Professional Documents
Culture Documents
ajog.org
OBSTETRICS
Cite this article as: Joneborg U, Eloranta S, Johansson ALV, et al. Hydatidiform mole and subsequent pregnancy outcome: a population-based cohort study. Am J Obstet
Gynecol 2014;211:681.e1-7.
M ATERIALS
AND
M ETHODS
Data sources
Data were obtained from 3 nationwide
Swedish population-based registers. Record linkage between registers is made
possible by use of an individually unique
national registration number assigned
681.e1
Research
Obstetrics
ajog.org
Study population
A total of 3,730,789 births were identied in the MBR between 1973 and
2009. From these, we excluded multiple
births (n 90,128), children with
missing data on maternal country of
origin (n 169), and children born to a
woman with a childbirth and a diagnosis
of HM registered the same date (suggesting a twin molar pregnancy or a
third-trimester partial molar pregnancy)
(n 17, referring to 7 unique women).
To be able to perform a complete case
analysis, children with missing data on
small for gestational age (SGA) or large
for gestational age (LGA) were also
FIGURE
1973
1958
2009
Molar pregnancy
HM exposure in
Statistical analysis:
Child1(MGR)
Not
included
Child 2 (MBR)
Index birth
1 birth between
HM and index
birth
Child 3 (MBR)
Index birth
1 birth between
HM and index
birth
Exposure variable
Our exposure variable was maternal
history of HM prior to childbirth. Information on exposure was extracted
from the SCR, using the International
Classication of Diseases (ICD)-7: 173
and pathoanatomical diagnosis: 801
codes to identify all recorded cases of
HM. In this way, a total of 4940 cases
of HM were identied in the SCR
since 1958, 20 of which were identied
as a repeat mole.
By means of record linkage between
the MBR and the SCR, we found 3709
unique women with a rst diagnosis of
HM during the period that was included
in the analysis. Of these, 3071 women
had a diagnosis of HM prior to at least 1
of their childbirths, with a total of 5186
exposed births. The study population
was further stratied into a maternal
history of HM prior to the index pregnancy, dened as no birth between the
HM and the index pregnancy (n
2867), and a maternal history of HM
and at least 1 birth between the HM
and the index pregnancy (n 2319).
Information on births occurring prior
to 1973 was retrieved from the MGR.
Outcome variables
Outcomes of interest included adverse
maternal pregnancy outcomes (maternal
preeclampsia [PE]; ICD-8: 63703,
63704, 63709, 63799; ICD-9: 642E, 642F,
642H; ICD-10: O11, O14; maternal hypertension; ICD-8: 63701; ICD-9: 642D,
642X; ICD-10: O13, O16; placental
abruption; ICD-8: 6514; ICD-9: 641C;
ICD-10: O45 and premature rupture of
membranes [PROM]; ICD-8: 6610;
ICD-9: 658B, 658C; ICD-10: O42; and
adverse offspring outcomes [congenital
malformations; ICD-9: 740-759; ICD10: Q30-Q99, preterm birth (delivery
<37 gestational weeks), stillbirth,
neonatal mortality (child died <28 days
postpartum), SGA, and LGA]).23,24
Statistical analysis
An unconditional logistic regression
analysis was used to estimate the
Obstetrics
ajog.org
association between maternal history of
HM and different adverse maternal and
offspring outcomes. Exposure to HM
was included in the regression models
using a dichotomous categorization
(ie, never exposed to HM vs exposed to
HM prior to index delivery) as well as a
3-level categorization (ie, never exposed
to HM, HM prior to index delivery, HM
followed by at least 1 childbirth prior to
index delivery). Births with no maternal
history of HM were used as the reference
group irrespective of categorization of
HM exposure.
Odds ratios (ORs) with 95% condence intervals (CIs) were calculated and
robust SEs were used in the logistic
regression analysis to account for the
dependency structure in the data (ie,
that the same woman could contribute
with >1 delivery).25 The strength of
the association between potential confounding variables (maternal age at index birth, country of origin, education,
smoking at rst visit to antenatal care,
and body mass index [BMI]) and HM
exposure prior to the index birth was
assessed using multinomial regression in
which the 3-level representation of HM
was included as the outcome variable
(Table 1). The unconditional logistic
regression models were adjusted for
maternal age at delivery (<20 years, 2029 years, 30-39 years, and 40 years old)
and region of birth (Europe, Africa, Asia,
and America) (Tables 2 and 3). P values
from Wald tests were calculated in all
regression models and a 5% signicance
level was used.
