Abstract
Background: The adjuvant chemotherapy trial of TS-1 for colon cancer phase III trial was designed to validate the non-inferiority of the oral fluoropyrimidine S-1 to uracil and tegafur/leucovorin as adjuvant chemotherapy for stage III colonic cancer. As a prospective biomarker study of this trial, DNA copy number was studied using formalin-fixed, paraffin-embedded specimens. Materials and Methods: FFPE blocks were obtained from 795 patients of the 1,535 patients enrolled in the study. The quality of extracted DNA was assessed using arbitrarily primed polymerase chain reaction and microfluidic analysis. Genomic copy-number alterations in cancer were analyzed by high-density single-nucleotide polymorphism arrays. Copy-number changes in Japanese patients with colonic cancer were compared with those in Western countries using data from a previously reported meta-analysis. We then compared genome-wide segment copy number and clinicopathological features of colorectal cancer. Results: Genome-wide copy number was analyzed in 161 samples and DNA copy-number alteration profiles showed frequent DNA copy-number gains at chromosome 7, 8q and 13, and losses at 4, 5q, 8p, 17p and 18q. The weighted kappa statistic from comparing copy-number alteration status with data from Western countries was 0.828 (95% confidence interval=0.786 -0.871). DNA copy-number alterations of 8,684 segments were compared with clinicopathological features in 161 patients. Location of the tumor correlated with genomic segments of chromosome 4, 5, 7, 8, 13, 14, 18 and 20. Differentiation of the tumor correlated with segments in chromosome 4, 6, 8, 11, 13, 14,15, 16, 17 and 20. Conclusion: Somatic copy-number alteration profiles of stage III colonic cancer in the Japanese ACTS-CC trial closely agreed with the results of previous Western studies. Location and differentiation of the tumor correlated with DNA copy-number alterations. Our findings will facilitate understanding the characteristics of colonic cancer. Further investigation may contribute to the exploration of valid biomarkers.
- Colonic cancer
- biomarker
- adjuvant chemotherapy
- DNA copy-number
- phase III
- S-1
- UFT
- ACTS-CC trial
Colorectal cancer (CRC) is a leading cause of cancer mortality worldwide. Studies performed in the 1990s demonstrated that postoperative adjuvant chemotherapy using intravenous administration of fluorouracil (FU) plus leucovorin (LV) improved the survival of patients with stage III (T, any N1-2, M0) colonic cancer (1, 2). Postoperative intravenous therapy with 5-FU/LV has come to be a standard adjuvant chemotherapy for colonic cancer. Subsequent phase III trials have demonstrated the non-inferiority of the oral fluoropyrimidines capecitabine and uracil plus tegafur (UFT)/LV owing to their easy administration (3, 4). Oral fluoropyrimidines have been widely employed as adjuvant therapy for colorectal cancer in Japan (5, 6). The randomized controlled phase III, the adjuvant chemotherapy trial of TS-1 for colon cancer (ACTS-CC) (TRICC0706), was designed to evaluate the non-inferiority of S-1 (an oral fluoropyrimidine consisting of tegafur, 5-chloro-2,4-dihydropyrimidine, and potassium oxonate) to UFT/LV in stage III colonic cancer and rectosigmoidal cancer (7). Screening of suitable patients for adjuvant chemotherapy is advisable, as response to treatment is diverse and increasingly appears to be dependent on aberrations in individual tumors.
Somatic DNA copy-number alterations (SCNAs; distinguished from germline copy-number variations, CNVs) are common in cancer (8, 9). Two major types of genomic variability, chromosomal instability (CIN) and microsatellite instability (MSI), have been recognized as alternative mechanisms of colorectal carcinogenesis with the more common, CIN, being present in about 65-70% of CRC cases (10, 11). Severity of loss of heterozygosity (LOH) also differs between tumors: some have LOH at many loci in different chromosomes, whereas others have less frequent LOH. These differences might reflect biological behavior of CRC (12).
Recently, the prognostic relevance of SCNAs in CRC has been suggested (13-16). However, in order to identify the major changes influencing clinical outcome, it is necessary to carry out detailed assessments such as array comparative genomic hybridization (CGH) based on large-scale prospective studies. Moreover, for the development of a biomarker study, it is important to include prospective translational studies with stringent quality control for reliable analysis (17). As the clinical outcome of CRC has been shown to differ between countries and regions (18), identification of biological differences between carcinomas is useful.
