Association between PD-1 and PD-L1 Polymorphisms and the Risk of Cancer: A Meta-Analysis of Case-Control Studies

A number of case-control studies regarding the association of the polymorphisms in the programmed cell death 1 (PD-1) and programmed cell death ligand 1 (PD-L1) genes with the risk of cancer have yielded inconsistent findings. Therefore, we have conducted a comprehensive, updated meta-analysis study to identify the impact of PD-1 and PD-L1 polymorphisms on overall cancer susceptibility. The findings revealed that PD-1 rs2227981 and rs11568821 polymorphisms significantly decreased the overall cancer risk (Odds Ratio (OR) = 0.82, 95% CI = 0.68–0.99, p = 0.04, TT vs. CT+CC; OR = 0.79, 95% CI = 0.67–0.94, p = 0.006, AG vs. GG, and OR = 0.82, 95% CI = 0.70–0.96, p = 0.020, AG+AA vs. GG, respectively), while PD-1 rs7421861 polymorphism significantly increased the risk of developing cancer (OR = 1.16, 95% CI = 1.02–1.33, p = 0.03, CT vs. TT). The PD-L1 rs4143815 variant significantly decreased the risk of cancer in homozygous (OR = 0.62, 95% CI = 0.41–0.94, p = 0.02), dominant (OR = 0.70, 95% CI = 0.50–0.97, p = 0.03), recessive (OR = 0.76, 95% CI = 0.60–0.96, p = 0.02), and allele (OR = 0.78, 95% CI = 0.63–0.96, p = 0.02) genetic models. No significant association between rs2227982, rs36084323, rs10204525, and rs2890658 polymorphisms and overall cancer risk has been found. In conclusions, the results of this meta-analysis have revealed an association between PD-1 rs2227981, rs11568821, rs7421861, as well as PD-L1 rs4143815 polymorphisms and overall cancer susceptibility.

. Locations and base pair positions of single nucleotide polymorphisms (SNPs) in PD-1 and PD-L1 genes.

Study Characteristics
A flow diagram of the study selection process is shown in Figure 1. For PD-1 polymorphisms, 54 case-control studies from a total of 26 articles [5][6][7]52] examining the associations of 6 widely studied polymorphisms in PD-1 gene and cancer risk were included in this meta-analysis. There were 16 studies involving 5622 cases and 5450 controls that reported the association between PD-1 rs2227981 polymorphism and cancer. Eleven studies including 4766 cases and 5839 controls investigated the relationship between PD-1 rs2227982 polymorphism and cancer. Nine studies with 1846 cases and 1907 cases reported the association between PD-1 rs11568821 variant and cancer risk. Seven studies including 3576 cancer cases and 5277 controls studied the correlation between PD-1 rs7421861 polymorphism and cancer. Seven studies involving 3589 cases and 4314 controls examined the association between PD-1 rs36084323 polymorphism and cancer risk. Six studies including 3366 cancer cases and 4391 controls studied the relationship between PD-1 rs10204525 polymorphism and cancer.  For PD-L1 polymorphisms, 13 case-control studies from 10 articles [27,38,[44][45][46][47][48][49][50][51] that assessed the impact of two polymorphisms of PD-L1 were included in the pooled analysis. Eight studies including 3030 cases and 4145 controls evaluated the association between PD-L1 rs4143815 polymorphism and cancer risk. Five studies with 1909 cases and 1970 controls assessed the correlation between PD-L1 rs2890658 variant and cancer risk. The characteristics of all these studies are shown in Table 2. For PD-L1 polymorphisms, 13 case-control studies from 10 articles [27,38,[44][45][46][47][48][49][50][51] that assessed the impact of two polymorphisms of PD-L1 were included in the pooled analysis. Eight studies including 3030 cases and 4145 controls evaluated the association between PD-L1 rs4143815 polymorphism and cancer risk. Five studies with 1909 cases and 1970 controls assessed the correlation between PD-L1 rs2890658 variant and cancer risk. The characteristics of all these studies are shown in Table 2.

