Recent evidence indicates a potential prognostic and predictive value for germline polymorphisms in genes involved in cell cycle control. We investigated the effect of cyclin D1 (CCND1) rs9344 G>A in stage II/III colon cancer patients and validated the findings in an independent study cohort. For evaluation and validation set, a total of 264 and 234 patients were included. Patients treated with 5-fluorouracil-based chemotherapy, carrying the CCND1 rs9344 A/A genotype had significantly decreased time-to-tumor recurrence (TTR) in univariate analysis and multivariate analysis (hazard ratio (HR) 2.47; 95% confidence interval (CI) 1.16–5.29; P=0.019). There was no significant association between CCND1 rs9344 G>A and TTR in patients with curative surgery alone. In the validation set, the A allele of CCND1 rs9344 G>A remained significantly associated with decreased TTR in univariate and multivariate analyses (HR 1.94; 95% CI 1.05–3.58; P=0.035). CCND1 rs9344 G>A may be a predictive and/or prognostic biomarker in stage II/III colon cancer patients, however, prospective trials are warranted to confirm our findings.
Colon cancer is the second leading cause of cancer death in Europe and the United States. Approximately 50% of patients with colon cancer develop synchronous or metachronous metastases. The 5-year survival rate of colon cancer patients with metastatic disease is <10%.1, 2
In current practice, the majority of colon cancer patients do not benefit from adjuvant treatment, either because they were cured by surgery alone or because they will relapse despite adjuvant treatment. It is therefore essential to identify those patients who will benefit from adjuvant therapy, sparing others needless toxicity and the financial burden of chemotherapy that will not work. There is significant clinical interest in the identification of prognostic and predictive biomarkers that will improve clinical outcome through patient selection.3, 4
Currently, the tumor-node-metastasis stage is the only proven prognostic marker to aid in the identification of patients with aggressive disease.5, 6, 7 The once prevailing ‘one size fits all’ approach of cancer therapy is now becoming a relict of the past. For example, microsatellite instability (MSI) is considered to be a strong and validated prognostic marker in stage II colon cancer, and it is currently the only such biomarker in this setting. In the appropriate clinical setting, it has been advocated that MSI data may be used in clinical decision making, particularly in stage II colon cancer, for which a favorable outcome of the patients with MSI-high tumors suggests that these patients should not receive adjuvant chemotherapy.8
As our knowledge of the molecular characteristics of patients and this disease has expanded, the significant heterogeneity that exists in both has become more apparent. The existence of prognostic and predictive biomarkers would provide a strategy for stratifying colon cancer patients for risk of tumor relapse and chemo resistance, which in turn would allow treatment options to be tailored to the individual.
Current evidence indicates potential prognostic and predictive value for germline polymorphisms in genes involved in cell cycle control.9 Essential regulators of cell cycle progression of the G1/S phase are cyclin-dependent kinases (CDKs).10 CDKs are heterodimeric complexes composed of a catalytic subunit and a regulatory subunit, called cyclin.11 D-type cyclins (D1, D2 and D3) and their catalytic partners CDK4 or CDK6 function as critical integrators of mitogenic signals for cells.11 The active CDK/cyclin complex phosphorylates and thereby inactivates the tumor-suppressor protein retinoblastoma leading to transcription of proteins necessary for progression through S phase.12 Cyclin D1 (CCND1) overexpression disrupts the normal cell cycle and leads to early passage through G1/S transition.13, 14 A common and functional single-nucleotide polymorphism (CCND1 rs9344 G>A [G870A]) at codon 242 affects splicing of CCND1 transcript and causes abnormal cell proliferation.15, 16
In this study, we investigated the prognostic and predictive effect of CCND1 rs9344 G>A in patients with stage II and III colon cancer. In a second step, the findings were validated in an independent study cohort.
Materials and methods
For the evaluation set, a total of 264 patients with histologically confirmed stage II and III colon cancer treated at the Division of Clinical Oncology, Department of Medicine, Medical University of Graz (MUG) from 2000 to 2009 were included in this study. Stage III and high-risk stage II patients were treated with 5-fluorouracil (5-FU)-based chemotherapy (n=192). For the validation set, a total of 234 patients with histologically confirmed stage III and high-risk stage II colon cancer treated at the Norris Comprehensive Cancer Center/University of Southern California (NCCC/USC) or the Los Angeles County/USC-Medical Center (LAC/USCMC) from 1987 to 2007 were included. All patients in the validation set were treated with 5-FU-based chemotherapy.
High-risk stage II colon cancer patients were defined if they presented with at least one of the following features: lymph node sampling <12; poorly differentiated tumor; vascular, lymphatic or perineural invasion; tumor presentation with obstruction or perforation and pT4.
All patients were included in the colon cancer surveillance program of MUG, NCCC/USC or LAC/USCMC, providing history and physical examination and carcinoembryonic antigen determination every 3 months for 3 years and every 6 months at years 4 and 5 after surgery, colonoscopy at year 1 and thereafter every 3–5 years and computed tomography scans of chest and abdomen every 6 months for the first 3 years.
