Search

PDF file - 48K, Multi�class SAM analysis revealed 503 probes which were found to significantly differ between the four subtypes in GSE22155 at FDR=0

Supplementary Table S5.

Supplementary Table S3. Table providing detailed survival results for univariate and multivariate analyses of association of subtype classification with distant metastasis-free survival.

Purpose: Primary lung adenocarcinoma remains a deadly disease. Gene-expression phenotypes (GEPs) in adenocarcinoma have potential to provide clinically relevant disease stratification for improved prognosis and treatment prediction, given appropriate clinical and methodologic validation.Experimental Design: 2,395 transcriptional adenocarcinoma profiles were assembled from 17 public cohorts and classified by a nearest centroid GEP classifier into three subtypes: terminal respiratory unit (TRU), proximal-proliferative, and proximal-inflammatory, and additionally scored by five transcriptional metagenes representing different biologic processes, including proliferation. Prognostic- and chemotherapy-predictive associations of the subtypes were analyzed by univariate and multivariate analysis using overall survival or distant metastasis-free survival as endpoints.Results: Overall, GEPs were associated with patient outcome in both univariate and multivariate analyses, although not in all individual cohorts. The prognostically relevant division was between TRU- and non-TRU–classified cases, with expression of proliferation-associated genes as a key prognostic component. In contrast, GEP classification was not predictive of adjuvant chemotherapy response. GEP classification showed stability to random perturbations of genes or samples and alterations to classification procedures (typically 20% of cases switching subtype) was observed when removing larger or entire fractions of a single subtype, due to gene-centering shifts not addressable by the classifier.Conclusions: In a large-scale evaluation, we show that GEPs add prognostic value to standard clinicopathologic variables in lung adenocarcinoma. Subject to classifier refinement and confirmation in prospective cohorts, GEPs have potential to affect the prognostication of adenocarcinoma patients through a molecularly driven disease stratification. Clin Cancer Res; 22(1); 218–29. ©2015 AACR.

Supplementary Methods. Document providing detailed information about cohorts and data analysis steps.

Supplementary Table S1. Table describing the six identified highly correlated gene expression modules in lung cancer.

XLS file - 193K, Patient and sample data

Supplementary Figure S6. Expression of proliferation-associated genes in adenocarcinoma subgroups defined by clinical smoking history or gene expression phenotypes (bronchioid, magnoid, squamoid), and the association of these phenotypes with smoking history.

Supplementary Data from Gene Expression Profiling–Based Identification of Molecular Subtypes in Stage IV Melanomas with Different Clinical Outcome

PDF file - 51K, Univariate analyses in the whole cohort and in melanomas with a Breslow thickness > 2mm

Purpose: The incidence of malignant melanoma is increasing worldwide in fair-skinned populations. Melanomas respond poorly to systemic therapy, and metastatic melanomas inevitably become fatal. Although spontaneous regression, likely due to immune defense activation, rarely occurs, we lack a biological rationale and predictive markers in selecting patients for immune therapy.Experimental Design: We performed unsupervised hierarchical clustering of global gene expression data from stage IV melanomas in 57 patients. For further characterization, we used immunohistochemistry of selected markers, genome-wide DNA copy number analysis, genetic and epigenetic analysis of the CDKN2A locus, and NRAS/BRAF mutation screening.Results: The analysis revealed four distinct subtypes with gene signatures characterized by expression of immune response, pigmentation differentiation, proliferation, or stromal composition genes. Although all subtypes harbored NRAS and BRAF mutations, there was a significant difference between subtypes (P < 0.01), with no BRAF/NRAS wild-type samples in the proliferative subtype. Additionally, the proliferative subtype was characterized by a high frequency of CDKN2A homozygous deletions (P < 0.01). We observed a different prognosis between the subtypes (P = 0.01), with a particularly poor survival for patients harboring tumors of the proliferative subtype compared with the others (P = 0.003). Importantly, the clinical relevance of the subtypes was validated in an independent cohort of 44 stage III and IV melanomas. Moreover, low expression of an a priori defined gene set associated with immune response signaling was significantly associated with poor outcome (P = 0.001).Conclusions: Our data reveal a biologically based taxonomy of malignant melanomas with prognostic effect and support an influence of the antitumoral immune response on outcome. Clin Cancer Res; 16(13); 3356–67. ©2010 AACR.

