Despite aggressive surgical resections and a combination of chemotherapy or radiotherapy, the prognosis for relapse in CRC patients is still dismal [5,16,17]. Disease-free interval and progression-free survival could be taken as the best predictors of long-term cure and prognosis in CRC [18]. To date, most studies regarding prognosis of postoperative CRC regional recurrence have focused on tumor characteristics, but less on host-related characteristics [16,17,19–21]. The factors increasing local recurrence rates of CRC should be clearly described. The highest quintile of Western diet (compared to the lowest quintile) can increase the hazard ratio of local/regional recurrence by about 2.5-fold [22]. Exercise can decrease the hazard ratio for colon cancer-specific mortality to about 0.50 [23,24]. On the other hand, in a study of 44,788 pairs of twins, statistically significant effects of heritable factors (polymorphisms, etc.) were found to account for 35% of the cases of colon cancer, so that polymorphisms are clearly important in colon cancer [25]. Mention should also be made of the particular importance of decreased expression of ERCC1 in progression to colon cancer, shown in recent studies. A 2012 report indicates that epigenetic repression of ERCC1 is found in 40% of about 1 million crypts in large field defects surrounding colon cancers (in which progression to cancer occurred) and in 100% of colon cancers themselves [26]. Local and systemic treatment modalities, like preoperative chemoradiotherapy, should be planned for patients carrying these risk factors.

Some host-related factors may predict the risk of metastasis after surgery of CRC [27]. There is accumulated evidence suggesting that the genetic polymorphisms involved in metabolizing enzymes, DNA repair, cell growth, differentiation and carcinogenesis have been linked to inter-individual differences of therapeutic effect in CRC [28]. This genetic polymorphism detection may permit better selection of patients suitable for adjuvant therapy [29]. Genetic analyses appear to be a promising tool in the development of personalized treatment plans for CRC [8,30]. Nevertheless, a comprehensive analysis of genetic polymorphisms in Taiwanese CRC patients with poor prognostic factors remains rare. This study aimed to examine the feasibility of developing a multigene predictor of poor prognosis in Taiwanese CRC patients, and it will be helpful in identifying those who may need aggressive treatment.

In the current study, we have attempted to move beyond single gene polymorphisms to a more comprehensive evaluation to identify genomic variants and patterns that may help predict CRC prognosis. The in vitro studies suggest a C to T transition at codon 118 of the ERCC1 gene associated with higher ERCC1 mRNA levels resulting in resistance to platinum drugs [31]. However, the clinical data regarding the assumed relationship between ERCC1 codon 118 polymorphism and platinum sensitivity is controversial [13,32,33]. Metastatic CRC patients with ERCC1 codon 118 C/T or T/T genotypes tended to have a marked increase of ERCC1 protein expression levels and had shorter progression-free survival and overall survival than patients with the other polymorphisms [7]. Codon 118 C→T polymorphism of ERCC1 was identified as an independent prognostic factor in Asian CRC patients [7]. In Caucasian advanced CRC patients who are treated with oxaliplatin, 5FU and leucovorin (LV), ERCC1 118T/T, XPD 751A/C and XPD 751C/C genotypes are independently associated with poor progression-free survival and short-term survival [12,13]. In the results of the present study, we also found that ERCC1 codon 118T/T combined with XPD codon 751A/A was associated with increased recurrence risk in postoperative Taiwanese CRC patients. The possible mechanism of the combination of ERCC1 codon 118T/T and XPD codon 751A/A and the increased recurrence risk relate to poor chemotherapy efficacy in combination with DNA repair system impairment, which led to cancer recurrence.

Enrolled in this study were originally 257 American Joint Commission on Cancer/International Union Against Cancer (AJCC/UICC) stage II–III CRC patients (median age, 61.5 ± 11.3 years) who were admitted to the Department of Surgery at Fooyin University Hospital, ZuoYing Armed Forces General Hospital and Kaohsiung Medical University Hospital for elective surgery from January 2006 to June 2009. Patients with other malignant diseases noted in the medical history were excluded. Patients, who were required to be at least 18 years of age with a life expectancy of 3 months, have an Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 2. Written informed consent was obtained from all subjects and/or guardians for use of their blood samples. This study was approved by the Institutional Review Board of Kaohsiung Medical University Hospital and was not supported by any commercial company. To avoid contamination of skin cells, sampled blood was taken via an intravenous catheter, and the first few milliliters of blood were discarded. Total RNA was immediately extracted from the peripheral whole blood, and then it served as a template for complementary DNA (cDNA) synthesis. The institutional review boards of the three hospitals approved the acquisition of samples, as well as their subsequent use. Written informed consent was obtained from all participants.

Pretreatment evaluation included a complete medical and clinical physical examination, baseline measurement of tumor size based on computed tomography scans, X-ray or other radiographic means and serum biochemistries. Patients were administered six 8-week cycles of adjuvant chemotherapy. Each cycle consisted of leucovorin 500 mg/m² administered as a 2-h infusion and given weekly for six doses and 5FU 500 mg/m² administered as an intravenous bolus 1 h after the start of leucovorin infusion and repeated weekly for 6 doses. This cycle was then repeated after a 2-week rest period. Postoperative surveillance consisted of medical history, physical examination and laboratory studies at 3-month intervals. Abdominal ultrasonography or computed tomography was performed at 6-month intervals, and chest radiography, bone scans and total colonoscopy were performed annually. Patients were followed up at 3-month intervals for 2 years and at 6-month intervals thereafter.

The follow-up endpoint was December 2010. The median follow-up time was 36.4 months (range, 24–46 months). This follow-up time is long enough to draw conclusions regarding regional recurrence [34]. Clinical stage and pathological features of primary tumors were defined according to the criteria of the American Joint Commission on Cancer/International Union Against Cancer (AJCC/UICC) [35]. Development of new post-operative tumor growth restricted to the anastomosis or the region of the primary operation was defined as postoperative regional recurrence.

In our previous study results, we found that XPD, ERCC1 and EGFR gene polymorphisms were related to stage II–III and metastatic CRC prognosis [36–38]. Constitutional gene polymorphisms were analyzed via DNA extraction from 4 mL peripheral blood using a PUREGENE^® DNA Isolation Kit (Gentra Systems, Inc., Minneapolis, MN, USA). All genomic DNA from patients were examined using the polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) approach to determine the genotypes of ERCC1, XPD and EGFR. Following digestion with suitable restriction enzymes, PCR fragments were separated on a 2.5%–3.0% agarose gel and visualized after ethidium bromide staining. This study also utilized the automated sequencing approach to confirm PCR-RFLP results. All primers utilized in this study were designed by using Primer 3 freeware [39]. Table 6 presents the primer sequences and restriction enzymes.

The observed numbers of each genotype were compared with those expected for Hardy-Weinberg equilibrium (HWE) using the χ² test. All data were analyzed using the Statistical Package for the Social Sciences Version 14.0 (SPSS, Inc., Chicago, IL, USA). The correlation between each polymorphism and survival was assessed using the hazard risk ratio and a 95% confidence interval (CI). To clarify the clinical significance of these combined genotypes as predictors of prognosis, a multivariate adjustment was performed by multivariate Cox proportional hazards regression analysis. Overall, statistical significance was defined as p-value < 0.05. In the tests with multiple comparisons, the Bonferroni correction was applied to the significance level in which a type I error of 0.0167 (0.05/3) was used.