Diffsig: Associating Risk Factors with Mutational Signatures.

BACKGROUND: Somatic mutational signatures elucidate molecular vulnerabilities to therapy, and therefore detecting signatures and classifying tumors with respect to signatures has clinical value. However, identifying the etiology of the mutational signatures remains a statistical challenge, with both small sample sizes and high variability in classification algorithms posing barriers. As a result, few signatures have been strongly linked to particular risk factors. METHODS: Here, we develop a statistical model, Diffsig, for estimating the association of one or more continuous or categorical risk factors with DNA mutational signatures. Diffsig takes into account the uncertainty associated with assigning signatures to samples as well as multiple risk factors' simultaneous effect on observed DNA mutations. RESULTS: We applied Diffsig to breast cancer data to assess relationships between five established breast-relevant mutational signatures and etiologic variables, confirming known mechanisms of cancer development. In simulation, our model was capable of accurately estimating expected associations in a variety of contexts. CONCLUSIONS: Diffsig allows researchers to quantify and perform inference on the associations of risk factors with mutational signatures. IMPACT: We expect Diffsig to provide more robust associations of risk factors with signatures to lead to better understanding of the tumor development process and improved models of tumorigenesis.

Diffsig: Associating Risk Factors With Mutational Signatures.

Somatic mutational signatures elucidate molecular vulnerabilities to therapy and therefore detecting signatures and classifying tumors with respect to signatures has clinical value. However, identifying the etiology of the mutational signatures remains a statistical challenge, with both small sample sizes and high variability in classification algorithms posing barriers. As a result, few signatures have been strongly linked to particular risk factors. Here we present Diffsig, a model and R package for estimating the association of risk factors with mutational signatures, suggesting etiologies for the pre-defined mutational signatures. Diffsig is a Bayesian Dirichlet-multinomial hierarchical model that allows testing of any type of risk factor while taking into account the uncertainty associated with samples with a low number of observations. In simulation, we found that our method can accurately estimate risk factor-mutational signal associations. We applied Diffsig to breast cancer data to assess relationships between five established breast-relevant mutational signatures and etiologic variables, confirming known mechanisms of cancer development. Diffsig is implemented as an R package available at: https://github.com/jennprk/diffsig.

Multiomics in primary and metastatic breast tumors from the AURORA US network finds microenvironment and epigenetic drivers of metastasis.

The AURORA US Metastasis Project was established with the goal to identify molecular features associated with metastasis. We assayed 55 females with metastatic breast cancer (51 primary cancers and 102 metastases) by RNA sequencing, tumor/germline DNA exome and low-pass whole-genome sequencing and global DNA methylation microarrays. Expression subtype changes were observed in ~30% of samples and were coincident with DNA clonality shifts, especially involving HER2. Downregulation of estrogen receptor (ER)-mediated cell-cell adhesion genes through DNA methylation mechanisms was observed in metastases. Microenvironment differences varied according to tumor subtype; the ER+/luminal subtype had lower fibroblast and endothelial content, while triple-negative breast cancer/basal metastases showed a decrease in B and T cells. In 17% of metastases, DNA hypermethylation and/or focal deletions were identified near HLA-A and were associated with reduced expression and lower immune cell infiltrates, especially in brain and liver metastases. These findings could have implications for treating individuals with metastatic breast cancer with immune- and HER2-targeting therapies.

DNA Damage Repair Classifier Defines Distinct Groups in Hepatocellular Carcinoma.

DNA repair pathways have been associated with variability in hepatocellular carcinoma (HCC) clinical outcomes, but the mechanism through which DNA repair varies as a function of liver regeneration and other HCC characteristics is poorly understood. We curated a panel of 199 genes representing 15 DNA repair pathways to identify DNA repair expression classes and evaluate their associations with liver features and clinicopathologic variables in The Cancer Genome Atlas (TCGA) HCC study. We identified two groups in HCC, defined by low or high expression across all DNA repair pathways. The low-repair group had lower grade and retained the expression of classical liver markers, whereas the high-repair group had more clinically aggressive features, increased p53 mutant-like gene expression, and high liver regenerative gene expression. These pronounced features overshadowed the variation in the low-repair subset, but when considered separately, the low-repair samples included three subgroups: L1, L2, and L3. L3 had high DNA repair expression with worse progression-free (HR 1.24, 95% CI 0.81-1.91) and overall (HR 1.63, 95% CI 0.98-2.71) survival. High-repair outcomes were also significantly worse compared with the L1 and L2 groups. HCCs vary in DNA repair expression, and a subset of tumors with high regeneration profoundly disrupts liver biology and poor prognosis.

RNA-Based Classification of Homologous Recombination Deficiency in Racially Diverse Patients with Breast Cancer.

