Main

Pregnancy loss is a common medical problem among reproductive age women.1 However, relatively few women having one pregnancy loss experience multiple or “recurrent” pregnancy losses (RPLs). Approximately 5% of such women experience a second pregnancy loss and only 1–2% three or more.1 Evaluation for RPL often includes ruling out parental chromosome abnormalities, identifying maternal exposures, and testing for underlying maternal conditions.1,2 Effective clinical interventions are available for some etiologies, such as correcting uterine anomalies surgically and treating antiphospholipid syndrome with aspirin and heparin.3,4,5

The evidence review summarized herein addressed the association of inherited thrombophilia with RPL, focusing on tests for two genetic variants that are frequently ordered: Factor V Leiden (“F5”) and prothrombin G20210A (“F2”). F5 defines a single-nucleotide substitution in the F5 gene (i.e., Factor V Leiden; F5 c.1691G>A; and p.Arg506Gln),3,6 whereas F2 defines a single-nucleotide substitution in an untranslated region of the F2 gene (i.e., prothrombin G20210A and F2 c.20210G>A).4,6 Associations between these (and other) heritable thrombophilia variants and serious pregnancy complications (e.g., RPL, fetal growth restriction, placental abruption, and preeclampsia) began to appear in the literature in 1996.6 By 2005, laboratories were offering clinical testing for F5/F2 as part of “infertility” or RPL evaluations.7 In some states, these tests continue to be available directly to consumers through the internet, without involvement of healthcare providers. The F5/F2 variants are the most common genetic variants associated with venous thromboembolism (VTE).3,4 Five to 7% of individuals of northern European white ancestry worldwide are F5 carriers (heterozygotes).8 In the United States, 5.1% of non-Hispanic whites, 2% of Hispanic/Mexican Americans, and 1.2% of African Americans are F5 carriers.9 F2 carriers are found in 2.2% of US non-Hispanic whites and Hispanic/Mexican Americans and in 0.6% of African Americans.9 The F5 variant is virtually absent in native Asian and African populations, and F2 is rare.8,10 Individuals having two copies of the F5 or F2 variants (approximately 1 in 1,000 and <1 in 8,000 whites, respectively), or one of each variant, are much less common.9 A thrombotic pathogenesis for pregnancy loss proposes that one or two copies of these variants could enhance the existing hypercoagulable state in pregnant women and that impaired fetal circulation or thrombotic vasculopathy in the placenta could result in placenta-mediated pregnancy complications and fetal loss.3,4,11,12

F5 and F2 are the most frequently ordered genetic variants associated with RPL13,14 but does the accumulated knowledge base support the broad-based use of these tests to screen women with RPL to identify those who may benefit from specific management or treatment? This review addresses the overarching question ( Table 1 ): “Does testing reproductive age women with recurrent pregnancy loss (RPL) for Factor V Leiden (F5 c.1691G>A) and prothrombin (F2 c.20210 G>A) variants improve outcomes, or are testing results useful in medical and personal decision-making?” If no studies directly answer the overarching question, a chain of evidence will be attempted, as shown in the analytic framework ( Figure 1 ). This review does not address: testing other markers associated with thrombophilia (e.g., MTHFR, proteins C and S); testing for other adverse pregnancy outcomes (e.g., growth restriction, placental abruption, and preeclampsia); or testing women with a personal history of thrombotic events.

Table 1 Key questions (KQ)
Full size table
Figure 1
figure 1

Analytic framework: Factor V Leiden G1691A and prothrombin G20210A testing in women with recurrent pregnancy loss. This schematic shows the analytic framework underlying the current review. The numbers indicate the four key questions in Table 1 .

Full size image

Materials and Methods

To guide the review, an analytic framework ( Figure 1 ) and key questions ( Table 1 ) were refined in consultation with a Technical Expert Panel (see Acknowledgments). Standard methods for evaluating analytic validity, clinical validity, and clinical utility included systematic searches of the published and gray literature (e.g., technology and policy reports and white papers not published in peer-reviewed journals), use of preset inclusion and exclusion criteria (eTable 1, Supplementary Data online), data abstraction and analyses, and grading of individual studies and overall strength of evidence.15,16,17 Electronic searches were conducted in PubMed, HuGE Navigator (version 2.0),18 and the Cochrane Library, using MeSH terms (e.g., “abortion, habitual,” “Factor V,” “factor V Leiden,” and “prothrombin”), key words from identified articles (e.g., F5/G1691A/factor V Leiden and RPL, recurrent miscarriage, stillbirth, and intrauterine fetal demise), and other related terms (e.g., genetic association and genetic testing). Initial searches included the time period January 1, 1990, to March 31, 2010; an update of the literature search on clinical validity and utility was conducted through April 30, 2011. Bibliographies were hand searched and information sought through targeted gray literature searches. One team member (L.A.B.) reviewed articles for inclusion and abstracted raw data into electronic spreadsheets; another (G.E.P.) reviewed the articles and spreadsheets to validate eligibility and the accuracy of data abstraction. Discrepancies were resolved by discussion.