All data sets were merged and dened
using the Statistical Analysis Software
version 9.2 (SAS Institute, Cary, NC) and
were then exported to Stata version 12.1
(StataCorp 2011, release 12; StataCorp
LP, College Station, TX).
The study was approved by the Research Ethics Committee at Karolinska
Institutet (Stockholm, Sweden).
Research
TABLE 1
Deliveries
HM prior to
index deliverya
HM and at least 1
childbirth prior to
index deliverya
Age at birth, y
<20
118,838
39
20-29
2,043,266
1440
872
30-39
1,387,749
1288
1336
72,559
100
106
3,402,706
2685
2189
40
P < .001
Maternal country of origin
Europe (including Russia)
Africa
44,663
25
23
137,946
124
83
37,097
33
24
512,207
383
394
10-13
1,738,707
1357
1114
More than13
1,306,811
1099
797
64,687
28
14
2,088,502
1747
1425
295,828
232
197
Missing
P < .001
Maternal smoking
No
1-9 cigarettes/day
10 cigarettes/day
164,775
117
106
1,073,307
771
591
134,012
104
56
19-24
1,321,924
1118
878
25-29
418,273
351
307
Missing
P .4592
Maternal BMI, kg/m2
<19
30
Missing
156,182
114
112
1,592,021
1180
966
3,622,412
2867
2319
P .0155
Total
R ESULTS
A total of 3,622,412 births between 1973
and 2009 were included in the analysis.
The distribution of HM exposure by
maternal characteristics is presented in
Table 1. A total of 0.14% of all births had
been exposed to a prior maternal molar
Note these numbers refer to deliveries, not unique women. Hence, women who have more than 1 exposed index delivery
contribute more than once to this table. The number of unique women with HM in this cohort was 3709. Of these, 3071 were
exposed to HM prior to at least 1 of their childbirths.
Joneborg. Hydatidiform mole and subsequent pregnancy outcome. Am J Obstet Gynecol 2014.
681.e3
Research
Obstetrics
ajog.org
TABLE 2
Outcome
Model 2a
No history
of HM
Previous history
of HM (ever)
HM and at least 1
childbirth prior to
index delivery
HM prior to
index delivery
Preeclampsia
OR (95% CI)
1.00 (reference)
0.75 (0.59e0.96)
P value
Number of events
0.79 (0.58e1.06)
.023
69,262
0.71 (0.50e1.02)
.074
77
44
33
Pregnancy hypertension
OR (95% CI)
1.00 (reference)
1.01 (0.74e1.38)
P value
Number of events
1.05 (0.70e1.57)
.947
27,577
0.96 (0.61e1.52)
.957
43
24
19
Placental abruption
OR (95% CI)
1.00 (reference)
1.10 (0.75e1.63)
P value
Number of events
1.05 (0.62e1.77)
.617
16,553
1.17 (0.68e2.02)
.841
27
14
13
PROM
OR (95% CI)
1.00 (reference)
0.88 (0.71e1.09)
P value
Number of events
Number of pregnancies
0.99 (0.76e1.30)
.247
0.74 (0.52e1.05)
.251
67,825
87
54
33
3,617,226
5186
2867
2319
CI, confidence interval; HM, hydatidiform mole; OR, odds ratio; PROM, premature rupture of membranes.
a
Both models were adjusted for maternal age at birth, maternal country of origin, and the correlation between siblings using a robust estimator of the SE.
Joneborg. Hydatidiform mole and subsequent pregnancy outcome. Am J Obstet Gynecol 2014.
C OMMENT
The results of this large nationwide
cohort study could not demonstrate an
association between a history of HM
and adverse maternal outcomes in subsequent pregnancies. Instead, women
exposed to a previous molar pregnancy
had a 25% lower risk of PE. However, we
found evidence of an increased risk of
LGA birth, stillbirth, and preterm birth
among women with a history of HM.
The observed increases in risk were,
although statistically signicant, small
and inconsistent across exposed groups
when stratied by the relationship between the molar exposure and the rank
order of subsequent index births. This
may indicate that the elevated risks of
adverse outcomes reect confounding
factors rather than the molar exposure.
