ARTICLES
Articles
Trial of Atorvastatin in Rheumatoid Arthritis (TARA):
double-blind, randomised placebo-controlled trial
David W McCarey, Iain B McInnes, Rajan Madhok, Rosie Hampson, Olga Scherbakova, Ian Ford, Hilary A Capell, Naveed Sattar
Summary
Background Rheumatoid arthritis is characterised by
inflammatory synovitis, articular destruction, and
accelerated atherogenesis. HMG-CoA (3-hydroxy-3-
methylglutarylcoenzyme A) reductase inhibitors (statins)
mediate clinically significant vascular risk reduction in
patients without inflammatory disease and might have
immunomodulatory function. We postulated that statins
might reduce inflammatory factors in rheumatoid arthritis
and modify surrogates for vascular risk.
Methods 116 patients with rheumatoid arthritis were
randomised in a double-blind placebo-controlled trial to
receive 40 mg atorvastatin or placebo as an adjunct to
existing disease-modifying antirheumatic drug therapy.
Patients were followed up over 6 months and disease
activity variables and circulating vascular risk factors were
measured. Coprimary outcomes were change in disease
activity score (DAS28) and proportion meeting EULAR
(European League Against Rheumatism) response criteria.
Analysis was by intention to treat.
Findings At 6 months, DAS28 improved significantly on
atorvastatin (–0·5, 95% CI –0·75 to –0·25) compared with
placebo (0·03, –0·23 to 0·28; difference between groups
–0·52, 95% CI –0·87 to –0·17, p=0·004). DAS28 EULAR
response was achieved in 18 of 58 (31%) patients
allocated atorvastatin compared with six of 58 (10%)
allocated placebo (odds ratio 3·9, 95% CI 1·42–10·72,
p=0·006). C-reactive protein and erythrocyte sedimentation
rate declined by 50% and 28%, respectively, relative to
placebo (p<0·0001, p=0·005, respectively). Swollen joint
count also fell (–2·69 vs –0·53; mean difference –2·16,
95% CI –3·67 to –0·64, p=0·0058). Adverse events
occurred with similar frequency in patients allocated
atorvastatin and placebo.
Interpretation These data show that statins can mediate
modest but clinically apparent anti-inflammatory effects
with modification of vascular risk factors in the context of
high-grade autoimmune inflammation.
Lancet 2004; 363: 2015–21
See Commentary page 2011
Centre for Rheumatic Diseases (D W McCarey MRCP,
Prof I B McInnes FRCP, R Madhok MD, R Hampson BSc, H A Capell MD)
and Department of Vascular Biochemistry (N Sattar MD), University
of Glasgow, Glasgow Royal Infirmary, Glasgow, UK; and Robertson
Centre for Biostatistics, University of Glasgow, Glasgow
(O Scherbakova PhD, I Ford PhD)
Correspondence to: Prof Iain McInnes, Centre for Rheumatic
Diseases, 3rd Floor Queen Elizabeth Building, Glasgow Royal
Infirmary, Glasgow G31 2ER, UK
(e-mail: i.b.mcinnes@clinmed.gla.ac.uk)
THE LANCET • Vol 363 • June 19, 2004 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Introduction
HMG-CoA (3-hydroxy-3-methylglutaryl coenzyme A)
reductase inhibitors (statins) reduce cardiovascular
morbidity and mortality.1–3 Although statins work in
part via lipid modulation, findings of studies indicate
they have broader properties, including alteration of
inflammatory pathways.4 Ex-vivo activities of statins
include suppression of adhesion molecule expression,
leucocyte cytokine release, MHC class II expression,
lymphocyte and macrophage cognate interactions,
and effects on reactive oxygen and nitrogen intermediate
production.5 Statins also modify apoptosis in smooth
muscle and endothelial cells leading to altered
vascular function and neovascularisation.5 These
properties offer the potential to modify chronic
inflammatory disease states with drugs that show
minimal toxic effects in both the short and long term.
However, as yet, no clinical data clearly show that
statins modulate autoimmune disease activity or indeed
modify vascular risk factors in the context of high-grade
inflammation.
Rheumatoid arthritis is a chronic inflammatory
arthropathy associated with rapid onset of clinically
significant functional impairment. It is also associated
with accelerated vascular risk with attendant early
mortality (pooled standardised mortality ratio 1·7)
and excess morbidity.6 We have proposed that
cytokine-mediated inflammatory pathways can in part
account for this increased vascular risk via accentuation
of both classic (lipids) and novel (endothelial function
or insulin resistance) pathways.7 Furthermore,
statins suppress articular inflammation in vivo in
collagen-induced arthritis in mice.8 Moreover, in-vitro
cytokine release by rheumatoid arthritis synovial
mononuclear cells and by synovial fibroblasts was
also reduced by statins.8 Similar anti-inflammatory
effects have been reported in neurological models
of inflammation.9 Statins therefore have a plausible
bioactivity profile in vitro and in vivo that makes
them possible therapeutic agents in rheumatoid
arthritis to target both vascular risk and synovial
inflammation. We therefore undertook a randomised
placebo-controlled study to test the clinical efficacy
and laboratory effects of these drugs in rheumatoid
arthritis.
Patients and methods
Patients
We screened patients attending Glasgow Royal
Infirmary and recruited those who gave informed
consent for a study approved by North Glasgow
University NHS Trust ethics committee. Recruitment
was from September, 2001, through November, 2002.
