Contributions of PARP‐1 rs1136410 C>T polymorphism to the development of cancer

Abstract Poly(ADP‐ribose) polymerase‐1 (PARP‐1) is a nuclear chromatin‐associated enzyme involved in the DNA damage response. SNP rs1136410 C>T, the most studied polymorphism in PARP‐1 gene, is highly implicated in the susceptibility of cancer. However, the roles of PARP‐1 rs1136410 C>T on cancer risk vary from different studies. We comprehensively screened all qualified publications from several databases, including PubMed, EMBASE, MEDLINE, CNKI and Wanfang. The searching was updated to April 2020. Our meta‐analysis included 60 articles with 65 studies, comprised of a total of 23 996 cases with cancer and 33 015 controls. Overall, pooled data showed that the PARP‐1 rs1136410 C>T polymorphism was significantly but a border‐line associated with an increased risk of overall cancer (CC vs. TT/TC: OR = 1.11, 95% CI = 1.00‐1.24; C vs T: OR = 1.07, 95% CI = 1.01‐1.14). Subgroup analysis indicated that rs1136410 C allele contributed to high risk among gastric, thyroid, and cervical cancer, but lower risk among brain cancer. Furthermore, increased cancer risk was detected in the subgroups of Asian, controls from population‐based design studies, and HWE ≤ 0.05 studies. Sensitivity analysis and Egger's test showed that results of the meta‐analysis were fairly stable. The current study indicated that PARP1 rs1136410 C>T polymorphism may have an impact on certain types of cancer susceptibility.

(Val762Ala) is a non-synonymous polymorphism that could change valine to alanine. PARP-1 rs1136410 T>C genetic polymorphism was previously investigated in various types of cancer. However, the results of epidemiological studies are inconsistent and contradictory. The present work aims to fill this gap in the literature by presenting the latest updated meta-analysis of the available evidence in elucidating the relationship of PARP-1 rs1136410 T>C and the risk of cancer.

| Publication search
PubMed, EMBASE, MEDLINE, Wanfang, and CNKI were searched for English/Chinese-language articles published from January 1990, through April 2020. The following syntaxes were used: (a) PARP-1 or poly (ADP-ribose) polymerase 1 or PARP1 or ADPRT or ADPRT1 or rs1136410 C>T; (b) SNPs or polymorphisms or polymorphism or variants; (c) cancer or cancers or carcinoma or tumour or neoplasm. To obtain other appropriate publications, we also manually examined the references of the selected articles. Eligible criteria were as follows: (a) case-control study; (b) assessing rs1136410 C>T and cancer risk; and (c) enough information in allele frequency. Editorials, reviews, meta-analysis, case-only studies, duplicate studies were ruled out.

| Data extraction
For each included study, a reviewer (H. Li) abstracted relevant study characteristics: (ie authors name, publication year, cancer type, ethnicity of the study subject, control source, genotype method, score, allelic frequency) and classified the data into a structured form.
Another reviewer (Y. Zha) checked all data for completeness and accuracy. Disagree parts (the conflicted information extraction by the two authors) were resolved through discussion until consensus was made.

| Statistical methods
We used a chi-square test to determine if genotype frequencies in controls of each study conformed to Hardy-Weinberg equilibrium (HWE). P > .05 indicates not violating HWE. The association between PARP-1 rs1136410 C>T and cancer risk were assessed by calculating ORs with the corresponding 95% CIs. Significant heterogeneity exists if I 2 > 50%. If so, the random-effect model was adopted, otherwise, the fixed-effect model was used. Subgroup analyses by ethnicity, cancer type, source of control, and HWE in controls were performed to detect the source of heterogeneity. We also assessed the quality of each included study, the detailed method was described elsewhere. 7 Sensitivity analysis was performed by re-calculating the overall ORs when each study is removed at a time.
Egger's regression test showing the funnel plot asymmetry was conducted to determine publication bias. STATA software version 11.0 (Stata Corporation, College Station, TX) was used for statistical analysis. All the statistics were two-sides with P value < .05 implies a significant finding.

| Study characteristics
Initial retrieval from PubMed, EMBASE, and MEDLINE databases got a total of 298 potentially relevant published records. We also obtained 12 articles from CNKI and Wanfang database. After titles and abstracts screening, 248 not relevant records were excluded. 4 articles were excluded due to being covered by others. A total of 58 case-control design articles with reporting frequency of genotypes were included. Additionally, 2 eligible articles were also identified from references cited in the obtained articles. Ultimately, 65 studies from 60 publications were included (Table S1). A total of 23 996 cases with cancer and 33 015 controls were enrolled into this study for analysing, shown in Table S2 and Figure S1.

| Quantitative analysis
The main results of the meta-analysis were shown in Table 1 and 95% CI = 1.01-1.14), but not in other models. However, we observed that polymorphism rs1136410 C>T could confer to increased risk in gastric cancer; thyroid cancer; cervical cancer, whereas it is associated with decreased risk of brain cancer. Subgroup analysis by ethnicity showed that rs1136410 C allele had a contributing effect on cancer in Asian. In Caucasian and African, no significant association was detected. In terms of source of controls, population-based controls were associated with the increased risk of cancer. Further subgroup analysis by HWE in controls revealed that rs1136410 C>T could not impact cancer risk in studies of HWE > 0.05, whereas studies of HWE ≤ 0.05 could impact cancer risk.

| Heterogeneity and sensitivity analysis
The Q test (P < .001) implied an existence of significant heterogeneity under all the genetic models. Therefore, a random-effect model was applied to produce ORs and 95% CIs. In addition, the sequential sensitivity analysis was conducted to give an evaluation of the impact of a single study on the pooled ORs. The data of meta-analysis is of great reliability, as no statistical fluctuation of the pooled ORs incurred after omitting in each study ( Figure S2).

| Publication bias
Begg's funnel plot and quantitative Egger's test were adopted to test the publication bias of the current meta-analysis. The outlines of the Begg's funnel plots were rather symmetric, indicating the absence of any significant publication bias ( Figure S3). Statistical evidence of Egger's test also confirms a none-existence of publication bias among the studies.

| D ISCUSS I ON
The findings of this study suggest that PARP-1 rs1136410 C>T polymorphism has a board-line significant relationship with overall cancer risk. However, further stratified analyses revealed that this polymorphism could predispose to gastric cancer, thyroid cancer, cervical cancer, but protects from brain cancer.

| CON CLUS ION
In conclusion, our meta-analysis showed that PARP-1 gene rs1136410 C>T polymorphism may contribute to increased cancer risk among Asian populations. More investigations are encouraged to provide more evidence regarding the role of PARP-1 rs1136410 C>T polymorphism to the aetiology of cancer predisposition.

ACK N OWLED G EM ENTS
This study was supported by grant from Guiding Project of Scientific Research Program of Hubei Education Department in 2019 (B2019106).

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
All the data were available upon request.