Artificial Intelligence in Oncology: Current Landscape, Challenges, and Future Directions.

Artificial intelligence (AI) in oncology is advancing beyond algorithm development to integration into clinical practice. This review describes the current state of the field, with a specific focus on clinical integration. AI applications are structured according to cancer type and clinical domain, focusing on the four most common cancers and tasks of detection, diagnosis, and treatment. These applications encompass various data modalities, including imaging, genomics, and medical records. We conclude with a summary of existing challenges, evolving solutions, and potential future directions for the field. SIGNIFICANCE: AI is increasingly being applied to all aspects of oncology, where several applications are maturing beyond research and development to direct clinical integration. This review summarizes the current state of the field through the lens of clinical translation along the clinical care continuum. Emerging areas are also highlighted, along with common challenges, evolving solutions, and potential future directions for the field.

Identifying Oncology Clinical Trial Candidates Using Artificial Intelligence Predictions of Treatment Change: A Pilot Implementation Study.

PURPOSE: Precision oncology clinical trials often struggle to accrue, partly because it is difficult to find potentially eligible patients at moments when they need new treatment. We piloted deployment of artificial intelligence tools to identify such patients at a large academic cancer center. PATIENTS AND METHODS: Neural networks that process radiology reports to identify patients likely to start new systemic therapy were applied prospectively for patients with solid tumors that had undergone next-generation sequencing at our center. Model output was linked to the MatchMiner tool, which matches patients to trials using tumor genomics. Reports listing genomically matched patients, sorted by probability of treatment change, were provided weekly to an oncology nurse navigator (ONN) coordinating recruitment to nine early-phase trials. The ONN contacted treating oncologists when patients likely to change treatment appeared potentially trial-eligible. RESULTS: Within weekly reports to the ONN, 60,199 patient-trial matches were generated for 2,150 patients on the basis of genomics alone. Of these, 3,168 patient-trial matches (5%) corresponding to 525 patients were flagged for ONN review by our model, representing a 95% reduction in review compared with manual review of all patient-trial matches weekly. After ONN review for potential eligibility, treating oncologists for 74 patients were contacted. Common reasons for not contacting treating oncologists included cases where patients had already decided to continue current treatment (21%); the trial had no slots (14%); or the patient was ineligible on ONN review (12%). Of 74 patients whose oncologists were contacted, 10 (14%) had a consult regarding a trial and five (7%) enrolled. CONCLUSION: This approach facilitated identification of potential patients for clinical trials in real time, but further work to improve accrual must address the many other barriers to trial enrollment in precision oncology research.

Multi-omics Analysis Reveals Immune Features Associated with Immunotherapy Benefit in Patients with Squamous Cell Lung Cancer from Phase III Lung-MAP S1400I Trial.

PURPOSE: Identifying molecular and immune features to guide immune checkpoint inhibitor (ICI)-based regimens remains an unmet clinical need. EXPERIMENTAL DESIGN: Tissue and longitudinal blood specimens from phase III trial S1400I in patients with metastatic squamous non-small cell carcinoma (SqNSCLC) treated with nivolumab monotherapy (nivo) or nivolumab plus ipilimumab (nivo+ipi) were subjected to multi-omics analyses including multiplex immunofluorescence (mIF), nCounter PanCancer Immune Profiling Panel, whole-exome sequencing, and Olink. RESULTS: Higher immune scores from immune gene expression profiling or immune cell infiltration by mIF were associated with response to ICIs and improved survival, except regulatory T cells, which were associated with worse overall survival (OS) for patients receiving nivo+ipi. Immune cell density and closer proximity of CD8+GZB+ T cells to malignant cells were associated with superior progression-free survival and OS. The cold immune landscape of NSCLC was associated with a higher level of chromosomal copy-number variation (CNV) burden. Patients with LRP1B-mutant tumors had a shorter survival than patients with LRP1B-wild-type tumors. Olink assays revealed soluble proteins such as LAMP3 increased in responders while IL6 and CXCL13 increased in nonresponders. Upregulation of serum CXCL13, MMP12, CSF-1, and IL8 were associated with worse survival before radiologic progression. CONCLUSIONS: The frequency, distribution, and clustering of immune cells relative to malignant ones can impact ICI efficacy in patients with SqNSCLC. High CNV burden may contribute to the cold immune microenvironment. Soluble inflammation/immune-related proteins in the blood have the potential to monitor therapeutic benefit from ICI treatment in patients with SqNSCLC.