Participants were 18–80 years of age, met 1987
American College of Rheumatology (ACR) criteria
(no limit on disease duration), and had active
2015
ARTICLES

inflammatory rheumatoid arthritis despite ongoing Atorvastatin Placebo

disease-modifying antirheumatic drug (DMARD) (n=58) (n=58)
therapy. We defined disease activity as six swollen Demographics
joints plus two of: 1) six tender joints; 2) early Age (years) 55·5 (11·8) 56·5 (10·0)
morning stiffness for more than 30 min; and 3) Women 51 (88%) 49 (84%)
erythrocyte sedimentation rate greater than 28 mm/h. Rheumatoid factor positive 52 (90%) 50 (86%)
Disease duration (years)
Exclusion criteria included inability to give informed <10 24 (41%) 25 (43%)
consent; diabetes; familial hypercholesterolaemia; 10–19 17 (29%) 19 (33%)
risk of coronary heart disease event of more 20–29 11 (19%) 9 (16%)
than 3% per year (joint British guidelines, British 30–41 6 (9%) 5 (7%)
National Formulary); current lipid-lowering Smoker 20 (34%) 16 (28%)
therapy; previous adverse reaction to statins; Clinical outcome measures
oral prednisolone greater than 10 mg/day; and Disease activity score 5·78 (1·10) 5·88 (0·97)
Erythrocyte sedimentation rate* (mm/h) 19·11 (2·26) 19·89 (2·36)
clinically significant renal disease or aspartate amino-
C-reactive protein* (mg/L) 11·60 (2·95) 13·18 (3·62)
transferase, alanine aminotransferase, or creatine Tender joint count 13·72 (8·14) 13·84 (8·61)
kinase more than twice the upper limit of laboratory Early morning stiffness* 65·36 (3·32) 71·52 (3·52)
normal range. Visual analogue score 58·62 (18·44) 66·33 (13·13)
Swollen joint count 12·19 (4·41) 11·29 (3·70)
Procedures Patient global assessment 58·67 (16·75) 67·44 (14·64)
Health assessment questionnaire 1·84 (0·56) 1·89 (0·56)
We recruited patients with active rheumatoid
arthritis as defined above if they had been on an Lipid concentrations
Cholesterol (mmol/L) 5·08 (0·81) 5·13 (0·99)
adequate dose of DMARD therapy for at least
Triglyceride (mmol/L) 1·34 (0·61) 1·39 (0·66)
3 months. Patients remained on all DMARDs, non- LDL-cholesterol (mmol/L) 3·11 (0·76) 3·36 (0·89)
steroidal anti-inflammatory drugs, and drugs for HDL-cholesterol (mmol/L) 1·36 (0·47) 1·32 (1·31)
comorbid conditions, and we asked them to maintain Haemostatic and endothelial factors
their DMARD dose from study entry. We allowed Fibrinogen (g/L) 4·45 (1·00) 4·58 (1·24)
and recorded intramuscular or intra-articular Plasma viscosity (mPa/s) 1·33 (0·10) 1·35 (0·11)
corticosteroid injection during the study, but this Von Willebrand factor (IU/dL) 150·55 (52·95) 166·59 (60·25)
procedure was forbidden within 4 weeks of clinical Intercellular adhesion molecule 1* 314·19 (1·43) 323·76 (1·45)
(ng/mL)
assessment. Interleukin 6* (pg/mL) 14·67 (3·16) 13·42 (4·05)
Data are mean (SD) or number of patients (%) unless otherwise indicated.
*Data are geometric means (SD); SDs calculated from log(natural)-transformed
177 assessed for eligibility distribution.
Table 1: Baseline characteristics
61 excluded
26 refused to participate
35 exclusion criteria
Patients were randomly allocated either atorvastatin
40 mg or placebo. Computerised randomisation
was done off-site (by independent study administrator)
116 randomised and both patients and doctors were unaware
of drug allocation. Atorvastatin (40 mg) and
identical placebo tablets were provided by Pfizer (NJ,
USA).
Our analysis was by intention to treat; patients
who did not complete 6 months of active treatment
58 allocated atorvastatin 58 allocated placebo or placebo were categorised as non-responders. The
58 received atorvastatin 58 received placebo primary outcome was change in DAS28, a validated
composite disease activity score,10,11 and we used
European League Against Rheumatism (EULAR)
criteria to calculate response in DAS28 (details at
1 lost to follow-up 2 lost to follow-up
http://www.das-score.nl). These criteria consider both
2 did not continue 2 did not continue
extent of DAS change and the final value achieved and
intervention intervention
2 adverse event§
have been validated against ACR criteria in clinical
1 protocol violation*
1 adverse event† 9 withdrew at 3-month trials.12 Components of DAS28 are erythrocyte
2 withdrew at 3-month assessment sedimentation rate, patient-assessed global score,
assessment 7 inefficacy visual analogue score (pain; 0–100), and swollen and
2 adverse event‡ 2 did not attend final tender joint counts (both 0–28). This score was
assessment calculated as previously described.11 Additional variables
included early morning stiffness, visual analogue
score pain score, and modified health assessment
questionnaire. Prespecified secondary endpoints
58 analysed for primary 58 analysed for primary
endpoint
included change in erythrocyte sedimentation rate;
endpoint
C-reactive protein; plasma viscosity, lipids, and
lipoproteins; thrombotic variables, fibrinogen, and von
Trial profile Willebrand factor; and the surrogate marker of
*Pregnant. †Creatine kinase raised at screening, entered endothelial activation, soluble intercellular adhesion
study but withdrew at doctor’s and patient’s preference at
week 2, creatine kinase unchanged. ‡Dry eyes and exacerbation
molecule 1.
of irritable bowel symptoms. §Dry eyes and mouth and gastrointestinal After screening, study visits were scheduled for 0, 3,
upset. and 6 months. The same researcher (RH), who was

2016 THE LANCET • Vol 363 • June 19, 2004 • www.thelancet.com

For personal use. Only reproduce with permission from The Lancet Publishing Group.
