An inherited defect in anticoagulation known as resistance to activated protein C has recently been described [2, 3]. In most cases, resistance to activated protein C results from a point mutation in the gene coding for coagulation factor V [4, 5]. This mutation, factor V Leiden, is associated with threefold to fivefold increased risks for first episodes [6, 7] and recurrent episodes [8] of venous thromboembolism. Moreover, risks for thromboembolism are markedly increased among persons with the factor V Leiden mutation who also have concomitant risk factors for thrombosis, such as hyperhomocystinemia, oral contraceptive use, older age, and deficiencies of protein C and protein S [9-12].

Whether the hypercoagulable state associated with factor V Leiden results in increased risk for recurrent spontaneous abortion is uncertain [13-15]. We therefore determined the prevalence of factor V Leiden mutation in a consecutive series of 113 women referred for evaluation of recurrent spontaneous abortion (case-patients) and 437 post-menopausal women with at least one successful pregnancy and no history of pregnancy loss (controls). We limited our investigation to white women to avoid the potential for bias and confounding by ethnicity [16]. We also evaluated the prevalence of factor V Leiden mutation in an additional group of 387 postmenopausal women who reported at least one pregnancy loss.

Methods

Top Methods Results Discussion Author & Article Info References

Case-patients were consecutive white women referred for evaluation at the Recurrent Miscarriage Clinic at Brigham and Women's Hospital, Boston, Massachusetts, between 1 July 1995 and 1 June 1996. Eligibility criteria were a history of three or more spontaneous abortions and no parental chromosomal abnormality. Of 141 women screened, 28 were excluded because they had had fewer than three spontaneous abortions before the 24th week of gestation. Thus, 113 case-patients were available for genetic analysis; none had a history of venous thromboembolism.

Controls were white women who provided a blood specimen before randomization into the Women's Health Study, a primary prevention trial being done in postmenopausal female health professionals [17]. These women reported no history of myocardial infarction, stroke, or cancer, and they completed a brief questionnaire that solicited demographic characteristics, medical history, and lifestyle information (for example, age at which menstrual periods began, number of pregnancies that reached term, and number of pregnancies that lasted less than 24 weeks). Of the women who reported at least one successful pregnancy and no history of pregnancy loss, 437 were randomly chosen as controls. None had a history of venous thromboembolism.

All women underwent analysis for factor V Leiden mutation. Testing involved 1) second-generation screening for resistance to activated protein C that used factor V-deficient plasma and 2) genetic confirmation with polymerase chain reaction techniques for all borderline and low-value results [7, 16]. Case-patients also underwent screening for anticardiolipin antibodies, antiphosphatidylserine antibodies, and lupus anticoagulant.

We also determined mutation status among the 387 white participants in the Women's Health Study who reported one or more pregnancy losses. This group included 294 women with one or two pregnancy losses and 93 women with three or more pregnancy losses. Thus, we evaluated a total of 206 women with three or more losses.

The odds ratio was used as a measure of the strength of the association, and the chi-square statistic was used to test the significance of any difference in the prevalence of factor V Leiden mutation between case-patients and controls. All P values were two-tailed, and 95% CIs were computed.

Results

Top Methods Results Discussion Author & Article Info References

Case-patients and controls had similar mean (4.6 compared with 4.0) and median (4.0 compared with 4.0) numbers of pregnancies. The number of viable deliveries was greater for controls (mean, 3.0; median, 3.0 [range, 1 to 6]) than for case-patients (mean, 0.5; median, 0.0 [range, 0 to 5]). The mean number of pregnancy losses among case-patients was 3.9 (median, 3.0 [range, 3 to 11]); of these women, 46 (40.7%) had at least one successful pregnancy.

Of the 437 controls, 16 (3.7%) carried the factor V Leiden mutation (Table 1). In contrast, 9 of 113 case-patients (8.0%) carried the mutation (odds ratio, 2.3 [95% CI, 1.0 to 5.2]; P = 0.050). In the subgroup of 67 case-patients with no history of successful pregnancy, 6 (9.0%) carried the factor V Leiden mutation (odds ratio, 2.6 [CI, 1.0 to 6.7]; P = 0.048). In contrast, in the subgroup of 46 case-patients who had had at least one successful pregnancy, 2 (4.3%) carried the mutation (odds ratio, 1.2; P > 0.2). None of the case-patients who carried the mutation tested positive for anticardiolipin or antiphosphatidylserine antibodies, had a positive result on testing for lupus anticoagulant, or had another known coagulation abnormality; 4 case-patients without the mutation had at least one of these abnormalities. Mean age at menarche did not vary according to factor V Leiden mutation status (12.6 compared with 12.5 years).

We further evaluated the prevalence of the factor V Leiden mutation among an additional 387 post-menopausal women participating in the Women's Health Study who reported one or more pregnancy losses. Compared with women who had no history of pregnancy loss (prevalence of mutation, 3.7%), the prevalence of mutation was 5.4% (16 of 294) among those with one to two losses, 7.2% (6 of 83) among those with three to five losses, and 10% (1 of 10) among those with six or more losses. The prevalence of the factor V Leiden mutation for the 93 women in this group who had three or more losses (7.5%) was almost identical to that seen in case-patients. Thus, the overall prevalence of the factor V Leiden mutation among all 206 women evaluated who had recurrent pregnancy losses was 7.8% (odds ratio, 2.2 [CI, 1.1 to 4.5]; P = 0.026).

