Thrombophilic risk factors for retinal vein occlusion

The aim is to study risk factors for retinal vein occlusion (RVO), such as thrombophilic and cardiovascular risk factors (CRF). A retrospective consecutive case series of 60 patients with RVO was made, tested for CRF, hyperhomocysteinemia, lupic anticoagulant, antiphospholipid antibody and 5 gene variants: factor V (FV) Leiden (G1691A), factor II (PT G20210A), 5,1-methylenetetra-hydrofolate reductase (MTHFR; 677 C > T and 1298 A > C), plasminogen activator inhibitor 1 (PAI-1; 4 G/5 G). More than 1 CRF were present in 36 patients (60%), which had a significantly higher mean age at diagnosis (66.7 ± 12.9 versus 59.5 ± 13.7 with ≤1 CRF, [t(57) = −2.05, p = 0.045, d = 0.54). Patients with thermolabile MTHFR forms with decreased enzyme activity (T677T or C677T/A1298C) had a significant lower mean age [57.6 ± 15.1; t (58) = 3.32; p = 0.002; d = 0.846] than patients with normal MTHFR enzyme activity (68.5 ± 10.2). Regarding CRF and thermolabile forms of MTHFR, the mean age at diagnosis could be significantly predicted [F(2,56) = 7.18; p = 0.002] by the equation: 64.8 − 10.3 × (thermolabile MTHFR) − 5.31 × ( ≤ 1CRF). Screening of MTHFR polymorphisms may be useful in younger RVO patients, particularly when multiple CRF are absent.

When considering CRF (>1 CRF versus ≤ 1 CRF) and thermolabile forms of MTHFR (T677T and C677T/ A1298C), the mean age at diagnosis of RVO may be significantly predicted [F(2,56) = 7.18; p = 0.002] by the equation For instance, a patient with ≤1 CRF and a thermolabile form of MTHFR would have a predicted mean age for a RVO event at age of 49 (49.2). On the other hand, a patient with >1 CRF and a non-thermolabile form of MTHFR would have a predicted mean age of diagnosis at age of 65 (64.8).