 ARTICLES

Atorvastatin Placebo Willebrand factor (Dako, Copenhagen, Denmark),

(n=58) (n=58) interleukin 6, and intercellular adhesion molecule 1
DMARD therapy (R&D systems, Oxon, UK) by ELISA techniques.
Methotrexate* 29 (50%) 15 (26%)
Sulfasalazine 35 (60%) 37 (64%) Statistical analysis
Hydroxychloroquine 7 (12%) 13 (22%) Assuming a 10% reduction from baseline in mean
Leflunomide 2 (3%) 5 (9%)
Gold† 4 (7%) 5 (9%)
DAS28 score on atorvastatin compared with no change
Penicillamine 1 (2%) 0 on placebo (equivalent to a 0·6 unit mean difference)
Minocycline 1 (2%) 0 and an underlying SD for the changes from baseline
Corticosteroid therapy of 1·0 unit, we calculated that 54 patients were
Prednisolone (oral) 2 (3%) 0 needed per group to provide 80% power at the 5%
Intra-articular triamcinolone 12 (21%) 17 (29%) significance level.12 Anticipating about 10% dropout,
Total dose (mg) 530 760 we aimed to enrol 59 patients per group. Baseline data
Intramuscular triamcinolone 6 (10%) 11 (19%) are summarised as mean (SD) for continuous variables
Total dose (mg) 640 880
and number of patients (%) for categorical variables.
Data are number of patients (%). 20 patients in the atorvastatin group and Distributions of C-reactive protein, intercellular
15 in the placebo group received combination DMARD therapy.
*More patients in the atorvastatin group received methotrexate (p=0·0074); no adhesion molecule 1, interleukin 6, erythrocyte
other significant differences between groups. †Intramuscular sodium sedimentation rate, and early morning stiffness were
aurothiomalate. markedly skewed and hence geometric means are given
Table 2: Concomitant DMARD therapy and corticosteroid as summary statistics.
administration according to study drug assignment Coprimary analyses were change in DAS28 and
EULAR DAS28 binary response. The EULAR
DAS28 binary response (moderate or good vs otherwise)
unaware of allocations, measured all clinical variables was analysed on an intention-to-treat basis, as
at study entry and at 3 and 6 months. We recorded described above. We analysed results with logistic
adverse events throughout the study and by specific regression to provide the estimated odds ratio for the
request at assessment timepoints. We did haematology treatment effect, the associated 95% CI, and the
and biochemical screening for liver function, creatine p value from the Wald statistic. To compare the two
kinase, and renal function at baseline and 3 and groups of patients with respect to the proportion
6 months. attending the 6-month visit, we measured the ␹2 statistic
We prepared plasma and serum samples immediately in those patients completing the study to 6 months.
after venepuncture. A sample was assayed straightaway All other analyses, including change in DAS28,
(lipids) and the remainder was stored at –70°C were done on the untransformed changes from
until assay. C-reactive protein was measured by baseline to 6 months in the outcome variable. We
fluorescence polarisation immunoassay with a used paired t tests and associated 95% CIs to examine
TDX analyser (Abbott Laboratories, Abbott Park, IL, within-group differences and two-sample t tests and
USA). Concentrations less than 6 mg/L were associated 95% CIs to compare changes between
subsequently measured by a high sensitivity assay groups. ANCOVA was used for adjustment for
with a sensitive double-antibody sandwich ELISA potential confounders. The assumptions of normality
with rabbit antihuman C-reactive protein and needed for these analyses seemed to be approximately
peroxidase conjugated rabbit antihuman C-reactive valid. We assessed associations between variables
protein, as described previously.13 We analysed plasma with Spearman’s rank-correlation coefficient. For
viscosity at 37°C in a Coulter-Harkness capillary change in laboratory variables, we accounted
viscometer (Beckman Coulter, High Wycombe, UK). for dropouts by carrying their baseline value forward
We measured fibrinogen by the Clauss assay and von to 6 months—ie, no change assumed.

