Over the past two decades, multi-population GWAS have identified dozens of risk sites for LC, and most of these sites have been concentrated in the TERT-CLPTM1L region in 5p15.33[3][4][5][6][7][8]. TERT, located at the short arm 15.33 (5p15.33) of chromosome 5, encodes the catalytic subunit of telomerase[12], regulates telomerase expression levels, and maintains telomere length [13][14] while longer telomere length helps increase the risk of LC [18][19][20]. CLPTM1L is also located on chromosome 5p15.33, which encodes a cleft lip and palate-associated transmembrane 1-like protein. Overexpression of CLPTM1L has been observed in LC cells [28][29], and it has been confirmed that the occurrence of LC is closely related to the anti-apoptotic function of CLPTM1L[30][31][32]. Current GWAS report that multiple polymorphisms in TERT increase LC risk and are strongly related to telomere length [15][16][17]. Multiple gene polymorphisms associated with LC risk have also been found in CLPTM1L [5][18][19], and can also affect telomere length[35]. However, in some studies, no associations between these sites and susceptibility to LC has been found. The reasons for these different results may also be related to the different ethnicities, countries, study methods, sample sizes, LC types, chain imbalance patterns, and smoking/non-smoking status of patients. Therefore, the results leading to the associations between TERT-CLPTM1L polymorphisms and LC currently lack a unified conclusion. This study included data from GWAS and case-control studies that have so far reported the associations of TERT-CLPTM1L polymorphisms with LC with the aim of clarifying the associations between TERT-CLPTM1L polymorphisms and LC and differences in these associations between LC patients of different ethnicities and different subtypes.
This meta-analysis included 54 studies, including 12 GWAS and 42 case-control studies, all with controls whose genotype distribution frequencies were consistent with the HWE tests. Our results confirmed that 6 polymorphisms in TERT-CLPTM1L were associated with the risk of LC and this association was also found across different ethnic/pathological subtypes of LC. Specifically, the C allele variant of rs2736100 in the TERT region was positively associated with LC, as well as in different ethnic populations. Previous GWAS[3][4][5][6][7][8][9][10][11][34] and multiple case-control studies[36][101] have also reported an increased frequency of rs2736100 [C] in LC patients of different ethnic populations, suggesting that mutations in the C allele of TERT rs2736100 can increase the risk of LC in different ethnic populations. A study has reported that telomere length increased by the C allele of rs2736100 was associated with cancer[23]. Therefore, mutations in the C allele of TERT rs2736100 can upregulate the expression of TERT, maintain and lengthen telomere length, thereby increase the risk of LC. From the genotype analysis results, it could be seen that the five genetic models of rs2736100 [C] were all associated with the risk of LC in Asians, but the CA vs. AA model wasn’t associated with the risk of LC in Caucasians. Furthermore, it was found that in addition to CA vs. AA, the LC risk associations in Asians were stronger than that in Caucasians in the other four genetic models. These results suggest that the LC risk association caused by the C allele mutation of rs2736100 in Asians is stronger than that in Caucasians. Evidence from Phase 3 of the 1000 Genomes Project [102] and the latest literature [62] suggested that rs2736100 was more strongly associated in Asians than in Caucasians (East Asians: r2 = 0.5789; Europeans: r2 = 0.4352), which confirms the credibility of our results.
In different LC subtypes, previous studies have found that increased C allele frequency of rs2736100 increases the risk of NSCLC [72][82], LUAD [103], SCLC[6], and LUSC [25][104] in different populations. Our study also showed that the rs2736100 [C] variant increased the risk of NSCLC and LUAD in Caucasians and NSCLC, LUAD, LUSC, and SCLC in Asians. These results suggest Ethnic differences in the incidence risk of rs2736100 and different subtypes of LC. From our results, the association between C allele variant and NSCLC was stronger than that of SCLC, and the association with LUAD was stronger than that of LUSC, which means that the association between C allele variant and LC risk is mainly concentrated in the LUAD subtype of NSCLC. Therefore, it can be assumed that C allele variant is more strongly associated with the risk of LUAD than other LC subtypes in Caucasians and Asians. However, we did not find a difference in the strength of the risk association for LUAD between the two populations, which means that the risk of C allele variation and LUAD is strong in both populations. In addition, our results showed that C allele variant can increase the risk of LUSC and SCLC in Asians, but not in Caucasians, suggesting that rs2736100 [C] is strongly associated with LC in Asians than in Caucasians, and mainly in the LUSC and SCLC subtypes.