Association of PD-1 Polymorphisms with Cancer Risk
The pooled analysis involving PD-1 rs2227981 polymorphism revealed that this variant significantly decreased the overall cancer risk in recessive (OR = 0.82, 95% CI = 0.68-0.99, p = 0.04, TT vs. CT+CC) genetic models (Table 3 and Figure 2).
We performed stratified analyses and the findings are summarized in Table 4. We observed that PD-1 rs2227981 significantly decreased the risk of gastrointestinal (GI) cancer (OR = 0.

PD-L1 Polymorphisms and Cancer Risk
The pooled ORs results for the relationship between the PD-L1 rs4143815 and rs2890658 polymorphisms and the risk of cancer are shown in Table 3 (Table 3 and Figure 3). The pooled analysis did not support an association between PD-L1 rs2890658 polymorphism and risk of cancer susceptibility (Table 3).

PD-L1 Polymorphisms and Cancer Risk
The pooled ORs results for the relationship between the PD-L1 rs4143815 and rs2890658 polymorphisms and the risk of cancer are shown in Table 3 (Table 3 and Figure 3). The pooled analysis did not support an association between PD-L1 rs2890658 polymorphism and risk of cancer susceptibility (Table 3).
We did stratified analysis (Table 4)

Heterogeneity
As shown in Table 3, heterogeneity between the studies regarding the PD-1 rs2227981, PD-1 rs36084323, PD-1 rs10204525, and PD-L1 rs4143815 was observed in all genetic models. For PD-1 rs2227982 polymorphism, our results showed no evidence of heterogeneity in the recessive model (TT vs. CT+CC). Regarding PD-1 rs11568821, heterogeneity was not observed in the heterozygous, homozygous, dominant, and recessive genetic models. Similarly, no evidence of heterogeneity in the heterozygous, homozygous, and recessive genetic models of PD-1 rs7421861 was found. Heterogeneity was not detected in the homozygous and recessive genetic models of the PD-L1 rs2890658.

Publication Bias
The potential publication bias of the studies included in the present meta-analysis was examined by Begg's funnel plot and Egger's test. The results of publication bias are summarized in Table 3. Based on the above analysis, no publication bias for the association of PD-1 rs2227982, PD-1 rs7421861, and PD-L1 rs4143815 variants in all genetic models and cancer risk was demonstrated (Table 3 and Figure 4).   As presented in Table 3 and Figure 5, no publication bias was observed in recessive genetic model of PD-1 rs2227981. Obvious publication bias was not found in the heterozygous, dominant, and allele genetic models of the PD-1 rs11568821 and PD-L1 rs2890658 (Table 3). Moreover, the publication bias was not observed in heterozygous, dominant, recessive, and allele genetic models of the PD-1 rs36084323 and PD-1 rs10204525. (Table 3).

Discussion
It has been proposed that environmental and genetic factors contribute to cancer development [53,54]. Single nucleotide polymorphisms (SNPs) can be considered as biological markers that help scientists to recognize genes that are related to cancer [55].
PD-1 and PD-L1 are involved in the regulation of programmed cell death, which is the regulator of cancer cell proliferation as well as primary response in many cancer therapy strategies. Several studies have investigated the association between PD-1 as well as PD-L1 polymorphisms and the risk of various types of cancers; however, the findings remain discrepant. This meta-analysis provides, for the first time a quantitative estimated of the association between six SNPs of PD-1 and two SNPs of PD-L1 gene and cancer susceptibility. The findings indicated that PD-1 rs2227981 and rs11568821 polymorphisms as well as PDL-1 rs4143815 variant significantly decreased the overall cancer risk, while PD-1 rs7421861 polymorphism significantly increased the risk of overall cancer. Our findings revealed no significant association between PD-1 rs2227982, PD-1 rs36084323, PD-1 rs10204525, and PD-L1 rs2890658 polymorphisms and overall cancer risk.

