Background

The correlation between schizophrenia and cancer has been investigated for over a century, but the research on this epidemiological puzzle has produced contradictory findings [1, 2]. According to recent studies, reduced risk and incidence for cancer were found to be associated with schizophrenia [3–5]. The explanations proposed for this finding have included genetic factors, neuroleptic medication, and environmental aspects [5–7]. The genetic basis of this lower risk has been ascribed to specific protective mechanism against cancer found in schizophrenia patients [7]. But until now p53 is the only gene to have been reported as being involved in this protective mechanism [8]. Recently, Park et al. [8] made a genetic association analysis of SNPs in the p53 gene between 104 lung cancer patients and 179 schizophrenia patients, and the results indicated that the p53 polymorphisms might be a genetic marker for lower susceptibility to lung cancer in schizophrenia patients.

The X-ray repair complementing defective repair in Chinese hamster cells 4 (XRCC4) gene is located on chromosome 5q14.2, which showed loss of heterozygosity (LOH) in sporadic colorectal cancer (Ratio 37.7%), and high ratio LOH indicates the presence of tumor suppressor locus. Moreover, in this region nearby allelic losses for tumor suppressor genes have been suggested as being associated with colorectal tumorigenesis [9]. The protein encoded by XRCC4 consists of 336 amino acid residues distributed among 8 exons, and has a long helical stem domain responsible for multimerization and interaction with DNA ligase IV [10]. By forming a complex with DNA ligase IV and DNA-dependent protein kinase, XRCC4 functions in the repair of DNA double-strand breaks (DSBs) by non-homologous end joining (NHEJ) and the completion of V(D)J recombination events [11].

The NHEJ pathway is required not only for normal development but also for suppression of tumors. Since it is one of the ubiquitous NHEJ components [11], and four other SNPs (rs6452526, rs2662238, rs35268 and rs2386275) from the HapMap project database http://www.hapmap.org and dbSNP http://www.ncbi.nlm.nih.gov/SNP/ to cover a ~205.7 kb region of XRCC4. All five markers are intronic SNPs (Figure 1). We genotyped these SNPs by the TaqMan® assay method using the ABI 7900 DNA detection system (Applied Biosystems, Foster City, California). All probes and primers were designed by the Assay-on-Design service of Applied Biosystems. The standard PCR was performed using the Taqman® Universal PCR Master Mix (Applied Biosystems) reagent.

Figure 1
figure 1

The five SNPs selected among the genomic region of XRCC4.

Full size image

Statistical analysis

We analyzed Hardy-Weinberg equilibrium, allelic and genotypic distributions on http://202.120.31.137/myanalysis.php[18], a user-friendly platform which integrates efficient analysis tools for association studies. Monte Carlo simulation strategy and χ2 test were used to compare the discrepancies of allele and genotype frequencies between CRC and schizophrenia patients [19]. We estimated linkage disequilibrium (LD) using "D" as the standardized measure, for all possible pairs of SNP loci. We corrected the P values of association analysis for multiplicity using a false discovery rate (FDR) controlling procedure [20]. Power calculations were performed using the G*Power program [21]. All the p values in the study were two-tailed and the significance level was set at p = 0.05. Results were also expressed in terms of the odds ratio (OR) and 95% confidence interval (CI), which were calculated on the website http://www.hutchon.net/ConfidOR.htm. The program UNPHASED was used to estimate haplotype frequencies [22].

Results

Genotype distributions of all five SNPs showed no significant deviations from Hardy-Weinberg equilibrium in either CRC or schizophrenia subjects. The allele and genotype frequencies of the SNPs are listed in Table 1. The power of rs6452526, rs2662238, rs963248, rs35268 and rs2386275 were of 0.767, 0.782, 0.852, 0.695 and 0.844 respectively, with an OR of 1.5 (95% CI). For genetic polymorphisms rs6452526, there were statistically significant differences in allele frequencies between the 312 CRC subjects and the 270 schizophrenia subjects (p = 0.004, p = 0.020, after the FDR correction). The T allele and TT genotype of rs6452526 were significantly less common in the CRC group compared to the schizophrenia group (allele, 13.4% versus 20.2%, OR 0.61, 95% CI 0.44-0.86; genotype, 2.3% versus 5.3%). Moreover, the allele frequency of rs35268 showed significant difference between the CRC and schizophrenia groups (p = 0.028, p = 0.070, after the FDR correction). The T allele of this marker was more common in CRC subjects than in schizophrenia subjects (90.9% versus 86.6%, OR 1.54, 95% CI 1.05-2.26).

Table 1 Allele and genotype frequencies of 5 SNPs among colorectal cancer and schizophrenia patients
Full size table

After analyzing linkage disequilibrium for each pair of the SNPs in the CRC and schizophrenia subjects (Table 2), two groups of markers (rs6452526-rs2662238 and rs963248-rs35268) were observed to be in strong LD respectively [23]. We therefore adopted the haplotype distributions for these SNPs in the later analysis. Haplotypes were omitted from analysis if the estimated haplotype probabilities were less than 3% in either the CRC or schizophrenia groups.