Programmatic Precision Oncology Decision Support for Patients With Gastrointestinal Cancer.

PURPOSE: With the growing number of available targeted therapeutics and molecular biomarkers, the optimal care of patients with cancer now depends on a comprehensive understanding of the rapidly evolving landscape of precision oncology, which can be challenging for oncologists to navigate alone. METHODS: We developed and implemented a precision oncology decision support system, GI TARGET, (Gastrointestinal Treatment Assistance Regarding Genomic Evaluation of Tumors) within the Gastrointestinal Cancer Center at the Dana-Farber Cancer Institute. With a multidisciplinary team, we systematically reviewed tumor molecular profiling for GI tumors and provided molecularly informed clinical recommendations, which included identifying appropriate clinical trials aided by the computational matching platform MatchMiner, suggesting targeted therapy options on or off the US Food and Drug Administration-approved label, and consideration of additional or orthogonal molecular testing. RESULTS: We reviewed genomic data and provided clinical recommendations for 506 patients with GI cancer who underwent tumor molecular profiling between January and June 2019 and determined follow-up using the electronic health record. Summary reports were provided to 19 medical oncologists for patients with colorectal (n = 198, 39%), pancreatic (n = 124, 24%), esophagogastric (n = 67, 13%), biliary (n = 40, 8%), and other GI cancers. We recommended ≥ 1 precision medicine clinical trial for 80% (406 of 506) of patients, leading to 24 enrollments. We recommended on-label and off-label targeted therapies for 6% (28 of 506) and 25% (125 of 506) of patients, respectively. Recommendations for additional or orthogonal testing were made for 42% (211 of 506) of patients. CONCLUSION: The integration of precision medicine in routine cancer care through a dedicated multidisciplinary molecular tumor board is scalable and sustainable, and implementation of precision oncology recommendations has clinical utility for patients with cancer.

MatchMiner: an open-source platform for cancer precision medicine.

Widespread, comprehensive sequencing of patient tumors has facilitated the usage of precision medicine (PM) drugs to target specific genomic alterations. Therapeutic clinical trials are necessary to test new PM drugs to advance precision medicine, however, the abundance of patient sequencing data coupled with complex clinical trial eligibility has made it challenging to match patients to PM trials. To facilitate enrollment onto PM trials, we developed MatchMiner, an open-source platform to computationally match genomically profiled cancer patients to PM trials. Here, we describe MatchMiner's capabilities, outline its deployment at Dana-Farber Cancer Institute (DFCI), and characterize its impact on PM trial enrollment. MatchMiner's primary goals are to facilitate PM trial options for all patients and accelerate trial enrollment onto PM trials. MatchMiner can help clinicians find trial options for an individual patient or provide trial teams with candidate patients matching their trial's eligibility criteria. From March 2016 through March 2021, we curated 354 PM trials containing a broad range of genomic and clinical eligibility criteria and MatchMiner facilitated 166 trial consents (MatchMiner consents, MMC) for 159 patients. To quantify MatchMiner's impact on trial consent, we measured time from genomic sequencing report date to trial consent date for the 166 MMC compared to trial consents not facilitated by MatchMiner (non-MMC). We found MMC consented to trials 55 days (22%) earlier than non-MMC. MatchMiner has enabled our clinicians to match patients to PM trials and accelerated the trial enrollment process.

OncoThreads: visualization of large-scale longitudinal cancer molecular data.

MOTIVATION: Molecular profiling of patient tumors and liquid biopsies over time with next-generation sequencing technologies and new immuno-profile assays are becoming part of standard research and clinical practice. With the wealth of new longitudinal data, there is a critical need for visualizations for cancer researchers to explore and interpret temporal patterns not just in a single patient but across cohorts. RESULTS: To address this need we developed OncoThreads, a tool for the visualization of longitudinal clinical and cancer genomics and other molecular data in patient cohorts. The tool visualizes patient cohorts as temporal heatmaps and Sankey diagrams that support the interactive exploration and ranking of a wide range of clinical and molecular features. This allows analysts to discover temporal patterns in longitudinal data, such as the impact of mutations on response to a treatment, for example, emergence of resistant clones. We demonstrate the functionality of OncoThreads using a cohort of 23 glioma patients sampled at 2-4 timepoints. AVAILABILITY AND IMPLEMENTATION: Freely available at http://oncothreads.gehlenborglab.org. Implemented in Java Script using the cBioPortal web API as a backend. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Network for Biomarker Immunoprofiling for Cancer Immunotherapy: Cancer Immune Monitoring and Analysis Centers and Cancer Immunologic Data Commons (CIMAC-CIDC).

