Transcriptome deconvolution reveals absence of cancer cell expression signature in immune checkpoint blockade response.

Immune-checkpoint therapy (ICB) has conferred significant and durable clinical benefit to some cancer patients. However, most patients do not respond to ICB, and reliable biomarkers of ICB response are needed to improve patient stratification. Here, we performed a transcriptome-wide meta-analysis across 1,486 tumors from ICB-treated patients and tumors with expected ICB outcomes based on microsatellite status. Using a robust transcriptome deconvolution approach, we inferred cancer and stroma-specific gene expression differences and identified cell-type specific features of ICB response across cancer types. Consistent with current knowledge, stromal expression of CXCL9, CXCL13, and IFNG were the top determinants of favorable ICB response. In addition, we identified a group of potential immune-suppressive genes, including FCER1A, associated with poor response to ICB. Strikingly, PD-L1 expression in stromal cells, but not cancer cells, is correlated with ICB response across cancer types. Furthermore, the unbiased transcriptome-wide analysis failed to identify cancer-cell intrinsic expression signatures of ICB response conserved across tumor types, suggesting that cancer cells lack tissue-agnostic transcriptomic features of ICB response.

Machine learning identifies activation of RUNX/AP-1 as drivers of mesenchymal and fibrotic regulatory programs in gastric cancer.

Gastric cancer (GC) is the fifth most common cancer worldwide and is a heterogeneous disease. Among GC subtypes, the mesenchymal phenotype (Mes-like) is more invasive than the epithelial phenotype (Epi-like). While gene expression of the epithelial-to-mesenchymal transition (EMT) has been studied, the regulatory landscape shaping this process is not fully understood. Here we use ATAC-seq and RNA-seq from a compendium of gastric cancer cell lines and primary tumors to detect drivers of regulatory state changes and their transcriptional responses. Using the ATAC-seq, we developed a machine learning approach to determine the transcription factors (TFs) regulating the subtypes of GC. We identified TFs driving the mesenchymal (RUNX2, ZEB1, SNAI2, AP-1 dimer) as well as the epithelial states (GATA4, GATA6, KLF5, HNF4A, FOXA2, GRHL2) in gastric cancer. We identified DNA copy number alterations associated with dysregulation of these TFs, specifically deletion of GATA4 and amplification of MAPK9 Comparisons with bulk and single-cell RNA-seq datasets identified activation toward fibroblast-like epigenomic and expression signatures in Mes-like GC. The activation of this mesenchymal fibrotic program is associated with differentially accessible DNA cis-regulatory elements flanking up-regulated mesenchymal genes. These findings establish a map of TF activity in GC and highlight the role of copy number driven alterations in shaping epigenomic regulatory programs as potential drivers of gastric cancer heterogeneity and progression.

A multimodal atlas of hepatocellular carcinoma reveals convergent evolutionary paths and 'bad apple' effect on clinical trajectory.

BACKGROUND & AIMS: Hepatocellular Carcinoma (HCC) is a highly fatal cancer characterized by high intra-tumor heterogeneity (ITH). A panoramic understanding of its tumor evolution, in relation to its clinical trajectory, may provide novel prognostic and treatment strategies. METHODS: Through the Asia-Pacific Hepatocellular Carcinoma (AHCC) trials group (NCT03267641), we recruited one of the largest prospective cohorts of HCC with over 600 whole genome and transcriptome samples from 123 treatment-naïve patients. RESULTS: Using a multi-region sampling approach, we revealed seven convergent genetic evolutionary paths governed by the early driver mutations, late copy number variations and viral integrations, which stratify patient clinical trajectories after surgical resection. Furthermore, such evolutionary paths shaped the molecular profiles, leading to distinct transcriptomic subtypes. Most significantly, although we found the coexistence of multiple transcriptomic subtypes within certain tumors, patient prognosis was best predicted by the most aggressive cell fraction of the tumor, rather than by overall degree of transcriptomic ITH level - a phenomenon we termed the 'bad apple' effect. Finally, we found that characteristics throughout early and late tumor evolution provide significant and complementary prognostic power in predicting patient survival. CONCLUSIONS: Taken together, our study generated a comprehensive landscape of evolutionary history for HCC and provided a rich multi-omics resource for understanding tumor heterogeneity and clinical trajectories. CLINICAL TRIAL NUMBER: NCT03267641 (Observational cohort) IMPACT AND IMPLICATIONS: This prospective study, utilizing comprehensive multi-sector, multi-omics sequencing and clinical data from surgically resected HCC, reveals critical insights into the role of tumor evolution and intra-tumor heterogeneity (ITH) in determining the prognosis of Hepatocellular Carcinoma (HCC). These findings are invaluable for oncology researchers and clinicians, as they underscore the influence of distinct evolutionary paths and the 'bad apple' effect, where the most aggressive tumor fraction dictates disease progression. These insights not only enhance prognostic accuracy post-surgical resection but also pave the way for developing personalized therapies tailored to specific tumor evolutionary and transcriptomic profiles. The co-existence of multiple sub-types within the same tumor prompts a re-appraisal of the utilities of depending on single samples to represent the entire tumor and suggests the need for clinical molecular imaging. This research thus marks a significant step forward in the clinical understanding and management of HCC, underscoring the importance of integrating tumor evolutionary dynamics and multi-omics biomarkers into therapeutic decision-making.