Among the Women's Health Study participants who had at least one successful pregnancy, age at the time of first pregnancy did not vary by mutation status (23.8 years compared with 24.3 years). One Women's Health Study participant was homozygous for factor V Leiden mutation and had had three pregnancy losses and two successful pregnancies.

Discussion

Top Methods Results Discussion Author & Article Info References

Our data are compatible with the hypothesis that the factor V Leiden mutation, the most common inherited predisposition to thrombosis, plays a role in some cases of unexplained recurrent pregnancy loss.

The possibility that placental thrombosis is an etiologic factor in some cases of recurrent spontaneous abortion has previously been raised in relation to patients with antiphospholipid syndromes and those with deficiencies of coagulation factor XII and intrinsic fibrinolysis. These abnormalities, however, are infrequent. In contrast, factor V Leiden mutation is common; thus, the possibility that this mutation contributes to the hypercoagulability of pregnancy raises issues of pathophysiologic and clinical interest. In this regard, it has been suggested that de novo resistance to activated protein C can develop during pregnancy among persons without the factor V Leiden mutation [15, 18] and that these persons may have an increased risk for pregnancy-related venous thromboembolism [18-21]. Furthermore, one report described a patient in whom the magnitude of resistance to activated protein C due to factor V Leiden mutation increased during pregnancy [15].

The risk estimates in our study are remarkably similar to those reported by Preston and colleagues [13], who found that the factor V Leiden mutation was associated with a 2.0-fold increase in risk for third-trimester pregnancy loss associated with intrauterine death. Similarly, Rai and colleagues [14] have reported an increased prevalence of resistance to activated protein C among women with second-trimester pregnancy loss. Two recent case series also suggest that the prevalence of factor V Leiden mutation is increased among women with several pregnancy losses [22, 23].

Despite the consistency of these observations, potential limitations of our study merit consideration. For example, it is possible that the case-patients, who were selected from a recurrent-miscarriage clinic, might have caused an overestimate of the true prevalence of the factor V Leiden mutation among women with several pregnancy losses. We believe that this possibility is unlikely, however, because 93 of the 387 additional women evaluated from the Women's Health Study reported a history of three or more pregnancy losses; in addition, the prevalence of factor V Leiden mutation in this group was almost identical to that in the case-patients. Furthermore, the sample size in the control group was large, greatly reducing the possibility that chance could have led to an underestimation of prevalence. Moreover, the prevalence of the factor V Leiden mutation in our control group is similar to that reported in population-based studies of the mutation [3, 5-710, 16]. We further believe that the comparison of mutation rates between case-patients selected from a hospital-based setting and controls derived from a population-based setting is unlikely to create any major biases because the exposure is genetic rather than acquired. Moreover, the potential for confounding by sex and ethnicity is unlikely because these factors were directly controlled for in the study design. Finally, by limiting controls to postmenopausal women, we eliminated the possibility that our observed results might be biased by different lengths of exposure. Indeed, because the controls had completed their reproductive years, our choice of this control group might lead to an underestimation of true effects.

Some cases of pregnancy loss are believed to have an anatomic, endocrinologic, infectious, or immunologic basis [1]; thus, the fact that our study did not specifically exclude such patients is a potential limitation. It is important to note, however, that the inclusion of patients with nonthrombotic causes of fetal loss would again tend to bias our study toward a false null result and could not account for our positive finding. Indeed, were it possible to study only patients with known placental thrombosis and exclude all other causes of loss, the relative risks associated with the factor V Leiden mutation may substantially increase in magnitude. On the other hand, our study design did exclude cases of recurrent pregnancy loss caused by parental chromosomal abnormalities.

The possibility that participants may have misclassified pregnancy losses also deserves consideration, particularly for controls, for whom data were gathered by questionnaire rather than physician interview. In addition, we cannot exclude the possibility that women may have overreported or underreported miscarriage rates. It is exceedingly unlikely, however, that any potential misclassification would have systematically occurred on the basis of factor V Leiden genotype because mutation status was unknown at study entry. Furthermore, the dilutional effect of random misclassification could again lead only to an underestimation of true risks.

Our study sample consisted of 206 women with recurrent pregnancy loss and 437 controls. Nonetheless, the 95% CIs around our summary risk estimate of 2.2 are relatively wide. Thus, large-scale studies are needed to confirm these observation.

Dr. Price: Section of Cardiology, Boston University Medical Center, 88 East Newton Street, Boston, MA 02118.

Dr. Ariyo: Cardiovascular Division, University of Sacramento Medical Center, 2315 Stockton Boulevard, Sacramento, CA 95817.

Dr. Miletich: Washington University Medical Center, 660 South Euclid Avenue, St. Louis, MO 63110.