Association of ABO Polymorphisms and Cancer/Cardiocerebrovascular Disease: a Meta-analysis

Background: ABO gene polymorphisms have been reported to be associated with the risk of multiple cancers and cardiocerebrovascular diseases. However, the results remained controversial. In this study, we conducted a systematic review and meta-analysis to clarify the association between two SNPs (rs505922 and rs657152) in ABO gene and cancers/cardiocerebrovascular diseases. Method: All eligible case-control studies come from PubMed, Embase and Web of Science up to Jan. 1, 2019. Pooled odds ratios (ORs) with 95% condence intervals (CIs) were used to assess the corresponding associations. Sensitivity analysis, publication bias assessment, and heterogeneity test were performed using STATA 12.0. Results: A total of eighteen articles involving twenty-nine case-control populations were included according to inclusion and exclusion criteria. Eleven populations (16,929 cases and 23,941 controls) were used to evaluate the relationship between rs505922 and overall cancers and nine populations (22,275 cases and 71,549 controls) were included to assess the association between rs505922 and cardiocerebrovascular diseases. The results showed a signicant association between the rs505922 polymorphism and cancers (CvsT: OR=1.13, 95% CI=1.04-1.22, P=0.003), and cardiocerebrovascular diseases (OR=1.36, 95%CI=1.19-1.57, P<0.001). Four populations (5,158 cases and 7,021 controls) were included to evaluate association between rs657152 and cancers and ve populations (8,105 cases and 6,712 controls) were included to estimate the relationship between rs657152 and cardiocerebrovascular diseases. The result of meta-analysis reveals that rs657152 was signicantly associated with cancers (OR=1.16, 95%CI=1.09-1.24, P<0.001) and cardiocerebrovascular diseases (OR=1.54, 95%CI=1.24-1.92, P<0.001). Conclusion: Our study suggested that ABO polymorphisms


Background
Epidemiological studies have demonstrated that ABO blood groups were associated with several chronic in ammation related diseases, including cancers and cardiocerebrovascluar diseases [1][2][3]. The ABO blood group system is composed of complex carbohydrate structures that are biosynthesized by A-and B-transferases encoded by ABO gene. ABO gene is located at the 9q34 region of the chromosome and encodes enzyme glycosyltransferase with two main allele (A and B), a speci c glycosyltransferase catalyzes the covalent attachment of N-acetylgalactosamine or D-galactose to a common precursor side chain (H determinant), ultimately to an A or B antigen [4]. However, the relationship between histoblood group antigens and in ammation related diseases remains unknown and the regulatory mechanisms underlying ABO expression was still unclear. Several SNPs in the ABO gene have been suggested to be associated with increased risks to the development of cancer and cardiocerebrovascluar disease by genome-wide association studies (GWAS) and candidate gene studies [5][6][7][8]. Particularly, the most widely investigated SNPs of the ABO gene were rs505922 and rs657152 [9][10][11][12]. Meta-analysis that evaluated the relationship between rs505922 SNP and overall cancer have already been reported by Duan et al [13] four years ago. However, several new ndings of ABO gene SNPs have been reported in recent years and the relationship between SNPs in ABO gene and cancer/cardiocerebrovascular diseases risk was still unclear. Here, we have performed a metaanalysis including the newly published studies and evaluated the associations between polymorphisms of ABO gene and cancers/cardiocerebrovascular diseases.

Search strategy
Two investigators (Yanxia Li and Luyang Liu) performed a systematic literature search in three databases: PubMed, Web of Science and Embase to identify relevant articles published from the initial to Jan 1, 2019. The following search terms were used either separately or in combination: "SNP" "Polymorphism" and "Polymorphism, Single Nucleotide" "rs505922, rs657152" and "neoplasms" "carcinoma" "tumor" "cancer" and "vascular diasease" "cardiovascular disease" "cardiac-cerebral disease" and "ABO Blood-Group System" "Lewis Blood-Group System" "ABO". Once suitable studies were singled out from the search results, other potentially relevant articles were identi ed by cross-references within eligible studies. The references of each identi ed articles were also searched manually to identify eligible studies.