Our results also showed that the T allele variant of rs2736098 in the TERT region was positively associated with LC, and this positive association was mainly present in Asians rather than Caucasians. Multiple previous case-control studies have also reported an increase in the frequency of rs2736098 [T][45][86] in LC patients in Asians. It has been reported that T allele variant of rs2736098 may lead to overexpression of TERT and increase telomerase activity, which regulates the development of LC by regulating unlimited cell division and carcinogenesis, and interact with activation of glycolytic pathways [105]. These evidences suggest that the T allele mutation of TERT rs2736098 regulates telomerase activity, thereby increasing the risk of LC in Asians. In addition, the rs2736098[T] variant also increased the risk of NSCLC, LUAD, LUSC, and SCLC in Asians. The increased frequency of rs2736098[T] has been reported in one study of increasing the risk of NSCLC and LUAD in Chinese population [92], and two other studies have confirmed that rs2736098[T] was associated with the risk of SCLC [100] and LUSC [99] in Chinese population. These results confirmed the reliability of our results. In different subtypes of LC in Asians, our results showed that the risk association between T allele mutations and NSCLC was the same as that of SCLC, and the risk association with LUAD was stronger than that with LUSC. This means that T allele variant and the risk of LC is mainly in LUAD and SCLC. However, due to the limited number of studies included in SCLC (n = 1), it cannot fully confirm the relationship between the T allele variant and the risk intensity of SCLC. Therefore, the T allele variant of rs2736098 and the risk of LC in Asians may mainly be in LUAD. In Caucasians, we did not find the associations between rs2736098 and the risk of LC due to the lack of sufficient evidence. Previous case-control studies have found the risk association between the T allele variant of rs2736098 and the incidence of LC in Caucasians in the United States and Europe [12][34], but we did not include their results in the analysis as these studies did not report complete genotype data. This means that rs2736098 may be associated with the risk of LC in Caucasians, but this association may not be very strong. Evidence from Phase 3 of the 1000 Genomes Project [102] and the latest literature [62] showed that rs2736098 was moderately linkage disequilibrium (LD) in Asians (East Asians: r2 = 0.3070) and weakly LD in Caucasians (Europeans: r2 = 0.1504). Combined with our findings, it can be concluded that the association between rs2736098 and the risk of LC mainly exists in Asians rather than in Caucasians.
The T allele variant of rs401681 in the CLPTM1L region was negatively associated with LC, as well as in both populations, these results were similar to previous GWAS results [5][10][11][34][106]. It has been reported that rs401681[C] may affect transcriptional regulation, leading to overexpression of the CLPTM1L gene and increasing the risk of LC [107], and rs401681[C] was associated with shorter telomeres [12]. These evidences suggest that mutations in the T allele of rs401681 reduce the risk of LC in Caucasians and Asians by regulating CLPTM1L gene expression and telomere length. From the analysis results of genotype, the negative associations between the five genetic models of rs401681[T] and LC existed in both Caucasians and Asians, and the negative associations of LC in Caucasians in five genetic models were the same as in Asians, suggesting that the T allele variant of rs401681 and the risk of LC were negatively associated in both populations. In different LC subtypes, rs401681[T] has been reported to reduce the risk of NSCLC, LUAD, and LUSC in Asians [27][92][106], but it did not reduce the risk of SCLC in Chinese populations [69]. Our study also found that T allele variant primarily reduced the risk of NSCLC (LUAD, LUSC) rather than SCLC in Asians. From our results, the negative association between T allele variant and LUAD was the same as that of LUSC, which means that the negative association between T allele variant and LC risk in Asians is mainly concentrated in the LUAD and LUSC subtypes of NSCLC. In addition, our findings suggested that T allele variant did not appear to reduce the risk of NSCLC in Caucasians, but previous studies have reported the reduction in the frequency of the T allele of rs401681 in NSCLC and LUAD patients in Caucasians [31][40][108], but we did not include their results in the analysis because these studies did not report complete genotype data. This means that there may be the negative association between rs401681 and the risk of NSCLC in Caucasians, but further studies are needed to confirm this.