The occurrence of repeat mole was low
Obstetrics
ajog.org
Research
TABLE 3
Outcome
Model 2a
No history
of HM
Previous history
of HM (ever)
HM and at least 1
childbirth prior to
index delivery
HM prior to
index delivery
1.00 (reference)
1.08 (0.92e1.27)
P value
Number of events
1.20 (0.98e1.46)
.332
110,283
0.94 (0.73e1.21)
.171
166
102
64
1.00 (reference)
1.19 (1.02e1.39)
P value
Number of events
1.06 (0.87e1.29)
.025
118,328
1.35 (1.10e1.67)
.019
215
103
112
Preterm birth
OR (95% CI)
1.00 (reference)
1.13 (1.00e1.27)
P value
Number of events
1.23 (1.06e1.43)
.059
184,171
0.99 (0.83e1.20)
.025
296
178
118
Congenital malformations
OR (95% CI)
1.00 (reference)
1.05 (0.91e1.21)
P value
Number of events
1.08 (0.91e1.31)
.473
136,911
1.01 (0.81e1.25)
.660
204
117
87
Neonatal death
OR (95% CI)
1.00 (reference)
0.88 (0.51e1.51)
P value
Number of events
0.97 (0.49e1.95)
.644
10,508
0.76 (0.32e1.83)
.828
13
0.56 (0.25e1.25)
1.82 (1.12e2.97)
Stillbirth
OR (95% CI)
1.00 (reference)
1.13 (0.74e1.71)
P value
Number of events
Number of pregnancies
.569
.021
13,283
22
16
3,617,226
5186
2867
2319
Both models were adjusted for maternal age at birth, maternal country of origin, and the correlation between siblings using a robust estimator of the SE.
Joneborg. Hydatidiform mole and subsequent pregnancy outcome. Am J Obstet Gynecol 2014.
681.e5
Research
Obstetrics
ajog.org
study population, allowing assessment
of rare adverse outcomes (PE, gestational
hypertension, PROM, placental abruption, stillbirth, neonatal mortality,
SGA, and LGA). The data at hand also
allowed the adjustment for known risk
factors for HM such as maternal age and
ethnicity.
Weaknesses included the previously
documented underreporting of HM to
the SCR,21,22 which may have diluted
possible differences in outcome between
the exposed group and the reference
group. Also, the absence of data on
postmolar GTN in the SCR might have
led to the inclusion of some women with
postmolar GTN treated with chemotherapy in the exposed cohort. For that
reason, it cannot be excluded that differences in adverse obstetric outcomes
reect chemotherapy related effects,
rather than the molar diagnosis. Because
complete and partial HM represent 2
different entities with possible different
effects on future pregnancies, an additional weakness was the lack of information on the subtypes of HM in the
SCR. In the case of selective underreporting of 1 of the 2 subtypes, the risk
of adverse outcomes might have been
over- or underestimated.
In conclusion, this large nationwide
cohort study found no association
between a history of HM and adverse
maternal outcomes in subsequent pregnancies. A slight but inconsistent
increased risk of LGA birth, stillbirth,
and preterm birth was observed, but for
the individual woman, the risk of adverse
offspring outcomes was low and barely
differed from that of the general population. The present ndings emphasize
that women treated for an HM can
expect normal future reproductive outcomes. Future studies should focus on
the possible association between levels
of perceived stress and premature birth
after a history of HM.
-
REFERENCES
1. Kim S. Epidemiology. In: Hancock BW,
Newlands ES, Berkowitz RS, eds. Gestational
trophoblastic disease. London: Chapman & Hall;
1997:27-42.
Obstetrics
ajog.org
Subsequent pregnancy outcome, including
repeat molar pregnancy. J Reprod Med 1998;43:
81-6.
18. Ludvigsson JF, Otterblad-Olausson P,
Pettersson BU, Ekbom A. The Swedish personal
identity number: possibilities and pitfalls in
healthcare and medical research. Eur J Epidemiol 2009;24:659-67.
19. Cnattingius S, Ericson A, Gunnarskog J,
Kallen B. A quality study of a medical birth registry. Scand J Soc Med 1990;18:143-8.
20. Barlow L, Westergren K, Holmberg L,
Talback M. The completeness of the Swedish
Cancer Register: a sample survey for year 1998.
Acta Oncol 2009;48:27-33.
21. Flam F, Rutqvist LE. Under-registration of
gestational trophoblastic disease in the Swedish
Cancer Registry. Eur J Epidemiol 1992;8:683-6.
22. Salehi S, Eloranta S, Johansson AL,
Bergstrom M, Lambe M. Reporting and incidence trends of hydatidiform mole in Sweden,
1973-2004. Acta Oncol 2011;50:367-72.
Research
681.e7