For this review, analytic validity is defined as the test’s ability to accurately and reliably identify the variant or genotype of interest in specimens consistent with those to be tested in clinical practice.15,16 Included articles would provide sufficient data to evaluate assay performance characteristics (e.g., repeatability of test results) and robustness (i.e., resistance to small changes in preanalytic and analytic variables) and to compute estimates of analytic sensitivity and specificity with 95% confidence intervals (CIs).15,16 Clinical validity is defined as the ability of the test to accurately and reliably identify or predict the disorder (or condition, phenotype) of interest.15,16 Included articles for clinical validity would assess the strength of association between test results and the disorder of interest and/or provide data from which to compute performance estimates (e.g., clinical sensitivity/specificity, odds ratio [OR], and relative risk). Data could be derived from published meta-analyses and/or analysis of primary reports of studies that met a minimum case definition for RPL (≥2 losses in the first or second trimester, or ≥1 stillbirth or intrauterine fetal demise in the third trimester); described selection criteria for cases and controls; performed F5 and/or F2 testing on all women; and provided sufficient raw data to compute effect sizes with 95% CIs.

Clinical utility is defined as evidence of measurable improvement in health outcomes, and/or the test’s added value to management or treatment decisions compared with current management without the testing.15,16 Included articles for clinical utility would address the benefits and harms related to F5/F2 testing among women diagnosed with RPL, such as treatments for women with RPL, trials assessing improved outcomes based on treatments, and evidence that decision-making processes or clinical outcomes are impacted by testing and/or the setting in which testing occurs (e.g., test ordered by a physician and direct-to-consumer purchase).

Because of the low prevalence of the homozygous F5/F2 genotypes (1 per 1000 or less), homozygotes and heterozygotes were grouped together (dominant model). Performance estimates and summary ORs with corresponding 95% CIs were derived and summarized using random effects modeling weighted by each data set’s total variance (Comprehensive Meta-analysis, version 2, Biostat, Englewood NJ).19 Between- study heterogeneity was evaluated using the Q test and I2 statistic.19,20 Potential for publication bias was assessed using a funnel plot and the Egger regression test.19,21 For each case-control study, the observed genotype frequencies in cases and controls were compared with expected frequencies based on Hardy-Weinberg equilibrium (χ2 test with 1 degree of freedom). All P values are two sided at the P = 0.05 level.

Before abstraction of clinical validity data, potential covariates were identified for stratified analyses should heterogeneity be identified. These covariates included method of testing (activated protein C resistance (APCR) and genotyping), publication year (effects of decreased APCR prescreening and/or improving test performance after 2002), study size, race/ethnicity of subjects, site of the study, number of losses (≥2 and ≥3), timing of losses (trimester), pattern of losses (primary, secondary), stated exclusion based on a personal history of VTE, and the extent to which the etiology of RPL was “unexplained” (reporting on the clinical evaluation to identify other causes). Data sources and study designs (Levels 1–4), quality of individual studies (good, fair, and marginal), and overall quality of evidence for analytic validity and clinical utility (convincing, adequate, or inadequate) were graded using a published methodology16 that is consistent with other approaches.22,23,24 The case-control and cohort studies reviewed for clinical validity were graded using the same method,16 with additional criteria based on the STROBE checklist.25

Results

Figure 2 provides a summary of the literature review. Abstracts were reviewed for 754 unique citations identified through electronic searches; articles were obtained for 261, and 41 others were identified through hand searches. Of 302 articles reviewed, 174 did not satisfy inclusion criteria, and 116 met the initial overall and key question-specific inclusion criteria.

Figure 2
figure 2

Overview of literature search and selection of key question-specific articles.