The T allele variant in rs402710 in the CLPTM1L region was also negatively associated with LC, as well as in both ethnic populations, and previous case-control studies have also reported the decrease in the frequency of rs402710[T] in LC patients in Caucasians and Asians[36]. It has been reported that rs402710 may block DNA damage-induced apoptosis by enhancing the accumulation of Bcl-xL, a member of the anti-apoptotic Bcl2 family, thereby affecting lung tissue tumorigenesis in vitro [109]. The rs402710 also maintain the telomere length, which may reduce the risk of LC in nonsmokers because the protective effects of rs2736100 were offset in patients with bladder cancer who currently smoke [36]. In addition, Zienolddiny et al. [39] found that rs402710 was associated with increased formation of DNA adducts in the lung, which may be a precursor to LC. These evidences suggest that mutations in the T allele of rs402710 reduce the risk of LC in Caucasians and Asians by regulating apoptosis and telomere length. From the analysis results of genotype, the negative associations between the five genetic models of rs402710 [T] and LC existed in both Caucasians and Asians, and the intensity of negative associations between LC in five genetic models in Caucasians was the same as in Asian populations, which suggest that the T allele variant of rs402710 and the risk of LC are negatively associated in both populations. In different LC subtypes, our findings showed that rs402710 [T] reduced the risk of NSCLC in Caucasians and Asians. Previous studies have also reported that rs402710[T] reduces the risk of NSCLC in Caucasians [33][40] and Asians [60][85]. From the perspective of pathogenesis association strength, the risk association of NSCLC in Caucasians was the same as in Asians, which shows that the negative association between T allele variation of rs402710 and NSCLC risk is strong in both populations. In addition, the results also showed that rs402710 [T] reduced the risk of LUAD rather than LUSC in Asians, which means that the negative association between the T allele variant of rs402710 and the risk of LC in the Asians may be mainly in the LUAD of NSCLC. However, due to the small sample size, these data were only reported in a Korean study [85], hence they cannot be representative of the risk association between LUAD and LUSC in all Asian populations. Furthermore, rs402710[T] was not found to be the risk association with the onset of LUAD and LUSC in Chinese populations in another study[69], and it wasn’t found to be the risk association with the onset of LUAD in a Japanese study [78]. Therefore, the negative association between the T allele variant of rs402710 and the risk of LUAD and LUSC in Asians needs to be verified by further expansion of the sample size.
In the CLPTM1L region, our results showed that the A allele variant of rs31489 was also negatively associated with LC, it was also roughly the same in Caucasians and Asians, and previous case-control studies have also reported the decrease in the frequency of rs31489 [A] in LC patients in both populations [34][40]. It has been reported that rs31489 [A] may affect telomere length, thereby reducing the risk of LC in nonsmokers because smoking counteracts the protective effects of the A allele, shortens telomere length, and enhances telomerase activity [110]. These evidences suggest that mutations in the A allele of rs31489 regulate telomerase activity, thereby reducing the risk of LC in Caucasians and Asians. From the analysis results of genotype, the five genetic models of rs31489[A] were negatively associated with the risk of LC in Caucasians, but A vs.C, AC vs.CC, AA + AC vs.CC model had no negative associations with LC in Asians, and these results suggest that the negative association of LC risk caused by allele A mutation of rs31489 may be stronger in Caucasians than in Asians. But in the AA vs. CC and AA vs. AC + CC models, we didn’t find stronger negative associations with LC risk in Caucasians than in Asians, suggesting that the results of rs31489 were unstable. It was revealed that there was heterogeneity in the results of the Asians after further careful analysis, which was mainly related to the small sample size of the Asian population included in the study (n = 2), and some false negative results would inevitably be generated due to the small sample size. Based on these factors, we cannot determine that the negative association of LC risk is stronger in Caucasians than in Asians, and further expansion of the sample size of Asians is needed to draw more reliable conclusions. In different LC subtypes, although our findings showed that rs31489 [A] reduced the risk of NSCLC in Asians rather than in Caucasians, two previous studies reported that rs31489 [A] reduced the risk of NSCLC, LUAD [40], and LUSC[25] in Caucasians (they were not included because they didn’t report the complete genotype). Therefore, we couldn’t confirm the difference between rs31489 and the risk of NSCLC in the two populations.