Discussion
The pathogenesis of RVO is not completely understood yet. The condition may be due to a combination of local or systemic factors known as the Virchow's triad: (i) hemodynamic changes (venous stasis), (ii) degenerative changes of the vessel wall, and (iii) blood hypercoagulability 26 . Our study found a lower mean age at diagnosis in males. This may be related to the role of CRF, since retinal venous obstruction may be associated with significant cardiovascular morbidity 3,4,26 . We found that arterial hypertension and dyslipidemia were present in the majority (64.4%) of patients with >1 CRF. Patients with >1 CRF had a significantly higher mean age at diagnosis, as previously reported 27 . An extensive and expensive workup for thrombophilic disease may be useless in the vast majority of patients with RVO. Several reviews concluded that TRF screening is not cost-effective and should not be performed in all RVO cases 23,28,29 . However, TRF screening may be relevant in specific "atypical" cases, such as RVO in patients aged less than 60, in patients without CRF, when there is involvement of both eyes or when a family history of systemic thrombosis is present 22,24 . The results of our study agree with data previously published where TRF are particularly important in younger patients without known CRF 9,10,30 .
We found an increased prevalence of HH in patients with RVO. HH has been reported as an independent predictor factor for atherosclerosis and for thrombosis 31,32 . A review demonstrated that HH and anticardiolipin antibodies are associated with RVO 16 . The prevalence of MTHFR polymorphisms and the proportion of HH in normal patients varies among the different studies 12,17,19,33 . Some factors, like the ethnicity and geography, may be a bias when comparing the prevalence of MTHFR polymorphisms in different RVO studies. MTHFR T677T homozygosity was previously reported to be significantly more prevalent in patients with RVO 24 , including ischemic CRVO 32 . Thermolabile 677 C > T polymorphisms, such as T677T and compound heterozygous C677T/A1298C, along with 1298 A > C homozygosity, were found to be related with significant MTHFR enzyme decreased activity, mild HH or decreased folate levels 17,[34][35][36] , including in large studies 37 . In our study MTHFR T677T was found in 11.7% of patients, which is in accordance with the prevalence of 11.8% found in the general population of nearby Spain 18 . However, MTHFR compound heterozygosity C677T/A1298C was found in 26.7% of patients, more than previously reported for the general population in Canada 34 , but less than in south-eastern Europe 38 . Homozygosity C1298C was found in 5% of patients which is a percentage in-between data previously described by the aforementioned groups 34,38 . More importantly, we found a significantly lower mean age at diagnosis in patients with the homozygous thermolabile form T677T. Furthermore, it was possible to build a model where lower mean age at diagnosis may be significantly predicted in the presence of MTHFR polymorphisms associated with decreased enzyme activity and <1 CRF. Overall, more than the relative prevalence of polymorphisms found in RVO, it is the association of thermolabile MTHFR polymorphisms and younger age at diagnosis that seems to be more relevant. This is important, because the ophthalmologist may be the first physician to identify patients suffering from thrombophilia and prone to systemic complications 39  www.nature.com/scientificreports www.nature.com/scientificreports/ Our results partially agree with Russo et al. in that we did not found FV Leiden G1691A or PAI-1 (4 G/4 G and 4 G/5 G) to be correlated with RVO. Conversely, we did not find prothrombin (Factor II) gene mutation PT G20210A to be correlated with RVO, either 24,40 . The prevalence obtained for FV Leiden G1691A, PT G20210A and PAI-1 mutations 4 G/4 G or 4 G/5 G was similar to those reported for healthy people 8,24 and age was not a differentiation factor between subgroups. However, the score obtained for PAI-1 mutations (64.7%) agrees with datum previously reported for RVO in a prospective work where it was found statistically significantly superior to a control group 40 . Our results do not agree with previous data indicating lupic anticoagulant and antiphospholipid antibody associated with RVO 8,16 .
This study has limitations. It is an institutional retrospective, consecutive case series report. We collected more data on TRF from CRVO than from BRVO patients' files. This happened with other studies as well 40 . There are several possible explanations for this discrepancy with the incidence of either form of RVO 2 : (i) CRVO patients are more prone to be referred from the emergency department to the medical retina department, where the searching for TRF is based on, due to its worse prognosis and increased treatment burden 2 , (ii) due to the former reason, CRVO patients are more prone to stay for longer at the hospital visiting scheduling, increasing the outpatient hospital relation CRVO/BRVO when comparing with the general population, (iii) systematic search for TRF began with CRVO cases in our department. It is possible that the results of this study cannot be universally extrapolated due to the aforementioned geographic variations of MTHFR polymorphisms and to the geographic variation of HH 41 . Nevertheless, we have made an effort to compare our results with the results of other studies, including studies on the general healthy population of our own area. The size of our sample of 60 RVO cases is relatively small, due to the interruption of systematic search for MTHFR polymorphisms in RVO patients in our department, based on costs and on disputable usefulness for its use. The strengths of this study are related to age stratification of the results and the comparison with data from RVO studies and studies on healthy population of different geographic areas, including our own.
To conclude, it seems important to consider searching for TRF, including MTHFR polymorphisms in 'atypical' RVO patients, younger at diagnosis or lacking multiple CRF.

Methods
Retrospective institutional consecutive case series of RVO patients (CRVO and BRVO) admitted between 2013 and 2018. CRVO and BRVO were diagnosed by characteristic fundus features. CRF, such as arterial hypertension, hyperlipidemia, atrial fibrillation and diabetes, were collected, as well as acute cardiovascular and cerebrovascular events. The exclusion criteria included renal disease, cancer and medication with supplements or drugs acting on homocysteine levels. TRF screening was performed using the CVD-StripAssay (ViennaLab Diagnostics GmbH, Vienna, Austria). Briefly, genomic DNA was extracted from whole blood; the different gene sequences were simultaneously amplified using a multiplex polymerase chain reaction (PCR) with biotin-labeling of the products; the products were hybridized on StripAssay ® test strips and then detected by streptavidin-alkaline phosphatase. Total L-homocysteine was quantified in serum using the ARCHITECT one-step immunoassay (Abbott Diagnostics, Abbott Laboratories, Abbott Park, IL, USA) with Chemiluminescent Microparticle Immuno Assay (CMIA) technology. The results of TRF screening were collected from data files: MTHFR 677 C > T and 1298 A > C polymorphisms, factor V1691A (FV Leiden), PT 20210 A, PAI-1, HH (homocysteine level > 15 µmol/L) 15 . The study follows the Declaration of Helsinki and was approved by the local ethics committee of the Leiria Hospital Center. All participants provided informed consent. Data analysis was performed with "Statistical Package for the Social Sciences-IBM SPSS Statistics24 ® ". Two-tailed p-values < 0.05 [Confidence Interval (CI) 95%] were considered to be statistically significant. After testing for normality with the Shapiro-Wilk test, one-way ANOVA or independent sample t-test were performed. Two different effect size measures were calculated: partial eta squared (η p 2 ) [small 0.01, medium 0.06, large 0.14] and Cohen's d scores [small 0.2, medium 0.5, large 0.8].