Prevalence and Geographical Variation of Prothrombin G20210A Mutation in Patients with Cerebral Vein Thrombosis: A Systematic Review and Meta-Analysis

Objectives To compare the prevalence of prothrombin G20210A in patients with objectively confirmed cerebral vein or cortical vein thrombosis against healthy controls, and evaluate geographical variations. Design Systematic review and meta-analysis of case control studies. Methods We conducted a systematic review of electronic databases including MEDLINE and EMBASE. The main outcome was the prevalence of prothrombin G20210A in patients with objectively confirmed cerebral vein or cortical vein thrombosis; we also analyzed individual country variations in the prevalence. The random-effects model OR was used as the primary outcome measure. Results In total 19 studies evaluated 868 cases of cerebral venous thrombosis and 3981 controls. Prothrombin G20210A was found in 103/868 of the patients with cerebral venous thrombosis and 105/3999 of the healthy controls [random effects pooled OR 5.838, 95% CI 3.96 to 8.58; I217.9%]. The prevalence of prothrombin G20210A was significantly elevated in Italian studies (OR 9.69), in Brazilian studies (OR 7.02), and in German studies (OR 3.77), but not in Iranian studies (OR 0.98). Conclusion Prothrombin G20210A is significantly associated with cerebral venous thrombosis when compared to healthy controls, although this association is highly dependent on the country of origin.


Introduction
Cerebral venous thrombosis is rare thrombotic condition commonly associated with the presence of thrombophilia [1].Prothrombin G20210A is one of the most common thrombophilias associated with venous thrombosis, including cerebral venous thrombosis [2].In patients with deep vein thrombosis or pulmonary embolism the prevalence of prothrombin G20210A is highly dependent on the country of origin or ethnicity [3][4][5][6][7] but this variation has not been evaluated in patients with cerebral venous thrombosis.To address this issue, we examined the prevalence and geographical variation of carriers of the prothrombin G20210A in patients with cerebral vein or cortical vein thrombosis.

Methods
We conducted a systematic review of electronic databases including MEDLINE and EMBASE to assess the prevalence of Prothrombin G20210A in patients with cerebral vein thrombosis (S1 Text).The timeframe of the search was from Jan 1995 to February 2015 and was designed with the support of a librarian from the Ottawa Hospital Health Services.The search was supplemented by hand-search of relevant articles, abstract books from international meetings and published reviews.

Study Selection
Case control studies were included if they reported the prevalence of prothrombin G20210A in patients with objectively confirmed cerebral vein or cortical vein thrombosis.All potentially relevant articles were reviewed in full length to ensure that they satisfied the inclusion criteria: 1) enrolment of non-paediatric patients with cerebral vein or cortical vein thrombosis; 2) the study reported the diagnostic test used to confirm the diagnosis (including digital subtraction angiography, MRI, CT angiography or autopsy); 3) Prothrombin G20210A genotyping was available for all participants; 4) the numbers of cases and controls with and without prothrombin G20210A were provided in the article.Studies were excluded if their subjects did not receive objective testing for the prothrombin G20210A mutation, included a paediatric population or included patients in which the diagnostic methods was not reported or patient with self-reported cerebral venous thrombosis.

Data Extraction and Quality Assessment
Two reviewers (J.G. and E.G.) independently assessed the eligibility of all articles identified in the initial search strategy and used the Newcastle-Ottawa Quality Assessment Scale for Observational Studies to assess the methodological quality of the selected studies.A third reviewer adjudicated all discrepancies if needed (A.B.)

Outcome Measure
The primary outcome measure was the odds ratio (OR) for the prevalence of prothrombin G20210A patients with cerebral vein or cortical vein thrombosis as compared with healthy controls.We also aimed to analyze the prevalence of prothrombin G20210A mutation across different countries, if two or more studies were available.

