Lung cancer is the leading cause of cancer-related deaths worldwide, accounting for 18.0% of all cancer deaths in 2020 [1]. Non-small cell lung cancer (NSCLC), which constitutes 85% of all lung cancer cases, is the major subtype of the cancer [2]. Platinum-based chemotherapy (PBC) is commonly used as first-line treatment and also as adjuvants to radiotherapy or surgery in patients with late-stage NSCLC [3–5]. However, only 30%–40% of patients show a good response to PBC, and tests for extreme drug resistance have shown high variability in chemoresistance (24%–88%) [6, 7].

The high variability in patient response to PBC has been attributed to individual genetic variants. In recent years, there has been a surge of publications on the pharmacogenetics of PBC in NSCLC patients. Many studies have reported that the genetic variants associated with platinum chemoresistance are involved in the DNA repair pathway, cellular trafficking and drug transport, and metabolic pathways [8–10]. However, the results of published studies are mostly inconsistent and inconclusive. The main reasons for the inconsistency of results are small sample sizes leading to low statistical power, heterogeneity of ethnicities in the different studies due to population stratification, or differences between histological subtypes [11]. A comprehensive meta-analysis is therefore necessary to yield a more precise estimate of the association between genetic variants and PBC responses and also to provide a field synopsis of research in this area. To our knowledge, the most recent comprehensive meta-analysis reporting the association of genetic variants with response to PBC was published in 2017, and only included studies up to 31 January 2016 [12]. As there have been numerous new studies published after January 2016, the inclusion of these studies may yield different results than the previous meta-analysis [13].

Here, we aim to identify, strengthen and interpret the associations of genetic variants with response to PBC in patients with NSCLC patients using a comprehensive systematic review and meta-analysis. We used the standardized guidelines including the Venice criteria to systematically assess the credibility of all relevant studies from publications [14–17]. In this work, a total of 184 genetic variants were examined based on the reported genotype frequencies for each variant corresponding to either chemoresistance or good response to PBC in NSCLC patients [18]. Results were also stratified by ethnicity to provide greater insight into the underlying factors that influence response to PBC.

To our knowledge, the most recent comprehensive meta-analysis examining associations between genetic variants and response to PBC in NSCLC patients was published in 2017, with the literature search last conducted on 31 January 2016 [12]. Since then, a large number of studies reporting the associations of genetic variants to response to PBC have been published. Here, we performed an updated meta-analysis that identified 12 genetic variants in 10 candidate genes that were statistically significantly associated with response to PBC in NSCLC patients with strong or moderate cumulative epidemiological evidence.

A large proportion of the genetic variants are involved in DNA synthesis and repair, consistent with the previous meta-analysis [12]. The mechanism of platinum drugs involves the formation of DNA lesions that lead to apoptosis of cancer cells. Therefore, genetic variants affecting DNA synthesis and repair would influence efficacy of PBC [26, 27]. ERCC1 is one of the genes involved in nucleotide excision repair (NER). High expression of ERCC1 was found in cisplatin-resistant cancer cells, and can be used to predict low clinical efficacy of cisplatin [28]. However, we demonstrated that the ERCC1 rs11615 was not significantly associated with PBC response under all genetic models. This is in contrast to with previous meta-analyses and gene association studies which showed significant associations with response to PBC [12, 20, 29–38]. The results may be due to a bias in favor skewed results towards of selected ethnicities, as new studies have been added since the previous meta-analysis. This postulation is supported by the results of our subgroup analysis, which showed that the Asian population has a higher OR compared to the European population. The ERCC1 rs11615 variant allele has been suggested by several studies to affect NER by reducing ERCC1 transcripts, which impairs DNA repair and leads to DNA damage accumulation [39, 40], but the functional studies revealed similar results for both the variant and wild type alleles in other studies [41]. The inconsistencies in the functional studies could be due to indirect mechanisms such as linkage disequilibrium or regulation by other factors. The variant allele of ERCC1 rs3212986 was also found to be associated with PBC response in the current meta-analysis, which is consistent with previous studies [12, 30, 34, 42–46]. Several studies have suggested the ERCC1 rs3212986 variant may modulate mRNA expression, because it is located in the 3′-UTR, is in linkage disequilibrium with the neighboring XRCC1 and XPD genes, which may affect mRNA stability, and the bioinformatics analysis of ERCC1 rs32123986 revealed the possible alteration of secondary structure in the 3′-UTR and post-transcriptional regulation via binding miRNAs [40, 47, 48]. The ERCC1 rs3212986 showed population stratification in the subgroup analysis, with the Asian population having a higher OR (1.556, 95% CI = 1.078–2.244) compared to the European population under the dominant genetic model (OR = 0.718, 95% CI = 0.510–1.010). The underlying mechanisms for the racial differences in PBC response are not clearly defined, but could be due to gene-gene or gene-environment interactions, which include differences in ethnicity or lifestyle [49–51].