The present study used genome-wide copy-number analysis to identify candidate loci influencing the prognosis of patients with CRC or the efficacy of oral fluoropyrimidines as a biomarker study of the ACTS-CC trial. This report outlines the biomarker study of the ACTS-CC trial and the analysis of SCNAs in colonic cancer tissues from Japanese patients.
Materials and Methods
Sample collection. The biomarker study was predefined in the ACTS-CC protocol. Formalin-fixed paraffin-embedded (FFPE) cancer specimens were obtained from patients with stage III colonic cancer who underwent curative surgery, and were randomly assigned to receive five courses of UFT/LV [300 600 mg/day according to body surface area (BSA)] and LV (75 mg/day) on days 1 28 followed by 7 days of rest, or four courses of S-1 (80, 100, or 120 mg/day according to BSA) on days 1 and 28 followed by 14 days of rest. This study was conducted in accordance with the ‘Declaration of Helsinki’ and the ‘Ethical Guidelines for Clinical Research’. The protocol was approved at each participating site by the relevant independent Ethics Committee or Institutional Review Board. All patients provided written informed consent before participating. The study protocol of ACTS-CC trial was registered at ClinicalTrials.gov (NCT00660894).
DNA preparation. Manual microdissection was performed to extract genomic DNA (gDNA) from 10-μm-thick FFPE specimens. After a review by a pathologist of representative hematoxylin and eosin (H&E)-stained slides to estimate the tumor load per sample, samples were stained with nuclear fast red (Sigma-Aldrich, St Louis, MO, USA) and prepared for manual microdissection. Malignant cells were selected under magnification (×5 to ×10), and dissected using a scalpel as described previously (17). After proteinase K treatment, phenol-chloroform extraction and ethanol precipitation, DNA concentrations were measured by spectrophotometry (NanoDrop1000: Thermo Fisher Scientific, Waltham, MA, USA).
Arbitrarily primed polymerase chain reaction (AP-PCR) amplification was performed in a total volume of 20 μl containing 3 μl DNA (30, 3, or 0.3 ng), 10 μl of 2 × HotStarTaq DNA polymerase (Qiagen, Hilden, Germany), and the primer 5’-AGGTCACTGA-3’ [19]. PCR cycling conditions were 95°C for 5 min followed by 40 cycles at 94°C for 1 min, 36°C for 1 min, and 72°C for 2 min. DNA quality was then assessed and classified into four grades according to PCR-amplified DNA fragment length in at least two of three input templates for each sample by microfluidic analysis (Bioanalyzer; Agilent Technology, Santa Clara, CA, USA): grade A, ≥700 bp; grade B, <700 bp and ≥400 bp; grade C, <400 bp and ≥200 bp; and grade D, <200 bp. Samples classified as grade A or grade B were used for DNA microarray, and the remainder were stored for future validation steps.
Microarray hybridization. gDNA copy-number alterations were analyzed by high-density single-nucleotide polymorphism (SNP) arrays (Human 250K StyI array; Affymetrix, Santa Clara, CA, USA) using samples of grade A or grade B quality. For GeneChip arrays (Affymetrix), 250 ng DNA was used according to the manufacturer's protocols. Briefly, gDNA was digested with StyI restriction enzyme, and adaptors were added to restriction fragment ends using T4 DNA ligase. Adaptor-modified samples were PCR-amplified using Clontech® Titanium Taq (Takara Bio Inc., Tokyo, Japan), and amplification products were purified using a Clontech® DNA-amplification cleanup kit (Clontech Laboratories, Inc., Mountain View, CA, USA). Purified DNA (90 μg) was fragmented using Affymetrix® Fragmentation Reagent.
We excluded from further analysis any samples that contained <90 μg DNA after PCR amplification, and stored them for future validation steps. Biotin-labeling of fragmented samples was accomplished using the GeneChip DNA-labeling reagent (Affymetrix). Biotin-labeled DNA was hybridized on StyI Affymetrix microarrays at 49°C for 16 18 h in an Affymetrix® rotation oven. After hybridization, a probe array wash and stain procedures were carried out on the automatic Affymetrix® Fluidics Stations according to the manufacturer's instructions. Microarrays were then scanned, and raw data were collected by the Affymetrix® GeneChip Scanner 3000.
