Keijiro Nozawa

Chromosomal Instability (CIN) Phenotype, CIN High or CIN Low, Predicts Survival for Colorectal Cancer

PURPOSE To examine whether chromosomal instability (CIN) phenotype, determined by the severity of CIN, can predict survival for stages II and III colorectal cancer (CRC). PATIENTS AND METHODS We determined microsatellite instability (MSI) and loss of heterozygosity (LOH) status in 1,103 patients (training [n = 845] and validation [n = 258] sets with stages II and III CRC). The LOH ratio was defined as the frequency of LOH in chromosomes 2p, 5q, 17p, and 18q. According to the LOH ratio, non-MSI high tumors were classified as CIN high (LOH ratio ≥ 33%) or CIN low (LOH ratio < 33%). CIN-high tumors were subclassified as CIN high (mild type; LOH ratio < 75%) or CIN high (severe type; LOH ratio ≥ 75%). We used microarrays to identify a gene signature that could classify the CIN phenotype and evaluated its ability to predict prognosis. RESULTS CIN high showed the worst survival (P < .001), whereas there was no significant difference between CIN low and MSI high. CIN high (severe type) showed poorer survival than CIN high (mild type; P < .001). Multivariate analysis revealed that CIN phenotype was an independent risk factor for disease-free and overall survival, respectively, in both the training (P < .001 and P = .0155) and validation sets (P < .001 and P = .0076). Microarray analysis also revealed that survival was significantly poorer in those with the CIN-high than in the CIN-low gene signature (P = .0203). In a validation of 290 independent CRCs (GSE14333), the CIN-high gene signature showed significantly poorer survival than the CIN-low signature (P = .0047). CONCLUSION The CIN phenotype is a predictive marker for survival and may be used to select high-risk patients with stages II and III CRC.

Heterogeneity of KRAS status may explain the subset of discordant KRAS status between primary and metastatic colorectal cancer.

BACKGROUND: KRAS status is a useful predictive marker for anti-epidermal growth factor receptor antibody therapy. OBJECTIVE: This study aimed to examine the concordance rate of KRAS mutation status between corresponding primary and metastatic colorectal cancer lesions, and also among multiple metastatic tumors. Furthermore, we examined the heterogeneity of KRAS mutations with respect to discordant KRAS status between primary and metastatic tumors. DESIGN AND SETTINGS: This study was retrospective in design. PATIENTS: Forty-three patients with primary tumors and 113 metastatic tumors were studied. MAIN OUTCOME MEASURES: The KRAS mutational status was determined by the peptide nucleic acid clamp real-time polymerase chain reaction TaqMan assay. We also performed sequencing analysis to validate the KRAS mutational status. When KRAS status differed between primary and metastatic tumors, we examined the heterogeneity of KRAS status within individual primary tumors by microdissecting multiple samples in each patient. RESULTS: The frequency of KRAS mutations in primary tumors was 34.9%. A high concordance rate of KRAS (88.4–91.7%) mutations was observed between primary and metastatic tumors. All 5 cases (11.6%) with discordant KRAS status had heterogeneous KRAS status in primary tumors. However, in 10 concordant cases all microdissected areas showed an identical KRAS mutational status within each patient. The KRAS mutational statuses in all multiple liver and/or lung metastatic tumors were the same as those of the primary tumor. LIMITATIONS: We could not validate KRAS status in microdissected samples by the direct sequence method that was used in the present study, because the quantity of DNA was not sufficient to perform direct sequencing. CONCLUSION: KRAS status in a primary site may be used for selecting patients who would benefit from anti-epidermal growth factor receptor therapy. However, KRAS status can be heterogeneous within a primary tumor, and thus different parts of such tumors should be examined for KRAS status to correctly predict the KRAS status in metastatic lesions.

Differential gene expression signatures between colorectal cancers with and without KRAS mutations: Crosstalk between the KRAS pathway and other signalling pathways