Discussion
It has been proposed that environmental and genetic factors contribute to cancer development [53,54]. Single nucleotide polymorphisms (SNPs) can be considered as biological markers that help scientists to recognize genes that are related to cancer [55].
PD-1 and PD-L1 are involved in the regulation of programmed cell death, which is the regulator of cancer cell proliferation as well as primary response in many cancer therapy strategies. Several studies have investigated the association between PD-1 as well as PD-L1 polymorphisms and the risk of various types of cancers; however, the findings remain discrepant. This meta-analysis provides, for the first time a quantitative estimated of the association between six SNPs of PD-1 and two SNPs of PD-L1 gene and cancer susceptibility. The findings indicated that PD-1 rs2227981 and rs11568821 polymorphisms as well as PDL-1 rs4143815 variant significantly decreased the overall cancer risk, while PD-1 rs7421861 polymorphism significantly increased the risk of overall cancer. Our findings revealed no significant association between PD-1 rs2227982, PD-1 rs36084323, PD-1 rs10204525, and PD-L1 rs2890658 polymorphisms and overall cancer risk.
We performed stratified analyses and our findings indicate that PD-1 rs2227981 significantly decreased the risk of gastrointestinal cancer, lung cancer and breast cancer. The PD-1 rs2227982 was associated with increased risk of cancer in hospital-based studies and lower risk of gastrointestinal and breast cancer. Similarly to PD-1 rs7421861, the PD-1 rs7421861 and PD-1 rs36084323 variants significantly increased the risk of cancer in hospital-based studies. The PD-1 rs11568821 was linked to reduce risk of cancer in population-based studies. Moreover, our findings revealed that PD-L1 rs4143815 polymorphism significantly reduced the risk of gastrointestinal cancer and hospital-based studies. A positive correlation between PD-L1 rs2890658 variant and the risk of lung cancer was observed.
Recently, Zou et al. [56] performed a meta-analysis of the association between PD-L1 rs4143815 polymorphism and the risk of cancer and found also a significant association between this variant and cancer risk, which is in line with our findings. Like our results, a meta-analysis conducted by Da et al. [57] revealed no significant association between PD-1 rs36084323 polymorphism and overall cancer susceptibility. Similar to previous meta-analysis conducted by Zhang et al. [58], we have also found that PD-1 rs2227981 and rs11568821 polymorphisms were associated with decreased cancer susceptibility. In another study, Dong et al. [59] conducted a meta-analysis aimed to inspect the associations between PD-1 rs2227981, rs2227982, rs7421861, and rs11568821 polymorphisms and cancer risk. There were seven studies involving 3395 cases and 2912 controls for PD-1 rs2227981, four studies including 1961 cases and 2390 controls for PD-1 rs2227982, four studies with 1975 cases and 2403 controls for PD-1 rs7421861, and four studies for PD-1 rs11568821 variant and cancer risk. They have found that rs2227981 and rs11568821 polymorphisms significantly decreased the risk of cancer. Mamat et al. [60] conducted a meta-analysis of six studies involving 1427 cases and 1811 controls and have observed no significant association between PD-1 rs2227981 polymorphism and the risk of cancer.
Nevertheless, the number of cases and controls as well as the number of polymorphisms in our meta-analysis is higher than in those previously published meta-analysis studies.
PD-1/PD-L1 axis impairs T cell activation by preventing Ras-Raf-MEK-ERK and PI3K-AKT signaling pathways, which are mainly believed to promote proliferation and differentiation of T cell [70]. The inhibitory regulation of PD-1/PD-L1 is typically compared to a brake in T cell activation [71]. PD-L1 is exerted by tumors to escape from immune system. Tumor-specific PD-L1-expression was not prognostic in colorectal cancer, while high immune cell-specific PD-1 expression was associated with a prolonged overall survival [72]. It has been revealed that high expression of PD-1 on peripheral blood T cell subsets is correlated with poor prognosis of metastatic gastric cancer [73]. Fang et al. [74] reported that the peripheral blood PD-1 expression was significantly higher in breast cancer patients than benign breast tumors. PD-1 and PD-L1 expression have been shown to be associated with adverse clinicopathological features in clear cell renal carcinoma [75].
This meta-analysis has however several limitations. Firstly, there are relatively small sample sizes of studies for some polymorphisms that should be expanded. Secondly, we have included in this meta-analysis only studies published in English, thus publication bias may have occurred. Thirdly, obvious heterogeneities were found in certain polymorphisms. Differences in ethnic background, type of cancer, and other baseline characteristics of participants may contribute to between-study heterogeneities. Lastly, gene-gene and gene-environment interactions which may affect cancer susceptibility were not evaluated in this meta-analysis due to lack of sufficient data. Therefore, the results of this meta-analysis should be cautiously interpreted.
In conclusion, the current meta-analysis suggests that rs2227981 and rs11568821 polymorphisms of PD-1 and the rs4143815 polymorphism of PD-L1 were associated with protection against cancer, while PD-1 rs7421861 polymorphism significantly increased cancer risk.