Patient data were collected retrospectively through chart review. Whole blood was collected at the time of diagnosis and stored at −80 °C. The study was approved by the Institutional Review Boards of MUG and USC and all study participants signed informed consent for the analysis of molecular correlates.
Genomic DNA was extracted from whole blood samples using the QIAmp-kit (Qiagen, Hilden, Germany). In the evaluation set, genotyping was performed using a 5 V-nuclease assay (TaqMan, Vienna, Austria) with primers and probes designed and manufactured using Applera’s ‘Assay by-Design’ custom service (Applied Biosystems, Vienna, Austria). PCR and evaluation of fluorescence data were performed as recently described.17 For each sample, one negative control containing water instead of DNA was added to check for contamination. In the validation set, genotyping was performed by direct DNA sequencing. For genotyping quality control purposes, a total of 10% of the samples were re-analyzed in both study sets. The investigators responsible for genotyping were blinded to the clinical data.
The end point of the study was time-to-tumor recurrence (TTR). TTR was calculated from the date of diagnosis of colon cancer to the date of the first observation of tumor recurrence. TTR was censored at the time of death or at the last follow-up if the patient remained tumor recurrence-free at that time. Allelic distribution of CCND1 rs9344 G>A was tested for deviation from Hardy–Weinberg equilibrium using χ2-test. The true mode of inheritance of CCND1 rs9344 G>A is not established yet and we assumed an additive, dominant or recessive genetic model where appropriate. The association of CCND1 rs9344 G>A with TTR was analyzed using Kaplan–Meier curves and log-rank test. In the multivariate Cox-regression analysis, the model was adjusted for stage and type of adjuvant therapy. Case-wise deletion for missing polymorphisms was used in univariate and multivariate analyses. All analyses were performed using SAS 9.2 (SAS Institute, Cary, NC, USA).
The baseline characteristics of the evaluation and validation set are summarized in Table 1. In the evaluation set, the median age at time of diagnosis was 62.2 years (range 25–83), with a median follow-up time of 53.5 months (range 7–125). In the validation set, the median age at time of diagnosis was 59 years (range 22–87), with a median follow-up time of 52.8 months (range 4.8–201.6). Median overall survival has not been reached yet in both study cohorts.
The genotyping quality control by TaqMan and direct DNA sequencing provided a genotype concordance of 99%. Genotyping was successful in 97% of MUG patients (wild type: 77, heterozygous mutant: 129 and homozygous mutant: 50) and 72% of USC patients (wild type: 51, heterozygous mutant: 73 and homozygous mutant: 44). In failed cases, genotyping was not successful because of limited quantity and/or quality of extracted genomic DNA. The allelic frequencies of CCND1 rs9344 G>A were within the probability limits of Hardy–Weinberg equilibrium (MUG cohort: P=0.96, USC cohort: P=0.09).
In the evaluation set, we found no significant association between CCND1 rs9344 G>A and TTR in both univariate and multivariate analysis (hazard ratio (HR) 0.87; 95% confidence interval (CI) 0.58–1.31; P=0.513; HR 0.84; 95% CI 0.55–1.28, P=0.414, respectively). However, in patients treated with 5-FU-based chemotherapy, the CCND1 rs9344 A/A genotype was significantly associated with decreased TTR in the univariate analysis. Patients with at least one CCND1 rs9344 G allele had a median TTR of 102.6 months. In contrast, patient carrying the CCND1 rs9344 A/A genotype showed a median TTR of 84.8 months (HR 2.14; 95% CI 1.02–4.47; P=0.044; Figure 1). There was no significant interaction between CCND1 rs9344 and clinical stage on TTR (P for interaction=0.12) In the multivariate analysis, the CCND1 rs9344 A/A genotype remained significantly associated with decreased TTR (HR 2.47; 95% CI 1.16–5.29; P=0.019). There was no significant association between CCND1 rs9344 G>A and TTR in patients with curative surgery alone in univariate or multivariate analysis (HR 0.70; 95% CI 0.2–2.52; P=0.587; HR 0.45; 95% CI 0.09–2.19; P=0.323, respectively; Figure 2).
In the validation set using a recessive genetic model (G/G and G/A vs A/A), we found no significant association between CCND1 rs9344 G>A and TTR in both univariate and multivariate analysis (HR 0.97; 95% CI 0.56–1.67; P=0.91; HR 0.93; 95% CI 0.53– 1.64, P=0.81, respectively). However, in a dominant model (G/G vs G/A and A/A) the A allele of CCND1 rs9344 G>A was significantly associated with decreased TTR in univariate analysis. Patients harboring the CCND1 rs9344 G/G genotype showed a median TTR of 128.4 months, compared with 60 months for patients carrying at least one A allele (HR 1.91; 95% CI 1.04–3.51; P=0.032; Figure 3). There was no significant interaction between CCND1 rs9344 and clinical stage on TTR (P for interaction=0.13) In the multivariate analysis, the CCND1 rs9344 A allele remained significantly associated with decreased TTR (HR 1.94; 95% CI 1.05–3.58; P=0.035).
