Safety of Pregnancy After Cerebral Venous Thrombosis
A Systematic Review
Background and Purpose—Pregnancy and puerperium are associated with an increased risk of venous thrombotic events (VTEs), including cerebral venous thrombosis (CVT). We aimed to systematically review, in pregnant woman with previous CVT, (1) the risk of recurrence of CVT or other VTE; (2) the result of pregnancy; and (3) the association of antithrombotic prophylaxis with these outcomes.
Methods—We searched MEDLINE, Cochrane Database of Systematic Reviews, clinicaltrials.gov (from inception to July 2015), and reference lists of included studies and review articles. We considered observational studies reporting original data on the frequency of CVT or other VTE associated with pregnancy or puerperium in women with history of CVT.
Results—Thirteen studies were included. A simple pooled analysis of individual patient data and meta-analysis of proportions using a random effect model were performed. (1) 1 CVT recurrences/217 pregnancies (9 per 1000; 95% confidence interval, 3–33) and 5 noncerebral VTE/186 pregnancies (27 per 1000; 95% confidence interval, 12–61). (2) Pregnancy outcome: 33 spontaneous abortions/186 pregnancies (17.7%; 95% confidence interval, 13–24). (3) Data on the risk of CVT/extracerebral VTE according to antithrombotic prophylaxis was limited. Miscarriage did not differ significantly in women undergoing antithrombotic therapy or not (11.3% versus 18.8%; P=0.34).
Conclusions—In women with previous CVT, the absolute risk of pregnancy-related venous thrombosis is low but the relative risk of noncerebral VTE is 16-fold higher and the recurrence of CVT is 80-fold higher than the baseline risk described in general population studies. The rate of miscarriage is not significantly different from that estimated for the general population.
- intracranial thrombosis
- sinus thrombosis
- systematic review
- venous thrombosis
Pregnancy and puerperium are associated with an increased risk of venous thrombotic events (VTEs), both in the cerebral territory and in other venous territories. Cerebral venous thrombosis (CVT) accounts for 27% to 57% of all pregnancy-related strokes.1,2
Women with a previous episode of noncerebral VTE have a 3- to 4-fold higher risk of VTE during subsequent pregnancies than outside pregnancy.3 This is particularly relevant given that the leading cause of maternal mortality in developed countries is thrombosis and thromboembolism.4 History of previous extracerebral VTE is also associated with adverse pregnancy outcome.5,6
Cohort studies evaluating the risk of recurrent VTE during pregnancy in women with a history of extracerebral VTE in whom no prophylaxis was given have shown variable results.7–9 This issue becomes more challenging because the use of anticoagulant therapy during pregnancy is potentially harmful for both the fetus and the mother.
There is a paucity of studies addressing the safety of pregnancy in women with previous CVT, namely concerning the risk of CVT recurrence and other VTEs and the incidence of spontaneous abortion. The risks and benefits of antithrombotics for prophylaxis of VTE (including CVT recurrence) in these women are also not established.
We performed a systematic review of all published evidence to estimate, in pregnant woman with previous CVT, the frequency and relative risk of (1) CVT recurrence, occurrence of noncerebral VTE, (2) adverse outcome of pregnancy, and (3) to explore the influence of antithrombotic prophylaxis in these outcomes.
For the purpose of this systematic review, we followed Meta-Analysis of Observational Studies in Epidemiology (MOOSE)10 and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)11 guidelines.
We considered published observational studies reporting original data on the frequency of CVT or noncerebral VTE associated with pregnancy and puerperium in adult women with history of previous CVT. All observational study designs were accepted with the exception of case reports or selected case series. Articles in English, French, Spanish, and Portuguese were accepted. Series of patients with previous CVT without data suitable to derive the relative risk of CVT recurrence or noncerebral VTE associated with subsequent pregnancy were excluded, but no study was dismissed a priori because of poor quality.
We considered maternal outcomes (CVT and noncerebral VTE) and fetal outcomes (spontaneous abortion, induced abortion, fetal death, and preterm birth). We accepted authors’ definitions of the outcomes. To assess the quality of such definitions, we compared them with the standards described in the supplemental material. We also used authors’ definitions of outcomes for data extraction and analysis.
A systematic search using combinations of keywords was performed in the following databases, from inception to 4 July 2015: Pubmed, Cochrane Database of Systematic Reviews and clinicaltrials.gov. The developed search strategy for all database combined the terms sinus, thrombosis, cerebral, vein/venous or intracranial and pregnancy, delivery or puerperium, as described in the supplemental content. Potential eligible studies/selected studies’ reference lists were crosschecked for additional studies and cited reference research was done using studies’ titles and authors.