Atorvastatin (n=58) Placebo (n=58) p

Primary outcome measure
Disease activity score –0·50 (–0·75 to –0·25) 0·03 (–0·23 to 0·28) 0·004
Secondary outcome measures
Erythrocyte sedimentation rate (mm/h) –5·03 (–8·4 to –1·67) 1·91 (–1·61 to 5·44) 0·005
C-reactive protein (log mg/L) –0·46 (–0·64 to –0·28) 0·12 (–0·09 to 0·34) <0·0001
Clinical measures
Tender joint count –1·21 (–3·28 to 0·86) 0·38 (–1·16 to 1·92) 0·22
Early morning stiffness (log) –0·47 (–0·82 to –0·12) –0·13 (–0·47 to 0·22) 0·17
Visual analogue score –5·07 (–10·5 to 0·36) 1·97 (–3·15 to 7·08) 0·06
Swollen joint count –2·69 (–3·81 to –1·57) –0·53 (–1·59 to 0·52) 0·0058
Patient global assessment –4·14 (–7·80 to –0·48) 0·15 (–4·25 to 4·56) 0·14
Health assessment questionnaire 0·02 (–0·12 to 0·16) 0·04 (–0·04 to 0·13) 0·75
Lipids
Cholesterol (mmol/L) –1·48 (–1·73 to –1·23) –0·01 (–0·14 to 0·12) <0·0001
Triglyceride (mmol/L) –0·24 (–0·34 to –0·14) 0·07 (–0·04 to 0·18) <0·0001
LDL-cholesterol (mmol/L) –1·40 (–1·63 to –1·17) –0·07 (–0·23 to 0·10) <0·0001
HDL-cholesterol (mmol/L) 0·03 (–0·03 to 0·09) –0·04 (–0·10 to 0·02) 0·097
Other risk markers
Fibrinogen (g/L) –0·38 (–0·69 to –0·07) 0·00 (0·19 to –0·20) 0·041
Plasma viscosity (mPa/s) –0·05 (–0·06 to –0·03) –0·00 (–0·02 to 0·01) 0·0004
Von Willebrand factor (IU/dL) –8·5 (–20·6 to 3·58) –4·53 (–16·7 to 7·6) 0·64
Intercellular adhesion molecule 1 (ng/mL) –22·6 (–41·6 to –3·7) 2·37 (–18·2 to 22·9) 0·076
Interleukin 6 (pg/mL) –6·6 (–13·2 to 0·01) 3·84 (–2·85 to 10·5) 0·028
Table 3: Differences after 6 months of treatment

THE LANCET • Vol 363 • June 19, 2004 • www.thelancet.com 2017

For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
Methotrexate status at baseline
Not taking (n=29) p Taking (n=29)
DAS28 –0·40 (–0·74 to –0·06) 0·02
Erythrocyte sedimentation rate (mm/h) –4·17 (–6·88 to –1·47) 0·004
C-reactive protein (log mg/L) –0·40 (–0·67 to –0·13) 0·006
Table 4: Changes in key study variables in the atorvastatin group according to baseline methotrexate status
Role of the funding source
Pfizer provided an educational grant and the
atorvastatin and placebo tablets. Otherwise, all
sponsors had no role in study design, data collection,
data analysis, data interpretation, or writing of the
report.
Results
177 patients with rheumatoid arthritis were screened,
of whom 116 were randomised to receive either
40 mg atorvastatin or placebo (figure). The median
age of the study group was 56 years (IQR 47–66)
with median disease duration of 11·5 years
(5–20). Atorvastatin and placebo groups were
generally comparable, although visual analogue score
and patient global assessment were slightly lower
in the atorvastatin group at baseline (table 1). By
chance, more patients allocated atorvastatin received
methotrexate compared with placebo (p=0·007),
whereas other drugs were equally represented
(table 2).
Table 3 documents outcomes for each study
arm. DAS28 was significantly reduced in patients
who were allocated atorvastatin compared with
placebo (difference between groups –0·52, 95% CI
–0·87 to –0·17; p=0·004). Adjustment for methotrexate
use did not change the significance of observed
reduction in this primary outcome measure (p=0·007).
Moderate or good DAS28 responses (EULAR
criteria) were achieved in 18 of 58 (31%) patients
allocated atorvastatin compared with six of 58
(10%) given placebo at 6 months (odds ratio for
DAS28 response 3·9, 95% CI 1·42–10·72, p=0·006).
More patients allocated atorvastatin completed the
study to 6 months compared with those on placebo
(53 vs 45, p=0·04; figure), further indicating
the beneficial effect of statin use and tolerability of
therapy.
Acute-phase reactants improved significantly in
patients allocated atorvastatin versus placebo
(table 3). At 6 months, C-reactive protein (including
detection by the high sensitivity assay) and erythrocyte
sedimentation rate declined by 50% and 28%,
respectively, relative to placebo. After adjustment (by
ANCOVA) for potential confounders that were
unbalanced at baseline, namely methotrexate use,
patient global assessment, and visual analogue pain
score, reduction in C-reactive protein and erythrocyte
sedimentation rate remained significant (p=0·0002 and
p=0·003, respectively). When we compared change in
disease activity score, erythrocyte sedimentation rate,
and C-reactive protein in patients allocated atorvastatin
and who were taking or not taking methotrexate we
noted that both groups showed similar significant
responses (table 4). A similar result was recorded
when data for individuals taking or not taking
sulfasalazine were compared (data not shown). No
difference in response was seen in individuals allocated
placebo and taking methotrexate compared with
those not taking this drug, suggesting that a delayed
2018
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Difference p
p
–0·59 (–0·97 to –0·21) 0·004 0·19 (–0·31 to 0·69) 0·46
–5·90 (–12·28 to 0·49) 0·069 1·72 (–5·06 to 8·51) 0·61
–0·52 (–0·77 to –0·27) 0·0002 0·12 (–0·24 to 0·49) 0·51
response to methotrexate was unlikely to account for our
findings.
