Integrating molecular profiles into clinical frameworks through the Molecular Oncology Almanac to prospectively guide precision oncology

Abstract

Tumor molecular profiling of single gene-variant ('first-order') genomic alterations informs potential therapeutic approaches. Interactions between such first-order events and global molecular features (for example, mutational signatures) are increasingly associated with clinical outcomes, but these 'second-order' alterations are not yet accounted for in clinical interpretation algorithms and knowledge bases. We introduce the Molecular Oncology Almanac (MOAlmanac), a paired clinical interpretation algorithm and knowledge base to enable integrative interpretation of multimodal genomic data for point-of-care decision making and translational-hypothesis generation. We benchmarked MOAlmanac to a first-order interpretation method across multiple retrospective cohorts and observed an increased number of clinical hypotheses from evaluation of molecular features and profile-to-cell line matchmaking. When applied to a prospective precision oncology trial cohort, MOAlmanac nominated a median of two therapies per patient and identified therapeutic strategies administered in 47% of patients. Overall, we present an open-source computational method for integrative clinical interpretation of individualized molecular profiles.

Extended Data Figure 1.. Illustrating a clinically relevant somatic variant matching to Molecular Oncology Almanac.

Molecular features whose gene is listed in Molecular Oncology Almanac (MOAlmanac) will at least be categorized as Biologically Relevant. Molecular features are then evaluated for assertions associated with therapeutic sensitivity, resistance, and prognosis independently. Consider the somatic variant EGFR p.T790M harbored by a non-small cell lung cancer (NSCLC) tumor being evaluated for associations to therapeutic sensitivity: a, If a gene and corresponding feature type are catalogued in MOAlmanac for the assertion type being evaluated, the molecular feature will at least be labeled as “Investigate Actionability”. b, Next, MOAlmanac will prioritize assertions of the same ontology and then match by additional feature details. While EGFR p.L858R is also a missense variant, the specific protein change p.T790M is catalogued by the database. EGFR p.T790M is thus reported as “Putatively Actionable” as it was able to fully match to a molecular feature catalogued in the database. c, Of the remaining database entries, those associated with the highest evidence tier are selected. The first returned result is selected, unless an entry marked as a preferred assertion is present, and the remaining are returned as equivalent matches, viewable within the produced report.

Extended Data Figure 2.. MOAlmanac investigates preclinical efficacy of nominated relationships.

If a nominated therapy has been characterized by the GDSC, MOAlmanac will investigate if cancer cell lines that are wild type and mutant for the associated molecular feature respond differently by comparing IC50 values using a two-sided Mann-Whitney-Wilcoxon test. For PIK3CA p.H1047R and response to Pictilisib, response data was available for 766 cancer cell lines. MOAlmanac investigated sensitivity for mutant and wild type cell lines for cell lines harboring either a PIK3CA somatic variant, copy number alteration, or fusion (n = 162 mutant cell lines, min IC50: 0.18, max: 93.92, median: 3.22, q1: 1.70, q2: 6.72; n = 604 wild type, min IC50: 0.04, max: 1616.65, median: 4.10, q1: 1.94, q3: 9.34), a PIK3CA somatic variant (n = 103 mutant cell lines, min IC50: 0.18, max: 50.01, median: 2.90, q1: 1.42, q2: 5.14; n = 653 wild type, min IC50: 0.037, max: 1616.65, median: 4.10, q1: 1.95, q3: 9.54), PIK3CA missense variants (n = 98 mutant cell lines, min IC50: 0.18, max: 50.01, median: 2.91, q1: 1.46, q2: 5.11; n = 668 wild type, min IC50: 0.037, max: 1616.65, median: 4.10, q1: 1.94, q3: 9.61), and the specific protein change PIK3CA p.H1047R (n = 21 mutant cell lines, min IC50: 0.54, max: 5.63, median: 1.86, q1: 0.865, q2: 3.25; n = 745 wild type, min IC50: 0.037, max: 1616.65, median: 3.92, q1: 1.90, q3: 9.15). Data is available as source data.

Extended Data Figure 3.. Number of features shared with nearest neighbors

MOAlmanac performs profile-to-cell line matchmaking by applying Similarity Network Fusion (SNF) on four distance matrices: Cancer Gene Census (CGC) genes altered by somatic variants, CGC genes altered by copy number alterations, CGC genes altered by fusions, and specific molecular features associated with FDA approvals. 154/205 cancer cell lines which harbor at least one FDA approval share at least one with their nearest neighbor. Data is available as source data.

Extended Data Figure 4.. Comparison to OncoKB and CIViC

Upset plots comparing PubMed ids, therapies, and genes catalogued by Molecular Oncology Almanac, OncoKB, and CIViC. No one knowledge base subsumes another. Data is available as source data.

Extended Data Figure 5.. Counts of clinically relevant molecular features observed in retrospective cohorts by MOAlmanac by cohort, feature type, evidence, and assertion type.

Counts of clinically relevant molecular features associated with therapeutic sensitivity, resistance, and prognosis categorized as putatively actionable (exactly matching a fully characterized genomic event catalogued in MOAlmanac) or investigate actionability (partial match) by evidence tier for metastatic melanomas (MEL, n = 110), metastatic castration-resistant prostate cancer (mCRPC, n = 150), kidney papillary renal-cell carcinoma (KIRP, n = 100), and osteosarcoma (OS, n = 59). Data is available as source data.