Reports retrieved through electronic identification were screened for potential eligibility by title and abstract analysis. The full text of potentially eligible studies was screened for appropriateness for inclusion by 2 independent authors (D.A.d.S. and J.M.F.). Disagreements were solved by consensus or by a third independent party (P.C.).
Data Collection Process
Two independent parties (D.A.d.S. and J.M.F.) extracted data from included studies full text to a specially designed form. Disagreements were solved by consensus or by a third independent party (P.C.). Corresponding authors were contacted for additional information if needed.
For the selected studies we extracted the following data: sample characteristics (country, study design, year of publication, study setting, number of patients and median/mean follow-up), pathogenesis of index CVT in women with subsequent pregnancy, total number of reported pregnancies in women with previous CVT, number of patients pregnant after CVT, specification of pregnancy as an assessed outcome in the methodology of study, CVT/noncerebral VTE definition, pregnancy outcome definitions, data suitable to derive the relative risk of CVT recurrence and noncerebral VTE in patients with previous CVT, data suitable to derive the relative risk of spontaneous abortion in patients with previous CVT and, when possible, timing of thrombotic event (gestational age or puerperium) and individual patient data on antithrombotic prophylaxis during pregnancy and puerperium.
Assessment of Study Quality
The studies eligible for inclusion in the meta-analysis were assessed for quality of study design and data reporting, according with the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach12 (Table I in the online-only Data Supplement).
The primary outcome was the frequency of CVT and noncerebral VTEs related to pregnancy among patients with previous CVT. The frequency of CVT and noncerebral VTE was estimated for different subgroup of patients according to antithrombotic prophylaxis, whenever possible. The secondary outcomes were the crude risk of abortion, stillbirth, and preterm birth in female patients with previous CVT, which was derived from raw data when no risk estimate was available. A sensitivity analysis excluding patients included in middle- and low-income countries was conducted (classified as developing countries according the World Bank Atlas method).13
Synthesis of Results/Statistical Analysis
We calculated the pooled frequencies of CVT, noncerebral VTE and spontaneous abortion and 95% confidence intervals (CIs). If no case–control studies were found, we a priori choose to compare the calculated pooled incidence with previously described incidences of CVT and VTE during pregnancy in the general population (Nationwide Databases from the Healthcare Cost and Utilization Project),14 using relative risk. A P value of 0.05 was considered statistically significant. 95% CI were calculated by the Wilson method.
We decided to perform as main statistical analysis simple pooled analysis of individual patient data, instead of meta-analysis of proportions, considering that the later method can produce misleading results for rare events, given its reliance on asymptotic statistical theory.15 Nevertheless, meta-analysis of proportions was also conducted, and it is available in the online-only Data Supplement. A random effect model for pooling proportions was used, with a continuity correction of 0.5. Heterogeneity among studies was assessed with Cochrane’s Q-test. The percentage of total variation across studies because of heterogeneity was evaluated by the I2 measure. A continuity correction was done and forest plots were drawn showing the variation of CVT and VTE rate among all studies, together with the pooled measure. All analyses were conducted using STATA version 13 (StataCorp, College Station, TX).
After removal of duplicate studies, we identified 523 studies using our search strategy in Pubmed, Cochrane Database of Systematic Reviews and clinicaltrials.gov (Data Supplement content). We excluded 447 after evaluation of titles and abstracts using the predefined inclusion and exclusion criteria. We retrieved 76 studies in full text for detailed evaluation and verification of overlaps in study populations. Additional studies were obtained from manually reviewing references of retrieved articles for full-text evaluation. Finally, we included 13 studies: 8 studies with long-term follow-up of patients with CVT16–23 and 5 studies evaluating complications during subsequent pregnancies in women with previous cerebrovascular disease.24–28 The remaining studies were excluded because of lack of information on the course of pregnancies occurring in women with previous CVT or for providing insufficient data for calculation of rates of thrombotic events and abortion. A MOOSE/PRISMA flowchart summarizes study selection (Figure).
These 13 studies were primarily conducted in Europe, apart from a single study from India and another from Brazil, which together contribute with the report of only 7 pregnancies. One study was multinational20 and included 7 patients from developing countries with pregnancy after CVT (Ferro JM et al, unpublished data, 2004). Most of the studies had a retrospective design (8/13). Details of the 13 included studies are displayed in Table 1.
Overall, 12 of 13 studies were scored as having moderate methodological quality (Table I in the online-only Data Supplement). None of the included studies was considered to have unacceptable quality. However, studies often lacked individual patient information about some features, such as pathogenesis of index CVT and antithrombotic prophylaxis. Even so, recurrence of CVT associated with pregnancy was consistently reported. More than half of the included publications were follow-up studies of established cohorts of patients with CVT, reducing the risk of reporting bias.