Several individual disease activity variables improved
in patients given atorvastatin (table 3). In particular, the
swollen joint count declined in individuals allocated
atorvastatin (–2·69 joints) compared with placebo
(–0·53 joints; mean difference –2·16, 95% CI –3·67 to
–0·64, p=0·0058), suggesting that the change in disease
activity score indicated not only a laboratory change but
also a true articular response.
Atorvastatin was well tolerated in the study
population. Events leading to withdrawal are shown
in the figure. Adverse events arose with similar
frequency in patients allocated atorvastatin and placebo.
In particular, no significant liver function or muscle
abnormality was detected in those given atorvastatin.
Frequency of corticosteroid administration is shown
in table 2. More patients in the placebo group than in
the atorvastatin group received intra-articular or
intramuscular administration of triamcinolone. No
patient received parenteral steroid within 4 weeks of
DAS28 assessment. Two patients were on oral
prednisolone at baseline—neither were clinical
responders (table 2).
Significant reductions in total cholesterol, LDL-
cholesterol, triglyceride, and VLDL-cholesterol
concentrations were recorded in atorvastatin versus
placebo groups (table 3). HDL-cholesterol was
slightly raised in atorvastatin recipients. By intention
to treat, plasma viscosity, fibrinogen, soluble
intercellular adhesion molecule 1, and interleukin 6
concentrations all declined significantly in the
atorvastatin group but were unchanged in the
placebo group. Reductions in plasma viscosity,
fibrinogen, and interleukin 6 remained significant when
change in patients allocated atorvastatin was compared
with that in placebo recipients (table 3).
We examined correlations with change over time in
patients allocated atorvastatin. When change over
6 months was examined in these individuals (n=53),
C-reactive protein reduction correlated with reduction
in interleukin 6 (r=0·31, p=0·03) and fibrinogen
(r=0·51, p=0·0009). By contrast, reduction in LDL-
cholesterol did not correlate with alteration in any other
study variable including disease activity score,
erythrocyte sedimentation rate, C-reactive protein, or
interleukin 6.
Discussion
We have shown marked suppression with atorvastatin
of acute-phase variables and a significant reduction
in swollen joint count in patients with rheumatoid
arthritis presenting with active disease despite
existing DMARD therapy. Significantly more patients
in the atorvastatin group remained on treatment
to completion. Although the magnitude of change
is modest, the significant reduction in DAS28
provides proof of concept that pathways targeted
by statins offer therapeutic opportunity in inflammatory
disease. However, no effect was seen on other
THE LANCET • Vol 363 • June 19, 2004 • www.thelancet.com

clinical measures of disease activity, particularly the
health assessment questionnaire, and larger studies
will be needed to establish the extent to which
these changes will prove meaningful in the clinical
context. In parallel, our data clearly show that statins
are effective in modification of classic and novel risk
factors of vascular disease despite the presence
of high-grade inflammation. Thus, potential biphasic
benefit in reduced inflammation and modified vascular
risk may accrue with use of statins in rheumatoid
arthritis.
Many data indicate effects for statins in innate
immune response, manifest on endothelial activation,5
macrophage, natural killer cell, and neutrophil
effector function.14 Similar effects on acquired immune
responses via suppression of antigen presentation15
and T-cell polarisation have been shown in vitro
and in vivo.8,16–20 However, there is a striking
absence of controlled human trial data to confirm
these observations in vivo. Synovial inflammation
in rheumatoid arthritis is characterised by activated
components of both innate and acquired immune
responses. Thus, postulated effects for statins might
reasonably operate within the synovial membrane.
Findings of our in-vitro study8 showed suppression of
synovial T cell and fibroblast-like synovocyte
cytokine release, which lends support to this notion.
We also showed immune suppressive effects of
simvastatin in rodent collagen-induced arthritis
in vivo, which prompted us to undertake the current
study. Our data now suggest that modulation
of inflammation can be achieved by atorvastatin
in a proportion of patients with rheumatoid
arthritis, suggesting that some of the above pathways
may indeed be tractable to HMG-CoA reductase
inhibition.
We tested atorvastatin in a stringent true-to-life
setting by giving the drug to patients with active
rheumatoid arthritis despite established DMARD
therapy (mean DAS28 score 5·8) and prolonged
disease duration. Although study designs render
direct comparison difficult, the improvement seen
is of the same order as that seen on addition of
ciclosporin to methotrexate (median reduction in
swollen joint count 3·8)21 or minocycline to established
DMARD therapy (median reduction in swollen
joint count 0·8).22 We recognise that the magnitude
of effect is modest, in particular when compared
with novel cytokine-targeting biological agents.23,24
Nevertheless, a clinically apparent anti-inflammatory
effect is remarkable given that atorvastatin was
developed mainly on the basis of lipid-lowering
properties. As such, we believe that our study should
provide impetus for development of statin analogues
on the basis of their anti-inflammatory profile. The
reduction in joint swelling distinct from tender
joint count or visual analogue scale for pain raises
the possibility of effects of statins on inflammation
distinct from pain pathways.
We recognise important limitations in our study.
The design chosen offers advantages in facilitating
single-centre recruitment over a short period.
However, the heterogeneous background of DMARD
use allows the possibility of statin-DMARD interactions
that could confound outcomes. The multicentre
design needed to recruit active rheumatoid arthritis
patients to a strictly defined DMARD background
was not readily achievable without proof of concept
data, which we have now generated. Future studies
THE LANCET • Vol 363 • June 19, 2004 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
should, however, be designed to test statin activity in a
strictly defined DMARD context.
