Role of the ABO Blood Groups as a Risk Factor for Retinal Vein Occlusion

Retinal vein occlusion (RVO) is the second most prevalent retinal vascular disorder with a prevalence of around 0.4% [1]. Its complications such as cystoid macular edema and retinal ischemia may cause significant visual impairment. Oxidative stress is known to be involved in the development and persistence of these complications [2].

The ABO blood group describes the presence of A and B antigens on the surface of erythrocytes. The expression of A or B antigens is the results of allelic combinations on chromosome 9q34.2. It is well known as an important factor for blood transfusion, which in case of a mismatch may cause a potentially fatal adverse reaction [2].

The ABO blood groups are also associated with blood coagulation. This association is at least partially mediated by the glycosyltransferase activity. Glycosyltransferase is an enzyme catalyzing the transfer of carbohydrate from nucleotide sugar substrates to incomplete glycolipid or glycoprotein acceptors. The O allele contains a deletion leading to a loss of enzymatic activity [2].

Additionally, the O blood group has been reported to reduce plasma levels and biologic activity of von Willebrand factor (vWF). Plasma levels are approximately 25% lower in O blood types. One possible explanation may be that A and B antigens reduce ADAMTS13 binding to vWF. ADAMTS13 cleaves high-molecular-weight vWF to low-molecular-weight vWF, leading to a reduced tendency of vWF-associated platelet aggregation and thrombus formulation [3]. VWF plays a major role in blood coagulation by binding to Factor VIII, thereby increasing its half-life from 1–2 h to 8–12 h, and binding collagen and multiple platelet receptors. Reduced vWF activity may thus prevent platelet aggregation and thrombosis [4].

The ABO blood group is a well-known risk factor for thrombotic vascular diseases. The non-O blood type has been shown to be an independent risk factor for stroke [5, 6], coronary artery disease [7, 8], myocardial infarction [9, 10], and venous thromboembolism [11‒13]. Interestingly non-O blood groups were also associated with hyperlipidemia and hypertension [2].

To our knowledge, there are only two studies reporting on ABO blood group and RVO. White [14] reported over 40 years ago a similar distribution of ABO blood groups in patients with RVO and control patients. A recent publication by Borella et al. [15] with a larger sample size found that non-O blood groups had an increased risk for the presence of RVO with an OR of 1.42 (95% CI: 1.01–2.01). In this study, we studied (i) an association between blood group and the presence of RVO and (ii) whether this association was associated with other RVO risk factors.

We included all 485 RVO patients and 295 control subjects who were recruited in this case-control study. RVO patients were diagnosed based on fundus examination, optical coherence tomography using the Spectralis® OCT device (Spectralis; Heidelberg Engineering, Heidelberg, Germany), and fluorescein angiography (FLA) using the Spectralis® angiography device (Spectralis; Heidelberg Engineering). FLA was not performed in patients with known allergies to fluorescein. Control subjects scheduled for cataract surgery at our outpatient department were recruited as controls. The presence of RVO in control patients was excluded by fundus examination and optical coherence tomography. Subjects with age-related macular degeneration, choroidal neovascularisation, history of retinal detachment, and diabetic macular edema were excluded.

After written informed consent was given, blood samples were drawn for determination of subjects’ polymorphism. Patients’ medical history was taken for arterial hypertension, presence of diabetes mellitus, stroke, myocardial infarction, deep vein thrombosis, pulmonary embolism, smoking history, height, weight, and current medication.

ABO genotypes rs8176719 (the O allele) and rs8176746 (B allele) were determined by fluorogenic 5′-exonuclease assays (TaqMan^TM) with primer and probe sets designed and manufactured by ThermoFisher (LifeTech Austria, Vienna, Austria). Endpoint fluorescence was measured in a FlexStation 3 plate reader (Molecular Devices, San Jose, CA, USA). Fluorescence data were exported into Excel format and analyzed as a scatter plot. The ABO serotype was obtained from the ABO genotype according to these two polymorphisms (shown in online suppl. Table 1; for all online suppl. material, see www.karger.com/doi/10.1159/000526874).

All statistical analyses were performed with Stata 15.1 (Stata Corp., Houston, TX, USA). Continuous variables were summarized as medians [25th–75th percentile], and count data as absolute frequencies (column %). The distribution of continuous and categorical variables between different groups (such as patients vs. controls or subjects with vs. without ABO blood group O) was evaluated with rank-sum tests, χ² tests, and Fisher’s exact tests, as appropriate. The main association between ABO serotype and patient/control status was evaluated within a univariable logistic regression model with patient/control status as the dependent variable and binary ABO blood group (O vs. others) as the explanatory variable. Subsequent multivariable adjustment was performed for age and sex (an approach chosen instead of age and sex matching of controls) and for variables that were statistically significantly distributed between the blood groups and/or between patients and controls, respectively. In a hypothesis generating exploratory analysis, the association between ABO blood group O and patient/control status was evaluated within subgroups defined by six pre-specified variables (hypertension, diabetes, gender, venous thromboembolism, C-reactive protein level, and age). Missing data were reported in Table 1, and a complete case analysis was performed.

Seven-hundred-eighty subjects were included in the analysis (shown in Table 1). Baseline covariables including age, selected comorbidities, and co-medication were generally well-balanced between RVO patients (n = 485, 62%) and control subjects (n = 295, 38%). However, patients had a significantly higher prevalence of a history of stroke and venous thromboembolism (VTE) than controls. In contrast, control subjects had a higher prevalence of co-medication with lipid-lowering agents.