Precision Endpoints for Contemporary Precision Oncology Trials.

SUMMARY: Traditional endpoints such as progression-free survival and overall survival do not fully capture the pharmacologic and pharmacodynamic effects of a therapeutic intervention. Incorporating mechanism-driven biomarkers and validated surrogate proximal endpoints can provide orthogonal readouts of anti-tumor activity and delineate the relative contribution of treatment components on an individual level, highlighting the limitation of solely relying on aggregated readouts from clinical trials to facilitate go/no-go decisions for precision therapies.

Oncogene-driven non-small cell lung cancers in patients with a history of smoking lack smoking-induced mutations.

Non-small cell lung cancers (NSCLCs) in non-smokers are mostly driven by mutations in the oncogenes EGFR, ERBB2, and MET and fusions involving ALK and RET. In addition to occurring in non-smokers, alterations in these "non-smoking-related oncogenes" (NSROs) also occur in smokers. To better understand the clonal architecture and genomic landscape of NSRO-driven tumors in smokers compared to typical-smoking NSCLCs, we investigated genomic and transcriptomic alterations in 173 tumor sectors from 48 NSCLC patients. NSRO-driven NSCLCs in smokers and non-smokers had similar genomic landscapes. Surprisingly, even in patients with prominent smoking histories, the mutational signature caused by tobacco smoking was essentially absent in NSRO-driven NSCLCs, which was confirmed in two large NSCLC datasets from other geographic regions. However, NSRO-driven NSCLCs in smokers had higher transcriptomic activities related to regulation of the cell cycle. These findings suggest that, while the genomic landscape is similar between NSRO-driven NSCLC in smokers and non-smokers, smoking still affects the tumor phenotype independently of genomic alterations.

Pushing the Frontiers of Cancer Research: Highlights from the Frontiers in Cancer Science Conference 2023.

The 15th annual Frontiers in Cancer Science (FCS) conference gathered scientific experts who shared the latest research converging upon several themes of cancer biology. These themes included the dysregulation of metabolism, cell death, and other signaling processes in cancer cells; using patient "omics" datasets and single-cell and spatial approaches to investigate heterogeneity, understand therapy resistance, and identify targets; innovative strategies for inhibiting tumors, including rational drug combinations and improved drug delivery mechanisms; and advances in models that can facilitate screening for cancer vulnerabilities and drug testing. We hope the insights from this meeting will stimulate further progress in the field.

Detection of circulating tumor DNA with ultradeep sequencing of plasma cell-free DNA for monitoring minimal residual disease and early detection of recurrence in early-stage lung cancer.

