Fontographer 4.1


        

Genetic influences on premature parturition in an
Australian twin sample
Susan A Treloar1, George A Macones2, Laura E Mitchell3 and Nicholas G Martin1

1Queensland Institute of Medical Research and Joint Genetics Program, The University of Queensland, Brisbane,
Australia
2The University of Pennsylvania Health System
3The Children’s Hospital of Philadelphia, Philadelphia, PA, USA

We investigated possible genetic influences on women’s liability to preterm birth, using data from
a large sample of Australian female twin pairs. In a 1988–90 questionnaire survey, both members
of 905 parous twin pairs (579 monozygotic and 326 dizygotic) reported on whether deliveries had
been more than two weeks preterm. Tetrachoric twin pair correlations for first birth were
rMZ = 0.20 ± 0.11 and rDZ = –0.03 ± 0.14, and for any birth were rMZ = 0.30 ± 0.08 and
rDZ = 0.03 ± 0.11. Best-fitting models to data contained only additive genetic influences and
individual environmental effects. Heritability was 17% for preterm delivery in first pregnancy,
and 27% for preterm delivery in any pregnancy. In the former case, however, we could not reject
a model without genetic influences. Although our data did not allow for differentiation of the
varying aetiologies of premature parturition, results from this exploratory analysis suggest that
further investigation of genetic influences on specific reasons for preterm birth is warranted. Twin
Research (2000) 3, 80–82.

Keywords: Preterm birth, pregnancy outcome, twins, genetics

The risk of preterm birth has been found to be higher
for women who were themselves born prematurely,1

suggesting a repeating pattern across generations.
The relevance of genetic influences on prematurity
had been highlighted by an earlier study using Old
Order Amish records in Lancaster County, Pennsyl-
vania, which suggested that prematurity was mostly
related to the maternal genotype.2

In order to explore the possibility of genetic
influences on a woman’s liability to preterm births
we have examined self-reported, retrospective data
on premature birth provided by a large number of
Australian, female twin pairs. These twins are
members of the Australian National Health and
Medical Research Council Twin Registry, and partic-
ipated in a health and lifestyle survey conducted
between 1988 and 1990.3–6 Responses to this survey
were obtained from both members of 1624 female
monozygotic (MZ) and dizygotic (DZ) twin pairs
aged 25 and older, for a pairwise response rate of
82%. The study questionnaire included questions on
their menstrual cycles and reproductive events,
including specific obstetric factors relating to each
delivery.3–6

Premature delivery (preterm birth) was defined in
the questionnaire as delivery ‘over 2 weeks early’,

rather than the standard definition of curtailment of
pregnancy before the 37th week of gestation. We did
not ask for number of weeks gestation. We investi-
gated parous twins’ independent reports of pre-
mature delivery of first baby, and also reports of
whether any birth was premature using the > 2 week
criterion. Premature births may have been sponta-
neous or induced, vaginal deliveries or performed by
Caesarean section. Data were also available on these
items, but analyses are not presented here.

Twin pair matrices of tetrachoric correlations, and
corresponding asymptotic covariance matrices, were
computed separately for MZ and DZ twin pairs,
using the Windows version of PRELIS 2.20.7 These
correlations apply appropriately to dichotomous
categorical data where the underlying distribution is
assumed to be continuous and normally distributed.
Genetic models were fitted by the method of asymp-
totic least squares to estimate the contributions of
additive genetic, shared and non-shared environ-
mental effects, using Mx.8 We proceeded by system-
atically testing the significance of dropping parame-
ters in turn. In addition to the likelihood ratio ø2 test
(LR), the Akaike Information Criterion (AIC, meas-
ured as ø2-2df) was used as an additional indicator of
fit. On the grounds of parsimony the model with the
least number of parameters that offered a fit not
significantly worse than the full model was chosen.
Data analysis methods are described more fully
elsewhere.9

Correspondence: Susan A Treloar, PhD, Queensland Institute of
Medical Research, PO Royal Brisbane Hospital, Queensland 4029,
Australia. Tel: 61 7 3362 0229; Fax: 61 7 3362 0101; E-mail:
sueT@qimr.edu.au
Received 15 November 1999; accepted 17 November 1999

Twin Research (2000) 3, 80–82
y © 2000 Macmillan Publishers Ltd All rights reserved 1369–0523/00 $15.00

www.nature.com/tr



Both members of 905 parous twin pairs (579 MZ
pairs and 326 DZ pairs) answered the question
concerning prematurity of their first and any sub-
sequent birth. Ages of parous twins ranged from 25
to 87 years, and mean age was 43.94 ± 0.25 years.
Mean ages of MZ twins (44.09 years) and DZ (43.71)
twins did not differ significantly. The prevalence of
premature first birth in the sample was 9.4% of
individuals in MZ pairs and higher at 12.9% in the
DZ twins, giving an overall prevalence of 10.7%.
Premature parturition relating to any birth using the
> 2 week criterion was reported by 17.3% of individ-
uals in MZ pairs and 21.6% of twins in DZ pairs. The
percentage of nulliparae was the same (26%) for
both MZ and DZ individuals.

Non-parametric one-way median score tests
showed that parous MZ and DZ twins did not differ
significantly on median total pregnancies. However,
they did differ on median number of full-term (vs
aborted) pregnancies (P < 0.05) and of offspring
(P < 0.01), the latter median being higher for DZ
twins (3) than for MZ twins (2). This suggested a
higher prevalence of multiple births in the DZ twins.
The median number of preterm babies reported by
DZ twins was significantly higher than for MZ twins
(P < 0.05), consistent also with a higher rate of
multiple births.

