MATERIALS AND METHODS
Computerized searches of the PubMed and Web of Science databases were conducted using a combination of the following keywords: “XRCC1” (“X-ray repair cross complementing 1” OR “X-ray repair cross complementing group 1” OR “X-ray cross complementing repair gene 1”), “AML” (“acute myeloid leukemia” OR “acute myeloblastic leukemia” OR “acute myelocytic leukemia”), “ALL” (“acute lymphocytic leukemia” OR “acute lymphoblastic leukemia”), “CLL” (“chronic lymphocytic leukemia” OR “chronic lymphoblastic leukemia”), “CML” (“chronic myeloid leukemia” OR “chronic myelocytic leukemia” OR “chronic myeloblastic leukemia”), “leukemia”, and “hematological malignancies”. All references cited in the selected studies were also reviewed to identify additional relevant work.
Published studies were included if they satisfied the following criteria:
- Only studies published in journals in English were included in the analysis;
- Used a case-control design;
- Focused on the association between the XRCC1 Arg399- Gln SNP and risk of AML, ALL, CML, or CLL;
- Provided sufficient data on the distribution of the XRCC1 Arg399Gln SNP in leukemia and in controls, or sufficient information for such data to be calculated.
The following information was extracted from each study: first author, publication year, subjects’ ethnicity, disease type, number of cases and controls, Hardy–Weinberg equilibrium (HWE) for the controls’ distribution of genotypes, and main results about associations between the XRCC1Arg399Gln SNP and risk of leukemia.
The association between the XRCC1 Arg399Gln SNP and risk of different types of leukemia in different populations was evaluated under the allele contrast (Gln versus Arg), homozygote contrast (Gln/Gln versus Arg/Arg), dominant (Gln/Gln + Arg/Gln versus Arg/Arg), and recessive (Gln/Gln versus Arg/Gln + Arg/Arg) models. Studies wherein the distribution of the XRCC1 Arg399Gln genotypes among the controls deviated from HWE were excluded from the meta-analysis.
Review Manager software (version 5.3) was used for the meta-analysis. Raw data of genotype distribution were used to calculate the study-specific estimates of odds ratios (ORs) and 95% confidence intervals (CIs). The heterogeneity of the studies was assessed using Cochran’s Q test, and was considered statistically significant at P<0.10. Heterogeneity was also quantified using the heterogeneity index (I2) statistic, with I2>50% indicating the presence of a high degree of heterogeneity.6
The strength of associations between the XRCC1 Arg399Gln SNP and different types of leukemia risk was assessed by the ORs and the corresponding 95% CIs. The significance of the pooled ORs was determined by the Z-test, and the threshold for significance was set at P<0.05. The fixed-effects model (Mantel–Haenszel methods) was used when there was no substantial heterogeneity. In the event of substantial heterogeneity, sensitivity analysis was performed by excluding individual studies; outlying studies were identified and excluded, and the I2 estimates for these different sets of studies were examined. The random-effects model (DerSimonian and Laird’s method)7 was used when removal of particular studies did not render the heterogeneity insignificant (ie, I2<50%).
Potential publication bias was estimated by constructing funnel plots. If most of the data appeared at the top of a funnel plot and was distributed roughly symmetrically, this would suggest the absence of obvious publication bias, and vice versa.8 There was no need to construct funnel plots when there were too few (ie, less than five) analyzed studies.