BACKGROUND: In early-stage non-small cell lung cancer (NSCLC), recurrence is frequently observed. Circulating tumor DNA (ctDNA) has emerged as a noninvasive tool to risk stratify patients for recurrence after curative intent therapy. This study aimed to risk stratify patients with early-stage NSCLC via a personalized, tumor-informed multiplex polymerase chain reaction (mPCR) next-generation sequencing assay. METHODS: This retrospective cohort study included patients with stage I-III NSCLC. Recruited patients received standard-of-care management (surgical resection with or without adjuvant chemotherapy, followed by surveillance). Whole-exome sequencing of NSCLC resected tissue and matched germline DNA was used to design patient-specific mPCR assays (Signatera, Natera, Inc) to track up to 16 single-nucleotide variants in plasma samples. RESULTS: The overall cohort with analyzed plasma samples consisted of 57 patients. Stage distribution was 68% for stage I and 16% each for stages II and III. Presurgery (i.e., at baseline), ctDNA was detected in 15 of 57 patients (26%). ctDNA detection presurgery was significantly associated with shorter recurrence-free survival (RFS; hazard ratio [HR], 3.54; 95% confidence interval [CI], 1.00-12.62; p = .009). In the postsurgery setting, ctDNA was detected in seven patients, of whom 100% experienced radiological recurrence. ctDNA positivity preceded radiological findings by a median lead time of 2.8 months (range, 0-12.9 months). Longitudinally, ctDNA detection at any time point was associated with shorter RFS (HR, 16.1; 95% CI, 1.63-158.9; p < .0001). CONCLUSIONS: ctDNA detection before surgical resection was strongly associated with a high risk of relapse in early-stage NSCLC in a large unique Asian cohort. Prospective studies are needed to assess the clinical utility of ctDNA status in this setting.

Landscape of germline pathogenic variants in patients with dual primary breast and lung cancer.

BACKGROUND: Cancer predisposition is most often studied in the context of single cancers. However, inherited cancer predispositions can also give rise to multiple primary cancers. Yet, there is a paucity of studies on genetic predisposition in multiple primary cancers, especially those outside of well-defined cancer predisposition syndromes. This study aimed to identify germline variants associated with dual primary cancers of the breast and lung. METHODS: Exome sequencing was performed on germline DNA from 55 Singapore patients (52 [95%] never-smokers) with dual primaries in the breast and lung, confirmed by histopathology. Using two large control cohorts: the local SG10K_Health (n = 9770) and gnomAD non-cancer East Asians (n = 9626); and two additional local case cohorts of early-onset or familial breast cancer (n = 290), and lung cancer (n = 209), variants were assessed for pathogenicity in accordance with ACMG/AMP guidelines. In particular, comparisons were made with known pathogenic or likely pathogenic variants in the ClinVar database, pathogenicity predictions were obtained from in silico prediction software, and case-control association analyses were performed. RESULTS: Altogether, we identified 19 pathogenic or likely pathogenic variants from 16 genes, detected in 17 of 55 (31%) patients. Six of the 19 variants were identified using ClinVar, while 13 variants were classified pathogenic or likely pathogenic using ACMG/AMP guidelines. The 16 genes include well-known cancer predisposition genes such as BRCA2, TP53, and RAD51D; but also lesser known cancer genes EXT2, WWOX, GATA2, and GPC3. Most of these genes are involved in DNA damage repair, reaffirming the role of impaired DNA repair mechanisms in the development of multiple malignancies. These variants warrant further investigations in additional populations. CONCLUSIONS: We have identified both known and novel variants significantly enriched in patients with primary breast and lung malignancies, expanding the body of known cancer predisposition variants for both breast and lung cancer. These variants are mostly from genes involved in DNA repair, affirming the role of impaired DNA repair in the predisposition and development of multiple cancers.

PUREE: accurate pan-cancer tumor purity estimation from gene expression data.

Tumors are complex masses composed of malignant and non-malignant cells. Variation in tumor purity (proportion of cancer cells in a sample) can both confound integrative analysis and enable studies of tumor heterogeneity. Here we developed PUREE, which uses a weakly supervised learning approach to infer tumor purity from a tumor gene expression profile. PUREE was trained on gene expression data and genomic consensus purity estimates from 7864 solid tumor samples. PUREE predicted purity with high accuracy across distinct solid tumor types and generalized to tumor samples from unseen tumor types and cohorts. Gene features of PUREE were further validated using single-cell RNA-seq data from distinct tumor types. In a comprehensive benchmark, PUREE outperformed existing transcriptome-based purity estimation approaches. Overall, PUREE is a highly accurate and versatile method for estimating tumor purity and interrogating tumor heterogeneity from bulk tumor gene expression data, which can complement genomics-based approaches or be used in settings where genomic data is unavailable.

