There have been previous studies investigating possible associations between XRCC1
SNPs and chemotherapy outcomes or overall survival of lung cancer [18
]. However these original results are inconsistent and until now the lack of systematic review evaluation failed to give further insights on this issue. We showed, by an extensive quantitative and systematic review of published reports, that XRCC1
194 C/T and XRCC1
399 G/A SNPs were associated with objective response and XRCC1
399 G/A genotype and A/A genotype could influence overall survival of lung cancer patients. Furthermore, interaction analysis suggested that compared with the patients carrying C/T+T/T genotype at XRCC1
194 and G/G genotype at XRCC1
399, the patients carrying 194 C/C and 399 G/A+A/A or 194 C/C and 399 G/G genotype showed much worse objective response.
Platinum agents are activated intracellularly to form reactive platinum complexes that bind to DNA, thereby inducing intrastrand and interstrand DNA cross-links, as well as DNA-protein cross-links. These platinum-induced DNA and protein effects result in apoptosis and cell growth inhibition. There are several signal transduction pathways involved in this process to exert antitumor effects, among which DNA damage recognition and repair are important. Cancer cells may be able to resist against the platinum-based chemotherapy when its DNA repair ability is enhanced to remove those DNA adducts caused by platinum agents. There is evidence that lung cancer patients with a lower DNA repair capacity had an increased overall survival after the first-line platinum-based chemotherapy [7
Genetic polymorphisms can contribute directly to the variety in phenotypic drug sensitivity by modifying functions of the related genes. SNPs as either prognostic or predictive biomarkers have many advantages, especially in the advanced cancer setting. Firstly, it is relatively easy to obtain the human specimen for detecting SNP. Secondly, the method to detect SNP is precise and practical. Finally, some biomarkers are detected by specialized and mostly body-harmed methods; otherwise SNP detecting could avoid these problems.
The XRCC1 protein is considered to play an important role in DNA damage repair. XRCC1
Arg194Trp and Arg399Gln polymorphisms were the commonest one among more than 60 validated SNPs in XRCC1
gene and showed no major variations by ethnicity [40
SNPs have been reported to be associated with an altered DNA repair activity [14
]. Previous reports have also suggested that XRCC1
polymorphisms might be risk factors for the development of lung cancer [41
] and promising predictive or prognostic makers for lung cancer patients [33
]. Therefore, functional SNPs in XRCC1
gene may relate with platinum sensitivity and have prognostic values among lung cancer patients. With a pooled dataset of 2926 patients, we performed a comprehensive and systematic evaluation of clinical outcomes by objective response and overall survival. We are delighted to find the statistically significant association between XRCC1
SNPs with objective response and overall survival of lung cancer, which were not significant in previous original reports.
Meanwhile, the interaction between these two SNPs of XRCC1 gene in the objective response was analyzed for the first time. It is important to analyze multiple SNPs and their interaction to find more reliable prognostic or predictive biomarkers, because it is hard to predict complex clinical outcomes of lung cancer patients by using only one SNP. Although there were only four original studies included in the present interaction analysis, the results are valuable. The results of interaction analysis showed that the patients carrying 194 C/C and 399 G/A+A/A or 194 C/C and 399 G/G genotype showed worse objective response, suggesting C allele at XRCC1 194 may be more significantly associated with poorer objective response than A allele at XRCC1 399. The interaction between these two SNPs in objective response of lung cancer patients was not found. The reason may be the interaction between these two SNPs does not exist at all, or false positive report probability (FPRP) is too higher due to the small sample size. So future studies with large sample size on analyzing multiple genes may be necessary to explain the SNPs interaction.
There are three important questions should be addressed in interpreting the results of meta-analysis: (1) were all relevant studies included in the analysis? This is an important question but difficult to assess. We made every effort to search and collect studies that were sufficient to estimate impact of XRCC1
SNPs on clinical outcomes and survival of lung cancer as of June 2011; (2) was there heterogeneity in the study? The differences in study population including age, gender, smoking status, cancer histopathology type and cancer stage, in chemotherapy schedule of patients, and in measurement of confounding factors and others may result in study heterogeneity. In the present study, the between-study heterogeneity was analyzed by the Cochran's Q test (P < 0.05) and quantified by I2
. Indeed, heterogeneous effects were observed in one subset in the present study. However, we could not separate studies further to obtain homogeneous groups because there was no information on confounding factors. So we performed initial analyses with a fixed-effect model and confirmatory analyses with random-effect model, if there was significant heterogeneity. At last we found the results were similar between fixed-effect model and random-effect model. On the other hand, when the studies, which are possible source of heterogeneity, were excluded from the analyses, there were similar results observed, which also showed that the heterogeneity did not appear to impact significantly on the results of our analyses; (3) publication bias is an important problem in meta-analysis and occurs if scientific studies with negative or null results fail to get published; this can happen due to bias in submitting, reviewing, accepting, publishing or aggregating scientific literature that fails to show positive results on a particular topic; and it could make scientific literature unrepresentative of the actual research studies [48
]. In our results, no evidence of publication bias was found using standard tests such as inverted funnel and the Begg's test. However, we should know that these methodologies did not completely exclude biases, because there might have been rejection or even non-submission of negative results.
Despite our efforts in performing a comprehensive analysis, limitations of our meta-analysis need to be stated. First, most of the included studies differed in their study designs, such as subject selection, chemotherapeutic protocol et al. The patients in some studies also had surgery or other treatment such as radiotherapy in addition to the chemotherapy. All these confounding factors may influence the homogeneity between studies. Stratified analyses by possible confounding factors such as smoking history, cancer histology, and treatment method, might be able to reduce the heterogeneity as stated above. However, few of these studies reported SNP genotype distribution by subgroups so such analyses were impossible to implement. Second, our analyses mostly used unadjusted estimates, because not all published studies stated adjusted estimates, or when they stated, the estimates in different studies were adjusted by different possible confounding factors. However, when the available adjusted estimates were used in our analyses, the conclusions have not been significantly changed (data not shown). Third, we did not analyze the association between XRCC1 SNPs and progression-free survival because there are only two studies involving progression-free survival. Fourth, the relationship between XRCC1 SNPs and platinum-based chemotherapy toxicities was not able to be analyzed in the present study, because few studies provided the related results.
In conclusion, our meta-analysis suggested that XRCC1 Arg194Trp and Arg399Gln polymorphisms may be associated with overall survival and response to platinum-based chemotherapy in lung cancer patients. Lung cancer patients with XRCC1 194 T allele or XRCC1 399 G allele may benefit from platinum-based chemotherapy. However, to address these issues, future prospective studies with large sample size and even randomized clinical trials may be necessary. Besides, efforts on analyzing multiple genes to find more reliable prognostic or predictive biomarkers and even on studying gene-environment interaction should be made, because it is hard to predict complex clinical outcomes of lung cancer patients by using only single gene.