Selection criteria
The inclusion criteria for these studies were as follows: 1) Studies evaluating the association between rs505922/rs657152 variants and cancers or cardiocerebrovascular diseases; 2) The study design was a case-control study in humans; 3) Researches containing universal allele and genotype data; 4) Studies written in English. The exclusion criteria were as follows: 1) Studies did not describe the association of ABO gene polymorphism with cancers or cardiocerebrovascular diseases; 2) Systematic reviews or articles focusing on animals; 3) Studies that did not provide usable data for meta-analysis.
Data extraction and quality assessment Two investigators (Yanxia Li and Luyang Liu) independently extract data and verify the accuracy of the data. The following information was extracted from each article: rst authors, publication years, ethnicity, cancer types, and study design (hospital-or population-based),sample size of subject, number of cases and controls for each genotype, OR and 95% CI in allele model. To assess the quality of the study, we used the Newcastle-Ottawa Scale (NOS) with a nine-star system [14]; this scale assesses the quality of cohort and case-control studies. The highest score of NOS is nine stars: four stars for the selection process, three stars for exposure/outcome, and two stars for comparability. A score of seven or above was considered to be high-quality study.

Statistical analysis
We used odds ratios (ORs) and 95% con dence intervals (CIs) to assess the relationship between each SNP and the risk of cancers or cardiocerebrovascular diseases. The association was examined using allele model. The signi cance of the pooled OR was determined by the Ztest. Subgroup analyses were conducted according to cancer types, ethnic groups and sources of controls. Heterogeneity between articles was identi ed with Q-test and I 2 index, I 2 values of >50% indicated heterogeneity among studies. If there was heterogeneity (I 2 >50%) between studies, we used a random effect model (DerSimonian-Laird method), otherwise we used a xed effect model (Mantel-Haenszel method). Sensitivity analysis was performed to assess the effects of individual studies on pooled results and the stability of the results. We used Egger's and Begg's test to evaluate the publication bias, with a P>0.05 considered to be evidence for no potential publication bias. Trim and ll method was also applied in detecting publication bias. All the tests were two-sided and P<0.05 was considered to be statistically signi cant. Stata version 12.0 SE was applied to carry out the statistical analysis.
Meta-analysis of rs505922 polymorphism Meta-analysis was conducted to estimate the associations between rs505922 and cancer risk (  Table 4; Suppl Figure 1). However, no signi cant association was observed in Caucasian population(OR = 1.05, 95%CI = 0.97-1.13, P = 0.232). Sensitivity analyses were conducted by omitting each individual article to measure its speci c effect on the pooled ORs (Suppl Figure 2a). Sensitivity analysis plot indicated that no single study signi cantly affected the combined OR of SNP loci. Because of the heterogeneity of the research, we use the random effect model (allele model: I 2 = 79.1%) (Suppl Table 3). No signi cant publication bias was observed in any studies of SNPs ( Fig. 4a; Suppl Table 3). After the applying the trim and ll method, there is no change in the OR value after the combination, also indicating that the original result is stable (Suppl Figure 3).

Meta-analysis of rs657152 polymorphisms
Four studies reported the association between rs657152 and cancers risk. Our results showed that rs657152 was signi cantly associated with increased cancers risk in allele model (OR = 1.16, 95%CI = 1.09-1.24, P<0.001) (Fig. 3a; Suppl Table 3). Sensitivity analysis revealed no signi cant in uence on the pooled OR by any individual study (Suppl Figure 6a). We conducted Begg's and Egger's tests to assess the publication bias for these studies (Fig. 4c; Suppl Table 3). No evidence of publication bias was detected among the enrolled studies as shown in Table 3. According to the results of trim and ll method, the result was stable (Suppl Figure 7).
We also evaluated the relationship between rs657152 and cardiocerebrovascular diseases. A signi cant association between rs657152 and cardiocerebrovascular diseases under allele model were observed (OR = 1.54, 95%CI = 1.24-1.92, P<0.001) ( Fig. 3b; Suppl Table 3). Sensitivity analysis suggested that there was no signi cant change in the overall outcomes after removing any of the studies (Suppl Figure 6b). The Begg's and Egger's tests also showed that no meaningful publication bias were found ( Fig. 4d; Suppl Table 3). The result of the trim and ll method proves that the result was stable (Suppl Figure 8).