Comprehensive molecular phenotyping of ARID1A-deficient gastric cancer reveals pervasive epigenomic reprogramming and therapeutic opportunities.

OBJECTIVE: Gastric cancer (GC) is a leading cause of cancer mortality, with ARID1A being the second most frequently mutated driver gene in GC. We sought to decipher ARID1A-specific GC regulatory networks and examine therapeutic vulnerabilities arising from ARID1A loss. DESIGN: Genomic profiling of GC patients including a Singapore cohort (>200 patients) was performed to derive mutational signatures of ARID1A inactivation across molecular subtypes. Single-cell transcriptomic profiles of ARID1A-mutated GCs were analysed to examine tumour microenvironmental changes arising from ARID1A loss. Genome-wide ARID1A binding and chromatin profiles (H3K27ac, H3K4me3, H3K4me1, ATAC-seq) were generated to identify gastric-specific epigenetic landscapes regulated by ARID1A. Distinct cancer hallmarks of ARID1A-mutated GCs were converged at the genomic, single-cell and epigenomic level, and targeted by pharmacological inhibition. RESULTS: We observed prevalent ARID1A inactivation across GC molecular subtypes, with distinct mutational signatures and linked to a NFKB-driven proinflammatory tumour microenvironment. ARID1A-depletion caused loss of H3K27ac activation signals at ARID1A-occupied distal enhancers, but unexpectedly gain of H3K27ac at ARID1A-occupied promoters in genes such as NFKB1 and NFKB2. Promoter activation in ARID1A-mutated GCs was associated with enhanced gene expression, increased BRD4 binding, and reduced HDAC1 and CTCF occupancy. Combined targeting of promoter activation and tumour inflammation via bromodomain and NFKB inhibitors confirmed therapeutic synergy specific to ARID1A-genomic status. CONCLUSION: Our results suggest a therapeutic strategy for ARID1A-mutated GCs targeting both tumour-intrinsic (BRD4-assocatiated promoter activation) and extrinsic (NFKB immunomodulation) cancer phenotypes.

Liquid Biopsies for Cancer Genotyping: Coming of Age?

Precision oncology is predicated on optimal molecular profiling that is "fit for purpose" to identify therapeutic vulnerabilities. Liquid biopsies may compensate for inadequate genotyping, but remain less sensitive and specific compared with tissue biopsies. The liquid biopsy toolbox is poised to expand through novel assays and insights from longitudinal profiling. See related article by Sugimoto et al., p. 1506.

PD-L1 score as a prognostic biomarker in asian early-stage epidermal growth factor receptor-mutated lung cancer.

AIM: To determine the prognostic value of programmed death-ligand 1 (PD-L1) score in early-stage epidermal growth factor receptor (EGFR)-mutated non-small cell lung cancer (NSCLC), contrasted against EGFR-wildtype NSCLC. METHODS: Consecutive patients with Stage IA-IIIA NSCLC diagnosed 1st January 2010-31st December 2019 at National Cancer Centre Singapore with evaluable EGFR and PD-L1 status were included. Co-primary end-points were 2-year disease-free survival (DFS) and 5-year overall survival (OS) by Kaplan-Meier method. RESULTS: 455 patients were included (267 EGFR-mutated, EGFR-M+; 188 EGFR-wildtype, wt). Median age at diagnosis was 65 years, 52.3% (238/455) of patients were males, 62.9% (286/455) of patients were never-smokers and 92.5% (421/455) of patients had R0 resection. Stage IA comprised 42.4% (193/455) of patients, Stage IB comprised 23.1% (105/455) of patients, Stage IIA comprised 10.8% of patients (49/455), Stage IIB comprised 5.1% of patients (23/455) and Stage IIIA comprised 18.7% (85/455) of patients. Among EGFR-M+, 45.3% (121/267) were Ex19del and 41.9% (112/267) were L858R. PD-L1 ≥1% among EGFR-M+ and EGFR-wt was 45.3% (121/267) and 54.8% (103/188) respectively (p = 0.047). At median follow-up of 47 months, 178 patients had relapsed. Among EGFR-M+, 2-year DFS comparing PD-L1 <1% and PD-L1 ≥1% was 78.1% and 67.6% (p = 0.007) while 5-year OS was 59.5% and 42.8% (p = 0.001), respectively. Controlling for age, gender, lymphovascular invasion, adjuvant therapy and resection margin status, PD-L1 ≥1% (hazard ratio, HR 2.18, 95% CI 1.04-4.54, p = 0.038), stage IIB (HR 7.78, 95% CI 1.72-35.27, p = 0.008) and stage IIIA (HR 4.45, 95% CI 1.44-13.80, p = 0.01) emerged as independent predictors of inferior OS on multivariable analysis. In exploratory analysis, genomic analysis of 81 EGFR-M+ tumours was performed. PD-L1 ≥1% tumours had significantly higher rates of TP53 mutations (36.1% versus 15.6%, p = 0.04), with predominantly missense mutations. CONCLUSION: PD-L1 ≥1% is an independent predictor of worse OS among early-stage EGFR-mutated NSCLC and is associated with inferior DFS regardless of EGFR status. PD-L1 score as a risk stratification factor should be evaluated in prospective adjuvant studies among EGFR-mutated NSCLC.