Full size image

Analytic validity

Six published articles6,26,27,28,29,30 and seven gray literature documents31,32,33,34,35,36,37 addressed Key Question 2. In the United States, F5/F2 testing is widely offered in a variety of laboratory settings (e.g., national, hospital, and genetic specialty) using US Food and Drug Administration cleared tests or laboratory developed tests.6,26,37 Five US Food and Drug Administration 510(k) Summary Documents provided information on assay performance characteristics (e.g., reproducibility and robustness) and reported high concordance between index and referent (bidirectional sequencing) test results.32,33,34,35,36 A published systematic review of 41 F5 and 23 F2 method comparison studies also found high (99%) concordance between index and referent tests.26,37 Limitations of the latter method comparisons included referent methods that did not include either a “gold standard “ (bidirectional sequencing) or test samples with genotype confirmed by broad consensus. Only 11 studies (27%) reported blinding samples,26 and none provided estimates of analytic sensitivity or specificity under “real world” conditions.

Two small, one-time interlaboratory comparison studies29,30 (Level 2, grade Fair)16 were excluded from computation of performance estimates due to concerns about generalizability (e.g., variation in sample types and protocols). However, experienced external proficiency testing programs for F5/F2 are available in the United States,31 United Kingdom/Europe,28 and Australasia.27 Three proficiency testing data sets from two sources had sufficient raw data for analysis of error rates (i.e., false positives and false negatives).27,31 Summary estimates of analytic sensitivity and specificity were 98.8% and 99.3% for F5 tests, respectively; results for F2 were 98.3% and 99.6%, respectively ( Table 2 ). Heterogeneity was high for F5 (I 2 > 70%, P < 0.05), possibly due an observed small proportion of poorly performing laboratories.27 Heterogeneity was low for F2 (I 2 ≤ 20%, P ≥ 0.30). These data sources were graded Level 1 (highest), quality Good27 and Fair31 (i.e., met quality standards but not published).16

Table 2 Estimates of analytic validity for the F5 and F2 variants
Full size table

Summary

The evidence on analytic validity is adequate (scale inadequate, adequate, and convincing)16 to reliably estimate the analytic sensitivity and specificity and reproducibility of F5/F2 test results. The rating was not convincing because results were not stratified by analytic method/platform, and data were collected between 1998 and 2003. These factors may limit generalizability to current testing methods and laboratories.

Clinical validity

Five reviews38,39,40,41,42 provided historical perspective on F5/F2 testing and RPL, and seven meta-analyses addressed case-control43,44,45,46,47,48 and cohort49 studies. No meta-analysis included recent studies, matched our inclusion criteria, and adequately addressed our key questions ( Table 1 ). Consequently, data from primary studies of different designs were analyzed to estimate effect sizes. Case-control studies were used to investigate an association between the presence of F5 or F2 variants and RPL. Cohort studies addressed two different questions, the subsequent recurrence of pregnancy loss in women with RPL and the occurrence of pregnancy loss in unselected cohorts of women (e.g., primigravidas combined with varied pregnancy histories, and diverse race/ethnicity).

Case-control studies

Of 73 identified case-control studies, 60 met initial criteria.50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109 Fifteen studies were excluded based on reported consanguinity in cases,51,65 overlap of data with another included study83,104,109 or inability to compute ORs (i.e., no observed carriers in cases and/or controls).53,55,56,59,60,67,74,84,94,107 All but three of the 45 remaining studies satisfied Hardy-Weinberg criteria (χ2; test; P ≥ 0.05); these three were excluded.77,103,108 Of interest, two of these studies77,108 were in a small subset of studies conducted in non-European white populations. Using a random effects model, the summary OR for 42 studies was 2.34 (95% CI: 1.88–2.92; P < 0.001), with moderate heterogeneity (I 2 = 45%; P < 0.001). To explore this heterogeneity, results were stratified by the preselected covariates.

Of the 10 covariates examined, only two were found to be associated with significant differences. Among five studies (conducted in Israel,95,100,110 Serbia,80 and Spain54), a phenotypic test, APCR, was used to determine which women would be tested for the F5 variant. The remaining 37 studies genotyped all women. The ORs in these two groups differed significantly (OR of 4.35 and 2.19 in the five and 37 studies, respectively; P = 0.008). The five studies did not technically meet the inclusion criterion of genotyping all patients, and the ability of the functional APCR test to accurately and reliably identify the presence of F5 has changed over time. Thus, these five studies were excluded from further analysis. Another group of four studies from India,102 Palestine,71 Lebanon/Bahrain,62 and Egypt,81 was compared with the 33 studies that reported on mainly white populations of northern European/Jewish ancestry. The OR in the first group was higher (ORs of 3.94 and 2.02, respectively; P = 0.11). Had the two Tunisian studies77,108 been included, the difference would have been statistically significant (ORs of 4.35 and 2.02, respectively; P = 0.005). Differences may have resulted from relatively high (15%62 and 11%71) or low (1%102) observed carrier rates, failure to satisfy Hardy-Weinberg criteria,77,108 and other variables (e.g., methodological differences, ascertainment, and other biases).