Purpose: For primary melanomas, tumor thickness, mitotic rate, and ulceration are well-laid cornerstones of prognostication. However, a molecular exposition of melanoma aggressiveness is critically missing. We recently uncovered a four-class structure in metastatic melanoma, which predicts outcome and informs biology. This raises the possibility that a molecular structure exists even in the early stages of melanoma and that molecular determinants could underlie histophenotype and eventual patient outcome.Experimental Design: We subjected 223 archival primary melanomas to a horizontally integrated analysis of RNA expression, oncogenic mutations at 238 lesions, histomorphometry, and survival data.Results: Our previously described four-class structure that was elucidated in metastatic lesions was evident within the expression space of primary melanomas. Because these subclasses converged into two larger prognostic and phenotypic groups, we used the metastatic lesions to develop a binary subtype-based signature capable of distinguishing between “high” and “low” grade forms of the disease. The two-grade signature was subsequently applied to the primary melanomas. Compared with low-grade tumors, high-grade primary melanomas were significantly associated with increased tumor thickness, mitotic rate, ulceration (all P < 0.01), and poorer relapse-free (HR = 4.94; 95% CI, 2.84–8.59), and overall (HR = 3.66; 95% CI, 2.40–5.58) survival. High-grade melanomas exhibited elevated levels of proliferation and BRCA1/DNA damage signaling genes, whereas low-grade lesions harbored higher expression of immune genes. Importantly, the molecular-grade signature was validated in two external gene expression data sets.Conclusions: We provide evidence for a molecular organization within melanomas, which is preserved across all stages of disease. Clin Cancer Res; 18(15); 4026–36. ©2012 AACR.

Figure legends for supplementary figures.

PDF file - 425K, Clusters of genes with coordinated expression across samples

Supplementary Figure Legends. Document containing figure legends for supplementary figures.

PDF file - 79K, Significant canonical pathways (p

The prognostic and treatment-predictive markers currently in use for breast cancer are commonly based on the protein levels of individual genes (e.g., steroid receptors) or aspects of the tumor phenotype, such as histological grade and percentage of cells in the DNA synthesis phase of the cell cycle. Microarrays have previously been used to classify binary classes in breast cancer such as estrogen receptor (ER)-α status. To test whether the properties and specific values of conventional prognostic markers are encoded within tumor gene expression profiles, we have analyzed 48 well-characterized primary tumors from lymph node-negative breast cancer patients using 6728-element cDNA microarrays. In the present study, we used artificial neural networks trained with tumor gene expression data to predict the ER protein values on a continuous scale. Furthermore, we determined a gene expression profile-directed threshold for ER protein level to redefine the cutoff between ER-positive and ER-negative classes that may be more biologically relevant. With a similar approach, we studied the prediction of other prognostic parameters such as percentage cells in the S phase of the cell cycle (SPF), histological grade, DNA ploidy status, and progesterone receptor status. Interestingly, there was a consistent reciprocal relationship in expression levels of the genes important for both ER and SPF prediction. This and similar studies may be used to increase our understanding of the biology underlying these markers as well as to improve the currently available prognostic markers for breast cancer.

Supplementary Methods. Document describing in detail the different cohorts and data processing steps used.

Purpose: Cigarette smoking is the major pathogenic factor for lung cancer. The precise mechanisms of tobacco-related carcinogenesis and its effect on the genomic and transcriptional landscape in lung cancer are not fully understood.Experimental Design: A total of 1,398 (277 never-smokers and 1,121 smokers) genomic and 1,449 (370 never-smokers and 1,079 smokers) transcriptional profiles were assembled from public lung adenocarcinoma cohorts, including matched next-generation DNA-sequencing data (n = 423). Unsupervised and supervised methods were used to identify smoking-related copy-number alterations (CNAs), predictors of smoking status, and molecular subgroups.Results: Genomic meta-analyses showed that never-smokers and smokers harbored a similar frequency of total CNAs, although specific regions (5q, 8q, 16p, 19p, and 22q) displayed a 20% to 30% frequency difference between the two groups. Importantly, supervised classification analyses based on CNAs or gene expression could not accurately predict smoking status (balanced accuracies ∼60% to 80%). However, unsupervised multicohort transcriptional profiling stratified adenocarcinomas into distinct molecular subgroups with specific patterns of CNAs, oncogenic mutations, and mutation transversion frequencies that were independent of the smoking status. One subgroup included approximately 55% to 90% of never-smokers and approximately 20% to 40% of smokers (both current and former) with molecular and clinical features of a less aggressive and smoking-unrelated disease. Given the considerable intragroup heterogeneity in smoking-defined subgroups, especially among former smokers, our results emphasize the clinical importance of accurate molecular characterization of lung adenocarcinoma.Conclusions: The landscape of smoking-related CNAs and transcriptional alterations in adenocarcinomas is complex, heterogeneous, and with moderate differences. Our results support a molecularly distinct less aggressive adenocarcinoma entity, arising in never-smokers and a subset of smokers. Clin Cancer Res; 20(18); 4912–24. ©2014 AACR.

Purpose: To evaluate performances of published gene signatures for the assessment of urothelial carcinoma.Experimental Design: We evaluated 28 published gene signatures designed for diagnostic and prognostic purposes of urothelial cancer. The investigated signatures include eight signatures for stage, five for grade, four for progression, and six for survival. We used two algorithms for classification, nearest centroid classification and support vector machine, and Cox regression to evaluate signature performance in four independent data sets.Results: The overlap of genes among the signatures was low, ranging from 11% among stage signatures to 0.6% among survival signatures. The published signatures predicted muscle-invasive and high-grade tumors with accuracies in the range of 70% to 90%. The performance for a given signature varied considerably with the validation data set used, and interestingly, some of the best performing signatures were not designed for the tested classification problem. In addition, several nonbladder-derived gene signatures performed equally well. Large randomly selected gene signatures performed better than the published signatures, and by systematically increasing signature size, we show that signatures with >150 genes are needed to obtain robust performance in independent validation data sets. None of the published survival signatures performed better than random assignments when applied to independent validation data.Conclusion: We conclude that gene expression signatures with >150 genes predict muscle-invasive growth and high-grade tumors with robust accuracies. Special considerations have to be taken when designing gene signatures for outcome in bladder cancer. Clin Cancer Res; 16(17); 4421–33. ©2010 AACR.

