West Nile Virus (WNV) is a mosquito-borne flavivirus that was introduced to the US in 1999 and is now the leading cause of arboviral encephalitis in the US.¹ Over the past 17 years, WNV has been responsible for more than 42,000 reported cases, of which 19,160 presented with neuroinvasive disease, 22,434 with West Nile (WN) fever, and 1,7957 had a fatal outcome.² Because WNV infection is asymptomatic in 80% of cases,¹ these numbers underrepresent the overall frequency of infection. A recent study projected that over 3 million persons have been infected with WNV in the US from 1999 to 2010, resulting in about 780,000 illnesses.^{3, 4}

To date there is no treatment and no vaccine to prevent the development of symptomatic infections in humans.⁵ Therefore it is essential to understand the reasons for symptomatic outcome and prognostic factors influencing severe disease outcome to better manage patients care. Underlying risk factors associated with severe WNV disease are older age, hypertension,⁶ immunosuppression,⁷ and genetic factors. Indeed host genetic background such as mutations or deletions in the CCR5,⁸ IRF-3 and OAS1 genes,⁹ and host HLA class I alleles, such as the HLA-A*68 and C*08 alleles,¹⁰ have been linked to the development of WNV neuroinvasive disease. Research has begun to clarify the relationship between immune protection and disease,¹¹ however the correlates of protective immunity in humans remain uncharacterized. It is, therefore, determinant to characterize aspects of the host-virus interaction to identify the host and viral parameters influencing WNV disease outcome.

Studies have suggested that WNV could bind to red blood cells (RBCs)¹² and WNV RNA was found to persist longer in whole blood than in plasma.^{13, 14} The mechanisms underlying compartmentalization of WNV in the RBC fraction of the blood are unclear and the consequences on pathogenesis are unknown. RBCs are nonnucleated biconcave discs that contain hemoglobin and have no cellular machinery to synthesize de novo virions, it is therefore unlikely that WNV would be able to replicate in RBCs. However, reticulocytes constitute 1% of the circulating RBCs and could support WNV replication. With approximately five million RBCs per cubic millimeter in adults and the ability to access vascularized tissues through the smallest capillaries, RBCs represent potent vehicles for pathogens to be carried around the body.

RBCs harbor glycosylated antigens that include the ABO blood group antigens. Those result from the expression of inherited specific genes which code for glycosyltransferases whose sequential action builds the H, A, or B antigens. RBCs also carry other type of antigens such as the rhesus (Rh) antigens. Those are non-glycosylated hydrophobic transmembrane proteins that are integral proteins of the RBC membrane and assembled into a multimeric complex with other glycoproteins such as the Rh50 glycoprotein, ¹⁵ CD47, glycophorin B, Duffy, and LW.¹⁶

Various studies have reported the association between specific blood groups and susceptibility to particular diseases or manifestations of disease caused by bacterial, parasitic, and viral infections such as SARS CoV,¹⁷ V. cholerae,^18–20 P.falciparum,^21–24 H.pylori and gastric ulcers or cancers,^25–28 norovirus,^29–32 rotavirus, dengue virus,³³ and hepatitis infections.^34–36 The present study investigates the association between ABO and Rh(D) blood groups and WNV disease outcome in a cohort of 374 WNV+ blood donors.

While race, gender, and age were not associated with WNV disease outcome in this cohort, blood group A was found at higher frequency in symptomatic than in asymptomatic WNV+ donors. Rh(D)-negativity was also more frequent in symptomatic than in asymptomatic WNV+ donors. The data suggest an association between blood groups and WNV disease outcome, with blood group A and Rh(D)-negativity being risk factors for the development of symptomatic WNV disease outcome. When the distribution of blood groups was compared within the WNV+ donor population and within the US blood donor population using the numbers previously reported by Garratty et al., an increased frequency of blood group A individuals was found within symptomatic WNV+ donors than within the US blood donor population⁴⁰.

While the comparison of blood group distribution within WNV+ groups experiencing different disease outcomes (asymptomatic versus symptomatic) is straightforward, the comparison of blood group distribution within WNV+ donor population and US blood donor population is interesting but limited. Indeed, comparing blood group distribution within WNV+ blood donor population enrolled by Blood Systems Inc. from 2004 to 2011 to blood group distribution as reported by Garratty et al. within US blood donors enrolled at five other blood centers between 1991 and 2000⁴⁰ may be biased by donor stratification resulting from collection based on blood group need that would potentially change over different period of time and geographic areas. Therefore, the comparison between blood group distribution within the WNV+ donor population and within the US blood donor population is worth reporting but the key findings are more related to the differences of blood group distribution identified within symptomatic versus asymptomatic WNV+ donors which would not be impacted by blood group recruitment biase.