Rate of Recurrent CVT Associated With Pregnancy
We found reports of 217 pregnancies in which CVT recurrence was assessed, with description of 2 cases of recurrent CVT (9 per 1000 pregnancies; 95% CI, 3–33 per 1000). A meta-analysis of proportions was also conducted as a secondary analysis (Data Supplement content), with an overall proportion of CVT associated with pregnancy of 7.3 CVT per 1000 pregnancies (95% CI, −11 to 26; I2=0.00%, Figure I and Table II in the online-only Data Supplement).
One CVT case occurred during the first trimester in a patient with sickle cell disease who did not receive antithrombotic prophylaxis. The other CVT recurrence occurred during the second trimester in a patient whose index CVT was associated with oral contraception and protein S deficiency (Ferro JM et al, unpublished data, 2004).
Because no case–control studies were found, we compared the calculated pooled incidence of recurrent CVT with the incidence of CVT associated with pregnancy in the general population,29 as described in the Methods section of this article. As shown in Table 2, recurrence of CVT associated with pregnancy was 80× more likely to occur compared with the incidence described in the general population (95% CI, 20–317). However, as the Healthcare Cost and Utilization Project14 is conducted in the United States, we also performed a sensitivity analysis excluding patients from low- or middle-income countries (Table 1), in which the rate of CVT is reported as being higher. After exclusion of 14 pregnancies,16,20,22 the relative risk was 85 (95% CI, 21–339) in patients with previous CVT.
Information about antithrombotic prophylaxis was incomplete or not available in several studies. Although 1 recurrent CVT occurred in a woman who was not receiving prophylaxis (1/57) while 77 pregnancies and 89 puerperal periods were completed without CVT recurrence in women receiving low-molecular-weight heparin, the CIs of these estimates are too wide. The use of antiplatelet was reported in few cases (Table III in the online-only Data Supplement). Besides, the information provided in the publications did not allow stratification about the risk of recurrent venous events during pregnancy in women with specific risk factors for the index CVT.
Rate of Noncerebral VTEs Associated With Pregnancy in Women With Previous CVT
In the included studies, the occurrence of noncerebral VTE was assessed in 186 pregnancies, among which 5 events were reported (Table 1), namely: 1 case of VTE during puerperium, 1 case of DVT on the 11th gestational week, 2 cases of nonspecified extracerebral location (Ferro JM et al, unpublished data, 2004) and 1 patient with pulmonary embolism (27 VTEs per 1000 pregnancies; 95% CI, 12–61).
As a secondary analysis a meta-analysis of proportions was also conducted, with an overall proportion of noncerebral VTE associated with pregnancy of 13.2 VTE per 1000 pregnancies (95% CI, 8–35; I2=0.00%, Figure II and Table IV in the online-only Data Supplement).
Because no case–control studies were found, we compared the calculated pooled incidence of VTE in women with previous CVT with the incidence of VTE associated with pregnancy in the general population.30 As shown in Table 2, VTE associated with pregnancy was 16× more likely to occur in patients with previous CVT, compared with the incidence described in the general population (95% CI, 7–37). After the sensitivity analysis excluding 7 cases from low- or middle-income countries,20 the relative risk was 16 (95% CI, 7–39) in patients with previous CVT.
Information about antithrombotic prophylaxis was incomplete or not available in several studies. Although 2 of 3 noncerebral VTEs occurred in women not receiving any antithrombotic prophylaxis, these estimates do not have statistical power to detect differences between groups (Table III in the online-only Data Supplement). Also, the information provided in the publications does not allow any stratification about the risk of recurrent venous events during pregnancy in women with specific risk factors for the index CVT.
Fetal Outcomes in Women With Previous CVT
Among 186 reported pregnancies, 33 resulted in spontaneous abortion, with a calculated overall crude risk of 17.7% (95% CI, 13–24). We found no case–control studies. Despite being highly variable across studies, the rate of spontaneous abortion is usually estimated to occur in 10% to 15% of clinically recognized pregnancies31 and a previous studies based on self-reported data reported a rate of ≈20%.32
Although this outcome is probably not adequately analyzed in most studies, we also found report of 13 voluntary interruptions of pregnancy and 2 induced abortions for medical reasons, of which one was related with CVT recurrence and another with teratogenic risk in a patient receiving warfarin therapy ≤10 weeks of pregnancy. We did not find reports of fetal death or preterm delivery in women with previous CVT in the included studies.
The rate of spontaneous abortion was slightly higher in patients without antithrombotic prophylaxis, comparing with patients receiving heparin (19% versus 11%). However, the number of pregnancies in which there was information about prophylaxis was low and the low-molecular-weight heparin dosage was unknown in most cases (Table V in the online-only Data Supplement).
In this systematic review, we could include 13 studies with data on pregnancy-related CVT/noncerebral VTE occurrence in women with previous CVT, mostly describing western European patients followed at tertiary hospitals.