PURPOSE: Immunoprofiling to identify biomarkers and integration with clinical trial outcomes are critical to improving immunotherapy approaches for patients with cancer. However, the translational potential of individual studies is often limited by small sample size of trials and the complexity of immuno-oncology biomarkers. Variability in assay performance further limits comparison and interpretation of data across studies and laboratories. EXPERIMENTAL DESIGN: To enable a systematic approach to biomarker identification and correlation with clinical outcome across trials, the Cancer Immune Monitoring and Analysis Centers and Cancer Immunologic Data Commons (CIMAC-CIDC) Network was established through support of the Cancer MoonshotSM Initiative of the National Cancer Institute (NCI) and the Partnership for Accelerating Cancer Therapies (PACT) with industry partners via the Foundation for the NIH. RESULTS: The CIMAC-CIDC Network is composed of four academic centers with multidisciplinary expertise in cancer immunotherapy that perform validated and harmonized assays for immunoprofiling and conduct correlative analyses. A data coordinating center (CIDC) provides the computational expertise and informatics platforms for the storage, integration, and analysis of biomarker and clinical data. CONCLUSIONS: This overview highlights strategies for assay harmonization to enable cross-trial and cross-site data analysis and describes key elements for establishing a network to enhance immuno-oncology biomarker development. These include an operational infrastructure, validation and harmonization of core immunoprofiling assays, platforms for data ingestion and integration, and access to specimens from clinical trials. Published in the same volume are reports of harmonization for core analyses: whole-exome sequencing, RNA sequencing, cytometry by time of flight, and IHC/immunofluorescence.

Cross-Site Concordance Evaluation of Tumor DNA and RNA Sequencing Platforms for the CIMAC-CIDC Network.

PURPOSE: Whole-exome (WES) and RNA sequencing (RNA-seq) are key components of cancer immunogenomic analyses. To evaluate the consistency of tumor WES and RNA-seq profiling platforms across different centers, the Cancer Immune Monitoring and Analysis Centers (CIMAC) and the Cancer Immunologic Data Commons (CIDC) conducted a systematic harmonization study. EXPERIMENTAL DESIGN: DNA and RNA were centrally extracted from fresh frozen and formalin-fixed paraffin-embedded non-small cell lung carcinoma tumors and distributed to three centers for WES and RNA-seq profiling. In addition, two 10-plex HapMap cell line pools with known mutations were used to evaluate the accuracy of the WES platforms. RESULTS: The WES platforms achieved high precision (> 0.98) and recall (> 0.87) on the HapMap pools when evaluated on loci using > 50× common coverage. Nonsynonymous mutations clustered by tumor sample, achieving an index of specific agreement above 0.67 among replicates, centers, and sample processing. A DV200 > 24% for RNA, as a putative presequencing RNA quality control (QC) metric, was found to be a reliable threshold for generating consistent expression readouts in RNA-seq and NanoString data. MedTIN > 30 was likewise assessed as a reliable RNA-seq QC metric, above which samples from the same tumor across replicates, centers, and sample processing runs could be robustly clustered and HLA typing, immune infiltration, and immune repertoire inference could be performed. CONCLUSIONS: The CIMAC collaborating laboratory platforms effectively generated consistent WES and RNA-seq data and enable robust cross-trial comparisons and meta-analyses of highly complex immuno-oncology biomarker data across the NCI CIMAC-CIDC Network.

Genomic Characterization of de novo Metastatic Breast Cancer.