ERCC2 (XPD) is another gene involved in NER, with XPD serving the DNA helicase subunit in TFIIH [52]. Genetic variants of ERCC2 have been shown to be associated with impaired DNA repair capacity, leading to accumulation of DNA adducts [53]. The variant allele of ERCC2 rs1052555 was also identified in this meta-analysis to be associated with response to PBC, consistent with Li et al. and Tan et al. [12, 54]. The ERCC2 rs1052555 is a silent substitution predicted to form a splicing abolish domain or exon splicing enhancer that affects post-transcriptional mRNA splicing [54]. The altered splicing could result in different expression levels of ERCC2 that contribute to PBC response. On the other hand, the variant allele of XRCC1 rs1799782 was found to be associated with a protective effect against PBC chemoresistance, which was in agreement with several studies, including a recent meta-analysis by Zhang et al. [12, 55–65]. One possible mechanism by which the variant allele of XRCC1 rs1799782 protects against PBC chemoresistance is by interfering with the repair of DNA breaks caused by platinum drugs, as the XRCC1 rs1799782 was predicted to be deleterious or damaging by bioinformatics tools. However, other functional studies suggest that allelic substitution of the variant may indirectly affect the stability of XRCC1 via interactions with miRNA and polymerases involved in DNA repair [66, 67]. Similarly, the variant allele of XRCC3 rs861539 was also demonstrated to be associated with a protective effect against chemoresistance, consistent with the meta-analysis by Tan et al. [12]. The protective effect of XRCC3 rs861539 against chemoresistance is supported by the bioinformatics prediction of a resulting benign and tolerated XRCC3 that contributes to the proper function of DNA repair.

Interestingly, after incorporating data from recent publications [21, 45, 68], we identified several new significant associations that were not found to be statistically significant in the previous meta-analysis by Tan et al. These include variants in genes involved in DNA repair (ERCC2 rs1799793, XRCC1 rs25487, and XPC rs77907221). The variant allele of ERCC2 rs1799793 showed associations with PBC chemoresistance, supporting the findings of a previous meta-analysis by Qiu et al. [69]. It is unlikely that ERCC2 rs1799793 causes PBC chemoresistance by affecting DNA repair, as our bioinformatics analysis predicted that the genetic variant is benign and tolerated. Instead, the ERCC2 rs1799793 may indirectly contribute to platinum chemoresistance by dysregulating apoptosis, as the resulting Asp-to-Asn substitution led to a 2.5-fold increased apoptotic response in a lymphoblastoid cell line [70]. Similarly, the variant allele of XRCC1 rs25487 also showed associations with PBC chemoresistance, consistent with a recent meta-analysis by Zhang et al. [61]. Bioinformatics analyses predicted that the XRCC1 rs25487 variants are benign and tolerated, suggesting that the genetic variant also does not directly affect DNA repair to result in PBC chemoresistance. On the other hand, XPC rs77907221 insertions were found to be associated with better PBC response here consistent with a few older studies [71, 72]. Since there are only a few studies on the functional effects of XPC SNPs, the causal mechanism of XPC rs77907221 is unclear. However, the strong linkage disequilibrium between XPC rs7790722, XPC rs2228001 and XPC intron 11 may make a synergistic contribution to the PBC response [73, 74].

Besides, whereas Tan et al. previously reported significant associations of XPA rs1800975, ERCC2 rs13181 and ERCC5 rs2296147 with PBC response, this result was not replicated in our work. For XPA rs1800975 and ERCC5 rs2296147, we could find only two eligible studies and therefore, classified the variants as having poor cumulative epidemiologic evidence. For ERCC2 rs13181, no statistically significantly association with PBC response was found in our meta-analysis, possibly due to the high heterogeneity score in the homozygous, recessive, dominant and allele models after the addition of recent publications. Subgroup analysis showed that Asians had a higher OR than Europeans, suggesting that population stratification contributes to the high heterogeneity score. The confounding effect of ethnicity is evident because ERCC2 rs13181 showed significant associations in Europeans but not in Asians in the dominant model.