The most relevant summary OR (2.02, 95% CI: 1.60–2.55; P < 0.001) was derived from the 33 case-control studies that focused on white populations of northern European/Jewish ancestry. Heterogeneity remained moderate (I 2 = 39%, P = 0.012). Figure 3 presents a Forest plot of these data. Reanalyses of the selected covariates in this subset identified no significant effects. The funnel plot19,111 was highly asymmetric with 10 studies imputed with low ORs (Egger regression test19,21 P = 0.058).

Figure 3
figure 3

Forest plot of the association of F5 genotype and recurrent pregnancy loss (RPL) in 33 case-control studies. Studies are ordered by effect size. The summary OR is 2.02 with moderate heterogeneity (I 2 = 39%, P = 0.012). CI, confidence interval; OR, odds ratio.

Full size image

Of 38 identified case-control studies on the association of F2 and RPL, 36 met initial inclusion criteria.50,52,55,57,58,59,61,63,65,68,69,72,73,75,76,78,80,82,83,86,87,88,89,91,92,96,97,98,99,101,102,104,105,112,113,114 Seven studies were excluded based on overlap of data with another included study83,104 or inability to compute ORs.55,59,65,102,114 One study did not initially satisfy Hardy-Weinberg criteria (P = <0.001).105 When a likely typographical error was corrected, the criteria were easily met, and the study was included with the correction. The summary OR for the remaining 29 studies was 2.07 (95% CI: 1.59–2.70; P < 0.001), with low heterogeneity (I2 = 0%; P = 0.84). Figure 4 presents a Forest plot of these data. The funnel plot19,111 was slightly asymmetric with three studies imputed with low ORs (Egger regression test19,21 P = 0.45).

Figure 4
figure 4

Forest plot of the association of F2 genotype and recurrent pregnancy loss (RPL) in 29 case-control studies. Studies are ordered by effect size. The summary OR is 2.07 with low heterogeneity (I 2 = 0%, P = 0.84).

Full size image

Overall, the included studies were Level 3 (lower quality case-control studies).16 Most studies were rated Fair, but 10 had one or more poor elements and were graded Mar- ginal. 50,55,62,64,73,77,79,81,112,113

Cohort studies

The overarching question in this review addresses how to inform F5/F2 carrier women with RPL about their recurrence risk for pregnancy loss and provide access to an effective treatment. From this perspective, an ideal study would identify a representative cohort of women with “unexplained “ RPL (i.e., other strong causal factors excluded) for which there is access to data on test results, clinical evaluation, treatments, and subsequent pregnancy outcomes. After genotyping, fetal loss rates in one or more subsequent pregnancy(ies) would be compared between F5/F2 carrier and noncarrier women. Of 23 identified cohort studies, only four met this initial inclusion criterion that study subjects meet the definition of RPL.115,116,117,118

Two studies prospectively followed pregnant white women with a history of ≥3 pregnancy losses occurring at ≤12 weeks’ gestation116,118 and a third117 reviewed pregnancy outcomes in a retrospective cohort of women having ≥3 pregnancy losses at ≤13 weeks. All showed an effect on loss rates in carrier women with RPL, and two individually reached significance ( Table 3 ). A fourth study retrospectively ascertained pregnancy histories for a cohort of primigravid women followed longitudinally as part of a large multicenter study on thrombosis (first-degree relatives of patients with VTE but no personal VTE events).115 Table 3 lists the data from third pregnancies of women with two previous losses. This study represents a high-risk scenario, with a higher rate of women with two variants included among “carriers” (7.4% versus an expected 1–2%). However, no significant effect on loss rate was observed. The summary OR for F5 carriers with RPL to have a subsequent event was 1.93 (95% CI: 1.21–3.09, P = 0.006; eFigure 1, Supplementary Data online), with low heterogeneity (I2 = 0%, P = 0.42) ( Table 3 ). Only one of these studies117 reported on F2 (OR: 3.29, P = 0.17; eFigure 2, Supplementary Data online). Overall, these studies were considered to be Level 2 and rated Good116,117,118 to Fair115 (based on lack of power).