PURPOSE KRAS mutation is an important predictive marker in determining resistance to anti-Epidermal Growth Factor Receptor (EGFR) antibody therapies. In order to clarify whether not only KRAS related signalling pathways but also other signalling pathways are altered in patients with colorectal cancers (CRCs) with KRAS mutations, we examined the differences in the gene expression signatures between CRCs with and without KRAS mutation. PATIENTS AND METHODS One-hundred and thirteen patients who underwent a surgical resection of a primary CRC were examined. KRAS mutational status was determined using the Peptide Nucleic Acid (PNA)-clamp real-time polymerase chain reaction (PCR) TaqMan assay. Gene expression profiles were compared between CRCs with and without KRAS mutation using the Human Genome GeneChip array U133. RESULTS Among 113 CRCs, KRAS mutations were present in 35 tumours (31%). We identified 30 genes (probes) that were differentially expressed between CRCs with and without KRAS mutation (False Discovery Rate (FDR), p<0.01), by which we were able to predict the KRAS status with an accuracy of 90.3%. Thirty discriminating genes included TC21, paired-like homeodomain 1 (PITX1), Sprouty-2, dickkopf homologue 4 (DKK-4), SET and MYND domain containing 3 (SMYD3), mitogen-activated protein kinase kinase kinase 14 (MAP3K14) and c-mer Proto-oncogene tyrosine kinase (MerTK). These genes were related to not only KRAS related signalling pathway but also to other signalling pathways, such as the Wnt-signalling pathway, the NF-kappa B activation pathway and the TGF-beta signalling pathway. CONCLUSIONS KRAS mutant CRCs exhibited a distinct gene expression signature different from wild-type KRAS CRCs. Using human CRC samples, we were able to show that there is crosstalk between the KRAS-mediated pathway and other signalling pathways. These results are necessary to be taken into account in establishing chemotherapeutic strategies for patients with anti-EGFR-refractory KRAS mutant CRCs.

Tumor location is a prognostic factor in poorly differentiated adenocarcinoma, mucinous adenocarcinoma, and signet-ring cell carcinoma of the colon

PurposeCancers which arise in the proximal and distal colon are suggested to be different clinically, pathologically, and genetically. The aim of this study is to clarify whether clinical behavior of colonic poorly differentiated adenocarcinoma, mucinous adenocarcinoma, and signet-cell carcinoma (Por/Muc/Sig cancers), minor and aggressive subpopulation in colonic cancers, differs in accordance with the tumor location.MethodsA total of 3,175 patients with curatively resected colonic cancers were studied. Clinical and pathological features were compared between Por/Muc/Sig cancers and well or moderately differentiated adenocarcinomas (Wel/Mod cancers) and between proximal and distal cancers in each histologic type.ResultsPor/Muc/Sig cancers (n = 213) were more advanced in the TNM stage and showed worse disease-specific survival than Wel/Mod cancers (n = 2,692). In Por/Muc/Sig cancers, but not in Wel/Mod cancers, proximal cancers showed significantly better disease-specific survival than distal cancers (88.9% vs. 76.5%, p = 0.0234), and a multivariate analysis showed that proximal tumor location was an independent predictor of fair prognosis (hazard ratio (HR), 0.458; 95% confidence interval (CI), 0.218–0.961; p = 0.0390). In addition, female gender also was an independent predictor of fair prognosis in Por/Muc/Sig cancers (HR, 0.373; 95% CI, 0.151–0.922) and not in Wel/Mod cancers.ConclusionsProximal Por/Muc/Sig cancers were suggested to be a distinct subpopulation with a favorable oncologic outcome. Tumor location and gender might be helpful in the risk stratification after curative surgery for Por/Muc/Sig cancers.

Prediction of response to preoperative chemoradiotherapy in rectal cancer by using reverse transcriptase polymerase chain reaction analysis of four genes.

BACKGROUND: Patients with rectal cancer exhibit a wide spectrum of responses to chemoradiotherapy. Several gene expression signatures have been reported to predict the response to chemoradiotherapy in rectal cancer, but the lack of practical assays has restricted the clinical use of this technique. OBJECTIVE: We aimed to identify a set of discriminating genes that can be used for the clinical prediction of response to chemoradiotherapy in rectal cancer. DESIGN AND SETTINGS: This study is a retrospective analysis of tumor samples in a single institute. PATIENTS: Sixty-two patients who underwent preoperative chemoradiotherapy were studied. MAIN OUTCOME MEASURES: Gene expression was initially studied in 46 training samples by microarray analysis, and the association between gene expression and response to chemoradiotherapy was evaluated. Quantitative reverse transcriptase polymerase chain reaction was performed to validate the microarray expression levels of the discriminating genes. We developed a gene expression model for the prediction of response to chemoradiotherapy based on the reverse transcriptase polymerase chain reaction findings and validated it by using 16 independent test samples. RESULTS: We identified 24 discriminating probes with expression levels that differed significantly between responders and nonresponders. Among 18 genes identified by Gene Symbol, real-time reverse transcriptase polymerase chain reaction showed significant differences in the expression of 16 genes between responders and nonresponders. We constructed a predictive model by using different sets of these 16 genes, and the highest accuracy rate (89.1%) was obtained by using LRRIQ3, FRMD3, SAMD5, and TMC7. The predictive accuracy rate of this 4-gene signature in the independent set of 16 patients was 81.3%. LIMITATIONS: Validation in a different and large cohort of patients is necessary. CONCLUSIONS: The 4-gene signature identified in this study is closely associated with response to chemoradiotherapy in rectal cancer.