Table 2 Estimation of linkage disequilibrium between the 5 SNPs
Full size table

We selected only those haplotypes with significant frequency discrepancies between CRC and schizophrenia subjects (Table 3) for presentation. Haplotype analysis of these polymorphisms revealed some significant global p values (Table 4). One two-SNP-based haplotype and one five-SNP-based haplotype showed significant frequency difference between the CRC and schizophrenia groups. The haplotype combining all five markers was the most significant, giving a global p = 0.0005 (p = 0.0141, after the FDR correction). As its frequency was greater in the CRC group than in the schizophrenia group, the haplotype C-G-A-T-G (rs6452526-rs2662238-rs963248-rs35268-rs2386275) was found to be correlated with an increased odds ratio for CRC (p = 0.001, OR 2.07, 95% CI 1.32-3.24, p = 0.025, after the FDR correction). However, none of other significant haplotypes survived the FDR correction.

Table 3 Estimated haplotype frequencies and association significance
Full size table
Table 4 Global p values of estimated haplotypes of the 5 SNPs within XRCC4
Full size table

In the power calculations using the G*Power 3 program, our sample size had greater than 90% power to detect a significant (α < 0.05) association for alleles, genotypes and haplotypes when an effect size index of 0.1 (corresponding to a "weak" gene effect) was used.

Discussion

The question whether schizophrenia is linked to a decreased risk of develo** cancer has been the subject of considerable research over a number of decades. The debate has centered on the incidence of malignancies in patients with schizophrenia which have been variously reported to be higher, lower, or similar to that in the general population. However, the majority of studies in the last decade have suggested that schizophrenia patients are protected against cancer in general [3–5].

Several possible explanations have been offered for the evidence that cancer risk and incidence are reduced in patients with schizophrenia. In one of the earlier studies, Mortensen observed a lower risk of prostate cancer among schizophrenia patients who had been exposed to prolonged neuroleptic medication [6]. More recent theories have focused on the possibility that increased apoptosis could account for the neurodevelopmental abnormalities as well as tumor resistance linked to schizophrenia [7], an explanation supported by Park et al. in a Korean cohort [8]. XRCC4 encodes a nuclear phosphoprotein that multimerizes and interacts with DNA Ligase 4 and DNA-dependent protein kinase, playing a critical role in the non-homologous end joining (NHEJ) pathway [11]. In mice, XRCC4-deficiency is embryonically lethal with a massive neuronal apoptosis, and XRCC4 has been observed to interplay with p53 in the regulation of apoptosis, indicating that XRCC4 is crucial for maintaining genomic stability and for the suppression of tumors [11], and four other SNPs selected from the HapMap project database http://www.hapmap.org and dbSNP http://www.ncbi.nlm.nih.gov/SNP/. At two of the five markers (rs6452526 and rs35268) there were statistically significant discrepancies of allele or genotype frequencies between CRC and schizophrenia subjects. We observed that the T allele of r6452526 was less frequent in CRC than in schizophrenia subjects (OR 0.61, 95% CI 0.44-0.86, p = 0.004, p = 0.020, after the FDR correction), indicating that it might be a protective factor for schizophrenia patients against CRC; whereas the higher frequency of the T allele for rs35268 in CRC than in schizophrenia subjects (OR 1.54, CI 1.05-2.26, p = 0.028, p = 0.070, after the FDR correction) implies that the T allele might be a risk allele for CRC.

Since haplotypes constructed from closely located markers will typically increase the statistical power for association with the disease, we performed haplotype analysis in SNPs with strong linkage disequilibrium (D' > 0.7). Our results indicated that one two-SNP-based haplotype and one five-SNP-based haplotype showed significant global frequency difference between the CRC and schizophrenia groups (Table 4). The most significant window spanned all five SNPs giving a global p = 0.0005 (p = 0.0141, after the FDR correction). In addition, we observed that the most significant haplotype C-G-A-T-G (rs6452526-rs2662238-rs963248-rs35268-rs2386275, OR 2.07, 95% CI 1.32-3.24, p = 0.001, p = 0.025, after the FDR correction) was twice as common in the CRC group (15.2%) as in the schizophrenia group (8.2%), suggesting that C-G-A-T-G is a risk haplotype for CRC. In addition, we have compared the frequencies of C-G-A-T-G haplotype between CRC patients and normal controls, but the individual P value showed there was no difference.

Park et al. using genetic association analysis, showed that p53, as a protective gene, induced a lower incidence of lung cancer among Korean patients with schizophrenia [8]. Using a similar research strategy, we found that polymorphisms of XRCC4 may confer genetically reduced susceptibility to CRC among Chinese schizophrenia patients. Compared to the study of Park et al. our sample was twice as large and involved better age-matched subjects, thus decreasing the possibility of false-positive results [24]. However, the genetic contribution to this reduced risk for CRC among schizophrenia subjects is likely to involve a series of susceptibility loci, each influencing but not determining overall risk. In addition, our sample size is relatively small, thus additional replication studies using more SNPs in large non-Asian samples are needed. Genetic, environmental, and pharmacological influences are all thought to be contributing factors, which poses a major challenge for solving this epidemiological puzzle [5–7].

Conclusions

In summary, our results provide a first indication that XRCC4 might be a potential protective gene with respect to schizophrenia, conferring decreased susceptibility to colorectal cancer in the Han Chinese population. Genetic studies to date have focused on subjects of Asian ethnicity, and further research needs to be undertaken in other ethnic groups. Replicating studies with more markers and with larger samples will be necessary to clarify the correlation between schizophrenia and cancer.