Table 3 Summary of observed pregnancy loss rates in cohort studies of women with RPL, stratified by F5/F2 carrier status
Full size table

Twenty cohort studies reported on fetal loss rates in unselected populations, stratified by genotype. Ten were excluded based on design or execution issues (see eTable 2, Supplementary Data online). Eight studies85,115,119,120,121,122,123,124 were analyzed to provide information about the occurrence rate of pregnancy losses among F5 carriers. The summary OR was 2.03 (95% CI: 1.29–3.17; P = 0.002; eFigure 1, Supplementary Data online), with high heterogeneity (I 2 = 66%; P = 0.004). Four studies121,123,124,125 in unselected cohorts provided information about the occurrence rate of pregnancy losses among F2 carriers (eFigure2, Supplementary Data online). The summary F2 OR was 1.77 (95% CI: 0.87–3.61; P = 0.11), with high heterogeneity (I 2 = 62%; P = 0.049). Three cohort studies85,115,123 were rated Level 2, Good; all others were rated Level 3, Fair.120,121,122,124,125,126

Summary

These case-control and cohort studies reflect methodological diversity and clinical heterogeneity. Studies have been conducted in more than 20 countries, using different study designs and in routine care settings and high-risk referral centers. Study limitations have included inadequately described and/or heterogeneous case and control groups and cohorts, insufficient information to adequately assess potential biases, and missing or incomplete information on important covariates such as maternal age and number and timing of losses and small studies (30% of F5 and 86% of F2 studies reported on fewer than 10 carriers among women with RPL).

The rationale for offering F5/F2 testing in clinical practice has been largely based on small case-control studies reporting higher odds that women with RPL will carry F5/F2 variants than women without RPL. These data confirmed an effect for both F5 and F2 (P < 0.001). More importantly, there was concern that the case-control data did not provide direct evidence on the risk for subsequent losses among carrier women with RPL. However, results from the case-control studies, as well as the cohort studies addressing occurrence of pregnancy loss and recurrence of loss in carrier women, are consistent. In summary, the evidence is adequate (scale inadequate, adequate, and convincing) to establish clinical validity of F5 and F2 testing in this scenario. The evidence for F5 is not convincing due to high heterogeneity and the potential for publication bias to reduce the effect size. The evidence for F2 is not convincing due to insufficient data (i.e., a single cohort study in women with RPL that showed an effect but did not reach significance).

Clinical utility

A test is said to have clinical utility, when results change clinical management, health outcomes improve, and the benefits outweigh known or potential harms.15,16 In women with RPL, the most likely outcome in the next pregnancy is a normal birth. Pregnancy loss rates vary from 10 to 50% based on multiple factors.127,128,129 Therefore, high-quality evidence is needed to show that a proposed intervention will improve the likelihood of a successful pregnancy in women with RPL who carry one or two F5/F2 variants. No studies were identified that directly addressed the balance of benefits and harms related to F5/F2 testing in this clinical scenario or the impact on decision making for either direct-to-consumer testing or testing ordered by a healthcare provider (overarching question, Key Question 4C). Treatments proposed for women with RPL, regardless of F5/F2 status (Key Question 4A), currently include low-dose aspirin, unfractionated or low molecular weight heparin (LMWH), and combined anticoagulant treatments.12,119,129,130,131,132,133,134,135,136 Two randomized controlled trials recently reported from the Netherlands133 and Scotland119 found that loss rates in women with RPL did not differ significantly among three study groups: low-dose aspirin only; LMWH plus low-dose aspirin; and placebo133 or intense surveillance.119 Three meta-analyses129,135,136 also reported that there is insufficient evidence that these treatments reduce loss rates in women with RPL who do not have antiphospholipid syndrome or support the routine use of these treatments in women with RPL. Other non-health-related benefits of F5/F2 testing have been proposed, including value of information (i.e., identifying a “cause” for the RPL or allowing for alternative reproductive options),137 but no relevant studies were identified. F5/F2 testing has also been suggested to have benefit in identifying risk for VTE in pregnant women without a personal history, but evidence of efficacy and cost-effectiveness is lacking.12,138

Potential and observed harms associated with testing women with RPL for F5/F2 and subsequent treatment options have been reported. Unfractionated heparin and LMWH do not cross the placenta, removing much of the risk to the fetus.12,129 For carrier women, proposed treatment during pregnancy presents measurable risks. Although rare, anticoagulant- related maternal risks can be serious and include bleeding (e.g., injection site bruising, nosebleeds, and hemorrhage/vaginal bleeding), osteopenia, thrombocytopenia, and allergic skin reactions.119,129,131,138 LMWH is associated with a lower risk for heparin-induced thrombocytopenia and osteopenia129,138 but significantly increases the cost of care.12,47 Other drawbacks include discomfort from daily injections and inconvenience.12,47 A potential harm of testing is a false-positive result, possibly leading to unneeded treatment. Information is not available on potential psychosocial and ethical harms (e.g., acceptability and impact of testing, implications of results for family members, and societal cost).