To investigate if the differences of allele frequencies between ethnicities (Ensembl genome browser, population genetics, rs9344) within the validation cohort cause the significant associations for different genetic models in the evaluation and validation cohort, we restricted the validation cohort to Caucasians only (n=123). However, we found no significant associations between CCND1 rs9344 G>A and TTR in an additive (HR 1.16; 95% CI 0.48–2.76; P=0.24), recessive (HR 0.79; 95% CI 0.41–1.55; P=0.49) and dominant genetic model (HR 1.50; 95% CI 0.72–3.13; P=0.27) among Caucasians in the validation set.
In this study, the prognostic and predictive effect of CCND1 rs9344 G>A in patients with stage II and stage III colon cancer was investigated. The results indicate a potential predictive effect of CCND1 rs9344 G>A in patients with colon cancer treated with adjuvant 5-FU-based chemotherapy.
Betticher et al.15 showed that the CCND1 rs9344 G>A polymorphism, located in the splicing region of exon 4, leads to alternate splicing of CCND1 mRNA into two transcripts. The altered transcript-b is primarily encoded by the variant allele A, whereas the wild-type mainly encodes transcript-a.15 The main difference affects the C-terminal region, which is responsible for rapid intracellular degradation.18 Variant transcript-b results in increased CCND1 because of a prolonged half-life.15 High CCND1 protein levels have been associated with decreased survival in various malignancies.19, 20, 21, 22, 23
There are limited data investigating CCND1 as a prognostic biomarker in colon cancer, however, overexpression of CCND1 protein was associated with decreased disease-free survival and overall survival in a small study including 123 patients with colorectal cancer.24 We found no association between CCND1 rs9344 G>A and TTR in patients treated with curative surgery alone, indicating no prognostic effect in our study cohort. Our results are in line with a study by McKay et al.25 showing no influence of CCND1 rs9344 G>A on clinical outcome in patients with colorectal cancer.
A growing number of studies evaluated the association between CCND1 rs9344 G>A and chemoresistance in various tumor entities. Li et al.26 investigated CCND1 isoforms in mediating response to DNA-damaging agents by treating human kidney cancer, colorectal cancer and breast cancer cells with doxorubicin or ionizing radiation and showed that transcript-a (encoded by the wild-type G-allele) was more likely to induce cell cycle arrest and double-stranded DNA breaks in contrast to transcript-b. Gautschi et al.27 investigated CCND1 rs9344 G>A in patients with non-small cell lung cancer and found that patients carrying the A-allele show a significant lack of response to platinum-based chemotherapy. In patients with metastatic colon cancer refractory to irinotecan, enrolled in a randomized controlled trial that compared bevacizumab plus cetuximab vs bevacizumab plus cetuximab plus irinotecan, the A/A genotype has been associated with a significant shorter TTR (6.8 vs 8.4 months; P=0.001) in the irinotecan group.28 In third-line therapy setting, patients with metastatic colon cancer treated with cetuximab alone harboring the A/A genotype showed a significant shorter median overall survival compared with those with any G-allele (2.3 vs 8.7 months; P=0.019).29 This is in line with our results showing a decreased TTR in patients treated with 5-FU-based chemotherapy carrying the CCND1 rs9344 A/A genotype.
Interestingly, patients treated at the MUG harboring a heterozygote genotype tended, in their clinical outcome, to be patients carrying the wild-type allele. In contrast, heterozygous patients treated at USC tended to be patients carrying the homozygous mutant genotype. When we validated our finding in an independent study cohort, we found no significant associations in the recessive genetic model, but in a dominant model. To clarify if this finding is dependent on differences of allele frequencies between ethnicities within the study cohort, we restricted the validation cohort to Caucasians only, but found no significant association between CCND1 rs9344 G>A and TTR in any genetic model. We therefore concluded that there was a selection bias in our retrospective cohort study and this may cause the significant associations in different genetic models (recessive model in the evaluation set and dominant model in USC cohort). This conclusion is further supported by the clinical outcome differences in both study cohorts (Figures 1 and 3).
This study has a number of limitations that need to be considered. MSI status was not available in our study cohorts, and therefore no conclusion can be drawn regarding the value of CCND1 rs9344 G>A above and beyond that offered by MSI status, which is now routinely performed in many centers for patients with colorectal cancer. Moreover, because of the retrospective design of our study, a selection bias cannot be fully excluded. Median OS has not been reached yet in either study cohort, therefore no statistical association between the polymorphism and OS could be performed.
In conclusion, CCND1 rs9344 G>A may be a predictive and/or prognostic biomarker in stage II/III colon cancer patients. Larger and prospective study cohorts are warranted to clarify these findings.
This work was supported by the START research grant of the Medical University of Graz and the Daniel Butler Memorial Fund.
About this article
Single Nucleotide Polymorphisms as Prognostic and Predictive Factors of Adjuvant Chemotherapy in Colorectal Cancer of Stages I and II
Gastroenterology Research and Practice (2016)