All the tissues investigated in this study were obtained from 397 consecutive, ethnic Korean, colorectal cancer patients who had undergone a curative resection between January 2003 and August 2006, at Kyungpook National University Hospital (Daegu, Korea). Written informed consent for gene expression analyses was received from all participating patients prior to surgery, and the study was approved by the Kyungpook National University Hospital Institutional Research Board. The diagnosis and staging of the colorectal cancer data was performed according to World Health Organization (WHO) classifications [18] and the tumor, node, and metastasis (TNM) classifications set by the American Joint Committee on Cancer (AJCC) [19].

Due to the enormous number of SNPs in the human genome, an appropriate strategy for efficient selection of those SNPs most likely to contribute phenotypic effects was our first challenge. Thus, a prioritization scheme was created using public databases providing diverse information on potential phenotypic risks associated with specific SNPs. First, candidate SNPs from CASP genes were collected from web-based databases which included information on the biologic pathways and potential biologic effects of these polymorphisms. Next, based on the allele frequencies recorded for East Asian populations obtained from FASTSNP, those SNPs with frequencies less than 0.1 were excluded. The remaining CASP gene SNPs were then scored according to particular phenotypic risks, and then, based on the algorithm suggested in a previous report [20], they were ordered according to the sum of their risk scores. Among the 13 polymorphisms in the CASP3, CASP6, CASP7, CASP8, CASP9, and CASP10 genes that have been reported to be potentially functional or otherwise associated with cancer risk [11-16], ten polymorphisms (CASP3rs2705897; CASP6 rs1042891, rs2301717; CASP7 rs2227310, rs11593766; CASP8 rs3769818, rs3834129; CASP9 rs1052571, rs4645978; CASP10 rs13006529) were examined. CASP8 rs1045485 and CASP10 rs13010627 were excluded as they are rare or nonexistent in Asian populations [11,21].

Genomic DNA was extracted from fresh colorectal mucosal tissue at the time of surgery using a Wizard genomic DNA purification kit (Promega, Madison, WI). The 10 selected CASP gene polymorphisms were then determined using a reverse transcription polymerase chain reaction (PCR) genotyping assay. For quality control, the genotyping analysis was performed blind as regards the subjects.

The selected, PCR-amplified DNA samples (n=2, for each genotype) were also examined by DNA sequencing to confirm the genotyping results.

The genotypes for each SNP were analyzed as a categorical variable for three-groups (reference model), and also grouped according to a dominant and recessive model. The survival estimates were calculated using the Kaplan-Meier method. As related to the appearance of SNPs of the CASP genes, the differences in patient overall survival (OS) or disease-free survival (DFS), were compared using log-rank tests. Cox's proportional hazard regression model was used for the multivariate survival analyses, whereby the analyses were adjusted for potential prognostic factors including age (median age, 63 years;≤63 years vs.>63 years), differentiation (well to poor differentiation), preoperative carcinoembryonic antigen (CEA) level (normal vs. elevated), and pathologic stage (0 to IV). The hazard ratio and 95% confidence interval were also estimated. A cutoff p-value of 0.05 was adopted for all statistical analyses. The statistical data were obtained using an SPSS ver. 11.5 (SPSS Inc., Chicago, IL) or SAS Genetic software (SAS Institute, Cary, NC).

The median age of the patients was 63 years (range, 21 to 85 years), and 217 (54.7%) of the total patients were male. Of the total patients, 218 (54.9%) were diagnosed with colon cancer and the remaining with rectal cancer. Laparoscopic surgery was performed on 271 (68.3%) patients, while the others received an open colorectal resection. The pathologic stages after the surgical resection were as follows: stage 0/I (n=86, 21.7%), stage II (n=146, 36.8%), stage III (n=145, 36.5%), and stage IV (n=20, 5.0%). Among the 291 patients with stage II or III disease, 275 (94.5%) received adjuvant chemotherapy using either 6 cycles of 5-fluorouracil/leucovorin ± radiotherapy (n=133), 12 cycles of oxaloplatin, leucovorin, and fluorouracil (FOLFOX-4) (n=12), 8 cycles of capecitabine (n=20), or doxifluridine during a treatment period of one year (n=110) (Table 1).

At the time of the last analysis, 80 patients had experienced disease relapse and 37 had died as a result of colorectal cancer disease progression. However, the deaths of 2 patients were unrelated to colorectal cancer. At the median follow-up duration of 34.8 months (range, 1.1 to 65.7 months), the estimated 5-year OS and DFS for all the patients was 89.3±1.6% and 72.3±4.3%, respectively. Survival results differed according to the pathologic stage (p<0.001).

The 10 CASP gene SNPs were successfully amplified in more than 95% of patients. The frequencies of occurrence for each genotype also conformed to the Hardy-Weinberg equilibrium (p>0.05). There were no sexual differences observed in relation to any genotype and allele. No correlation was observed between any given frequency of the genotype or allele and clinicopathologic parameters, including for T, N, or M stage patients (data not shown). A multivariate survival analysis including pathologic stage, patient age, differentiation, and CEA level demonstrated there was no association between any genotypes with outcomes for DFS or OS (Table 2). In haplotype analysis of the CASP6, CASP8 and CASP9 genes, none were associated with colorectal cancer prognosis. In terms of clinicopathologic parameters in a Cox model, patient age, CEA level, and TNM stage were all significant prognostic factors for DFS and OS (Table 3).