In summary, the evidence is adequate to support the lack of safe and effective treatments that will lead to a reduction in future pregnancy loss among women with RPL.47,119,129,133,135,136 The RCTs have generally been underpowered for women with F5/F2-related RPL, and, therefore, the evidence is inadequate on the efficacy of these treatments in this subgroup. Given the lack of benefit, and the potential harms, F5/F2 testing and treatment in women with RPL is likely to be associated with a net harm. The certainty of evidence was rated as moderate. The certainty was not high because limitations in data quantity, quality, consistency, and generalizability reduce confidence, and new evidence could alter the conclusion.16

Discussion

Genetic and epidemiologic research suggests that RPL of maternal etiology is multifactorial, with risk determined by interactions of multiple genetic variants and environmental factors. Factors implicated in RPL recurrence risk include maternal age, race/ethnicity, number of previous miscarriages, and whether care is received early in pregnancy.127,128 The F5 and F2 variants have a low prevalence in the population, and women with these variants can have normal pregnancies, so have relatively small impact on the absolute risk for pregnancy loss. It is not surprising, therefore, that designing and conducting studies to measure the relatively small effects associated with these specific variants has proved challenging. Methodological problems in case-control and cohort studies on F5/F2 and RPL are described herein and in previous reviews44,46,47,48 and include variable case definitions and quality of diagnosis for RPL, ascertainment bias in obtaining cases/controls and cohorts, lack of power, poorly described and/or heterogeneous patient and control populations, and lack of information to assess bias and confounding variables.

Most analyses have been conducted in whites of northern European and Jewish ancestry. As our preliminary data suggest, results in Asian and Middle Eastern populations may differ in prevalence of at risk genotypes, effect size, or both. Kist et al.48 previously reported APCR pretesting and ethnicity (Israeli whites) as “confounders” (heterogeneity and significance not reported). However, APCR was used in four of the nine Israeli studies in that analysis, and we did not observe a difference between Jewish Israeli women and other whites of northern European ancestry. Most studies provide insufficient information to accurately assess maternal age or the number and timing of pregnancy losses as covariates. Data from the first trimester are particularly problematic, as few studies differentiate between very early anembryonic losses and postembryonic losses, and this may impact effect size.72 This and other meta-analyses have been unable to find significant differences in F5 or F2 ORs based on trimester or number of losses.43,45,46,59 Results from a recent meta-analysis49 of unselected cohort studies reported a lower effect size (F5 OR: 1.52; F2 OR: 1.13) but were generally consistent with our findings (ORs 2.03 and 1.68, respectively) on fetal loss occurrence rates in carrier women.

In summary, the data support an association between F5/F2 carrier status and pregnancy loss. However, important gaps in knowledge remain.

  • What information can be provided to women with RPL and a positive genotype about their risk for loss in subsequent pregnancies? Well-designed and adequately powered studies are needed in women with RPL to estimate the clinical sensitivity and specificity of F5/F2 testing for subsequent pregnancy loss, with stratification by important covariates (e.g., race/ethnicity, maternal age, gestational age, and number of losses).

  • Is there a safe and effective intervention that can be offered to women at risk that will increase the likelihood of a successful pregnancy? Regardless of F5/F2 genotype, there is no evidence for an effective intervention for women with RPL. Studies are needed to demonstrate a safe and effective treatment for women with RPL who carry a F5/F2 variant(s).

  • Information is also needed on the current utilization of tests, impact of positive test results on physician and/or patient/consumer decision making, and perception of risk related to treatment versus routine supportive care. Of concern are preliminary clinical surveys reporting that thrombophilia panels are ordered in as many as three quarters of pregnant women with RPL13,14 and that management is commonly modified in women who test positive.13,14,139,140 From a practitioner’s perspective, patients who have experienced RPL are more likely to request treatment, once found to have a genetic “abnormality” that could potentially explain the losses. The risks associated with treatment may be readily accepted without full consideration by patients desperate for a successful pregnancy. F5/F2 testing is also available direct to consumers, exposing an unknown number of women to potential harms.

Disclosure

The authors declare no conflict of interest.