Data analysis and software. To detect SCNAs, the scanned image CEL files were imported into Partek Genomics Suite v6.6 (Partek Inc., St. Louis, MO, USA). We filtered out probes that detect SNPs on Sty I fragments >700 bp, then normalized individual probe signals from HapMap project (http://hapmap.ncbi.nlm.nih.gov/) base-line data to 1.0 (haploid genome). The genomic segmentation algorithm used three criteria to find a segment: firstly, breakpoints (region boundaries) were chosen to give optimal statistical significance (p<0.001); secondly, detected regions contained a minimum of 10 probes; and thirdly, segments had a copy number >2.3 or <1.7, as previously described (20). Depending on the differences in sample frequency (delta; % of gain – % of loss), we classified 293 cytogenetic bands as “gain” (Delta >2%), “loss” (Delta <2%), or “unchanged” (2% > Delta >2%). Concordance was calculated by weighted kappa statistics using the R-software package (https://www.r-project.org/). Detected DNA segments, representing potential deletions or highly polymorphic regions, were plotted along chromosomes to produce copy-number genome view. Either Pearson's correlation test or the Wilcoxon test was used to assess correlations between segmental copy number and clinicopathological parameters. All statistical analysis used the SAS software package ver.9.3, JMP software version 8.11 (SAS Institutes Inc., Cary, NC, USA) and R software package.
Results
Patients and sample preparation. In total, 1,535 patients from 358 hospitals were enrolled in the ACTS-CC trial between April 2008 and June 2009. Among the 1,518 patients included in the efficacy analysis, FFPE samples were obtained from 795 patients (52.4%). Genomic DNAs were successfully extracted from 777 samples. The characteristics of the patients are shown in Table I. There were no significant differences between the characteristics of the patients in the efficacy analysis set, the DNA analysis set, and the SNP array analysis set.
Quality assessment of samples and SNP array analysis. Quality assessments were performed for 777 samples (Figure 1). Of these, 482 passed the AP-PCR assessment, and 161 were successfully amplified. These pre-determined quality control processes allowed 616 samples to be prepared for future validation steps. The success rate for SNP array detection was affected by the quality of DNA (data not shown): gDNA samples of grade A quality had a success rate of 45.1%, which was significantly higher than that of grade B samples (28.6%). The average SNP call rate was 86.2% [95% confidence interval (CI)=85.4-87.0%].
Copy-number alterations. Figure 2 shows the frequency of copy-number alterations across all samples. The fraction of samples with copy-number gain (red) or loss (blue) was plotted along each chromosome. The most common changes were losses of chromosomes 4, 5q, 8p, 15q, 17p, and 18q, and gains of chromosomes 7, 8q, 13, and 20. Concordance of the copy-number gain or loss of each cytogenetic band with meta-analysis data reported by Diep et al. (21) was assessed by weighted kappa statistics (Table II, Figure 3). A contingency table is shown for both the present study data (Japan, primary carcinomas, n=161) and meta-analysis data (Western countries, primary carcinomas, n=720; liver metastases, n=139). The weighted kappa statistic was 0.828 (95% CI=0.786-0.871) for primary colonic cancer and 0.735 (95% CI=0.668-0.802) for liver metastases, whereas that between primary carcinomas and liver metastases in the meta-analysis was 0.748 (95% CI=0.692-0.804).
Correlation between copy-number alterations and clinicopathological parameters. DNA copy-number alterations of 8,684 segments were compared with clinicopathological features in 161 patients. Location and differentiation of the tumor correlated with the frequency of DNA copy-number alterations (Figure 4). Namely, the frequency of DNA copy-number alterations of tumors in left-sided colon was significantly higher than that of right-sided colon, while copy-number alterations frequency of well or moderately differentiated tumors was significantly higher than that of poorly differentiated or mucinous tumors. Median copy numbers and p-values of each DNA segment were plotted along genomic coordinates according to the tumor location (Figure 5) and differentiation (Figure 6). Location of the tumor correlated with genomic segments of chromosome 4, 5, 7, 8, 13, 14, 18 and 20. Differentiation of the tumor correlated with segments in chromosome 4, 6, 8, 11, 13, 14,15, 16, 17 and 20. Age and gender had no correlation with copy-number alterations (data not shown).
Discussion
Herein we report on an outline of the biomarker study in the ACTS-CC trial. Although strict quality control processes limited the number of samples subjected to array-CGH measurements, we obtained precise data from 161 out of 777 (20.7%) patients with CRC regarding cancer chromosomal alteration, and 616 gDNA samples that will be used for validation steps in PCR-based measurements. SCNA profiles of stage III colonic cancer in Japan were found to agree closely with the results of previous studies in Western countries. SCNAs were significantly frequent in left-sided tumors and well-differentiated tumors.