Fig. 1 |. Molecular Oncology Almanac, a clinical interpretation framework.

a, The Molecular Oncology Almanac (MOAlmanac) is a paired clinical interpretation algorithm and underlying knowledge base to enable integrative interpretation of multimodal genomics data for point-of-care decision making and translational-hypothesis generation. b, A literature review was performed to grow MOAlmanac’s underlying knowledge base from TARGET. c, Assertions catalogued in MOAlmanac, categorized by evidence (left) and therapy types (right). d, MOAlmanac matches molecular features to its own knowledge base and several others to prioritize somatic variants for clinical and biological relevance. MSigDB, Molecular Signatures Database; VUS, variant of unknown significance. e, Germline variants are evaluated for pathogenicity and allele frequency and reported if the gene is related to the American College of Medical Genetics and Genomics (ACMG), hereditary cancers, or somatic cancers. Vignettes of how MOAlmanac annotates molecular features of each feature type can be found in Supplementary Table 1. TARGET and MOAlmanac as present in the study are available as Supplementary Table 2. Data for b,c are available as source data.

Fig. 2 |. MOAlmanac increases the number of nominated clinically relevant molecular features in four retrospective cohorts.

MOAlmanac was benchmarked against PHIAL and TARGET using the molecular profiles of 110 patients with metastatic melanoma, 150 patients with mCRPC, 100 patients with KIRP, and 59 patients with OS. a, Molecular Oncology Almanac increased the number of patients with a clinically relevant somatic variant or copy number alteration from 295 to 365 relative to results from PHIAL; patients are aligned across feature types vertically. b, Molecular features not routinely used in clinical sequencing were utilized to expand translational hypotheses. c, Counts of clinically relevant somatic variants or copy number alterations by ontology. Amp, amplification; del, deletion. d, Counts of clinically relevant molecular features from expanded feature types. WGD, whole-genome doubling. Data are available as source data.

Fig. 3 |. Counts of clinically relevant molecular features observed in retrospective cohorts by method and feature type.

Counts of molecular features labeled as either “putatively actionable” or “investigate actionability” by PHIAL and TARGET versus MOAlmanac. MEL, melanoma. Data are available as source data.

Fig. 4 |. MOAlmanac increases the number of patients with at least one clinically relevant alteration in four retrospective cohorts.

MOAlmanac was benchmarked against PHIAL and TARGET using the molecular profiles of 110 patients with metastatic melanoma, 150 patients with mCRPC, 100 patients with KIRP, and 59 patients with OS. a, MOAlmanac reduces the number of patients with at least one clinically relevant alteration over PHIAL-TARGET and reduces the number of otherwise variant-negative patients by considering additional feature types. CNA, copy number alteration; SNV, single-nucleotide variant. b, Including preclinical evidence for evidence for therapeutic sensitivity provides an additional 68 patients with a molecularly matched therapeutic hypothesis. Data are available as source data.

Fig. 5 |. Profile-to-cell line matchmaking.

MOAlmanac leverages preclinical data from cancer cell lines which have been molecularly characterized and subject to high-throughput therapeutic screens to provide supplemental hypotheses through profile-to-cell line matchmaking. a, Somatic SNVs, CNAs, and fusions of cancer cell lines are formatted, annotated with MOAlmanac and the CGC, and vectorized into sample x feature boolean DataFrames. Feature sets and similarity metrics were evaluated by their ability to sort cell lines relative to one another based on shared genomic features, such that cell lines that shared therapeutic sensitivity were deemed more similar. Metrics from information retrieval were used for evaluation (Methods). b, Models were evaluated on cancer cell lines using a hold-one-out approach. The chosen model utilized Similarity Network Fusion (SNF) to combine networks of somatic variants, copy number alterations, and fusions in CGC genes with specific MOAlmanac features associated with an FDA approval. Nonsyn., nonsynonymous; PCA, principle-component analysis. c, Recurrent nearest neighbors and their sensitive therapies for four patient cohorts. CNS, central nervous system; NB, neuroblastoma. Data for panels b, c are available as source data.

Fig. 6 |. Application of MOAlmanac to a prospective clinical trial.

We investigated if MOAlmanac could highlight similar therapeutic strategies that were utilized by real-world evidence. MOAlmanac was applied to the I-PREDICT trial, which evaluated efficacy of molecularly matched therapies in 83 patients. Therapies and corresponding molecular features were mapped to therapeutic strategies for those administered in I-PREDICT and highlighted by MOAlmanac. a, A shared therapeutic strategy was observed in 39 (47%) of patients, 31 of which involved a therapy most prioritized for the patient by MOAlmanac. b, MOAlmanac nominated therapeutic strategies applied for a given patient more often for those based on well-established evidence (that is, FDA approvals; 60% of therapy-patient pairs) relative to less-established evidence, such as preclinical evidence (18%). c, Therapeutic strategies, individual therapies, and genes and molecular features as administered or targeted by I-PREDICT and highlighted by MOAlmanac. TMB-Int, tumor mutational burden intermediate. Data for panels are available as source data.