Discussion
In this study, we conducted a meta-analysis to clarify the relationship between ABO SNPs (rs505922 and rs657152) and cancer/cardiocerebrovascular diseases risk. Our results showed that these two SNPs were associated with cancer risk and also increased the risk of cardiocerebrovascular diseases.
Regarding to rs505922 SNP, our results showed that the variant type of rs505922 could increase the risk of overall cancer, suggesting a potential predictive ability of this SNP for cancer risk. When we conducted a subgroup analysis of rs505922 based on cancer sites and ethnicity, we found that there was no signi cant association between rs505922 and Non-pancreatic cancer subgroup or Caucasian subgroup. This result may be due to heterogeneity of cancer types or insu cient statistical power. Therefore, further studies with large samples size are warrant to evaluate the association between the rs505922 polymorphisms in Non-pancreatic cancers. On the other side, our data also showed that rs505922 was related to cardiocerebrovascular diseases. In subgroup analysis, we found that there was no association between rs505922 and cardiocerebrovascular diseases in Asian subgroup. Only two studies with small population were included in this analysis, further studies are needed in Asian population.
The SNP is located in the rst intron region of ABO gene, the protective T allele of rs505922 is in complete linkage disequilibrium (r 2 = 1.0) with the O allele, is marks of O allele. However, the regulatory mechanism underlying the expression of histoblood group antigens was unclear.
For rs657152, our study demonstrated that this SNP was associated with cancer/ cardiocerebrovascular diseases risk. However, only seven studies [9-12, 19, 21, 24] were reported and most of studies were conducted in Caucasian population. Therefore, more studies with different ethnic background and larger sample size are needed in the future. rs657152 is located in the intron area of ABO, thus it may affect gene transcription and shearing [23]. In addition, rs657152 has been found to be associated with several biological molecule, including LDL cholesterol [26], liver derived alkaline phosphatas [27], and IL-6 [28]. This implied that rs657152 may affect the occurrence and development of disease by in uencing these biological molecule. Further subsequent functional studies are warrant.
The underlying mechanism for the relationship between ABO blood group and cancer risk is still poorly understood [29]. It is reported that blood type may affect the progression and expansion of malignant tumors by altering the systemic in ammatory response [30]. Recent studies reported an association between polymorphisms at the ABO gene locus and circulating levels of tumour necrosis factor-alpha [31], soluble intercellular adhesion molecule (ICAM)-1 [32,33], E-selectin [34,35], and P-selectin [33]. These adhesion molecules were important mediators of chronic in ammation and immune cell recruitment [36]. They may provide a biological basis for the postulated in uence of ABO on cancer survival, by linking ABO blood group and tumour initiation and spread [29]. In addition, some research have shown that the structure of certain tumor antigens was similar to the structure of antigens of ABO blood group system. Smith and Prieto [37] suggested the Forssmann antigen which predominant in stomach and colon tumors, was almost structurally identical to the A antigen determinant. Blood group A carrier may have diminished tumor immune response due to reduced ability to recognize and attack tumor cells [38].
There was some evidence linking ABO blood group and cardiocerebrovascular diseases. Jenkins, P. V. et al. reported an association between ABO blood types and von Willebrand factor(vWF) and factor VIII(FVIII), both of which play crucial roles in the coagulation pathway [39]. Higher levels of vWF and FVIII has been observed in non-O blood type than O blood type [40]. Therefore, type O blood may be a risk factor for bleeding [41]. In addition, the non-O blood group has been shown to be correlated with higher total cholesterol and LDL-C levels [42], and the latest study proposed that approximately 10% of the effect of ABO blood group on coronary artery disease (CAD) susceptibility was mediated by plasma cholesterol levels [43].
Limitations in this study should be mentioned. First, the studies included in our meta-analysis were limited to published reports and English language studies. Unpublished reports or those published in non-English language studies were not included in the analysis. It would limit our sample size and publication bias might be exist. Second, both of the hospital based and population based case-control studies were included in our study. Therefore, selection bias would be exist compared to the meta-analysis only included population based case-control studies. Third, the limited number of published studies may in uence the reliability of our results. Finally, the lack of original data limited further evaluations of the potential gene-gene and gene-environment interactions.

Conclusion
In summary, the results of our meta-analysis revealed that rs505922 and rs657152 were correlated with an increased cancers risk. Due to most of the studies were conducted in pancreatic cancer type and Caucasian populations, further studies in multiple cancer types and multiple ethnic populations are needed. In addition, our meta-analysis also revealed that rs505922 and rs657152 associated with cardiocerebrovascular diseases.
However, owing to limited number of studies, further studies with larger samples size are warrant. This study can provide clues for further exploration of novel biomarkers with cancer/cardiocerebrovascular early-warning function.

Declarations
Ethics approval and consent to participate: Not applicable.

Consent for publication:
Not applicable.
Availability of data and material: Not applicable.

Competing interests:
The authors declare no con icts of interest.   Forest plot of the relationship between rs505922 polymorphisms and cancer (a) and cardiocerebrovascular disease (b) risk (allele model and random-effect model).