Literature Search
We searched PubMed, Web of Science, Scopus, and Google Scholar databases for publications that studied the association between PD-1 and PD-L1 polymorphisms and cancer risk. The last search was updated on 18 December 2019. The following search terms were used; "programmed cell death 1 or PDCD1 or PD-1, or CD279, or programmed death-1-ligand 1 or CD274 or B7-H1" and "polymorphism or single nucleotide polymorphism or SNP or variation" and "cancer or carcinoma, or tumor".
The process of recognizing eligible studies is presented in Figure 1. The inclusion and exclusion criteria were as follows. (1) The studies evaluated the association between the PD-1 and PD-L1 polymorphisms and cancer risk, (2) studies with necessary information on genotype or allele frequencies to estimate ORs and 95% Cis, (3) studies with human subjects, and (4) case-control design. We excluded reviews, conference papers, and other studies that were published as abstracts only.

Data Extraction
The data were recovered from eligible articles independently by two authors. Disagreements were discussed with the third investigator. The following information was recorded for each study: first author's name, publication year, patient's nationality, genotypes, and allele frequencies.

Statistical Analysis
We performed a meta-analysis to assess the association between PD-1 and PD-L1 polymorphisms and cancer susceptibility. The observed genotype frequencies in the controls were tested for Hardy-Weinberg equilibrium (HWE) using the chi-squared test.
Odds ratio (OR) and 95% confidence interval (CI) were calculated to evaluate the association between PD-1 and PD-L1 polymorphisms and cancer risk in five genetic models, which were heterozygous, homozygous, dominant, recessive, and allele. The strength of the association between each polymorphism and cancer risk was assessed by pooled odds ratios (ORs) and their 95% confidence intervals (CIs). The Z-test was used for statistical significance of the pooled OR. We estimated the between-study heterogeneity by the Q-test and I2 test: If I2 < 50% and P > 0.1, the fixed effects model was used to estimate the ORs and the 95% CI; otherwise, the random effects model was applied.
We evaluated publication bias using funnel plots for visual inspection and conducting quantitative estimations with Egger's test.
Sensitivity analysis was achieved by excluding each study in turn to assess the stability of the results. All analyses were achieved by STATA 14.1 software (Stata Corporation, College Station, TX, USA).

Conclusions
The findings of our meta-analysis proposed that PD-1 rs2227981, rs11568821, rs7421861, as well as PD-L1 rs4143815 polymorphisms associated with overall cancer susceptibility. Further well-designed studies with large sample sizes are warranted to confirm our findings.

Conflicts of Interest:
The authors declare no conflicts of interest.