Regulatory enhancer profiling of mesenchymal-type gastric cancer reveals subtype-specific epigenomic landscapes and targetable vulnerabilities.

OBJECTIVE: Gastric cancer (GC) comprises multiple molecular subtypes. Recent studies have highlighted mesenchymal-subtype GC (Mes-GC) as a clinically aggressive subtype with few treatment options. Combining multiple studies, we derived and applied a consensus Mes-GC classifier to define the Mes-GC enhancer landscape revealing disease vulnerabilities. DESIGN: Transcriptomic profiles of ~1000 primary GCs and cell lines were analysed to derive a consensus Mes-GC classifier. Clinical and genomic associations were performed across >1200 patients with GC. Genome-wide epigenomic profiles (H3K27ac, H3K4me1 and assay for transposase-accessible chromatin with sequencing (ATAC-seq)) of 49 primary GCs and GC cell lines were generated to identify Mes-GC-specific enhancer landscapes. Upstream regulators and downstream targets of Mes-GC enhancers were interrogated using chromatin immunoprecipitation followed by sequencing (ChIP-seq), RNA sequencing, CRISPR/Cas9 editing, functional assays and pharmacological inhibition. RESULTS: We identified and validated a 993-gene cancer-cell intrinsic Mes-GC classifier applicable to retrospective cohorts or prospective single samples. Multicohort analysis of Mes-GCs confirmed associations with poor patient survival, therapy resistance and few targetable genomic alterations. Analysis of enhancer profiles revealed a distinctive Mes-GC epigenomic landscape, with TEAD1 as a master regulator of Mes-GC enhancers and Mes-GCs exhibiting preferential sensitivity to TEAD1 pharmacological inhibition. Analysis of Mes-GC super-enhancers also highlighted NUAK1 kinase as a downstream target, with synergistic effects observed between NUAK1 inhibition and cisplatin treatment. CONCLUSION: Our results establish a consensus Mes-GC classifier applicable to multiple transcriptomic scenarios. Mes-GCs exhibit a distinct epigenomic landscape, and TEAD1 inhibition and combinatorial NUAK1 inhibition/cisplatin may represent potential targetable options.

Brief Report: Droplet Digital Polymerase Chain Reaction Versus Plasma Next-Generation Sequencing in Detecting Clearance of Plasma EGFR Mutations and Carcinoembryonic Antigen Levels as a Surrogate Measure.