In a previous study, higher WNV viral loads were observed over the three months post-index in blood group A WNV+ subjects than in blood group O WNV+ subjects.¹⁴ This was not the case for Rh(D)-negative donors, suggesting the mechanisms that underlie the association between blood group A and WNV disease outcome are different from the ones underlying the association between blood group Rh(D)-negative and WNV disease outcome. From an evolutionary stand point, it is however interesting to note that the frequency of Rh(D)-negative individuals is higher in white non-Hispanic, than in African, and Asian populations.⁴⁰ Therefore the world distribution of Rh(D)-negative humans is concentrated in geographic areas where WNV outbreaks have been documented for the increased severity of symptoms, as reported in the 90’s around the Mediterranean’s, and since 1999 in North America. WNV was first isolated in 1937 in Uganda from a women with fever and studies have documented the widespread distribution of WNV in Africa through seroprevalence studies conducted in the 50’s and revealing a high frequency of young adults immunized during childhood.⁴² The symptoms associated with WNV infection seemed more severe after WNV introduction in North America. Whether this is due to virus and human co-evolution and immunization earlier during childhood in endemic areas, or underreporting of more severe symptoms is unclear but it is worth to note that the frequency of Rh(D)-negative subjects is the highest in Causian populations of Europe and North America and the lowest in the Asian population where WNV infection remains unnoticed.

Numerous studies have been published to date on the association between blood groups and diseases.^{43, 44} However, most studies are based on statistical analyses and some are limited by sample size, racial heterogeneity, methodological approaches.⁴⁵ Therefore it is important to reproduce the findings in different cohorts. Noteworthy, similar results were recently presented by Politis et al.⁴⁶ Indeed, an association was reported between increased frequencies of blood group A and Rh(D)-negative individuals within 132 patients previously infected with lineage 2 WNV in Greece compared to healthy blood donors from the same WNV affected areas. This is supporting the findings of the present study.

The body of literature reporting an association between blood groups and diseases is growing. Some studies have clearly established the association between blood group O and reduced risk of severe malaria, with the theory that Plasmodium falciparum could have exerted a selective pressure on the human population in favor of individuals with blood group O in malaria-endemic regions.²² The mechanisms underlying the association between blood groups and malaria outcome are unclear but reports have suggested that blood group A antigen could serve as a co-receptor for Plasmodium falciparum,²¹ potentially explaining the association between blood group A and severe malaria.²² It was also recently reported that human macrophages have a higher avidity for Plasmodium falciparum-infected O erythrocytes than for Plasmodium falciparum-infected A or B erythrocytes and therefore subjects with blood group O might have accelerated parasite clearance compared to others.²⁴ In the present study, similar mechanisms could underlie the association between blood group A and symptomatic WNV disease. Indeed, it was shown that WNV interacts with LBP and αVβ3 integrin at the cell surface of host cells through its E glycoprotein,^{47, 48} and blood group A antigen could serve as a co-receptor for WNV. This would explain the findings that subjects with blood group A demonstrated higher viral loads in whole blood than subjects with other blood groups,¹⁴ however, this would not explain why the same assoication is not observed for subjects with blood group AB, whose erythrocytes express approximately half of the number of A sites than A erythrocytes⁴⁹ and this would not explain the association with Rh(D)-negativity. An alternative hypothesis would be that other glycosylated structures expressed differentially at the surface of erythrocytes from blood group A and Rh(D)-negative donors would facilitate WNV binding through glycan-glycan or lectin-glycan interactions in a Velcro-like interaction or would serve as receptors or co-receptors. While the potential mechanisms underlying such associations have been previously hypothesized and discussed,¹⁴ further investigation is needed to clearly understand the association between blood group and WNV pathogenesis.

The present study suggests an association between blood groups and WNV pathogenesis with A and Rh(D)-negative blood groups being potential new risk factors for the development of symptoms after WNV infection. This study adds to our previous report of WNV RNA persisting at higher levels in whole blood from blood group A WNV+ individuals. The findings reported in the present study are based on the statistical analysis of epidemiologic data collected from a cohort of individuals infected with lineage 1 WNV and are strengthened by the findings obtained on a different cohort of individuals infected with lineage 2 WNV.⁴⁶ Therefore the present findings support the interest in further pursuing the investigation of the mechanisms underlying the association between blood group A and Rh(D)-negativity and symptomatic WNV disease outcome.