PDF file - 91K, Flow chart describing the analysis

Analysis + 1 Figure and 3 Tables

PDF file - 392K, Expression of cell adhesion genes

PDF file - 95K, The expression values for these genes were averaged across GSE22155 samples in the two classes thereby providing us with two distinct centroids

Supplementary Figure S1. Figure describing extended classification stability analyses for biased removal of proximal-proliferative or proximal-inflammatory classified tumors

PDF file - 116K, Among the set of 36 BRAF-mutated cases, high-grade tumors showed worse survival than low-grade lesions

Purpose: Even though urothelial cancer is the fourth most common tumor type among males, progress in treatment has been scarce. A problem in day-to-day clinical practice is that precise assessment of individual tumors is still fairly uncertain; consequently efforts have been undertaken to complement tumor evaluation with molecular biomarkers. An extension of this approach would be to base tumor classification primarily on molecular features. Here, we present a molecular taxonomy for urothelial carcinoma based on integrated genomics.Experimental Design: We use gene expression profiles from 308 tumor cases to define five major urothelial carcinoma subtypes: urobasal A, genomically unstable, urobasal B, squamous cell carcinoma like, and an infiltrated class of tumors. Tumor subtypes were validated in three independent publically available data sets. The expression of 11 key genes was validated at the protein level by immunohistochemistry.Results: The subtypes show distinct clinical outcomes and differ with respect to expression of cell-cycle genes, receptor tyrosine kinases particularly FGFR3, ERBB2, and EGFR, cytokeratins, and cell adhesion genes, as well as with respect to FGFR3, PIK3CA, and TP53 mutation frequency. The molecular subtypes cut across pathologic classification, and class-defining gene signatures show coordinated expression irrespective of pathologic stage and grade, suggesting the molecular phenotypes as intrinsic properties of the tumors. Available data indicate that susceptibility to specific drugs is more likely to be associated with the molecular stratification than with pathologic classification.Conclusions: We anticipate that the molecular taxonomy will be useful in future clinical investigations. Clin Cancer Res; 18(12); 3377–86. ©2012 AACR.

Purpose: Lung cancer is the worldwide leading cause of death from cancer. DNA methylation in gene promoter regions is a major mechanism of gene expression regulation that may promote tumorigenesis. However, whether clinically relevant subgroups based on DNA methylation patterns exist in lung cancer remains unclear.Experimental Design: Whole-genome DNA methylation analysis using 450K Illumina BeadArrays was performed on 12 normal lung tissues and 124 tumors, including 83 adenocarcinomas, 23 squamous cell carcinomas (SqCC), 1 adenosquamous cancer, 5 large cell carcinomas, 9 large cell neuroendocrine carcinomas (LCNEC), and 3 small-cell carcinomas (SCLC). Unsupervised bootstrap clustering was performed to identify DNA methylation subgroups, which were validated in 695 adenocarcinomas and 122 SqCCs. Subgroups were characterized by clinicopathologic factors, whole-exome sequencing data, and gene expression profiles.Results: Unsupervised analysis identified five DNA methylation subgroups (epitypes). One epitype was distinctly associated with neuroendocrine tumors (LCNEC and SCLC). For adenocarcinoma, remaining four epitypes were associated with unsupervised and supervised gene expression phenotypes, and differences in molecular features, including global hypomethylation, promoter hypermethylation, genomic instability, expression of proliferation-associated genes, and mutations in KRAS, TP53, KEAP1, SMARCA4, and STK11. Furthermore, these epitypes were associated with clinicopathologic features such as smoking history and patient outcome.Conclusions: Our findings highlight one neuroendocrine and four adenocarcinoma epitypes associated with molecular and clinicopathologic characteristics, including patient outcome. This study demonstrates the possibility to further subgroup lung cancer, and more specifically adenocarcinomas, based on epigenetic/molecular classification that could lead to more accurate tumor classification, prognostication, and tailored patient therapy. Clin Cancer Res; 20(23); 6127–40. ©2014 AACR.

Supplementary Figures 1-6. PDF containing six supplementary figures. Figure S1: Identification and validation of epitypes in the discovery cohort. Figure S2: Reproducibility of epitypes in the Sandoval et al. cohort. Figure S3: Reproducibility of epitypes in the TCGA cohort. Figure S4: Mutation transversions across epitypes for TCGA adenocarcinomas stratified by smoking status. Figure S5: Copy number alterations (CNAs) and genomic complexity in adenocarcinoma epitypes. Figure S6: Promoter and global hypermethylation with respect to smoking status and adenocarcinoma gene expression subtypes.

XLS file - 22K, Summary of patient and sample data

PDF file - 1.8MB, Validation of tumor clusters in external data sets