Prediction of liver metastasis after colorectal cancer using reverse transcription-polymerase chain reaction analysis of 10 genes

PURPOSE Liver metastasis is one of the major types of recurrence after surgery for colorectal cancer. Traditional methods of predicting liver metastasis are limited in their accuracy, suggesting the need to develop new predictors. We developed a 10-gene signature that is closely associated with the development of liver metastasis after colorectal cancer. PATIENTS AND METHODS We examined a total of 189 frozen specimens of primary colorectal cancers using both microarray and quantitative real-time reverse transcriptase-polymerase chain reaction (RT-PCR) analysis. Initially, we studied gene expression in colorectal cancer tissue from 160 randomly selected patients who had undergone surgical resection of colorectal cancer and evaluated the association between the level of gene expression and the occurrence of liver metastasis. We developed a gene-expression model for the prediction of liver metastasis based on the RT-PCR findings. We then used specimens from 29 other patients for validation. RESULTS The expression of 14 genes was correlated with liver metastasis according to both microarray and RT-PCR analysis. We constructed an accurate predictive model based on the results for 10 of these genes, which included epiregulin (EREG), amphiregulin (AREG), cyclooxygenase 2 (COX-2) and lymphocyte-specific protein tyrosine kinase (LCK). The 10-gene signature was an independent predictor of liver metastasis. The model was validated in the independent set of 29 patients. The predictive accuracy of the model in a test set of patients was 86.2%. CONCLUSION The 10-gene signature identified in this study is closely associated with the occurrence of liver metastasis in colorectal cancer patients.

Gene expression of mesenchyme forkhead 1 (FOXC2) significantly correlates with the degree of lymph node metastasis in colorectal cancer.

In stage III colorectal cancer, patients with N1 stage tumors show poorer outcome than patients with N2 stage tumors. Our objective was to identify genes that are predictive for the presence of lymph node metastasis, and to characterize the aggressiveness of lymph node metastases. Gene expression profiles of colorectal cancer were determined by microarray in training (n = 116) and test (n = 25) sets of patients. We identified 40 discriminating probes in patients with and without lymph node metastases. Using these probes, we could predict the presence of lymph node metastasis with an accuracy of 87.1% (training set) and 76.0% (test set). Among discriminating probes, FOXC2 expression was significantly correlated with the degree of lymph node metastasis. FOXC2 was expressed significantly and disparately in patients with N1 and N2 stage tumors as analyzed by real-time reverse transcriptase-polymerase chain reaction. FOXC2 appears to be involved in determining the aggressiveness of lymph node metastasis in colorectal cancer.

Gene expression of vascular endothelial growth factor A, thymidylate synthase, and tissue inhibitor of metalloproteinase 3 in prediction of response to bevacizumab treatment in colorectal cancer patients.

BACKGROUND: Regimens containing bevacizumab and 5-fluorouracil have achieved substantial progress in the treatment of colorectal cancer. However, individual responses to bevacizumab vary widely in regard to efficacy and toxicity. OBJECTIVE: To be able to select patients who would benefit from bevacizumab, we aimed to establish a predictor model for response to bevacizumab therapy based on gene expression profiles. DESIGN AND SETTING: Retrospective analysis of tumor samples in the laboratory. PATIENTS: The patient population comprised 25 patients with metastatic colorectal cancer treated with bevacizumab with either modified FOLFOX6 or FOLFIRI, from whom tumor samples were available for gene expression analysis. MAIN OUTCOME MEASURES: Response Evaluation Criteria in Solid Tumors were used to classify patients as responders or nonresponders to chemotherapy. Gene-expression profiles were determined with both microarray analysis and quantitative, real-time reverse-transcriptase polymerase chain reaction, and responders were compared with nonresponders, correcting for multiple comparisons. Genes that discriminated between groups on both analyses with the greatest accuracy were selected for the predictive model. Between-group differences in protein expression were confirmed with polymerase chain reaction and immunohistochemical staining. RESULTS: From 19 probes that differentiated between responders and nonresponders on microarray analyses, we identified 13 genes that were differentially expressed between responders and nonresponders on both microarray and real-time reverse-transcriptase polymerase chain reaction. A model using the genes for vascular endothelial growth factor-A, thymidylate synthase, and tissue inhibitor of metalloproteinase 3 predicted response to bevacizumab therapy with an accuracy of 96%, sensitivity of 90.9% (10/11), specificity of 100% (14/14), positive predictive value of 100% (10/10), and negative predictive value of 93.3% (14/15). The protein expression of vascular endothelial growth factor-A, thymidylate synthase, and tissue inhibitor of metalloproteinase 3 correlated with the findings of mRNA expression analyses. LIMITATIONS: Validation of the model in a different cohort of patients is necessary. CONCLUSIONS: The present predictive model based on quantitative, real-time, reverse-transcriptase polymerase chain reaction assessment of vascular endothelial growth factor-A, thymidylate synthase, and tissue inhibitor of metalloproteinase 3 may enable selection of colorectal cancer patients who would benefit from bevacizumab therapy.

Gene expression signature and response to the use of leucovorin, fluorouracil and oxaliplatin in colorectal cancer patients.

PurposeFOLFOX (a combination of leucovorin, fluorouracil and oxaliplatin) has achieved substantial success in the treatment of colorectal cancer (CRC) patients. However, about half of all patients show resistance to this regimen and some develop adverse symptoms such as neurotoxicity. In order to select patients who would benefit most from this therapy, we aimed to build a predictor for the response to FOLFOX using microarray gene expression profiles of primary CRC samples.Patients and methodsForty patients who underwent surgery for primary lesions were examined. All patients had metastatic or recurrent CRC and received modified FOLFOX6. Responders and nonresponders were determined according to the best observed response at the end of the first-line treatment. Gene-expression profiles of primary CRC were determined using Human Genome GeneChip arrays U133. We identified discriminating genes whose expression differed significantly between responders and nonresponders and then carried out supervised class prediction using the k-nearest-neighbour method.ResultsWe identified 27 probes that were differentially expressed between responders and nonresponders at significant levels. Based on the expression of these genes, we constructed a FOLFOX response predictor with an overall accuracy of 92.5%. The sensitivity, specificity, positive and negative predictive values were 78.6%, 100%, 100% and 89.7%, respectively.ConclusionThe present model suggests the possibility of selecting patients who would benefit from FOLFOX therapy both in the metastatic and the adjuvant setting. To our knowledge, this is the first study to establish a prediction model for the response to FOLFOX chemotherapy based on gene expression by microarray analysis.

Predicting Ulcerative Colitis-Associated Colorectal Cancer Using Reverse-Transcription Polymerase Chain Reaction Analysis

BACKGROUND Widespread genetic alterations are present not only in ulcerative colitis (UC)-associated neoplastic lesions but also in the adjacent normal colonic mucosa. This suggests that genetic changes in nonneoplastic mucosa might be effective markers for predicting the development of UC-associated cancer (UC-Ca). This study aimed to build a predictive model for the development of UC-Ca based on gene expression levels measured by reverse-transcription polymerase chain reaction (RT-PCR) analysis in nonneoplastic rectal mucosa. PATIENTS AND METHODS Fifty-three UC patients were examined, of which 10 had UC-Ca and 43 did not (UC-NonCa). In addition to the 40 genes and transcripts previously shown to be predictive for developing UC-Ca in our microarray studies, 149 new genes, reported to be important in carcinogenesis, were selected for low density array (LDA) analysis. The expression of a total of 189 genes was examined by RT-PCR in nonneoplastic rectal mucosa. RESULTS We identified 20 genes showing differential expression in UC-Ca and UC-NonCa patients, including cancer-related genes such as CYP27B1, RUNX3, SAMSN1, EDIL3, NOL3, CXCL9, ITGB2, and LYN. Using these 20 genes, we were able to build a predictive model that distinguished patients with and without UC-Ca with a high accuracy rate of 83% and a negative predictive value of 100%. CONCLUSION This predictive model suggests that it is possible to identify UC patients at a high risk of developing cancer. These results have important implications for improving the efficacy of surveillance by colonoscopy and suggest directions for future research into the molecular mechanisms of UC-associated cancer.