Metabolic and detoxification regulators also account for a large proportion of the genetic variants identified as having significant associations with PBC responses in this meta-analysis. The cytidine deaminase (CDA) enzyme plays a critical role in the metabolism and inactivation of gemcitabine, which is commonly used in combination with platinum drugs to treat NSCLC [75, 76]. CDA rs1048977 is a newly identified genetic variant that shows a significant association with PBC response, which was not reported in the previous meta-analysis by Tan et al. [12] The variant allele of CDA rs1048977 showed a significant association with better response to PBC in the allele and recessive models, which is consistent with Hu et al. and Ludovini et al., although in the recessive model it was considered as having weak cumulative epidemiologic evidence [77, 78]. Ludovini et al. suggested that the CDA rs1048977 is associated with better response due to lower enzyme activity which results in high drug availability. However, a more recent study by Ciccolini et al. showed that the polymorphism did not alter CDA activity, suggesting indirect regulatory mechanisms [78, 79].

CYP1A1 is another gene involved in the metabolism of antineoplastic drugs and has been shown to influence responses to PBC [80]. CYP1A1 rs1048943 is another newly identified genetic variant in this current meta-analysis that shows a significant association with response to PBC. The variant allele of CYP1A1 rs1048943 was also newly identified in this meta-analysis to be significantly associated with PBC chemoresistance and classified as having moderate cumulative epidemiological evidence in the heterozygous, dominant, and allele models. Interestingly, CYP1A1 rs1048943 had a slight tendency to deleterious functional effects based on the Mutation Assessor bioinformatics score, which assesses the evolutionary conservation of affected amino acids in protein homologs. The CYP1A1 rs1048943 is another newly identified genetic variant that has not been reported to show associations with PBC response in any previous meta-analysis, but was previously shown to be significantly associated with lung cancer susceptibility [81]. Furthermore, the CYP1A1 rs1048943 variant allele has previously been shown to have higher enzyme activity that promotes DNA adduct accumulation [82–84].

ABCC2 and ABCG2 are drug transporters that play critical roles for the influx and efflux of platinum drugs [85]. Overexpression of ABCC2 during cisplatin treatment has been shown to contribute to cisplatin chemoresistance due to less DNA-cisplatin adduct formation, and lower intracellular accumulation of cisplatin [85–90]. In this meta-analysis, we identified two significant variants in these genes that had not been reported to be associated with response to PBC in previous meta-analyses, i.e., ABCC2 rs717620 and ABCG2 rs2231142. The drug transporters ABCC2 rs717620 variant allele and ABCG2 rs2231142 variant allele were both found to be significantly associated with protective effects against PBC chemoresistance. The better response of NSCLC patients to the ABCC2 rs717620 variant allele is consistent with the reports by Han et al. and Qiao et al [91, 92]. The ABCC2 rs717620 may have indirect effects on PBC response, as Zhang et al. reported no effects on mRNA expression or downstream open reading frame translation, whereas Nguyen et al. reported that the ABCC2 rs717620, together with ABCC2 rs18885301 and ABCC2 rs2804402, increased promoter activity by 35% [93, 94]. Similarly, the variant allele of ABCG2 rs2231142 was also reported to be associated with better PBC response by Qiao et al. [95]. The improved PBC response by ABCG2 rs2231142 may be due to reduced efflux of platinum drugs from tumor cells using ABCG2. Since ABCG2 rs2231142 was predicted to be benign or tolerated in our bioinformatics analysis, the decreased ABCG2 activity may occur indirectly via regulation of protein levels. Previous studies have suggested that ABCG2 rs2231142 elicits its good PBC response by having lower protein levels in the lungs and increased platinum bioavailability in cell lines [96–101]. Previous meta-analyses by Tan et al. and Wei et al. also identified MTHFR rs18001133 and MDR rs1045642 as genetic variants that showed significant associations with PBC response [12, 33], but no statistically significant association was observed in this current meta-analysis. The lack of significant associations may be due to the significant heterogeneity reported for most genetic models after the addition of new publications.