By chance, more patients on methotrexate entered
the atorvastatin arm of our study. Correction for
drug use did not alter the significance of the primary
outcomes. Moreover subgroup analysis suggested
that subgroups treated with methotrexate or
sulfasalazine were similar in terms of response
(table 4 and data not shown). The measures of
DAS28—namely patient global assessment and visual
analogue scale—that were lower in the atorvastatin
group by chance would probably bias against a
clinical response in this group.25 Thus, minor
discrepancies in baseline characteristics if anything
will bias against finding a treatment effect for
atorvastatin.
Other limitations should be considered. Our
study group is small, reflecting the single-centre design;
large high-powered studies will be important to confirm
the relevance of our clinical dataset to the wider
rheumatoid arthritis population. Direct effects
of statins on hepatic C-reactive protein synthesis
could exaggerate the impression of disease modification.
However, we used erythrocyte sedimentation rate
for our composite DAS28, and the noted reduction
in interleukin 6 argues against an effect mediated
only on the liver. Finally, we allowed steroid injection
within our protocol, reflecting real practice, but
forbade this within 28 days of DAS28 evaluation.
Importantly, intra-articular and intramuscular
steroid were used more frequently in the placebo-treated
group, rendering a significant positive confounding
effect on primary outcomes unlikely. Frequency of
oral steroid use was very low in our cohort, indicating
local practice, and as such is unlikely to have affected
our data.
Accelerated atherogenesis is a major cause of
morbidity and mortality in rheumatoid arthritis.
Although initial mean lipid values lay within the
reference range, atorvastatin substantially reduced
LDL-cholesterol and triglyceride concentrations,
despite the context of high-grade inflammation.
Since lipid lowering even within the so-called
normal range is of vascular benefit,2 this fact provides
a direct coronary heart disease-protective pathway
in rheumatoid arthritis for atorvastatin via traditional
risk factor modification. Equally in rheumatoid
arthritis, in which inflammation-driven vascular risk
has been proposed,6,7 anti-inflammatory mechanisms
with accompanying downstream improvements in
novel risk factor pathways can also operate for
atorvastatin. Thus, atorvastatin reduced C-reactive
protein, itself an established independent risk factor for
coronary heart disease.26–32 We also measured reduction
in serum interleukin 6, plasma viscosity, erythrocyte
sedimentation rate, soluble intercellular adhesion
molecule 1, and fibrinogen. Chronic elevation of
interleukin 6 might promote atherogenesis directly
through effects in the vessel wall and indirectly
via secondary effects on insulin resistance, for
example.33 Moreover, plasma viscosity and fibrinogen
and intercellular adhesion molecule 1 are markers
of hypercoagulability and endothelial activation,
respectively. Finally, atorvastatin could modulate lipid
moieties—particularly oxidised LDL—that in turn
regulate local inflammation.34
Most rheumatoid arthritis patients now receive a
combination of DMARDs as treatment emphasis
moves to disease amelioration and remission induction
2019
ARTICLES
together with tissue protection. A significant effect
on disease activity was noted in our study. Although
not indicative for first line use, this effect of atorvastatin
could prove beneficial in the context of DMARD
combination design, in which a statin offers both
vascular protective and adjunctive immune modulatory
potential. Whereas existing DMARD therapy might
partly reverse vascular risk,35 our findings suggest
that statins can work to reduce traditional and
novel vascular risk factors. Further studies are now
needed to establish what benefit is conferred by
inclusion of statins on long-term outcomes in
rheumatoid arthritis. Since findings indicate endothelial
dysfunction in even early rheumatoid arthritis,36
intervention at the outset deserves consideration.
Finally, statin-sensitive biochemical pathways offer
further opportunity for the generation of novel
disease-modifying drugs to treat chronic inflammatory
diseases.
Contributors
D W McCarey and I B McInnes contributed equally to this report.
I B McInnes (principal investigator), D W McCarey, R Madhok,
H A Capell, and N Sattar designed and coordinated the trial and
drafted the paper. D W McCarey, I B McInnes, R Madhok, and
H A Capell recruited study patients. D W McCarey and R Hampson
ran the study clinic, assessed the patients, and acquired data.
O Scherbakova, I Ford, and N Sattar analysed and interpreted
data and provided statistical support. All authors critically reviewed
the manuscript and approved the final version.
Conflict of interest statement
IBM and NS were recipients of an educational grant from Pfizer to fund
ex-vivo analyses in this report. None of the authors has a relationship
with Pfizer in terms of employment, consultancies, stock ownership,
paid expert testimony, or patent applications. DWM, IBM, HAC, and
NS have been recipients of travel grants or honoraria from Pfizer within
3 years of beginning the work submitted. RM, RH, OS, and IF have no
known conflict of interest.
Acknowledgments
We thank the Chief Scientist Office, Scotland, and the
Arthritis Research Campaign (UK) for research grant support; and
Lynne Cherry for excellent laboratory and technical assistance.
References
1 Shepherd JS, Cobbe SM, Ford I, et al. Prevention of coronary
heart disease with pravastatin in men with hypercholesterolemia.
N Engl J Med 1995; 333: 1301–07.