Introduction: To compare the performance of droplet digital polymerase chain reaction (ddPCR) and plasma next-generation sequencing (NGS) in detecting clearance of plasma EGFR (pEGFR) mutations. Methods: Patients with treatment-naive advanced EGFR-mutated lung cancer treated with first-line tyrosine kinase inhibitors (TKIs) were included. pEGFR were measured at baseline and first response assessment using ddPCR and NGS. Clearance of pEGFR was defined as undetectable levels after a positive baseline result. Results were correlated with time-to-treatment failure (TTF). In exploratory analysis, corresponding change in carcinoembryonic antigen (CEA) levels was evaluated. Results: Between January 1, 2020, and December 31, 2021, 27 patients were recruited. Ex19del comprised 74% (20 of 27) and L858R 26% (seven of 27). Osimertinib was used in 59% (16 of 27), dacomitinib 4% (one of 27), and gefitinib/erlotinib 37% (10 of 27). Sensitivity of ddPCR and NGS in detecting pEGFR mutation at baseline was 70% (19 of 27) and 78% (21 of 27), respectively (p = 0.16). All patients with detectable pEGFR by ddPCR were detected by NGS.At a median of 8 (range 3-24) weeks post-TKI initiation, clearance of pEGFR was achieved in 68% (13 of 19) and 71% (15 of 21) using ddPCR and NGS, respectively. Concordance between ddPCR and NGS was 79% (kappa = 0.513, p = 0.013). Clearance of pEGFR was associated with longer median TTF (not reached versus 6 months, p = 0.03) and median decrease in CEA levels by 70% from baseline.In another cohort of 124 patients, decrease in CEA levels by greater than 70% within 90 days of TKI initiation was associated with doubling of both TTF and overall survival. Conclusions: Plasma NGS trended toward higher sensitivity than ddPCR in detecting pEGFR, although both had similar concordance in detecting pEGFR clearance. Our results support using NGS at diagnosis and interchangeability of NGS and ddPCR for monitoring, whereas CEA could be explored as a surrogate for pEGFR clearance.

A Randomized Phase 2 Trial of Nivolumab Versus Nivolumab-Ipilimumab Combination in EGFR-Mutant NSCLC.

Introduction: Although immune checkpoint inhibitors (ICIs) have dramatically improved outcomes for nononcogene-addicted NSCLC, monotherapy with programmed cell death protein-1 (PD1) inhibition has been associated with low efficacy in the EGFR-mutant setting. Given the potential for synergism with combination checkpoint blockade, we designed a trial to test the activity of combination nivolumab (N)-ipilimumab (NI) in EGFR-mutant NSCLC. Methods: This is a randomized phase 2 study (NCT03091491) of N versus NI combination in EGFR tyrosine kinase inhibitor (TKI)-resistant NSCLC, with crossover permitted on disease progression. The primary end point was the objective response rate, and the secondary end points included progression-free survival, overall survival, and safety of ICI after EGFR TKI. Results: Recruitment ceased owing to futility after 31 of 184 planned patients were treated. A total of 15 patients received N and 16 received NI combination. There were 16 patients (51.6%) who had programmed death-ligand (PDL1) 1 greater than or equal to 1%, and 15 (45.2%) harbored EGFR T790M. Five patients derived clinical benefits from ICI with one objective response (objective response rate 3.2%), and median progression-free survival was 1.22 months (95% confidence interval: 1.15-1.35) for the overall cohort. None of the four patients who crossed over achieved salvage response by NI. PDL1 and tumor mutational burden (TMB) were not able to predict ICI response. Rates of all grade immune-related adverse events were similar (80% versus 75%), with only two grade 3 events. Conclusions: Immune checkpoint inhibition is ineffective in EGFR TKI-resistant NSCLC. Whereas a small subgroup of EGFR-mutant NSCLC may be immunogenic and responsive to ICI, better biomarkers are needed to select appropriate patients.

Clinical and Genomic Features of HER2 Exon 20 Insertion Mutations and Characterization of HER2 Expression by Immunohistochemistry in East Asian Non-Small-Cell Lung Cancer.

