Thromb Haemost 2006; 95: 922–3 Warfarin and other coumarin derivatives acting as vitamin K antagonists (VKA) are teratogenic and may induce the well-defined coumarin or warfarin embryopathy. There is no doubt about the developmental toxic potency of this drug group. However, as with many other drugs it took years or decades to find out more precisely when and how and to what extent a suspected drug acts as a developmental toxicant. Valproic acid for example was first seen as a novel antiepileptic without teratogenic effects in humans until spina bifida (1) and other birth defects were found to be associated with its exposure 25 years ago, and finally it became clear that it is the strongest teratogen among antiepileptic drugs (2), including mental effects in the child. Vice versa, lithium was suspected to be a strong teratogen decades ago, while today it is evident that less than one of 100 exposed foetuses develop teratogenic defects, like Ebstein anomaly of the heart (3). A similar change in risk estimation can be observed with coumarin derivatives. In handbooks of pharmacology one may still find a risk rate of 15–30% for birth defects after its use in the 1st trimester, more recent studies summarize some 5%. The “teratogenic window” is usually defined in the early 1st trimester during weeks 6–9, allowing the interpretation that the high-risk phase starts four weeks after conception, and a restart of anticoagulation may be safe four weeks later. The difficulties with analyzing drug effects during human pregnancy are due to the fact that we do not investigate in a laboratory setting, neither before licensing nor post marketing, since there are ethical obstacles against enrolling women of reproductive age in randomized controlled pregnancy outcome studies to test the embryotoxic potential of a drug. Therefore, we need other data sources, and these are very rare. To put it simply, there are two options: i) Data from birth (defect) registries with (retrospective) drug exposure ascertainment allowing case-control studies valuable for quantifying the relative risk for a particular birth defect in association with a particular drug exposure. However, this approach is useless to study abortion risk. ii) Prospective observational studies of exposed pregnancies identified before the outcome is known. One important source for the latter approach are follow up-data from “Teratology Information Services” (TIS). Data collection by TIS has the advantage of low costs and motivated responders, because the counselling is offered free of charge. Most callers to a TIS are highly satisfied and grateful for the information they received, so that they are willing to respond to a questionnaire sent to them to report on pregnancy outcome. TIS studies use an already existing infrastructure that selects “cases” with potentially teratogenic drugs, because the majority of TIS users mainly ask about insufficiently tested or potentially problematic drugs and not about those with evidence for low or no risk. In other words, TIS users and TIS research focus on the same spectrum of agents. In this issue of Thrombosis and Haemostasis, Schaefer et al. (see pages 949-57) (4) compared in a multi-centre TIS study 666 pregnant women to a non-exposed control group. It is the largest prospective cohort study on VKA in pregnancy and the first presenting large case numbers apart from acenocoumarol and warfarin or phenprocoumon and fluindione as well. Furthermore, in contrast to other published data almost all pregnancies were exposed from the beginning.The rate of major birth defects after 1st trimester exposure was significantly increased (OR 3.86, 95% CI: 1.86–8.00). However, there were only two coumarin embryopathies among 356 live births (0.6%). Apart from two children with diaphragmatic hernias – one was exposed beyond the sensitive period and therefore definitely unrelated to VKA exposure – the majority of birth defects were heterogeneous and neither indicative of well established symptoms of coumarin embryopathy nor resembling to one of the various published case reports discussing additional defects in context with coumarins. One could argue that a higher proportion of prenatally exposed newborns diagnosed as normal develop VKA-induced health problems only later in life, and are therefore missed by this study. However, other investigations suggest that the risk is remote that a healthy newborn develops late onset teratogenic effects from 1st trimester VKA exposure (5). Schaefer et al. correctly point out that the increased rate of spontaneous abortions they observed may indicate a drug-related (embryotoxic) effect which requires careful further investigation and evaluation. What is unique about this study is the introduction of the proportional hazard model for calculating the
Read full abstract
Sep 30, 2019
Duygun Altıntaş Aykan + 1
Open Access