Regardless of antithrombotic prophylaxis, the pooled estimate for recurrent CVT and noncerebral VTE associated with pregnancy was 9 per 1000 pregnancies and 27 per 1000 pregnancies, respectively. Sample size, duration of follow-up, and antithrombotic prophylaxis were variable but data heterogeneity was low. These figures indicate a low absolute risk. However, in comparison with the general population, women with history of CVT seem to have an increased relative risk of extracerebral VTE (>10×) and recurrent CVT (80×) during subsequent pregnancies.
The frequency of spontaneous abortion was 18%. Although there was a trend toward lower rates of spontaneous abortion in women on prophylaxis with heparin, miscarriage rate does not differ significantly in women undergoing antithrombotic therapy or not.
This systematic review has limitations related to included studies and the method of analysis. Research on pregnancy complications and outcome in women with previous CVT is scarce, and it is hindered by the diversity of study designs and data collection methods. However, this bias is minimized by the fact that these are objective and definite outcomes (hard outcomes) and, therefore, less susceptible to recall bias.
We used as reference for incidence rates of pregnancy-related VTEs data from the United States concerning CVT and extracerebral VTE diagnosis in hospital discharges.29,30 This may also be subject to bias. However, most of the studies included in the review have been conducted in developed countries in Western Europe and a sensitivity analysis excluding studies performed in low- and middle-income countries was also performed.
In addition, it was difficult to account for antithrombotic prophylaxis in the analysis because of the lack of information in some studies and loss of power in the stratified analysis. Finally, a low number of studies contained individual patient data on the risks factors for the index CVT in patients with subsequent pregnancy. We consider that the current evidence concerning the use of antithrombotic prophylaxis in this setting is insufficient to draw firm conclusions applicable to current practice.
The most important implication of this review is the finding of the low absolute risk of recurrent venous thrombosis associated with pregnancy in women with previous CVT in developed countries, although the relative risk is much higher than those described for the general population, particularly about recurrent CVT. However, it should be considered that these results probably underestimate the true incidence because of the effect of antithrombotic prophylaxis in a large proportion of women.
This review also showed that, contrary to what has been suggested by previous series,26 the fetal outcome of pregnancy after a CVT seems to be favorable; with a rate of spontaneous abortion similar to others previously described in the general population.
Our results suggest that pregnancy should not be contraindicated in women with previous CVT. However, women at fertile age with history of CVT should be informed on the risks associated with future pregnancies
This review also points the need to perform larger studies of these outcomes in women with previous CVT, gathering data concerning the efficacy and complications of antithrombotic prophylaxis in this setting and its relation with particular risk factors identified at the index CVT.
Sources of Funding
This study was supported by Fundação para a Ciência e Tecnologia, Doctoral grant SFRH/SINTD/92677/2013.
Presented in part at the European Stroke Organization Conference (ESOC), Glasgow, United Kingdom, April 17–19, 2015.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.115.011955/-/DC1.
- Received October 27, 2015.
- Revision received December 7, 2015.
- Accepted December 17, 2015.
- © 2016 American Heart Association, Inc.
- Cantu-Brito C,
- Arauz A,
- Aburto Y,
- Barinagarrementeria F,
- Ruiz-Sandoval JL,
- Baizabal-Carvallo JF.
- Pabinger I,
- Grafenhofer H,
- Kyrle PA,
- Quehenberger P,
- Mannhalter C,
- Lechner K,
- et al
- Lewis G
- Brill-Edwards P,
- Ginsberg JS,
- Gent M,
- Hirsh J,
- Burrows R,
- Kearon C,
- et al
- 13.↵The World Bank. Classification of Economies. http://data.worldbank.org/about/country-classifications. Retrieved May 23, 2015.
- 14.↵Agency for Healthcare Research and Quality. Healthcare Cost and Utilization Project (HCUP)-HCUP Databases. http://www.hcup-us.ahrq.gov/databases.jsp. Retrieved December 29, 2014.
- Srinivasan K.
- Preter M,
- Tzourio C,
- Ameri A,
- Bousser MG.
- Ferro JM,
- Canhão P,
- Stam J,
- Bousser MG,
- Barinagarrementeria F
- Martinelli I,
- Bucciarelli P,
- Passamonti SM,
- Battaglioli T,
- Previtali E,
- Mannucci PM.
- Mehraein S,
- Ortwein H,
- Busch M,
- Weih M,
- Einhäupl K,
- Masuhr F.
- Lanska DJ,
- Kryscio RJ.
- Brigham SA,
- Conlon C,
- Farquharson RG.
- Buss L,
- Tolstrup J,
- Munk C,
- Bergholt T,
- Ottesen B,
- Grønbaek M,
- et al