PURPOSE: In contrast to recurrence after initial diagnosis of stage I-III breast cancer [recurrent metastatic breast cancer (rMBC)], de novo metastatic breast cancer (dnMBC) represents a unique setting to elucidate metastatic drivers in the absence of treatment selection. We present the genomic landscape of dnMBC and association with overall survival (OS). EXPERIMENTAL DESIGN: Targeted DNA sequencing (OncoPanel) was prospectively performed on either primary or metastatic tumors from 926 patients (212 dnMBC and 714 rMBC). Single-nucleotide variants, copy-number variations, and tumor mutational burden (TMB) in treatment-naïve dnMBC primary tumors were compared with primary tumors in patients who ultimately developed rMBC, and correlated with OS across all dnMBC. RESULTS: When comparing primary tumors by subtype, MYB amplification was enriched in triple-negative dnMBC versus rMBC (21.1% vs. 0%, P = 0.0005, q = 0.111). Mutations in KMTD2, SETD2, and PIK3CA were more prevalent, and TP53 and BRCA1 less prevalent, in primary HR+/HER2- tumors of dnMBC versus rMBC, though not significant after multiple comparison adjustment. Alterations associated with shorter OS in dnMBC included TP53 (wild-type: 79.7 months; altered: 44.2 months; P = 0.008, q = 0.107), MYC (79.7 vs. 23.3 months; P = 0.0003, q = 0.011), and cell-cycle (122.7 vs. 54.9 months; P = 0.034, q = 0.245) pathway genes. High TMB correlated with better OS in triple-negative dnMBC (P = 0.041). CONCLUSIONS: Genomic differences between treatment-naïve dnMBC and primary tumors of patients who developed rMBC may provide insight into mechanisms underlying metastatic potential and differential therapeutic sensitivity in dnMBC. Alterations associated with poor OS in dnMBC highlight the need for novel approaches to overcome potential intrinsic resistance to current treatments.

The Human Tumor Atlas Network: Charting Tumor Transitions across Space and Time at Single-Cell Resolution.

Crucial transitions in cancer-including tumor initiation, local expansion, metastasis, and therapeutic resistance-involve complex interactions between cells within the dynamic tumor ecosystem. Transformative single-cell genomics technologies and spatial multiplex in situ methods now provide an opportunity to interrogate this complexity at unprecedented resolution. The Human Tumor Atlas Network (HTAN), part of the National Cancer Institute (NCI) Cancer Moonshot Initiative, will establish a clinical, experimental, computational, and organizational framework to generate informative and accessible three-dimensional atlases of cancer transitions for a diverse set of tumor types. This effort complements both ongoing efforts to map healthy organs and previous large-scale cancer genomics approaches focused on bulk sequencing at a single point in time. Generating single-cell, multiparametric, longitudinal atlases and integrating them with clinical outcomes should help identify novel predictive biomarkers and features as well as therapeutically relevant cell types, cell states, and cellular interactions across transitions. The resulting tumor atlases should have a profound impact on our understanding of cancer biology and have the potential to improve cancer detection, prevention, and therapeutic discovery for better precision-medicine treatments of cancer patients and those at risk for cancer.

A harmonized meta-knowledgebase of clinical interpretations of somatic genomic variants in cancer.

Precision oncology relies on accurate discovery and interpretation of genomic variants, enabling individualized diagnosis, prognosis and therapy selection. We found that six prominent somatic cancer variant knowledgebases were highly disparate in content, structure and supporting primary literature, impeding consensus when evaluating variants and their relevance in a clinical setting. We developed a framework for harmonizing variant interpretations to produce a meta-knowledgebase of 12,856 aggregate interpretations. We demonstrated large gains in overlap between resources across variants, diseases and drugs as a result of this harmonization. We subsequently demonstrated improved matching between a patient cohort and harmonized interpretations of potential clinical significance, observing an increase from an average of 33% per individual knowledgebase to 57% in aggregate. Our analyses illuminate the need for open, interoperable sharing of variant interpretation data. We also provide a freely available web interface (search.cancervariants.org) for exploring the harmonized interpretations from these six knowledgebases.

Identifying ERBB2 Activating Mutations in HER2-Negative Breast Cancer: Clinical Impact of Institute-Wide Genomic Testing and Enrollment in Matched Therapy Trials.

