of 219
patients and guidelines for diagnosis and treat-
ment. Blood 95:3653-3661, 2000
23. Huber W, von Heydebreck A, Sultmann H,
et al: Variance stabilization applied to microarray
data calibration and to the quantification of
differential expression. Bioinformatics 18:S96-
104, 2002 (suppl)
24. Benjamini Y, Hochberg Y: Controlling the
false discovery rate: A practical and powerful
approach to multiple testing. J R Statist Soc B
57:289-300, 1995
25. Toyota M, Kopecky KJ, Toyota MO, et al:
Methylation profiling in acute myeloid leukemia.
Blood 97:2823-2829, 2001
26. Garcia-Manero G, Daniel J, Smith TL, et al:
DNA methylation of multiple promoter-associ-
ated CpG islands in adult acute lymphocytic
leukemia. Clin Cancer Res 8:2217-2224, 2002
27. Kaltoft K, Bisballe S, Dyrberg T, et al:
Establishment of two continuous T-cell strains
from a single plaque of a patient with mycosis
fungoides. In Vitro Cell Dev Biol 28A:161-167,
1992 (3 pt 1)
28. Yan PS, Chen CM, Shi H, et al: Dissecting
complex epigenetic alterations in breast cancer
using CpG island microarrays. Cancer Res 61:
8375-8380, 2001
29. Wei SH, Chen CM, Strathdee G, et al:
Methylation microarray analysis of late-stage
ovarian carcinomas distinguishes progression-
free survival in patients and identifies candidate
epigenetic markers. Clin Cancer Res 8:2246-
2252, 2002
30. Zhu X, Deng C, Kuick R, et al: Analysis of
human peripheral blood T cells and single-cell-
www.jco.org
Downloaded from www.jco.org at WALAEUS LIBRARY on May 20, 2005 .
Copyright © 2005 by the American Society of Clinical Oncology. All rights reserved.
Epigenetic Profiling of CTCL
derived T cell clones uncovers extensive clonal
CpG island methylation heterogeneity through-
out the genome. Proc Natl Acad Sci U S A
96:8058-8063, 1999
31. Rush LJ, Dai Z, Smiraglia DJ, et al: Novel
methylation targets in de novo acute myeloid
leukemia with prevalence of chromosome 11
loci. Blood 97:3226-3233, 2001
32. Sengupta PK, Smith EM, Kim K, et al: DNA
hypermethylation near the transcription start site
of collagen alpha2(I) gene occurs in both cancer
cell lines and primary colorectal cancers. Cancer
Res 63:1789-1797, 2003
33. Fraga MF, Esteller M: DNA methylation: A
profile of methods and applications. Biotech-
niques 33:632,634,636-49, 2002
34. Kawano S, Miller CW, Gombart AF, et al:
Loss of p73 gene expression in leukemias/
lymphomas due to hypermethylation. Blood
94:1113-1120, 1999
35. Esteller M, Gaidano G, Goodman SN, et al:
Hypermethylation of the DNA repair gene O(6)-
methylguanine DNA methyltransferase and
survival of patients with diffuse large B-cell
lymphoma. J Natl Cancer Inst 94:26-32, 2002
36. Chim CS, Wong KY, Loong F, et al: SOCS1
and SHP1 hypermethylation in mantle cell
lymphoma and follicular lymphoma: Implications
for epigenetic activation of the Jak/STAT path-
way. Leukemia 18:356-358, 2004
37. Conway KE, McConnell BB, Bowring CE,
et al: TMS1, a novel proapoptotic caspase
recruitment domain protein, is a target of
methylation-induced gene silencing in human
breast cancers. Cancer Res 60:6236-6242, 2000
38. Toyota M, Sasaki Y, Satoh A, et al:
Epigenetic inactivation of CHFR in human
tumors. Proc Natl Acad Sci U S A 100:7818-
7823, 2003
39. Roman-Gomez J, Jimenez-Velasco A, Cas-
tillejo JA, et al: Promoter hypermethylation of
cancer-related genes: A strong independent
prognostic factor in acute lymphoblastic leuke-
mia. Blood 104:2492-2498, 2004
40. Zani VJ, Asou N, Jadayel D, et al: Molec-
ular cloning of complex chromosomal trans-
location t(8;14;12)(q24.1;q32.3;q24.1) in a
Burkitt lymphoma cell line defines a new gene
(BCL7A) with homology to caldesmon. Blood
87:3124-3134, 1996
41. Tracey L, Villuendas R, Dotor AM, et al:
Mycosis fungoides shows concurrent deregula-
tion of multiple genes involved in the TNF
signaling pathway: An expression profile study.