2 Heart Protection Study Collaborative Group. MRC/BHF Heart
Protection Study of cholesterol lowering with simvastatin in 20 536
high-risk individuals: a randomised placebo-controlled trial. Lancet
2002; 360: 7–22.
3 Sever PS, Dahlöf B, Poulter NR, et al. Prevention of coronary and
stroke events with atorvastatin in hypertensive patients who have
average or lower-than-average cholesterol concentrations, in the
Anglo-Scandinavian Cardiac Outcomes Trial-Lipid Lowering Arm
(ASCOT-LLA): a multicentre randomised controlled trial. Lancet
2003; 361: 1149–58.
4 Blake GJ, Ridker PM. Are statins anti-inflammatory?
Curr Control Trials Cardiovasc Med 2000; 1: 161–65.
5 Palinski W, Napoli C. Unraveling pleiotropic effects of
statins on plaque rupture. Arterioscler Thromb Vasc Biol 2002; 22:
1745–50.
6 Van Doornum S, McColl G, Wicks IP. Accelerated atherosclerosis:
an extraarticular feature of rheumatoid arthritis? Arthritis Rheum
2002; 46: 862–73.
7 Sattar N, McCarey DW, Capell H, McInnes IB. Explaining how
“high-grade” systemic inflammation accelerates vascular risk in
rheumatoid arthritis. Circulation 2003; 108: 2957–63.
8 Leung BP, Sattar N, Crilly A, et al. A novel anti-inflammatory role
for simvastatin in inflammatory arthritis. J Immunol 2003; 170:
1524–30.
9 Youssef S, Stuve O, Patarroyo JC, et al. The HMG-CoA reductase
inhibitor, atorvastatin, promotes a Th2 bias and reverses paralysis in
2020
For personal use. Only reproduce with permission from The Lancet Publishing Group.
central nervous system autoimmune disease. Nature 2002; 420:
78–84.
10 van der Heijde DMFM, van Riel PLCM, Leeuwen MA,
van ‘t Hof MA, van Rijswijk MH, van de Putte LBA. Prognostic
factors for radiographic damage and physical disability in early
rheumatoid arthritis: a prospective follow-up study of 147 patients.
Br J Rheumatol 1992; 31: 519–25.
11 Prevoo ML, van ’t Hof MA, Kuper HH, van Leeuwen MA,
van de Putte LB, van Riel PL. Modified disease activity scores
that include twenty-eight-joint counts: development and validation
in a prospective longitudinal study of patients with rheumatoid
arthritis. Arthritis Rheum 1995; 38: 44–48.
12 van Gestel AM, Anderson JJ, van Riel PL, et al. ACR and EULAR
improvement criteria have comparable validity in rheumatoid
arthritis trials. J Rheumatol 1999; 26: 705–11.
13 Packard CJ, O’Reilly DS, Caslake MJ, et al. Lipoprotein-associated
phospholipase A2 as an independent predictor of coronary heart
disease. N Engl J Med 2000; 343: 1148–55.
14 Choi M, Rolle S, Rane M, Haller H, Luft FC, Kettritz R.
Extracellular signal-regulated kinase inhibition by statins
inhibits neutrophil activation by ANCA. Kidney Int 2003; 63:
96–106.
15 Kwak B, Mulhaupt F, Myit S, Mach F. Statins as a newly
recognized type of immunomodulator. Nat Med 2000; 6:
1399–402.
16 Weitz-Schmidt G, Welzenbach K, Brinkmann V, et al.
Statins selectively inhibit leukocyte function antigen-1 by
binding to a novel regulatory integrin site. Nat Med 2001; 7:
687–92.
17 Hakamada-Taguchi R, Uehara Y, Kuribayashi K, et al. Inhibition of
hydroxymethylglutaryl-coenzyme a reductase reduces Th1
development and promotes Th2 development. Circ Res 2003; 93:
948–56.
18 Sparrow CP, Burton CA, Hernandez M, et al. Simvastatin has
anti-inflammatory and antiatherosclerotic activities independent
of plasma cholesterol lowering. Arterioscler Thromb Vasc Biol 2001;
21: 115–21.
19 Yokota N, O’Donnell M, Daniels F, et al. Protective effect of
HMG-CoA reductase inhibitor on experimental renal ischemia-
reperfusion injury. Am J Nephrol 2003; 23: 13–17.
20 Shimizu K, Aikawa M, Takayama K, Libby P, Mitchell RN. Direct
anti-inflammatory mechanisms contribute to attenuation of
experimental allograft arteriosclerosis by statins. Circulation 2003;
108: 2113–20.
21 Tugwell P, Pincus T, Yocum D, et al. Combination therapy with
cyclosporine and methotrexate in severe rheumatoid arthritis.
N Engl J Med 1995; 333: 137–41.
22 Kloppenburg M, Breedveld FC, Terwiel JP, Mallee C,
Dijkmans BA. Minocycline in active rheumatoid arthritis: a
double-blind, placebo-controlled trial. Arthritis Rheum 1994; 37:
629–36.
23 Klareskog L, van der Heijde D, de Jager JP, et al. Therapeutic
effect of the combination of etanercept and methotrexate
compared with each treatment alone in patients with rheumatoid
arthritis: double-blind randomised controlled trial. Lancet 2004;
363: 675–81.
24 Feldmann M, Maini RN. Anti-TNF alpha therapy of rheumatoid
arthritis: what have we learned? Annu Rev Immunol 2001; 19:
163–96.