PURPOSE: The yield of comprehensive genomic profiling in recruiting patients to molecular-based trials designed for small subgroups has not been fully evaluated. We evaluated the likelihood of enrollment in a clinical trial that required the identification of a specific genomic change based on our institute-wide genomic tumor profiling. PATIENTS AND METHODS: Using genomic profiling from archived tissue samples derived from patients with metastatic breast cancer treated between 2011 and 2017, we assessed the impact of systematic genomic characterization on enrollment in an ongoing phase II trial (ClinicalTrials.gov identifier: NCT01670877). Our primary aim was to describe the proportion of patients with a qualifying ERBB2 mutation identified by our institutional genomic panel (OncoMap or OncoPanel) who enrolled in the trial. Secondary objectives included median time from testing result to trial registration, description of the spectrum of ERBB2 mutations, and survival. Associations were calculated using Fisher's exact test. RESULTS: We identified a total of 1,045 patients with metastatic breast cancer without ERBB2 amplification who had available genomic testing results. Of these, 42 patients were found to have ERBB2 mutation and 19 patients (1.8%) were eligible for the trial on the basis of the presence of an activating mutation, 18 of which were identified by OncoPanel testing. Fifty-eight percent of potentially eligible patients were approached, and 33.3% of eligible patients enrolled in the trial guided exclusively by OncoPanel testing. CONCLUSION: More than one half of eligible patients were approached for trial participation and, significantly, one third of those were enrolled in NCT01670877. Our data illustrate the ability to enroll patients in trials of rare subsets in routine clinical practice and highlight the need for these broadly based approaches to effectively support the success of these studies.

Interactive or static reports to guide clinical interpretation of cancer genomics.

Objective: Misinterpretation of complex genomic data presents a major challenge in the implementation of precision oncology. We sought to determine whether interactive genomic reports with embedded clinician education and optimized data visualization improved genomic data interpretation. Materials and Methods: We conducted a randomized, vignette-based survey study to determine whether exposure to interactive reports for a somatic gene panel, as compared to static reports, improves physicians' genomic comprehension and report-related satisfaction (overall scores calculated across 3 vignettes, range 0-18 and 1-4, respectively, higher score corresponding with improved endpoints). Results: One hundred and five physicians at a tertiary cancer center participated (29% participation rate): 67% medical, 20% pediatric, 7% radiation, and 7% surgical oncology; 37% female. Prior to viewing the case-based vignettes, 34% of the physicians reported difficulty making treatment recommendations based on the standard static report. After vignette/report exposure, physicians' overall comprehension scores did not differ by report type (mean score: interactive 11.6 vs static 10.5, difference = 1.1, 95% CI, -0.3, 2.5, P = .13). However, physicians exposed to the interactive report were more likely to correctly assess sequencing quality (P < .001) and understand when reports needed to be interpreted with caution (eg, low tumor purity; P = .02). Overall satisfaction scores were higher in the interactive group (mean score 2.5 vs 2.1, difference = 0.4, 95% CI, 0.2-0.7, P = .001). Discussion and Conclusion: Interactive genomic reports may improve physicians' ability to accurately assess genomic data and increase report-related satisfaction. Additional research in users' genomic needs and efforts to integrate interactive reports into electronic health records may facilitate the implementation of precision oncology.

Institutional implementation of clinical tumor profiling on an unselected cancer population.

BACKGROUND. Comprehensive genomic profiling of a patient's cancer can be used to diagnose, monitor, and recommend treatment. Clinical implementation of tumor profiling in an enterprise-wide, unselected cancer patient population has yet to be reported. METHODS. We deployed a hybrid-capture and massively parallel sequencing assay (OncoPanel) for all adult and pediatric patients at our combined cancer centers. Results were categorized by pathologists based on actionability. We report the results for the first 3,727 patients tested. RESULTS. Our cohort consists of cancer patients unrestricted by disease site or stage. Across all consented patients, half had sufficient and available (>20% tumor) material for profiling; once specimens were received in the laboratory for pathology review, 73% were scored as adequate for genomic testing. When sufficient DNA was obtained, OncoPanel yielded a result in 96% of cases. 73% of patients harbored an actionable or informative alteration; only 19% of these represented a current standard of care for therapeutic stratification. The findings recapitulate those of previous studies of common cancers but also identify alterations, including in AXL and EGFR, associated with response to targeted therapies. In rare cancers, potentially actionable alterations suggest the utility of a "cancer-agnostic" approach in genomic profiling. Retrospective analyses uncovered contextual genomic features that may inform therapeutic response and examples where diagnoses revised by genomic profiling markedly changed clinical management. CONCLUSIONS. Broad sequencing-based testing deployed across an unselected cancer cohort is feasible. Genomic results may alter management in diverse scenarios; however, additional barriers must be overcome to enable precision cancer medicine on a large scale. FUNDING. This work was supported by DFCI, BWH, and the National Cancer Institute (5R33CA155554 and 5K23CA157631).

The landscape of kinase fusions in cancer.