The current meta-analysis provides the most up-to-date field synopsis and assessment of the associations between genetic variations and response to PBC. In addition, there are sufficient data from recent publications to allow the meta-analyses to be performed for all genetic models (whereas the previous meta-analysis by Tan et al. was limited to the homozygous, heterozygous, and dominant models). The allele model here combined both the homozygous and heterozygous into a single allele model. One advantage provided by the allele model is to distinguish the biological effects when heterozygous and homozygous genotypes are similar. Furthermore, by focusing on allele frequencies rather than genotypes in our meta-analysis, we take advantage of the increased number of observations, essentially doubling our sample size since each individual contributes two alleles. This approach not only increases the statistical power of our study and allows for more precise detection of genetic associations, but also shifts our focus more directly to the individual effects of the alleles, allowing for a clearer understanding of their role in the disease under study. Our study findings provide the impetus for clinical applications in precision medicine, where complementary data on gene-gene, gene-environment, functional studies, and validation in local population of patients will drive the development of potential prognostic biomarkers in the clinical setting. Identification of variants that showed a robust association with response to PBC could contribute to a more accurate assessment of the potential target population for optimizing the efficacy of PBC [102]. These findings could be helpful in the development of genetic tests that can predict the response of patients to PBC and thus enable a more targeted selection of therapy [103]. For example, patients with genotypes associated with improved respond to PBC could be prioritized for the treatment, which would improve treatment efficacy and patient outcomes. In addition, understanding the genetic basis of PBC response may lead to optimization of drug dosing. Personalized dosing based on individual genetic profiles may maximize therapeutic efficacy while minimizing adverse effects, which could significantly improve the quality of life of NSCLC patients undergoing treatment. The identified genetic variants also provide insights into the molecular mechanisms underlying PBC resistance [104]. This knowledge is invaluable for drug development as it can guide the development of new agents that can overcome resistance or improve the efficacy of existing treatments. This genetic knowledge could also facilitate the development of combination therapies. By understanding the genetic profiles that drive resistance to PBC, clinicians can better select complementary treatments that may overcome resistance and improve overall treatment efficacy. In addition, these findings also have implications for health economics [105]. By identifying patients who are more likely to benefit from PBC, healthcare resources can be allocated more efficiently, reducing the cost of ineffective treatments and the management of side effects. The discovery of these genetic variants also opens up opportunities for further research. For example, investigating the interaction between these genetic factors and environmental or lifestyle factors could lead to a more comprehensive understanding of the response to PBC in NSCLC, ultimately leading to improved patient care and better outcomes.

However, there are several limitations to this meta-analysis. First, more than 80% of the genetic polymorphisms identified in the eligible publications were reported in only one or two studies, resulting in small sample sizes and led to having poor cumulative epidemiologic evidence for a number of genetic variations. The limited number of studies on individual variants also prevented us from performing sensitivity analyses, subgroup analyses or tests for publication bias, which may be a source of heterogeneity. We also did not examine the gene-gene or gene-environment interactions that might influence the efficacy of PBC. It is also possible that individual genetic variants identified to be associated with response to PBC here are not the causal variants. The genetic variants could be in linkage disequilibrium with other SNPs or alter the stability of the resulting mRNA. The gene-gene and gene-environment interactions, along with the possible linkage disequilibrium with causal variants, need to be further investigated before the significant variants are used as clinical prognostic biomarkers. For instance, the combined effects of ERCC1 rs11615, ERCC1 rs3212986 and ERCC2 rs1799793 were shown to decrease the overall survival in non-squamous NSCLC patients undergoing pemetrexed/platinum-based chemotherapy but the individual effects of SNPS towards OS were not significant [106]. The OS of the patient decreased with the presence of every unfavorable allele from approximately 30 months with two to three unfavorable alleles to 11.8 months (p = 0.01) with four unfavourable alleles [106]. In addition, genome-wide scans should also be conducted in more populations to identify novel genetic variants associated with response to PBC in an unbiased manner. It would also be interesting to investigate whether these genetic variants are associated with lncRNA dysregulations and mitochondrial DNA alterations, which have recently been shown to influence cisplatin response [107, 108]. Finally, functional predictions of significant genetic variants were only made using in silico methods, which use different algorithms that assess the degree of conservation of amino acids across different species, physiochemical properties, or the combination of functional data and variant annotations, leading to discrepancies in the predictions [109]. Thus, the results of functional predictions should be interpreted with caution. Despite possible discrepancies in functional prediction, the use of multiple algorithms is advantageous as it allows for a comprehensive analysis, leading to more robust and reliable conclusions [110]. This approach allows for a balanced interpretation of the results and ensures that conclusions are not overly reliant on a single predictive model, but are instead informed by a spectrum of computational insights, which increases confidence in the pathogenicity prediction of each genetic variant. In the future, additional in vitro or in vivo work is needed to confirm and validate the function of these genetic variants [111].

In conclusion, our meta-analyses identified 12 genetic variants in 10 candidate genes (ABCC2 rs717620, ABCG2 rs2231142, CDA rs1048977, CYP1A1 rs1048943, ERCC1 rs3212986, ERCC2 rs1799793 and rs1052555, GSTM1 rs36631, XPC rs77907221, XRCC1 rs1799782 and rs25487, and XRCC3 rs861539) that showed statistically significant associations with response to PBC in NSCLC patients with strong or moderate cumulative epidemiological evidence. We also identified 172 genetic variants that were not associated with PBC response in at least one genetic model. Our results provide the most up-to-date summary and field synopsis of the genetic variants associated with response to PBC in NSCLC patients.

This work represents an advance in biomedical science because it offers the most comprehensive, up-to-date synthesis of genetic variants influencing platinum-based chemotherapy response in NSCLC patients.