Fifty percent, 12%, 7%, and 31% of the study population had ABO blood groups A, B, AB, and O, respectively. Baseline characteristics were highly similarly distributed between patients with ABO blood group O (n = 242, 31%) and patients with all other blood groups (n = 538, 69%), respectively (shown in Table 2).

We did not find an association between a non-O blood group and being an RVO patient. In detail, the proportion of RVO patients was similar among subjects with ABO blood group O versus subjects with non-O blood groups (63% vs. 61%, p = 0.693). In logistic regression, non-O blood group was associated with 1.1-fold higher odds of being a patient (odds ratio = 1.06; 95% CI: 0.78–1.45, p = 0.693), and this lack of association prevailed upon multivariable adjustment (shown in Table 3). The non-O blood group was neither associated with uni- or bilateral RVO, an age ≤ and >55 years, nor with the type of RVO.

In an exploratory analysis, we gauged the potential of five pre-specified variables to modify the association between ABO blood group and patient/control status. Here, the lack of association between ABO blood group and patient/control was similar in subjects with and without arterial hypertension (interaction p = 0.439), with and without diabetes mellitus (interaction p = 0.554), males and females (interaction p = 0.417), with and without a history of VTE (interaction p = 0.127), and an age ≤ and >55 years (interaction p = 0.164), respectively.

In this study ABO blood group was not associated with RVO. This lack of association remained after multivariable adjustment for age, gender, history of stroke and venous thromboembolism, and lipid-lowering co-medication.

These results are in contrast to the association found by Borella et al. [15]. This disparity may be due to different inclusion criteria. Borella included only patients referred to the Thrombotic and Hemorrhagic Diseases Unit. These patients may have additional thrombotic risk factors, and thus the patient selection may be potentially biased. Our study included all RVO patients presenting at the department of ophthalmology, and thus our patient selection might be less biased. Further, the study from Borella adjusted for the most common thrombophilic diseases but not for cardiovascular diseases. As we know, especially cardiovascular risk factors are associated with the presence of RVO.

This study did not investigate thrombophilic factors associated with the ABO blood group such as factor VIII and vWF. Previous studies demonstrated that the blood-clotting protein factor VIII was increased in patients with RVO [16, 17] and vWF increased in CRVO patients [18]. However, another study by Boyd [19] could not confirm these findings. It has been difficult to determine to what extent thrombophilia plays a role in RVO, and thus it was postulated that risk factors should be screened in selected patient groups such as younger patients, patients with no systemic risk factor, or bilateral RVO [17, 20].

Therefore, we investigated in a subgroup analysis, whether ABO blood group played a role in younger patients (≤55 years), bilateral RVO and type of RVO (i.e., BRVO, CRVO, or hemiretinal RVO). This study found no significant association in these selected patient groups.

A recent study could show that ABO blood group has an additive effect with other thrombophilic factors in thromboembolism [21]. Thrombophilic factors such as factor VIII and vWF have been associated with RVO, but no investigated in this study. Future studies should investigate a possible additive effect of ABO blood group with thrombophilic factors in RVO.

To ensure data quality, we compared the ABO distribution in this study to the Austrian-blood donor registry with around quarter of a million blood donors. The distribution of the blood groups (O 37.4%, A 42.7%, B 13.7%, and AB 6.3%) was similar to ours (p = 0.47) [22].

For SNP analysis, our patient number is quite small, as genetic polymorphism often have very subtle influences. In a post-hoc power analysis, we found that the difference in ABO blood group found in this study would require the inclusion of 37.524 patients in order to become significant. Further, as Borella et al. [15] found a significant association, further studies are required to clarify the role of ABO in RVO.

This study did not find an association between non-O blood groups, VTE, and cardiovascular disease. These diseases were not our end points, so we can make no conclusion about non-O blood groups as a risk factor. However, this might hint towards a too small sample size.

Further, this study used two SNPs, namely rs8176719 and rs8176746, to determine ABO blood group. If there is a G at the rs8176719 site, it encodes blood group type A or type B. If the G nucleotide is removed in both copies, so homozygous for this deletion, it encodes type O allele. However theoretically, a person with G/G genotype can still be type O due to brought about mutations in other SNPs.

This largest study to date does not support the concept that ABO blood group is associated with the presence of RVO.

We want to thank our biomedical analysts for their help in processing the blood samples.

This study complies with the Declaration of Helsinki. Written informed consent was obtained from all participants. This study protocol was reviewed and approved by the Ethics Committee, Medical University Graz, approval number 1243/2016.

The authors have no conflicts of interest to declare.

The authors acknowledge the financial support by the Medical University of Graz for open access publishing fees.

The authors have fulfilled the following criteria for authorship. Laura Posch-Pertl and Martin Weger: conception and design of work, data acquisition, and drafting the work. Wolfgang List, Monja Michelitsch, Felix Innauer, and Silke Pinter-Hausberger: data acquisition and critical revision of the work. Florian Posch: data interpretation and drafting and critical revision of the work. Wilfried Renner: data analysis and drafting and critical revision of the work. All the authors have approved the final version of the manuscript and are accountable for all aspects of the work.

Research data are not publicly available but are made readily available to all by Laura Posch-Pertl ([email protected]).