Blood 102:1042-1050, 2003
42. Martinez-Delgado B, Melendez B, Cuadros
M, et al: Expression profiling of T-cell lympho-
mas differentiates peripheral and lymphoblastic
lymphomas and defines survival related genes.
Clin Cancer Res 10:4971-4982, 2004
91
43. Alizadeh AA, Eisen MB, Davis RE, et al:
Distinct types of diffuse large B-cell lymphoma
identified by gene expression profiling. Nature
403:503-511, 2000
44. Motiwala T, Ghoshal K, Das A, et al:
Suppression of the protein tyrosine phosphatase
receptor type O gene (PTPRO) by methylation in
hepatocellular carcinomas. Oncogene 22:6319-
6331, 2003
45. Wang Z, Shen D, Parsons DW, et al:
Mutational analysis of the tyrosine phosphatome
in colorectal cancers. Science 304:1164-1166,
2004
46. LaForgia S, Morse B, Levy J, et al: Re-
ceptor protein-tyrosine phosphatase gamma is
a candidate tumor suppressor gene at human
chromosome region 3p21. Proc Natl Acad Sci
USA 88:5036-5040, 1991
47. Ahuja N, Mohan AL, Li Q, et al: Association
between CpG island methylation and micro-
satellite instability in colorectal cancer. Cancer
Res 57:3370-3374, 1997
48. Esteller M, Garcia-Foncillas J, Andion E,
et al: Inactivation of the DNA-repair gene MGMT
and the clinical response of gliomas to alkylating
agents. N Engl J Med 343:1350-1354, 2000
49. Rossi D, Capello D, Gloghini A, et al:
Aberrant promoter methylation of multiple
genes throughout the clinico-pathologic spec-
trum of B-cell neoplasia. Haematologica 89:154-
164, 2004
50. Reddy AN, Jiang WW, Kim M, et al: Death-
associated protein kinase promoter hypermethy-
lation in normal human lymphocytes. Cancer Res
63:7694-7698, 2003
51. Lissy NA, Davis PK, Irwin M, et al: A
common E2F-1 and p73 pathway mediates cell
death induced by TCR activation. Nature 407:
642-645, 2000
52. Martinez-Delgado B, Melendez B, Cuadros
M, et al: Frequent inactivation of the p73 gene
by abnormal methylation or LOH in non-Hodg-
kin’s lymphomas. Int J Cancer 102:15-19, 2002
53. Siu LL, Chan JK, Wong KF, et al: Aberrant
promoter CpG methylation as a molecular
marker for disease monitoring in natural killer
cell lymphomas. Br J Haematol 122:70-77, 2003
54. Issa JP, Garcia-Manero G, Giles FJ, et al:
Phase 1 study of low-dose prolonged exposure
schedules of the hypomethylating agent 5-aza-
2#-deoxycytidine (decitabine) in hematopoietic
malignancies. Blood 103:1635-1640, 2004
55. Cheng JC, Yoo CB, Weisenberger DJ,
et al: Preferential response of cancer cells to
zebularine. Cancer Cell 6:151-158, 2004
56. Davis AJ, Gelmon KA, Siu LL, et al: Phase I
and pharmacologic study of the human DNA
methyltransferase antisense oligodeoxynucleo-
tide MG98 given as a 21-day continuous infusion
every 4 weeks. Invest New Drugs 21:85-97,
2003
11
92