PURPOSE: HER2-altered non-small-cell lung cancer (NSCLC) represents a diverse subgroup, including mutations, amplifications, and overexpression. However, HER2 exon 20 insertion mutations are emerging as a distinct molecular subtype with expanding therapeutic options. We describe the molecular epidemiology and genomic features of HER2-altered NSCLC in an Asian tertiary cancer center. METHODS: We identified patients with HER2-mutated NSCLC in our institutional database, collating clinicopathological features and treatment outcomes. The genomic landscape of human epidermal growth factor receptor 2 (HER2)-mutated NSCLC was further evaluated using whole-exome sequencing (WES) data from combined local and publicly available data sets. HER2 amplification and overexpression as selection biomarkers in NSCLC were further interrogated using HER2 immunohistochemistry and correlations with WES and RNA sequencing data. RESULTS: Among 1,252 patients with consecutive lung adenocarcinoma undergoing routine next-generation sequencing, the prevalence of HER2 mutations was 3.1%-exon 20 insertion mutations comprised 2.7%. We examined the clinicopathological features in 55 patients with HER2-mutated NSCLC comprising 40 exon 20 insertion and 15 nonexon 20 insertion mutations. The most common exon 20 insertion mutation was HER2Y772_A775dup in 30 (75%), followed by HER2G776delinsVC in five patients (13%). There were limited responses to HER2-directed therapies apart from trastuzumab-deruxtecan, and no responses were seen with immunotherapy monotherapy. Evaluating the genomics features of HER2 exon 20 insertion mutations using WES data revealed low tumor mutational burden (TMB), low incidence of cancer driver comutations, and a predominance of aging mutational signature-similar to EGFR-mutated tumors. In contrast, uncommon (or nonexon 20 insertion) HER2-mutated tumors resembled EGFR wild-type tumors with higher TMB, higher frequency of cancer driver comutations, and greater presence of smoking and APOBEC mutational signature. Finally, in evaluating HER2 immunohistochemistry in all lung adenocarcinoma, there was significant discordance comparing different scoring systems and poor correlation with HER2 RNA expression and HER2 amplification. CONCLUSION: The incidence of HER2 mutations is 3.1% in East Asian nonsquamous NSCLC. HER2 exon 20 insertion-mutated tumors appear genomically distinct from uncommon (nonexon 20 insertion) HER2 mutations, the latter demonstrating higher TMB, co-occurring drivers, and predominant nonaging mutational signature. The therapeutic implications of the genomic and clinical features of HER2-mutated NSCLC warrant further investigation.

Heat selection enables highly scalable methylome profiling in cell-free DNA for noninvasive monitoring of cancer patients.

Genome-wide analysis of cell-free DNA methylation profile is a promising approach for sensitive and specific detection of many cancers. However, scaling such assays for clinical translation is impractical because of the high cost of whole-genome bisulfite sequencing. We show that the small fraction of GC-rich genome is highly enriched in CpG sites and disproportionately harbors most of the cancer-specific methylation signature. Here, we report on the simple and effective heat enrichment of CpG-rich regions for bisulfite sequencing (Heatrich-BS) platform that allows for focused methylation profiling in these highly informative regions. Our novel method and bioinformatics algorithm enable accurate tumor burden estimation and quantitative tracking of colorectal cancer patient's response to treatment at much reduced sequencing cost suitable for frequent monitoring. We also show tumor epigenetic subtyping using Heatrich-BS, which could enable patient stratification. Heatrich-BS holds great potential for highly scalable screening and monitoring of cancer using liquid biopsy.

Accurate somatic variant detection using weakly supervised deep learning.

Identification of somatic mutations in tumor samples is commonly based on statistical methods in combination with heuristic filters. Here we develop VarNet, an end-to-end deep learning approach for identification of somatic variants from aligned tumor and matched normal DNA reads. VarNet is trained using image representations of 4.6 million high-confidence somatic variants annotated in 356 tumor whole genomes. We benchmark VarNet across a range of publicly available datasets, demonstrating performance often exceeding current state-of-the-art methods. Overall, our results demonstrate how a scalable deep learning approach could augment and potentially supplant human engineered features and heuristic filters in somatic variant calling.

Single-cell and bulk transcriptome sequencing identifies two epithelial tumor cell states and refines the consensus molecular classification of colorectal cancer.

The consensus molecular subtype (CMS) classification of colorectal cancer is based on bulk transcriptomics. The underlying epithelial cell diversity remains unclear. We analyzed 373,058 single-cell transcriptomes from 63 patients, focusing on 49,155 epithelial cells. We identified a pervasive genetic and transcriptomic dichotomy of malignant cells, based on distinct gene expression, DNA copy number and gene regulatory network. We recapitulated these subtypes in bulk transcriptomes from 3,614 patients. The two intrinsic subtypes, iCMS2 and iCMS3, refine CMS. iCMS3 comprises microsatellite unstable (MSI-H) cancers and one-third of microsatellite-stable (MSS) tumors. iCMS3 MSS cancers are transcriptomically more similar to MSI-H cancers than to other MSS cancers. CMS4 cancers had either iCMS2 or iCMS3 epithelium; the latter had the worst prognosis. We defined the intrinsic epithelial axis of colorectal cancer and propose a refined 'IMF' classification with five subtypes, combining intrinsic epithelial subtype (I), microsatellite instability status (M) and fibrosis (F).