25 Anderson JJ, Wells G, Verhoeven AC, Felson DT. Factors
predicting response to treatment in rheumatoid arthritis:
the importance of disease duration. Arthritis Rheum 2000;
43: 22–29.
26 Ridker PM, Rifai N, Rose L, Buring JE, Cook NR. Comparison of
C-reactive protein and low-density lipoprotein cholesterol levels in
the prediction of first cardiovascular events. N Engl J Med 2002;
347: 1557–65.
27 Kobayashi S, Inoue N, Ohashi Y, et al. Interaction of oxidative
stress and inflammatory response in coronary plaque instability:
important role of C-reactive protein. Arterioscler Thromb Vasc Biol
2003; 23: 1398–404.
28 Devaraj S, Xu DY, Jialal I. C-reactive protein increases
plasminogen activator inhibitor-1 expression and activity in
human aortic endothelial cells: implications for the metabolic
syndrome and atherothrombosis. Circulation 2003; 107:
398–404.
29 Venugopal SK, Devaraj S, Yuhanna I, Shaul P, Jialal I.
Demonstration that C-reactive protein decreases eNOS expression
and bioactivity in human aortic endothelial cells. Circulation 2002;
106: 1439–41.
30 Zwaka TP, Hombach V, Torzewski J. C-reactive protein-mediated
low density lipoprotein uptake by macrophages: implications for
atherosclerosis. Circulation 2001; 103: 1194–97.
THE LANCET • Vol 363 • June 19, 2004 • www.thelancet.com

31 Marz W, Winkler K, Nauck M, Bohm BO, Winkelmann BR.
Effects of statins on C-reactive protein and interleukin-6 (the
Ludwigshafen Risk and Cardiovascular Health study). Am J Cardiol
2003; 92: 305–08.
32 McCarty MF. Reduction of serum C-reactive protein by statin
therapy may reflect decreased isoprenylation of Rac-1, a mediator of
the IL-6 signal transduction pathway. Med Hypotheses 2003; 60:
634–39.
33 Yudkin JS, Kumari M, Humphries SE, Mohamed-Ali V.
Inflammation, obesity, stress and coronary heart disease: is
Uses of error
Do not dismiss intense pain
Federico Balagué
Regardless of experience and expertise, we all make
errors. In 30 years of medical practice, I have made
errors, possibly more than I realise. Some were due to
limitations of my scientific knowledge or insufficient
clinical experience. Communication problems and
misunderstandings may also have contributed. I hope
that each has made me a better doctor, and that sharing
them will help to prevent others from making the same
mistakes.
Early in my postgraduate training I was called to
examine a patient in his sixties, who had developed
sudden pain in his thigh after turning over in bed.
Despite the fact that he reported very intense pain (at
that time, visual analogue scales were not part of my
routine daily practice), I was convinced that such a
mechanism of onset could not explain any major organic
lesion. Strong in my beliefs, I did not order a radiograph
and simply prescribed analgesics. Throughout the night,
I was called by the ward to request increasingly strong
analgesia. Early in the morning, the gentleman very
calmly and sincerely told me that, if he had had his gun
in his night-table drawer, he would have shot himself
during the night. A radiograph taken later that morning
revealed a femoral fracture at the site of a bone
metastasis, which required immediate surgery.
I have never forgotten that man and ever since I have
tried to give absolute priority to what a patient says or
Department of Rheumatology, Physical Medicine and Rehabilitation, Hôpital Cantonal, 1708 Fribourg, Switzerland (F Balagué MD)
THE LANCET • Vol 363 • June 19, 2004 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
interleukin-6 the link? Atherosclerosis 2000; 148: 209–14.
34 Palinski W, Tsimikas S. Immunomodulatory effects of statins:
mechanisms and potential impact on arteriosclerosis.
J Am Soc Nephrol 2002; 13: 1673–81.
35 Choi HK, Hernan MA, Seeger JD, et al. Methotrexate and mortality
in patients with rheumatoid arthritis: a prospective study. Lancet
2002; 359: 1173–77.
36 Vaudo G, Marchesi S, Gerli R, et al. Endothelial dysfunction in
young patients with rheumatoid arthritis and low disease activity.
Ann Rheum Dis 2004; 63: 31–35.
feels over any biomechanical hypothesis of mine.
However, many years later, this experience did not
prevent us—in this case I was supervising a young
resident—from recommending an intensive physical
rehabilitation programme for a 41-year-old man
complaining of disabling lower back pain. This patient
was unable to complete the 3 week, demanding
programme and dropped out. A few weeks later, a
colonic tumour with hepatic metastases was diagnosed.
The patient died within 6 months despite aggressive
therapy. As a rheumatologist with a special interest in
spinal disorders, I know that most back patients fall into
the category of non-specific lower back pain, and that
only about 1% involve specific inflammatory diseases,
including neoplasm. Even now, I remain afraid to miss a
life-threatening disease, which can happen despite our
best efforts to make correct diagnoses.
It can be very difficult to apply current
epidemiological information, classification criteria, or
therapeutic recommendations to an individual patient.
All this information may be perfectly useful from a
public health standpoint, but because today’s truth
might become tomorrow’s heresy, I think that physicians
should keep in mind that a priori our patients are always
right when they complain. It is our task to understand
the cause of their pain. This is one of the principles of the
art of medicine.
2021