BACKGROUND: Aberrant expression of DNA repair pathways such as homologous recombination (HR) can lead to DNA repair imbalance, genomic instability, and altered chemotherapy response. DNA repair imbalance may predict prognosis, but variation in DNA repair in diverse cohorts of breast cancer patients is understudied. METHODS: To identify RNA-based patterns of DNA repair expression, we performed unsupervised clustering on 51 DNA repair-related genes in the Cancer Genome Atlas Breast Cancer [TCGA BRCA (n = 1,094)] and Carolina Breast Cancer Study [CBCS (n = 1,461)]. Using published DNA-based HR deficiency (HRD) scores (high-HRD ≥ 42) from TCGA, we trained an RNA-based supervised classifier. Unsupervised and supervised HRD classifiers were evaluated in association with demographics, tumor characteristics, and clinical outcomes. RESULTS: : Unsupervised clustering on DNA repair genes identified four clusters of breast tumors, with one group having high expression of HR genes. Approximately 39.7% of CBCS and 29.3% of TCGA breast tumors had this unsupervised high-HRD (U-HRD) profile. A supervised HRD classifier (S-HRD) trained on TCGA had 84% sensitivity and 73% specificity to detect HRD-high samples. Both U-HRD and S-HRD tumors in CBCS had higher frequency of TP53 mutant-like status (45% and 41% enrichment) and basal-like subtype (63% and 58% enrichment). S-HRD high was more common among black patients. Among chemotherapy-treated participants, recurrence was associated with S-HRD high (HR: 2.38, 95% confidence interval = 1.50-3.78). CONCLUSIONS: HRD is associated with poor prognosis and enriched in the tumors of black women. IMPACT: RNA-level indicators of HRD are predictive of breast cancer outcomes in diverse populations.

Genomic characterization of rare molecular subclasses of hepatocellular carcinoma.

Primary liver cancer, consisting of both cholangiocarcinoma (CCA) and hepatocellular carcinoma (HCC), is the second leading cause of cancer deaths worldwide. Our goal is to genomically characterize rare HCC subclasses to provide insight into disease biology. Leveraging The Cancer Genome Atlas (TCGA) to perform a combined analysis of CCA (n = 36) and HCC (n = 275), we integrated multiple genomic platforms, to assess transcriptional profiles, mutational signatures, and copy number patterns to uncover underlying etiology and linage specific patterns. We identified two molecular classes distinct from prototypical HCC tumors. The first, CCA-Like, although histologically indistinguishable from HCC, had enrichment of CCA mutations (IDH1, BAP1), mutational signatures, and transcriptional patterns (SOX9, KRT19). CCA-Like, however, retained a copy number landscape similar to HCC, suggesting a hepatocellular linage. The second, Blast-Like, is enriched in TP53 mutations, HBV infection, exposure related mutational signatures and transcriptionally similar to hepatoblasts. Although these subclasses are molecularly distinct, they both have a worse progression-free survival compared to classical HCC tumors, yet are clinically treated the same. The identification of and characterization of CCA-Like and Blast-Like subclasses advance our knowledge of HCC as well as represents an urgent need for the identification of class specific biomarkers and targeted therapy.

Identification of a Novel Inflamed Tumor Microenvironment Signature as a Predictive Biomarker of Bacillus Calmette-Guérin Immunotherapy in Non-Muscle-Invasive Bladder Cancer.

PURPOSE: Improved risk stratification and predictive biomarkers of treatment response are needed for non-muscle-invasive bladder cancer (NMIBC). Here we assessed the clinical utility of targeted RNA and DNA molecular profiling in NMIBC. EXPERIMENTAL DESIGN: Gene expression in NMIBC samples was profiled by NanoString nCounter, an RNA quantification platform, from two independent cohorts (n = 28, n = 50); targeted panel sequencing was performed in a subgroup (n = 50). Gene signatures were externally validated using two RNA sequencing datasets of NMIBC tumors (n = 438, n = 73). Established molecular subtype classifiers and novel gene expression signatures were assessed for associations with clinicopathologic characteristics, somatic tumor mutations, and treatment outcomes. RESULTS: Molecular subtypes distinguished between low-grade Ta tumors with FGFR3 mutations and overexpression (UROMOL-class 1) and tumors with more aggressive clinicopathologic characteristics (UROMOL-classes 2 and 3), which were significantly enriched with TERT promoter mutations. However, UROMOL subclasses were not associated with recurrence after bacillus Calmette-Guérin (BCG) immunotherapy in two independent cohorts. In contrast, a novel expression signature of an inflamed tumor microenvironment (TME) was associated with improved recurrence-free survival after BCG. Expression of immune checkpoint genes (PD-L1/PD-1/CTLA-4) was associated with an inflamed TME, but not with higher recurrence rates after BCG. FGFR3 mutations and overexpression were both associated with low immune signatures. CONCLUSIONS: Assessment of the immune TME, rather than molecular subtypes, is a promising predictive biomarker of BCG response. Modulating the TME in an immunologically "cold" tumor warrants further investigation. Integrated transcriptomic and exome sequencing should improve treatment selection in NMIBC.