Human cancer genomes harbour a variety of alterations leading to the deregulation of key pathways in tumour cells. The genomic characterization of tumours has uncovered numerous genes recurrently mutated, deleted or amplified, but gene fusions have not been characterized as extensively. Here we develop heuristics for reliably detecting gene fusion events in RNA-seq data and apply them to nearly 7,000 samples from The Cancer Genome Atlas. We thereby are able to discover several novel and recurrent fusions involving kinases. These findings have immediate clinical implications and expand the therapeutic options for cancer patients, as approved or exploratory drugs exist for many of these kinases.

Genomic analyses of gynaecologic carcinosarcomas reveal frequent mutations in chromatin remodelling genes.

Malignant mixed Müllerian tumours, also known as carcinosarcomas, are rare tumours of gynaecological origin. Here we perform whole-exome analyses of 22 tumours using massively parallel sequencing to determine the mutational landscape of this tumour type. On average, we identify 43 mutations per tumour, excluding four cases with a mutator phenotype that harboured inactivating mutations in mismatch repair genes. In addition to mutations in TP53 and KRAS, we identify genetic alterations in chromatin remodelling genes, ARID1A and ARID1B, in histone methyltransferase MLL3, in histone deacetylase modifier SPOP and in chromatin assembly factor BAZ1A, in nearly two thirds of cases. Alterations in genes with potential clinical utility are observed in more than three quarters of the cases and included members of the PI3-kinase and homologous DNA repair pathways. These findings highlight the importance of the dysregulation of chromatin remodelling in carcinosarcoma tumorigenesis and suggest new avenues for personalized therapy.

PiHelper: an open source framework for drug-target and antibody-target data.

MOTIVATION: The interaction between drugs and their targets, often proteins, and between antibodies and their targets, is important for planning and analyzing investigational and therapeutic interventions in many biological systems. Although drug-target and antibody-target datasets are available in separate databases, they are not publicly available in an integrated bioinformatics resource. As medical therapeutics, especially in cancer, increasingly uses targeted drugs and measures their effects on biomolecular profiles, there is an unmet need for a user-friendly toolset that allows researchers to comprehensively and conveniently access and query information about drugs, antibodies and their targets. SUMMARY: The PiHelper framework integrates human drug-target and antibody-target associations from publicly available resources to help meet the needs of researchers in systems pharmacology, perturbation biology and proteomics. PiHelper has utilities to (i) import drug- and antibody-target information; (ii) search the associations either programmatically or through a web user interface (UI); (iii) visualize the data interactively in a network; and (iv) export relationships for use in publications or other analysis tools. AVAILABILITY: PiHelper is a free software under the GNU Lesser General Public License (LGPL) v3.0. Source code and documentation are at http://bit.ly/pihelper. We plan to coordinate contributions from the community by managing future releases.

Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal.

The cBioPortal for Cancer Genomics (http://cbioportal.org) provides a Web resource for exploring, visualizing, and analyzing multidimensional cancer genomics data. The portal reduces molecular profiling data from cancer tissues and cell lines into readily understandable genetic, epigenetic, gene expression, and proteomic events. The query interface combined with customized data storage enables researchers to interactively explore genetic alterations across samples, genes, and pathways and, when available in the underlying data, to link these to clinical outcomes. The portal provides graphical summaries of gene-level data from multiple platforms, network visualization and analysis, survival analysis, patient-centric queries, and software programmatic access. The intuitive Web interface of the portal makes complex cancer genomics profiles accessible to researchers and clinicians without requiring bioinformatics expertise, thus facilitating biological discoveries. Here, we provide a practical guide to the analysis and visualization features of the cBioPortal for Cancer Genomics.

Using MEMo to discover mutual exclusivity modules in cancer.

Although individual tumors show surprisingly diverse genomic alterations, these events tend to occur in a limited number of pathways, and alterations that affect the same pathway tend to not co-occur in the same patient. While pathway analysis has been a powerful tool in cancer genomics, our knowledge of oncogenic pathway modules is incomplete. To systematically identify such modules, we have developed a novel method, Mutual Exclusivity Modules in Cancer (MEMo). The method searches and identifies modules characterized by three properties: (1) member genes are recurrently altered across a set of tumor samples; (2) member genes are known to or are likely to participate in the same biological process; and (3) alteration events within the modules are mutually exclusive. MEMo integrates multiple data types and maps genomic alterations to biological pathways. MEMo's mutual exclusivity uses a statistical model that preserves the number of alterations per gene and per sample. The MEMo software, source code and sample data sets are available for download at: http://cbio.mskcc.org/memo.