The ratio of MZ to DZ twin pair correlations
( > 2:1) shown in Table 1 suggests that genetic influ-
ences may be operating on premature parturition.10

For both prematurity variables, using the AIC as the
indicator, best-fitting most parsimonious models
contained only additive genetic variation (A) and
individual environmental effects (E). For having had
a premature delivery for any birth, the best fitting
model (AIC = –0.924) included a heritability of
27%. A model including non-additive genetic influ-
ences resulted in a better fit than a model containing
shared environment (C), but was not a significant
improvement on the AE model. A CE model contain-
ing no additive genetic variation was rejected on
significance grounds. For premature delivery, the AE
model was again the model of choice (AIC = –1.30)
with a heritability of 17%; however, in this case a
model containing no genetic influences could not be
rejected on significance grounds.

Evidence suggests that in Australia the impact of
changes in obstetric care that might influence pre-
mature labour were minimal up to 1975.11 We would

not expect the impact of obstetric practices to differ
for individuals on the basis of their zygosity.
Increased prevalence of multiple births to DZ twins
may have influenced pairwise concordance for pre-
term birth, but any effect would have been too small
to detect in this sample. Data were retrospective and
collected in 1988–90, before the widespread availa-
bility of assisted reproductive technologies, which
are increasing the prevalence of non-monozygotic
multiple births, and hence may influence prevalence
of premature delivery.

We cannot unequivocally separate out the differ-
ent aetiologies for preterm birth in our data. The
aetiologies of spontaneous premature membrane
rupture or preterm labour may well differ, and those
of indicated preterm births relating to foetal or
maternal morbidity clearly may vary even further.
Retrospectivity may have influenced accuracy of
recall of preterm delivery. Data on length of gestation
or date of last menstrual period (LMP) were not
available.

Furthermore, genetic heterogeneity may be impli-
cated in the predisposing conditions.1 For example,
the prevalence of pre-eclampsia, which may be
associated with indicated premature delivery, is
higher for first than subsequent births. If the aetiol-
ogy is foetal in origin rather than maternal, which
may well be the case,12 the effect may be to reduce
the (maternal) twin pair correlations. This might
suggest that the heritability we have observed is
occurring in the spontaneous rather than the indi-
cated preterm deliveries, but to test this hypothesis
would require a very large sample indeed. It would
be fascinating to know whether, for example, there is
a genetic contribution to factors such as cervical
incompetence per se. Nevertheless, our data do
suggest that genetic influences may be important in
influencing preterm birth, and further investigation
of this hypothesis may well be warranted.

Acknowledgements

We thank the National Health and Medical Research
Council of Australia for support (No. 941177,
No. 971232) and the twins for their co-operation.

References

1 Porter T, Fraser A, Hunter C, Ward R, Varner M: The risk of
preterm birth across generations. Obstet Gynecol 1997; 90:
63–67.

2 Khoury M, Cohen B: Genetic heterogeneity of prematurity and
intrauterine growth retardation: clues from the Old Order
Amish. Am J Obstet Gynecol 1987; 157: 400–410.

Table 1 Tetrachoric twin pair correlations for prematurity of first
delivery and of any delivery

MZ pairs (n=579) DZ pairs (n=326)
Premature by Twin pair Twin pair
>2 weeks correlation SE correlation SE

First birth 0.20 0.11 –0.03 0.14
Any birth 0.30 0.08 –0.03 0.11

Twin Research

Genetic influences on preterm birth in Australian twin mothers
SA Treloar et al y

81



3 Do K-A, Treloar SA, Pandeya N, Purdie D, Green A, Heath AC,
Martin NG: Predictive factors of age at menopause in a large
Australian twin study. Hum Biol 1998; 70: 1073–1091.

4 Heath A, Cloninger C, Martin N: Testing a model for the
genetic structure of personality: A comparison of the person-
ality systems of Cloninger and Eysenck. J Pers Soc Psychol
1994; 66: 762–775.

5 Treloar S, Martin N, Heath A: Longitudinal genetic analysis of
menstrual flow, pain and limitation in a sample of Australian
twins. Behav Genet 1998; 28: 107–116.

6 Treloar S, Martin N, Dennerstein L, Raphael B, Heath A:
Pathways to hysterectomy: insights from longitudinal twin
research. Am J Obstet Gynecol 1992; 167: 82–88.

7 Jöreskog K, Sörbom D: PRELIS 2.20 for Windows. Scientific
Software International Inc: Chicago, 1998.

8 Neale M: Mx: Statistical Modeling. Department of Psychiatry:
MCV, Richmond, VA, 1997.

9 Neale M, Cardon L: Methodology for Genetic Studies of Twins
and Families NATO ASI Series. Kluwer Academic Publishers:
Dordrecht, 1992.

10 Heath AC, Martin NG, Eaves LJ, Loesch D: Evidence for
polygenic epistatic interactions in man? Genet 1984; 106:
719–727.

11 Stanley F: Perinatal outcome in Western Australia, 1968 to
1976. Perinatal mortality and birthweight. Med J Aust 1981; 1:
370–374.

12 Thornton J, Macdonald A: Twin mothers, pregnancy hyper-
tension and pre-eclampsia. Br J Obstet Gynaecol 1999; 106:
570–575.

Twin Research

Genetic influences on preterm birth in Australian twin mothers
y SA Treloar et al
82