In the TERT-CLPTM1L region, the G allele variant of rs4975616 was negatively associated with LC, presumably in Caucasians and Asians. Previous studies have also reported a decrease in the frequency of rs4975616 [G][31][34] in LC patients in both populations. These data demonstrate that the G allele variant of rs4975616 reduces the risk of LC in Caucasians and Asians. From the analysis results of genotype, the five genetic models of rs4975616[G] were negatively associated with the risk of LC in Caucasians, but in Asians, except for the allele model (G vs. A), the other 4 models didn’t have the negative associations with LC risk, and these results suggest that the negative association of LC risk caused by G allele mutation of rs4975616 may be stronger in Caucasians than in Asians. However, in G vs. A model, we did not find a stronger negative association with LC risk in Caucasians than in Asians, suggesting that the results of rs4975616 may be biased. Further careful analysis showed that there was a certain publication bias in the results of rs4975616, but there was no significant sensitivity or heterogeneity in these included data. Based on these factors, we cannot fully determine the difference in this negative risk association between Caucasians and Asians, and more studies need to be included for further analysis. The G allele variant of rs4975616 appeared to reduce the risk of NSCLC and LUSC in different LC subtypes, but no such association was found in each ethnic subgroup. In addition, its variants appeared to reduce the risk of LUAD in Caucasians rather than in Asians, and did not appear to reduce the risk of SCLC, however, due to factors such as small sample size, large heterogeneity, and high sensitivity, the reliability of these results is not high and further studies are needed to confirm it.
In summary, it can be generally concluded that their associations with the risk of LC are: (1) The rs2736100[C] variant can increase the risk of LC, and the risk association in Asians is stronger than that in Caucasians. The rs2736100 [C] variant is strongly associated with the risk of LUAD than with other LC subtypes in different LC subtypes, and the rs2736100 [C] variant is strongly associated with the risk of LUAD in both populations. In addition, rs2736100[C] is associated with LUSC and SCLC risk in Asians to a greater extent than in Caucasians. (2) The correlation between the rs2736098 [T] variant and the risk of LC mainly exists in Asians rather than in Caucasians, and this risk association mainly exists in the LUAD of the Asians. (3) rs401681[T] variant can reduce the risk of LC, and this negative association is strong in both Caucasians and Asians. Among the different LC subtypes, the rs401681[T] variant mainly reduce the risk of LUAD and LUSC in NSCLC of Asians, and its risk of NSCLC in Caucasians needs to be confirmed by further studies. (4) rs402710[T] variant can reduce the risk of LC, and this negative association is strong in both Caucasians and Asians. In different LC subtypes, this negative association mainly exists in NSCLC in both populations. (5) rs31489 [A] and rs4975616 [G] variants can reduce the risk of LC in Caucasians and Asians, however, the differences in LC risk(including various LC subtypes)between the two populations need further studies.
Limitations of this study: ① This meta-analysis is based on the research reports of two ethnic groups and different types of LC, which will inevitably produce some heterogeneity and publication bias; ② Most studies used different methods of genetic testing and genotyping, which also may lead to bias in the analyses; ③ The sample size of the study is generally sufficient, but after subgroup analysis according to different LC subtypes and ethnic groups, the results showed that the sample size of NSCLC (LUAD, LUSC) and SCLC is relatively small. The reason for the low sample size in this subgroup is related to our exclusion of some studies with incomplete genotype reporting and control group non-HWE. Therefore, this inevitably produces some false negatives, heterogeneity, sensitivity, and publication bias for the results of NSCLC (LUAD, LUSC) and SCLC; ④ Due to the insufficient sample size of smoking cases in the included studies and the lack of separate reports on male or female studies, this study cannot further discuss the impact of factors such as smoking, gender, etc. on LC; ⑤Although some polymorphisms(such as rs2242652, rs2736108, rs2853669, rs2853676, rs2853677, rs31490, rs380286, rs451360, rs465498, etc.) were reported to be associated with the risk of LC, they were not included in the statistics due to incomplete data and insufficient studies.