Data Synthesis and Analysis
The meta-analysis was conducted in accordance with the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement (S1 and S2 Tables presents the meta-analysis of genetic association studies checklist).The random-effects model OR was used as the outcome measure, along with the corresponding 95% confidence intervals (CIs).The I 2 statistic was used to quantify heterogeneity among the pooled estimates across studies.An I 2 value less than 25% was considered low-level heterogeneity, 25% to 50% as moderate-level, and greater than 50% as high-level.Homozygote and heterozygote carriers of prothrombin G20210A were analyzed as one group due to the rarity of homozygote carriers.A funnel plot and test for bias using the Harbord and Egger test [8].We also added a L'Abbé plot which is useful for exploring heterogeneity and identifying outlying trials in a meta-analysis [9].The statistical analysis and graphs were performed using MedCalc Statistical Software version 12.7.7 (MedCalc Software, Ostend, Belgium; http://www.medcalc.org;2013) and the bias analysis was performed with STATS direct (StatsDirect Ltd.StatsDirect statistical software.http:// www.statsdirect.com.England: StatsDirect Ltd. 2013).
In total the 19 studies evaluated 868 cases of cerebral venous thrombosis and 3999 controls (see Table 3 for a more comprehensive review of the number of patients included in each study individual OR and bias assessment).Prothrombin G20210A was found in 103/868 of the patients with cerebral venous thrombosis and 105/3999 of the controls [random effects pooled OR 5.838, 95% CI 3.96 to 8.58; I 2 17.9%] (Fig 2) with low evidence of bias (Figs 3 and 4).As shown in Table 3 the addition of the two relevant studies not reporting the diagnostic methods used did not modify the random effects pooled OR [13,26].

Discussion
Our systematic review confirmed a significant association of prothrombin G20210A with cerebral venous thrombosis but this association is highly dependent to the country of origin.Individual country meta-analysis suggested significant association between prothrombin G20210A and cerebral venous thrombosis in studies conducted in Italy [17,21,27,31], Brazil [18,25,28] and Germany [10,20,24].Consistently with prior literature [3] suggesting a low prevalence of the condition in those from Asian or African descent, none of the studies included from India  [33], Iran [16,23] or Tunisia [14]reported an association between prothrombin G20210A and cerebral venous thrombosis.For the first time, and to the best of our knowledge, information about the geographical variations of prothrombin G20210A with cerebral venous thrombosis has been presented.Similar to deep vein thrombosis and pulmonary embolism the question that remains unanswered is the value of screening for prothrombin G20210A with cerebral venous thrombosis in order to predict who is at high risk of recurrent venous thromboembolism?Our results would suggest that given it high prevalence in countries such as Italy or Brazil, it would be reasonable to screen for this condition, especially in those with unprovoked events [34].On the other hand, evidence suggests that the rate of recurrent venous thromboembolism associated with prothrombin G20210A gene mutation is low in those with deep vein thrombosis or pulmonary embolism [35,36].Two our knowledge only two studies, conducted in Italy, have reported the rate of recurrent venous thromboembolism in patients with prothrombin G20210A and cerebral venous thrombosis who discontinue anticoagulation [26,37], none of them showed an increased risk for recurrent venous thromboembolism in carriers of prothrombin G20210A diagnosed with cerebral venous thrombosis.Other systematic reviews have addressed the prevalence of prothrombin G20210A in patients diagnosed with cerebral venous thrombosis [2,38].Our results confirms prior findings in other systematic reviews [2,38] suggesting a significant association between prothrombin G20210A diagnosed with cerebral venous thrombosis and provide.Also for the first time information about geographical variations associated with the prevalence of prothrombin G20210A in patients with cerebral vein thrombosis is presented.As previously suggested in patients with deep vein thrombosis or pulmonary embolism [3][4][5][6][7], we found that the prevalence of prothrombin gene mutation was higher in southern European countries, and no association in Asian countries.
Our systematic review has limitations.First, we did could not evaluate the prevalence of prothrombin gene mutation in patients with unprovoked cerebral venous thrombosis.Second, not all the studies used matching and, as such, potentially introduced bias.Third, homozygous prothrombin G20210A could not be assessed owing to the small number of patients affected.Fourth, our individual country analysis did not take into different genetic populations who might live in a country [39].Fifth, we did not adjust for important risk factors (such as provoked cerebral vein thrombosis vs. unprovoked; female vs. male) which could introduce some bias.Finally, most of the studies included did not reported on the Hardy-Weinberg equilibrium and given that only two of the studies reported on it we did not conducted a sensitivity analysis [40,41].
In conclusion, our systematic review confirmed a significant association of prothrombin G20210A with cerebral venous thrombosis but suggest that this association is highly dependent to the country of origin.More studies should evaluate the role of prothrombin G20210A as a predictor of recurrent venous thromboembolism in patients with cerebral venous thrombosis, especially in those countries where its prevalence high prevalence.

Table 1 .
Characteristics of the studies included.

Table 3 .
Meta-analysis and individual study data.