In this issue of the American Journal of Obstetrics and Gynecology , Khalil et al present very provocative data. They obtained images of low-risk women with magnetic resonance venography (MRV) 1-4 days after spontaneous vaginal delivery and found definite evidence of pelvic vein thrombosis in 30% of subjects, probable evidence in 27% of subjects, and possible evidence in another 10% of subjects. (None of the thrombi was occlusive.) Of the 30 participants, only one-third of the women had no evidence of thrombosis. In a previous article, the same investigators reported a similarly high rate of pelvic vein thrombosis after cesarean delivery. In that study, the investigators used MRV at the time of hospital discharge to obtain images of women who were at moderate-to-high risk for venous thromboembolism (VTE) and found definite evidence of pelvic vein thrombosis in 47% of subjects, probable evidence in 7% of subjects, and possible evidence in another 13% of subjects. (One of the thrombi was occlusive.) Of the 15 participants, only one-third of the women had no evidence of thrombosis. Therefore, in both studies, two-thirds of the subjects had evidence of thrombosis.

On reading these results, an obstetrician’s first reaction has to be incredulity, and he or she is prompted to ask several questions. Were the MRV studies performed correctly? The imaging was conducted according to accepted protocols in a large, university medical center that was known for conducting studies in thromboembolism. Therefore, one has to assume that the imaging was performed correctly. Is MRV actually sensitive and specific for the detection of VTE in the pelvis? Studies of MRV in nonpregnant patients, with contrast venography as the gold standard for the detection of pelvic vein thrombosis, revealed sensitivities of 100% and specificities of 95-98%. Dupas et al found a negative predictive value of 100%. Is postpartum asymptomatic pelvic vein thrombosis of any clinical consequence? Because none of these subjects went on to experience symptomatic VTE, we can assume that, in almost all cases, asymptomatic pelvic vein thrombosis that is detected by MRV after delivery is not of any clinical consequence; however, we know that postpartum pelvic vein thrombosis can be potentially lifethreatening. Unlike VTE outside of pregnancy, which, 99% of the time, arises in the deep veins of the lower extremities, 11% of diagnosed postpartum VTE originates in the pelvic veins and, without any warning signs, such as lower extremity pain or swelling, can embolize.

After delivery, women have a 20- to 80-fold increased risk of VTE compared with women who are not pregnant; in the first week after delivery, the risk is 100-fold higher. Pregnancy-related VTE accounts for 1.1 deaths per 100,000 deliveries or 10% of all maternal deaths in the United States. One-third to one-half of pregnancy-related VTE occurs after delivery. Forty percent to 50% of the postpartum events are pulmonary emboli. Therefore, 20-25% of all pregnancy-related VTE is postpartum pulmonary embolism; a substantial portion arises from the pelvic veins.

The best explanation for the investigators’ observation of a high incidence of nonocclusive thrombosis in the early postpartum period is that they have provided us with a window into a physiologic process that previously has been described poorly. We know that changes in circulation to the uterus are part of the process of involution. In the postpartum period, blood flow to the uterus declines from 15% of cardiac output to <1% of cardiac output. Because venous bleeding from uterine veins can contribute to bleeding from the placental bed, perhaps we can infer that thrombosis of pelvic veins is part of the physiologic process to interrupt blood flow at the placental site and facilitate involution. We recently presented data that levels of both fibrinogen and factor VIII fall in the first 24 hours after delivery, before returning to above-baseline levels by 72 hours after delivery. This initial nadir in clotting factor levels after delivery suggests consumption that is associated with clot formation. Perhaps these changes in clotting factor levels are biochemical evidence of the same physiologic process that Khalil et al have described.

If two-thirds of women have evidence of pelvic vein thrombosis after delivery, then what are the additional intermediary steps that result in the development of significant, symptomatic potentially life- or limb-threatening clot? Further studies are needed to delineate how long thrombosis remains or is visible in the pelvis after delivery. From Khalil et al, we can anticipate that thrombosis will be visible from 1-4 days after delivery. What about 3 weeks after delivery, when clotting factor levels have returned to normal? Or in 4-6 weeks after delivery, when lochia has ceased and the uterus has involuted? Are there biochemical markers that are correlated with the presence of clot?

What are the implications for clinical research? Pregnancy-related VTE and postpartum pulmonary embolism may or may not be amenable to prevention, with the use of the same thromboprophylactic strategies that are applied to other medical and surgical patients. Can the presence of asymptomatic pelvic vein thrombosis be used as a marker for symptomatic deep vein thrombosis in clinical trials of the efficacy of thromboprophylaxis or anticoagulation management? Clearly, what is needed, first, is an understanding of those additional intermediary steps that result in the development of significant, symptomatic clot.

In the meantime, what is a practitioner supposed to do if he or she suspects a pelvic vein thrombosis? After delivery, when signs or symptoms suggest deep vein thrombosis, the recommended initial diagnostic test is still compression ultrasonography of the proximal veins. When results are negative or equivocal and pelvic vein thrombosis is suspected, imaging with MRV is recommended. Computed tomography venography has not been shown to be superior to MRV in the diagnosis of deep vein thrombosis; compression ultrasonography, which cannot detect the full extent of thrombosis in the pelvis, actually has been shown to be inferior to MRI in the diagnosis of pelvic vein thrombosis. If thrombosis is detected on MRV after delivery, is one compelled to treat the patient? If a practitioner had a sufficient index of suspicion to seek imaging for the patient, then I believe the practitioner is obligated to treat. If the MRV is negative, one can at least have confidence in the negative predictive value of the test. What if a thrombosis, either occlusive or nonocclusive, was an incidental finding on magnetic resonance imaging that had been ordered for another reason? In all probability, the patient is hospitalized, is febrile, or has other risk factors for thrombosis and at least deserves some form of thromboprophylaxis.

Ideally, although MRV is expensive, the findings of this study should be replicated in other centers and in other populations. At present, none of these findings should change current practice; however, thanks to the investigators, we have new insights into postpartum physiologic conditions and have a number of additional questions to answer regarding the evolution of postpartum VTE.