Previous articles have reported the usefulness of genomic analysis of FFPE cancer tissue samples using SNP arrays (22-24). FFPE samples could be advantageous for biomarker studies in clinical trials as they are readily available and are routinely examined in general municipal hospitals. Because the molecular profiles of FFPE samples might be affected by fixation and preservation methods, stringent quality control is important for reliable analysis. Our microarray data were obtained from approximately 20% of the total number of samples. The rate-limiting process was the PCR amplification step during target preparation prior to microarray hybridization. Although the successful execution rate rises if the number of PCR cycles or number of independent PCR reactions increases, as described by Thompson et al. (25), we carried out this process according to standard protocols so that PCR bias would not affect copy-number data.
Many analyses have attempted to demonstrate new factors that are independently prognostic or predictive, but they have often failed to establish a role in therapeutic decision making. The five levels of evidence recommended by The American Society of Clinical Oncology Tumor Markers Guidelines Committee and the REMARK criteria have been used to establish the clinical utility of new biomarkers (26, 27). However, the methodological difficulties involved in implementing these guidelines led Simon et al. to suggest a refined grading system to assist in establishing the clinical utility of biomarkers (28). Although the ACTS-CC trial was not designed to evaluate the medical utility of biomarkers, the present study might determine those with high levels of evidence according to the grading system proposed by Simon et al. based on the following considerations: firstly, biomarker utility was accommodated in the design of the ACTS-CC trial; secondly, patients were prospectively enrolled, treated, and followed-up in a prospective controlled trial; thirdly, specimens were collected prospectively; fourthly, analysis focusing on markers was applied before the assays; and fifthly, an independent validation step was used to confirm the candidates identified by the microarray screening step. The current Japanese biomarker study of the ACTS-CC phase III trial could provide insights into the biology and can be used for the development of new drugs.
CRC has been shown to have molecular heterogeneity that results from different combinations of genetic and epigenetic alterations. Recently, molecularly defined subtypes of stage III colonic cancer were shown to have associations in clinical and pathological features and survival outcomes (29). These sub-types had significant relationship with patient characteristics, such as age, gender and race, and the pathological features, such as tumor location, histological grade, N stage and T stage. In the present study, location and differentiation of the tumor correlated with DNA copy-number alterations. Our study might provide new molecular markers to establish a new classifier.
Data from the Organization for Economic Co-operation and Development Health Care Quality Indicators showed that Japan has the highest 5-year relative survival rate for CRC (18), and a favorable 5-year overall survival (87.5%) for stage III colonic cancer treated with adjuvant UFT/LV was reported in a Japanese randomized phase III study (30). This result is an improvement on the findings of the NSABP-C06 trial, with a 5-year overall survival of 69.6%. Improved outcomes in Japanese populations are speculated to be caused by differences in colonic cancer surgery (D3 resection) (31), stage migration in pathological examination, or racial/ethnic bias (32). For the progress of treatment of CRC, investigations of these differences are important. The present study, suggesting differences in tumor biology between Japan and Western countries, could contribute to those investigations as well as to the surgical treatment and pathological management of colonic cancer.
Although the efficacy of oral fluoropyrimidines for the adjuvant treatment of colonic cancer has been demonstrated (33), most related phase III trials are being carried out in Japan (5). Our findings could be useful when these phase III findings are introduced to Western countries and will help explore the basic rationale of adjuvant chemotherapy for colonic cancer. Survival data of ATCS-CC trail has been opened and adjuvant chemotherapy using S-1 for stage III colonic cancer was confirmed to be non-inferior in disease-free survival compared to UFT/LV (34). Our future study will provide important prognostic information in stage III colonic cancer with implications for patient management.
Acknowledgements
The Authors would like to thank all the patients and the co-investigators for their cooperation in the ACTS-CC trial. The Authors also thank the following additional investigators for their contributions to this study: Yasuyo Kusunoki in data management, Yoko Takagi, Aiko Saito, Yasuyo Okamoto and Sachiko Kishiro as the translational study office staff, and Masanori Fukushima, M.D., Ph.D. as a director of Translational Research Inforamatics Center.
Footnotes
This article is freely accessible online.
- Received December 20, 2015.
- Revision received January 26, 2016.
- Accepted January 29, 2016.
- Copyright© 2016 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved