Prognosis of stage II colon cancer by non-neoplastic mucosa gene expression profiling

Prognostic Nomogram of Prognosis-Related Genes and Clinicopathological Characteristics to Predict the 5-Year Survival Rate of Colon Cancer Patients

Background: The Cancer Genome Atlas (TCGA) has established a genome-wide gene expression profile, increasing our understanding of the impact of tumor heredity on clinical outcomes. The aim of this study was to construct a nomogram using data from the TCGA regarding prognosis-related genes and clinicopathological characteristics to predict the 5-years survival rate of colon cancer (CC) patients. Methods: Kaplan-Meier and Cox regression analyses were used to identify genes associated with the 5-years survival rate of CC patients. Cox regression was used to analyze the relationship between the clinicopathological features and prognostic genes and overall survival rates in patients with CC and to identify independent risk factors for the prognosis of CC patients. A nomogram for predicting the 5-years survival rate of CC patients was constructed by R software. Results: A total of eight genes (KCNJ14, CILP2, ATP6V1G2, GABRD, RIMKLB, SIX2, PLEKHA8P1, and MPP2) related to the 5-years survival of rate CC patients were identified. Age, stage, and PLEKHA8P1 were independent risk factors for the 5-years survival rate in patients with CC. The accuracy, sensitivity and specificity of the nomogram model constructed by age, TNM staging, and PLEKHA8P1 for predicting the 5-years survival of rate CC patients were 83.3, 83.97, and 85.79%, respectively. Conclusion: The nomogram can correctly predict the 5-year survival rate of patients with CC, thus aiding the individualized decision-making process for patients with CC.

Identification and Clinical Validation of a Novel 4 Gene-Signature with Prognostic Utility in Colorectal Cancer

Colorectal cancer (CRC) is a high burden disease with several genes involved in tumor progression. The aim of the present study was to identify, generate and clinically validate a novel gene signature to improve prediction of overall survival (OS) to effectively manage colorectal cancer. We explored The Cancer Genome Atlas (TCGA), COAD and READ datasets (597 samples) from The Protein Atlas (TPA) database to extract a total of 595 candidate genes. In parallel, we identified 29 genes with perturbations in > 6 cancers which are also affected in CRC. These genes were entered in cBioportal to generate a 17 gene panel with highest perturbations. For clinical validation, this gene panel was tested on the FFPE tissues of colorectal cancer patients (88 patients) using Nanostring analysis. Using multivariate analysis, a high prognostic score (composite 4 gene signature-DPP7/2, YWHAB, MCM4 and FBXO46) was found to be a significant predictor of poor prognosis in CRC patients (HR: 3.42, 95% CI: 1.71-7.94, p < 0.001 *) along with stage (HR: 4.56, 95% CI: 1.35-19.15, p = 0.01 *). The Kaplan-Meier analysis also segregated patients on the basis of prognostic score (log-rank test, p = 0.001 *). The external validation using GEO dataset (GSE38832, 122 patients) corroborated the prognostic score (HR: 2.7, 95% CI: 1.99-3.73, p < 0.001 *). Additionally, higher score was able to differentiate stage II and III patients (130 patients) on the basis of OS (HR: 2.5, 95% CI: 1.78-3.63, p < 0.001 *). Overall, our results identify a novel 4 gene prognostic signature that has clinical utility in colorectal cancer.

Standardising RNA profiling based biomarker application in cancer-The need for robust control of technical variables

Histopathology-based staging of colorectal cancer (CRC) has utility in assessing the prognosis of patient subtypes, but as yet cannot accurately predict individual patient's treatment response. Transcriptomics approaches, using array based or next generation sequencing (NGS) platforms, of formalin fixed paraffin embedded tissue can be harnessed to develop multi-gene biomarkers for predicting both prognosis and treatment response, leading to stratification of treatment. While transcriptomics can shape future biomarker development, currently <1% of published biomarkers become clinically validated tests, often due to poor study design or lack of independent validation. In this review of a large number of CRC transcriptional studies, we identify recurrent sources of technical variability that encompass collection, preservation and storage of malignant tissue, nucleic acid extraction, methods to quantitate RNA transcripts and data analysis pipelines. We propose a series of defined steps for removal of these confounding issues, to ultimately aid in the development of more robust clinical biomarkers.

Evaluation of frozen tissue-derived prognostic gene expression signatures in FFPE colorectal cancer samples

Defining molecular features that can predict the recurrence of colorectal cancer (CRC) for stage II-III patients remains challenging in cancer research. Most available clinical samples are Formalin-Fixed, Paraffin-Embedded (FFPE). NanoString nCounter® and Affymetrix GeneChip® Human Transcriptome Array 2.0 (HTA) are the two platforms marketed for high-throughput gene expression profiling for FFPE samples. In this study, to evaluate the gene expression of frozen tissue-derived prognostic signatures in FFPE CRC samples, we evaluated the expression of 516 genes from published frozen tissue-derived prognostic signatures in 42 FFPE CRC samples measured by both platforms. Based on HTA platform-derived data, we identified both gene (99 individual genes, FDR < 0.05) and gene set (four of the six reported multi-gene signatures with sufficient information for evaluation, P < 0.05) expression differences associated with survival outcomes. Using nCounter platform-derived data, one of the six multi-gene signatures (P < 0.05) but no individual gene was associated with survival outcomes. Our study indicated that sufficiently high quality RNA could be obtained from FFPE tumor tissues to detect frozen tissue-derived prognostic gene expression signatures for CRC patients.

Personalized identification of altered pathways in cancer using accumulated normal tissue data

MOTIVATION

Identifying altered pathways in an individual is important for understanding disease mechanisms and for the future application of custom therapeutic decisions. Existing pathway analysis techniques are mainly focused on discovering altered pathways between normal and cancer groups and are not suitable for identifying the pathway aberrance that may occur in an individual sample. A simple way to identify individual's pathway aberrance is to compare normal and tumor data from the same individual. However, the matched normal data from the same individual are often unavailable in clinical situation. Therefore, we suggest a new approach for the personalized identification of altered pathways, making special use of accumulated normal data in cases when a patient's matched normal data are unavailable. The philosophy behind our method is to quantify the aberrance of an individual sample's pathway by comparing it with accumulated normal samples. We propose and examine personalized extensions of pathway statistics, overrepresentation analysis and functional class scoring, to generate individualized pathway aberrance score.

RESULTS

Collected microarray data of normal tissue of lung and colon mucosa are served as reference to investigate a number of cancer individuals of lung adenocarcinoma (LUAD) and colon cancer, respectively. Our method concurrently captures known facts of cancer survival pathways and identifies the pathway aberrances that represent cancer differentiation status and survival. It also provides more improved validation rate of survival-related pathways than when a single cancer sample is interpreted in the context of cancer-only cohort. In addition, our method is useful in classifying unknown samples into cancer or normal groups. Particularly, we identified 'amino acid synthesis and interconversion' pathway is a good indicator of LUAD (Area Under the Curve (AUC) 0.982 at independent validation). Clinical importance of the method is providing pathway interpretation of single cancer, even though its matched normal data are unavailable.

AVAILABILITY AND IMPLEMENTATION

The method was implemented using the R software, available at our Web site: http://bibs.snu.ac.kr/ipas.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Prognostic impact of genomic instability in colorectal cancer

BACKGROUND

The prognostic impact of an indication of chromosomal instability (CIN) is evaluated in a consecutive series of 952 colorectal cancer patients treated at Aker University Hospital, Norway, during 1993-2003. Microsatellite instability (MSI) in this case series has recently been reported and made it possible to find the co-occurrence and compare the prognostic significance of CIN and MSI.

METHODS

Data sets for overall survival (OS; n=855) and time to recurrence (TTR; n=579) were studied. To reveal CIN we used automated image cytometry (ICM). Non-diploid histograms were taken as indicative of the presence of CIN. PCR-based measures of MSI in this material have already been described.

RESULTS

As with MSI, CIN was found to be an independent predictor of early relapse and death among stage II patients (TTR: n=278: HR 2.19 (95% CI: 1.35-3.55), P=0.002). Of the MSI tumours (16%), 71% were found to be DNA diploid, 21% were DNA tetraploid and 8% were DNA aneuploid. Among microsatellite stable tumours, 24% were DNA diploid, 15% were DNA tetraploid and 61% were DNA aneuploid.

CONCLUSION

For patients presenting with stage II disease, genomic instability as detected by DNA image cytometry has the potential to provide a useful biomarker for relapse and cancer-related death following surgery with curative intent.

Activation of the PI3K/AKT pathway correlates with prognosis in stage II colon cancer

Background:

Patients with UICC/AJCC stage II colon cancer have a high 5-year overall survival rate after surgery. Nevertheless, a significant subgroup of patients develops tumour recurrence. Currently, there are no clinically established biomarkers available to identify this patient group. We applied reverse-phase protein arrays (RPPA) for phosphatidylinositide-3-kinase pathway activation mapping to stratify patients according to their risk of tumour recurrence after surgery.

Methods:

Full-length proteins were extracted from formalin-fixed, paraffin-embedded tissue samples of 118 patients who underwent curative resection. RPPA technology was used to analyse expression and/or phosphorylation levels of six major factors of the phosphatidylinositide-3-kinase pathway. Oncogenic mutations of KRAS and BRAF, and DNA microsatellite status, currently discussed as prognostic markers, were analysed in parallel.

Results:

Expression of phospho-AKT (HR=3.52; P=0.032), S6RP (HR=6.3; P=0.044), and phospho-4E-BP1 (HR=4.12; P=0.011) were prognostic factors for disease-free survival. None of the molecular genetic alterations were significantly associated with prognosis.

Conclusions:

Our data indicate that activation of the PI3K/AKT pathway evidenced on the protein level might be a valuable prognostic marker to stratify patients for their risk of tumour recurrence. Beside adjuvant chemotherapy targeting of upregulated PI3K/AKT signalling may be an attractive strategy for treatment of high-risk patients.

Anticipating the clinical use of prognostic gene expression-based tests for colon cancer stage II and III: is Godot finally arriving?

PURPOSE

According to current recommendations for adjuvant treatment, patients with colon cancer stage II are not routinely offered chemotherapy, unless considered to have a high risk of relapse based on specific clinicopathological parameters. Following these criteria, it is challenging to identify the subgroup of patients that will benefit the most from adjuvant treatment. Contrarily, patients with colon cancer stage III are routinely offered chemotherapy, but due to expected adverse effects and frailty, elderly patients are often excluded from standard protocols. Colon cancer is a disease of the elderly and accordingly, there is a large subgroup of patients for which guidelines for adjuvant treatment remain less clear. In these two clinical settings, improved risk stratification has great potential impact on patient care, anticipating that high-risk patients will benefit from chemotherapy. However, microsatellite instability is the only molecular prognostic marker recommended for clinical use.

EXPERIMENTAL DESIGN

In this perspective, we provide an updated view on the status and clinical potential of the many proposed prognostic gene expression-based tests for colon cancer stage II and III.

RESULTS

The main limitation for clinical implementation is lack of prospective validation. For patients with stage II, highly promising tests have been identified and clinical trials are ongoing. For elderly patients with stage III, the value of such tests has received less focus, but promising early results have been shown.

CONCLUSION

Although awaiting results from prospective trials, improved risk assessment for patients with stage II and III is likely to be achieved in the foreseeable future.

Molecular markers in colorectal cancer: clinical relevance in stage II colon cancer

SUMMARY Colorectal cancer is the second most common cause of cancer death in developed countries. Adjuvant chemotherapy is standard for stage III colorectal cancer but its use in stage II is controversial. Several clinicopathological factors have been described to define a high-risk group among stage II colon cancers, which can aid the selection of patients who may benefit from chemotherapy. Local tumor invasion (T4), high histological grade, obstruction and perforation at diagnosis, and number of lymph nodes removed are the most widely accepted factors. Several molecular factors have been also investigated as prognostic candidate biomarkers. DNA ploidy, KRAS and TP53 mutations, thymidylate synthase, dihydropyrimidine dehydrogenase, thymidine phosphorylase, loss of heterozygosity on chromosome 18q and microsatellite instability have been widely investigated. The aim of this review is to analyze the current evidence and clinical applications of the classical molecular biomarkers as well as new ones such as BRAF, circulating tumor cells, genome expression signatures and DNA methylation.

Guidelines for biomarker testing in colorectal carcinoma (CRC): a national consensus of the Spanish Society of Pathology (SEAP) and the Spanish Society of Medical Oncology (SEOM)

This consensus statement, conceived as a joint initiative of the Spanish Society of Pathology and the Spanish Society of Medical Oncology, makes diagnostic and treatment recommendations for the management of patients with hereditary, localised and advanced CRC based on the current scientific evidence on biomarker use. This consensus statement thus provides an opportunity to improve healthcare efficiency and resource use, which will benefit these patients. Based on the currently available data on this subject, this expert group recommends testing for microsatellite instability (MSI) in patients with localised CRC, as this is a strong predictive factor for deciding on adjuvant treatment. However, although the ColoPrint(®) and Oncotype Dx(®) gene expression signatures have been shown to have prognostic value, no consensus yet exists concerning their use in clinical practice. For advanced CRC, it is essential to test for KRAS mutation status before administering an anti-EGFR treatment, such as cetuximab or panitumumab. However, testing for other biomarkers, such as BRAF, EGFR, PI3K and PTEN mutations, should not be done routinely, because this does not influence treatment planning at the present time. Other important issues addressed include organisational requirements and the quality controls needed for proper testing of these biomarkers as well as the legal implications to be borne in mind when testing some biomarkers.

A network-based gene expression signature informs prognosis and treatment for colorectal cancer patients

Background

Several studies have reported gene expression signatures that predict recurrence risk in stage II and III colorectal cancer (CRC) patients with minimal gene membership overlap and undefined biological relevance. The goal of this study was to investigate biological themes underlying these signatures, to infer genes of potential mechanistic importance to the CRC recurrence phenotype and to test whether accurate prognostic models can be developed using mechanistically important genes.

Methods and Findings

We investigated eight published CRC gene expression signatures and found no functional convergence in Gene Ontology enrichment analysis. Using a random walk-based approach, we integrated these signatures and publicly available somatic mutation data on a protein-protein interaction network and inferred 487 genes that were plausible candidate molecular underpinnings for the CRC recurrence phenotype. We named the list of 487 genes a NEM signature because it integrated information from Network, Expression, and Mutation. The signature showed significant enrichment in four biological processes closely related to cancer pathophysiology and provided good coverage of known oncogenes, tumor suppressors, and CRC-related signaling pathways. A NEM signature-based Survival Support Vector Machine prognostic model was trained using a microarray gene expression dataset and tested on an independent dataset. The model-based scores showed a 75.7% concordance with the real survival data and separated patients into two groups with significantly different relapse-free survival (p = 0.002). Similar results were obtained with reversed training and testing datasets (p = 0.007). Furthermore, adjuvant chemotherapy was significantly associated with prolonged survival of the high-risk patients (p = 0.006), but not beneficial to the low-risk patients (p = 0.491).

Conclusions

The NEM signature not only reflects CRC biology but also informs patient prognosis and treatment response. Thus, the network-based data integration method provides a convergence between biological relevance and clinical usefulness in gene signature development.

Multiwalled carbon nanotube-induced gene signatures in the mouse lung: potential predictive value for human lung cancer risk and prognosis

Concerns over the potential for multiwalled carbon nanotubes (MWCNT) to induce lung carcinogenesis have emerged. This study sought to (1) identify gene expression signatures in the mouse lungs following pharyngeal aspiration of well-dispersed MWCNT and (2) determine if these genes were associated with human lung cancer risk and progression. Genome-wide mRNA expression profiles were analyzed in mouse lungs (n = 160) exposed to 0, 10, 20, 40, or 80 μg of MWCNT by pharyngeal aspiration at 1, 7, 28, and 56 d postexposure. By using pairwise statistical analysis of microarray (SAM) and linear modeling, 24 genes were selected, which have significant changes in at least two time points, have a more than 1.5-fold change at all doses, and are significant in the linear model for the dose or the interaction of time and dose. Additionally, a 38-gene set was identified as related to cancer from 330 genes differentially expressed at d 56 postexposure in functional pathway analysis. Using the expression profiles of the cancer-related gene set in 8 mice at d 56 postexposure to 10 μg of MWCNT, a nearest centroid classification accurately predicts human lung cancer survival with a significant hazard ratio in training set (n = 256) and test set (n = 186). Furthermore, both gene signatures were associated with human lung cancer risk (n = 164) with significant odds ratios. These results may lead to development of a surveillance approach for early detection of lung cancer and prognosis associated with MWCNT in the workplace.

A systems biology approach identifies SART1 as a novel determinant of both 5-fluorouracil and SN38 drug resistance in colorectal cancer

Chemotherapy response rates for advanced colorectal cancer remain disappointingly low, primarily because of drug resistance, so there is an urgent need to improve current treatment strategies. To identify novel determinants of resistance to the clinically relevant drugs 5-fluorouracil (5-FU) and SN38 (the active metabolite of irinotecan), transcriptional profiling experiments were carried out on pretreatment metastatic colorectal cancer biopsies and HCT116 parental and chemotherapy-resistant cell line models using a disease-specific DNA microarray. To enrich for potential chemoresistance-determining genes, an unsupervised bioinformatics approach was used, and 50 genes were selected and then functionally assessed using custom-designed short interfering RNA (siRNA) screens. In the primary siRNA screen, silencing of 21 genes sensitized HCT116 cells to either 5-FU or SN38 treatment. Three genes (RAPGEF2, PTRF, and SART1) were selected for further analysis in a panel of 5 colorectal cancer cell lines. Silencing SART1 sensitized all 5 cell lines to 5-FU treatment and 4/5 cell lines to SN38 treatment. However, silencing of RAPGEF2 or PTRF had no significant effect on 5-FU or SN38 sensitivity in the wider cell line panel. Further functional analysis of SART1 showed that its silencing induced apoptosis that was caspase-8 dependent. Furthermore, silencing of SART1 led to a downregulation of the caspase-8 inhibitor, c-FLIP, which we have previously shown is a key determinant of drug resistance in colorectal cancer. This study shows the power of systems biology approaches for identifying novel genes that regulate drug resistance and identifies SART1 as a previously unidentified regulator of c-FLIP and drug-induced activation of caspase-8.

Specific activity of cyclin-dependent kinase I is a new potential predictor of tumour recurrence in stage II colon cancer

Background:

There are no established biomarkers to identify tumour recurrence in stage II colon cancer. As shown previously, the enzymatic activity of the cyclin-dependent kinases 1 and 2 (CDK1 and CDK2) predicts outcome in breast cancer. Therefore, we investigated whether CDK activity identifies tumour recurrence in colon cancer.

Methods:

In all, 254 patients with completely resected (R0) UICC stage II colon cancer were analysed retrospectively from two independent cohorts from Munich (Germany) and Leiden (Netherlands). None of the patients received adjuvant treatment. Development of distant metastasis was observed in 27 patients (median follow-up: 86 months). Protein expression and activity of CDKs were measured on fresh-frozen tumour samples.

Results:

Specific activity (SA) of CDK1 (CDK1SA), but not CDK2, significantly predicted distant metastasis (concordance index=0.69, 95% confidence interval (CI): 0.55–0.79, P=0.036). Cutoff derivation by maximum log-rank statistics yielded a threshold of CDK1SA at 11 (SA units, P=0.029). Accordingly, 59% of patients were classified as high-risk (CDK1SA ⩾11). Cox proportional hazard analysis revealed CDK1SA as independent prognostic variable (hazard ratio=6.2, 95% CI: 1.44–26.9, P=0.012). Moreover, CKD1SA was significantly elevated in microsatellite-stable tumours.

Conclusion:

Specific activity of CDK1 is a promising biomarker for metastasis risk in stage II colon cancer.

Prediction of metastasis and recurrence in colorectal cancer based on gene expression analysis: ready for the clinic?

Cancers of the colon and rectum, which rank among the most frequent human tumors, are currently treated by surgical resection in locally restricted tumor stages. However, disease recurrence and formation of local and distant metastasis frequently occur even in cases with successful curative resection of the primary tumor (R0). Recent technological advances in molecular diagnostic analysis have led to a wealth of knowledge about the changes in gene transcription in all stages of colorectal tumors. Differential gene expression, or transcriptome analysis, has been proposed by many groups to predict disease recurrence, clinical outcome, and also response to therapy, in addition to the well-established clinico-pathological factors. However, the clinical usability of gene expression profiling as a reliable and robust prognostic tool that allows evidence-based clinical decisions is currently under debate. In this review, we will discuss the most recent data on the prognostic significance and potential clinical application of genome wide expression analysis in colorectal cancer.

Deletion of chromosome 4q predicts outcome in stage II colon cancer patients

BACKGROUND

Around 30% of all stage II colon cancer patients will relapse and die of their disease. At present no objective parameters to identify high-risk stage II colon cancer patients, who will benefit from adjuvant chemotherapy, have been established. With traditional histopathological features definition of high-risk stage II colon cancer patients is inaccurate. Therefore more objective and robust markers for prediction of relapse are needed. DNA copy number aberrations have proven to be robust prognostic markers, but have not yet been investigated for this specific group of patients. The aim of the present study was to identify chromosomal aberrations that can predict relapse of tumor in patients with stage II colon cancer.

MATERIALS AND METHODS

DNA was isolated from 40 formaldehyde fixed paraffin embedded stage II colon cancer samples with extensive clinicopathological data. Samples were hybridized using Comparative Genomic Hybridization (CGH) arrays to determine DNA copy number changes and microsatellite stability was determined by PCR. To analyze differences between stage II colon cancer patients with and without relapse of tumor a Wilcoxon rank-sum test was implemented with multiple testing correction.

RESULTS

Stage II colon cancers of patients who had relapse of disease showed significantly more losses on chromosomes 4, 5, 15q, 17q and 18q. In the microsatellite stable (MSS) subgroup (n = 28), only loss of chromosome 4q22.1-4q35.2 was significantly associated with disease relapse (P < 0.05, FDR < 0.15). No differences in clinicopathological characteristics between patients with and without relapse were observed.

CONCLUSION

In the present series of MSS stage II colon cancer patients losses on 4q22.1-4q35.2 were associated with worse outcome and these genomic alterations may aid in selecting patients for adjuvant therapy.

Consensus pathways implicated in prognosis of colorectal cancer identified through systematic enrichment analysis of gene expression profiling studies

Background

A large number of gene expression profiling (GEP) studies on prognosis of colorectal cancer (CRC) has been performed, but no reliable gene signature for prediction of CRC prognosis has been found. Bioinformatic enrichment tools are a powerful approach to identify biological processes in high-throughput data analysis.

Principal Findings

We have for the first time collected the results from the 23 so far published independent GEP studies on CRC prognosis. In these 23 studies, 1475 unique, mapped genes were identified, from which 124 (8.4%) were reported in at least two studies, with 54 of them showing consisting direction in expression change between the single studies. Using these data, we attempted to overcome the lack of reproducibility observed in the genes reported in individual GEP studies by carrying out a pathway-based enrichment analysis. We used up to ten tools for overrepresentation analysis of Gene Ontology (GO) categories or Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways in each of the three gene lists (1475, 124 and 54 genes). This strategy, based on testing multiple tools, allowed us to identify the oxidative phosphorylation chain and the extracellular matrix receptor interaction categories, as well as a general category related to cell proliferation and apoptosis, as the only significantly and consistently overrepresented pathways in the three gene lists, which were reported by several enrichment tools.

Conclusions

Our pathway-based enrichment analysis of 23 independent gene expression profiling studies on prognosis of CRC identified significantly and consistently overrepresented prognostic categories for CRC. These overrepresented categories have been functionally clearly related with cancer progression, and deserve further investigation.

A 12-gene genomic instability signature predicts clinical outcomes in multiple cancer types

BACKGROUND AND AIMS

Genomic instability, as reflected in specific chromosomal aneuploidies and variation in the nuclear DNA content, is a defining feature of human carcinomas. It is solidly established that the degree of genomic instability influences clinical outcome. We have recently identified a 12-gene expression signature that discerned genomically stable from unstable breast carcinomas. This gene expression signature was also useful to predict, with high accuracy, the clinical course in independent multiple published breast cancer cohorts. From a biological point of view, this result confirmed the central role of genomic instability for a tumor's ability to adapt to external challenges and selective pressure, and hence for continued survival fitness. This prompted us to investigate whether this genomic instability signature could also predict clinical outcome in other cancer types of epithelial origin, including colorectal tumors, non-small cell lung carcinomas, and ovarian cancer.

RESULTS

The results show that the gene expression signature that defines genomic instability and poor outcome in breast cancer contributes significantly more accurate (p<0.05 compared with random prediction) prognostic information in multiple cancer types independent of established clinical parameters. The 12-gene genomic instability signature stratified patients into high- and low-risk groups with distinct postoperative survival in three non-small cell lung cancer cohorts (n=637) in Kaplan-Meier analyses (log-rank p<0.05). It predicted recurrence in colon cancer patients (n=92) with an overall accuracy greater than 69% (p=0.04) in cross-cohort validation. It quantified relapse-free survival in ovarian cancer (n=124; log-rank p<0.05). Functional pathway analysis revealed interactions between the 12 signature genes and well-known cancer hallmarks.

CONCLUSION

The degree of genomic instability has diagnostic and prognostic implications. It is tempting to speculate that pursuing genomic instability therapeutically could provide entry points for a target that is unique to cancer cells.

Pretreatment transcriptional profiling for predicting response to neoadjuvant chemoradiotherapy in rectal adenocarcinoma

PURPOSE

Patients presenting with locally advanced rectal cancer currently receive preoperative radiotherapy with or without chemotherapy. Although pathologic complete response is achieved for approximately 10% to 30% of patients, a proportion of patients derive no benefit from this therapy while being exposed to toxic side effects of treatment. Therefore, there is a strong need to identify patients who are unlikely to benefit from neoadjuvant therapy to help direct them toward alternate and ultimately more successful treatment options.

EXPERIMENTAL DESIGN

In this study, we obtained expression profiles from pretreatment biopsies for 51 rectal cancer patients. All patients underwent preoperative chemoradiotherapy, followed by resection of the tumor 6 to 8 weeks posttreatment. Gene expression and response to treatment were correlated, and a supervised learning algorithm was used to generate an original predictive classifier and validate previously published classifiers.

RESULTS

Novel predictive classifiers based on Mandard's tumor regression grade, metabolic response, TNM (tumor node metastasis) downstaging, and normal tissue expression profiles were generated. Because there were only 7 patients who had minimal treatment response (>80% residual tumor), expression profiles were used to predict good tumor response and outcome. These classifiers peaked at 82% sensitivity and 89% specificity; however, classifiers with the highest sensitivity had poor specificity, and vice versa. Validation of predictive classifiers from previously published reports was attempted using this cohort; however, sensitivity and specificity ranged from 21% to 70%.

CONCLUSIONS

These results show that the clinical utility of microarrays in predictive medicine is not yet within reach for rectal cancer and alternatives to microarrays should be considered for predictive studies in rectal adenocarcinoma.

Extracting very simple diagnostic rules from microarray data

We present an approach to deriving very simple classification rules from microarray data by first selecting very small gene subsets that can ensure highly accurate classification of cancers. Finding such minimum gene subsets can greatly reduce the computational load and "noise" arising from irrelevant genes. The derived simple classification rules allow for accurate diagnosis without the need for any classifiers. This work can simplify gene expression tests by including only a very small number of genes rather than thousands or tens of thousands of genes, which can significantly bring down the cost for cancer testing. These studies also call for further investigations into possible biological relationship between these small number of genes and cancer development and treatment. For example, we report the following simple, and yet 100% accurate, diagnostic rules involving only 2 genes to separate the 3 types of lymphoma patients: the patient has diffuse large B-cell lymphoma (DLBCL), if and only if the expression level of gene GENE1622X is greater than -0.75; the patient has chronic lymphocytic leukaemia (CLL), if and only if the expression level of gene GENE540X is less than -1; and the patient has follicular lymphoma (FL) otherwise, i.e., if and only if the expression level of gene GENE1622X is less than -0.75 and the expression level of gene GENE540X is greater than -1.

Using gene expression profiling to predict response and prognosis in gastrointestinal cancers-the promise and the perils

Cancer treatment is now moving toward a personalized approach, promising improved rates of response and survival. A number of studies have employed the use of microarrays to investigate the predictive potential of expression profiling in gastrointestinal (GI) cancer patients. However while many robust predictive classifiers relating to response and prognosis have been generated for GI cancer patients, these have yet to make the transition to the clinic. The main obstacle is the limited cross validation between predictive gene lists identified for the same tumor type and outcome. Differences in the experimental design, analysis, and interpretation of results all contribute to this variation, with numerous factors influencing which genes are highlighted as predictive. While predictive genomics shows immense potential, it is still a relatively new field and the validation of predictive gene lists derived from microarray data remains a challenge. Future studies must carefully consider all aspects of experimental design to ensure a clinically applicable predictive test can be developed. With this in mind, more extensive and collaborative research must be undertaken before microarray-based platforms can be used routinely in tailoring GI cancer treatment and change clinical practice. Larger cohorts and consistency in methodology will enable the findings from this research to make the transition to the clinic.

Gene expression signature to improve prognosis prediction of stage II and III colorectal cancer

PURPOSE

This study aims to develop a robust gene expression classifier that can predict disease relapse in patients with early-stage colorectal cancer (CRC).

PATIENTS AND METHODS

Fresh frozen tumor tissue from 188 patients with stage I to IV CRC undergoing surgery was analyzed using Agilent 44K oligonucleotide arrays. Median follow-up time was 65.1 months, and the majority of patients (83.6%) did not receive adjuvant chemotherapy. A nearest mean classifier was developed using a cross-validation procedure to score all genes for their association with 5-year distant metastasis-free survival.

RESULTS

An optimal set of 18 genes was identified and used to construct a prognostic classifier (ColoPrint). The signature was validated on an independent set of 206 samples from patients with stage I, II, and III CRC. The signature classified 60% of patients as low risk and 40% as high risk. Five-year relapse-free survival rates were 87.6% (95% CI, 81.5% to 93.7%) and 67.2% (95% CI, 55.4% to 79.0%) for low- and high-risk patients, respectively, with a hazard ratio (HR) of 2.5 (95% CI, 1.33 to 4.73; P = .005). In multivariate analysis, the signature remained one of the most significant prognostic factors, with an HR of 2.69 (95% CI, 1.41 to 5.14; P = .003). In patients with stage II CRC, the signature had an HR of 3.34 (P = .017) and was superior to American Society of Clinical Oncology criteria in assessing the risk of cancer recurrence without prescreening for microsatellite instability (MSI).

CONCLUSION

ColoPrint significantly improves the prognostic accuracy of pathologic factors and MSI in patients with stage II and III CRC and facilitates the identification of patients with stage II disease who may be safely managed without chemotherapy.

A breast cancer prognostic signature predicts clinical outcomes in multiple tumor types

Epidemiological studies indicate an increased risk of subsequent primary ovarian cancer from women with breast cancer. We have recently identified a 28-gene expression signature that predicts, with high accuracy, the clinical course in a large population of breast cancer patients. This prognostic gene signature also accurately predicts response to chemotherapy commonly used for treating breast cancer, including CMF, Tamoxifen, Paclitaxel, Docetaxel and Doxorubicin (Adriamycin), in a panel of 60 cancer cell lines of nine different tissue origins. This prompted us to investigate whether this prognostic gene signature could also predict clinical outcome in other cancer types of epithelial origins, including ovarian cancer (n=124), colon tumors (n=74) and lung adenocarcinomas (n=442). The results show that the gene expression signature contributes significantly more accurate (P<0.05; compared with random prediction) prognostic information in multiple cancer types independent of established clinical parameters. Furthermore, the functional pathway analysis with curated database delineated a biological network with tight connections between the signature genes and numerous well established cancer hallmarks, indicating important roles of this prognostic gene signature in tumor genesis and progression.

Shrinkage-based diagonal discriminant analysis and its applications in high-dimensional data

High-dimensional data such as microarrays have brought us new statistical challenges. For example, using a large number of genes to classify samples based on a small number of microarrays remains a difficult problem. Diagonal discriminant analysis, support vector machines, and k-nearest neighbor have been suggested as among the best methods for small sample size situations, but none was found to be superior to others. In this article, we propose an improved diagonal discriminant approach through shrinkage and regularization of the variances. The performance of our new approach along with the existing methods is studied through simulations and applications to real data. These studies show that the proposed shrinkage-based and regularization diagonal discriminant methods have lower misclassification rates than existing methods in many cases.

Genetic prognostic and predictive markers in colorectal cancer

Despite many studies of the likely survival outcome of individual patients with colorectal cancer, our knowledge of this subject remains poor. Until recently, we had virtually no understanding of individual responses to therapy, but the discovery of the KRAS mutation as a marker of probable failure of epidermal growth factor receptor (EGFR)-targeted therapy is a first step in the tailoring of treatment to the individual. With the application of molecular analyses, as well as the ability to perform high-throughput screens, there has been an explosive increase in the number of markers thought to be associated with prognosis and treatment outcome in this disease. In this Review, we attempt to summarize the sometimes confusing findings, and critically assess those markers already in the public domain.

Preoperative chemoradiotherapy in locally advanced rectal cancer: correlation of a gene expression-based response signature with recurrence

Preoperative chemoradiotherapy is recommended for locally advanced rectal cancer (UICC stage II/III). We recently demonstrated that responsive and nonresponsive tumors showed differential expression levels of 54 genes. In this follow-up study, we investigated the relationship between this gene set and disease-free (DFS) and overall survival (OS). Pretherapeutic biopsies from 30 participants in the CAO/ARO/AIO-94 trial of the German Rectal Cancer Study Group were analyzed using gene expression microarrays. Statistical analysis was performed to identify differentially expressed genes between recurrent and nonrecurrent tumors and to correlate these changes with disease recurrence and outcome. After a median follow-up of 59 months, seven of eight patients with recurrent disease was a nonresponder, and one responsive tumor recurred. Response to chemoradiotherapy was significantly correlated with an improved DFS (log rank P=0.028), whereas OS did not differ significantly (P=0.11). Applying a class comparison analysis, we identified 20 genes that were differentially expressed between recurrent and nonrecurrent tumors (P<0.001). Analyzing the first two principal components of the 54 genes previously identified to predict response, we observed that this response signature correlated with an increased risk of cancer recurrence. These data suggest that the genetic basis of local response also affects the genetic basis of tumor recurrence. Genes that are indicative of nonresponse to preoperative chemoradiotherapy might also be linked to an increased risk of tumor recurrence.

DNA methylation markers predict outcome in node-positive, estrogen receptor-positive breast cancer with adjuvant anthracycline-based chemotherapy

PURPOSE

We have shown that DNA methylation of the PITX2 gene predicts risk of distant recurrence in steroid hormone receptor-positive, node-negative breast cancer. Here, we present results from a multicenter study investigating whether PITX2 and other candidate DNA methylation markers predict outcome in node-positive, estrogen receptor-positive, HER-2-negative breast cancer patients who received adjuvant anthracycline-based chemotherapy.

EXPERIMENTAL DESIGN

Using a microarray platform, we analyzed DNA methylation in regulatory regions of PITX2 and 60 additional candidate genes in 241 breast cancer specimens. Using Cox regression analysis, we assessed the predictive power of the individual marker/marker panel candidates. Clinical endpoints were time to distant metastasis, disease-free survival, and overall survival. A nested bootstrap/cross-validation strategy was applied to identify and validate marker panels.

RESULTS

DNA methylation of PITX2 and 14 other genes was correlated with clinical outcome. In multivariate models, each methylation marker added significant information to established clinical factors. A four-marker panel including PITX2, BMP4, FGF4, and C20orf55 was identified that resulted in improvement of outcome prediction compared with PITX2 alone.

CONCLUSIONS

This study provides further evidence for the PITX2 biomarker, which has now been successfully confirmed to predict outcome among different breast cancer patient populations. We further identify new DNA methylation biomarkers, three of which can be combined into a panel with PITX2 to increase the outcome prediction performance in our anthracycline-treated primary breast cancer population. Our results show that a well-defined panel of DNA methylation markers enables outcome prediction in lymph node-positive, HER-2-negative breast cancer patients treated with anthracycline-based chemotherapy.

Clinical determinants of response to irinotecan-based therapy derived from cell line models

PURPOSE

In an attempt to identify genes that are involved in resistance to SN38, the active metabolite of irinotecan (also known as CPT-11), we carried out DNA microarray profiling of matched HCT116 human colon cancer parental cell lines and SN38-resistant cell lines following treatment with SN38 over time.

EXPERIMENTAL DESIGN

Data analysis identified a list of genes that were acutely altered in the parental cells following SN38 treatment as well as constitutively altered in the SN38-resistant cells.

RESULTS

Independent validation of 20% of these genes by quantitative reverse transcription-PCR revealed a strong correlation with the microarray results: Pearson's correlation was 0.781 (r(2) = 0.61, P < 0.000001) for those genes that were acutely altered in the parental setting following SN38 treatment and 0.795 (r(2) = 0.63, P < 0.000002) for those genes that were constitutively altered in the SN38-resistant cells. We then assessed the ability of our in vitro-derived gene list to predict clinical response to 5-fluorouracil/irinotecan using pretreatment metastatic biopsies from responding and nonresponding colorectal cancer patients using both unsupervised and supervised approaches. When principal components analysis was used with our in vitro classifier gene list, a good separation between responding and nonresponding patients was obtained, with only one nonresponding and two responding patients separating with the incorrect groups. Supervised class prediction using support vector machines algorithm identified a 16-gene classifier with 75% overall accuracy, 81.8% sensitivity, and 66.6% specificity.

CONCLUSIONS

These results suggest that in vitro-derived gene lists can be used to predict clinical response to chemotherapy in colorectal cancer.

p16INK4a gene promoter hypermethylation in mucosa as a prognostic factor for patients with colorectal cancer

Low gene expression of folylpolyglutamate synthase (FPGS) in colorectal mucosa correlates with low folate levels and poor survival of colorectal cancer (CRC) patients. Because gene-specific hypermethylation is affected by the folate level, the hypermethylation status in mucosa may also be linked to clinical outcome of CRC patients. The tumor suppressor gene p16INK4a (p16) regulates the cell cycle and angiogenic switch. In human neoplastic tissues, the main mechanism of p16 inactivation is promoter methylation. The aim of the study was to determine whether hypermethylation of the p16 promoter could be detected in mucosa of CRC patients (n = 181) and to analyze if hypermethylation was related to survival. The relation between p16 hypermethylation and expression of FPGS and two other folate-associated genes, reduced folate carrier 1 (RFC-1), and thymidylate synthase (TS), was analyzed (n = 63). The results showed that p16 was hypermethylated in 65 (36%) of the mucosa samples and that hypermethylation was age-related (P = 0.029). After adjustment for known risk factors, Cox regression analysis showed that Dukes' A-C patients with p16 hypermethylation in mucosa had an increased risk of cancer-related death (hazard ratio = 2.9, P = 0.007) and shorter disease-free survival (hazard ratio = 2.5, P = 0.015) compared with patients with no p16 hypermethylation. RFC-1 and FPGS gene expression levels were significantly correlated in patients lacking p16 hypermethylation in mucosa (P = 0.0003), but not at all correlated in patients having hypermethylation in mucosa (P = 1.0). In conclusion, p16 hypermethylation in mucosa of CRC patients was identified as an independent prognostic parameter for cancer-specific survival as well as an independent predictor of DFS. The results suggest that there might be a connection between folate-associated gene expression and p16 methylation status.

Progress in the development of prognostic and predictive markers for gastrointestinal malignancies

Gastrointestinal cancers remain a significant cause of morbidity and mortality. While increasing therapeutic options have improved outcomes for many patients, they have also complicated treatment decision-making. Unfortunately, most patients with advanced gastrointestinal malignancies die from their disease. Prognostic and predictive markers could improve treatment significantly by identifying patients who may or may not require a given therapy, and determining those most likely to benefit from a therapy. Candidates for such markers include blood antigens and circulating tumor cells, tumor enzyme and gene expression, and pharmacodynamic endpoints. In this review, we summarize reported and ongoing research to define and validate prognostic and predictive markers in gastrointestinal malignancies, with an emphasis on colorectal cancer and brief overview of pancreatic and neuroendocrine tumors.

Systemic and targeted therapy for advanced colon cancer

Colorectal cancer (CRC) is the third most common malignancy and the second most common cause of cancer-related death in the USA. Prognosis is best when the disease is detected early; however, nearly two-thirds of newly diagnosed cases of CRC have lymph node involvement or metastatic disease. For years, 5-fluorouracil (FU)-based regimens represented the only viable treatment option for patients with metastatic CRC. However, in recent years several newer agents have been added to the treatment arsenal for this disease. These agents include oxaliplatin, irinotecan and newer monoclonal antibodies targeting the epidermal growth factor receptor (cetuximab and panitumumab) and VEGF (bevacizumab). This review aims to discuss current systemic and targeted therapies for metastatic colon cancer with a focus on mechanism of action, indications, toxicity and efficacy.

Accurate cancer classification using expressions of very few genes

We aim at finding the smallest set of genes that can ensure highly accurate classification of cancers from microarray data by using supervised machine learning algorithms. The significance of finding the minimum gene subsets is three-fold: 1) It greatly reduces the computational burden and "noise" arising from irrelevant genes. In the examples studied in this paper, finding the minimum gene subsets even allows for extraction of simple diagnostic rules which lead to accurate diagnosis without the need for any classifiers. 2) It simplifies gene expression tests to include only a very small number of genes rather than thousands of genes, which can bring down the cost for cancer testing significantly. 3) It calls for further investigation into the possible biological relationship between these small numbers of genes and cancer development and treatment. Our simple yet very effective method involves two steps. In the first step, we choose some important genes using a feature importance ranking scheme. In the second step, we test the classification capability of all simple combinations of those important genes by using a good classifier. For three "small" and "simple" data sets with two, three, and four cancer (sub)types, our approach obtained very high accuracy with only two or three genes. For a "large" and "complex" data set with 14 cancer types, we divided the whole problem into a group of binary classification problems and applied the 2-step approach to each of these binary classification problems. Through this "divide-and-conquer" approach, we obtained accuracy comparable to previously reported results but with only 28 genes rather than 16,063 genes. In general, our method can significantly reduce the number of genes required for highly reliable diagnosis.