Enhancing anticancer activity of macrophages through rational drug combinations.

Targeting tumor-associated macrophages (TAMs) is an emerging approach being tested in multiple clinical trials. TAMs, depending on their differentiation state, can exhibit pro- or antitumorigenic functions. For example, the M2-like phenotype represents a protumoral state that can stimulate tumor growth, angiogenesis, metastasis, therapy resistance, and immune evasion by expressing immune checkpoint proteins. In this issue of the JCI, Vaccaro and colleagues utilized an innovative drug screen approach to demonstrate that targeting driver oncogenic signaling pathways concurrently with anti-CD47 sensitizes tumor cells, causing them to undergo macrophage-induced phagocytosis. The combination treatment altered expression of molecules on the tumor cells that typically limit phagocytosis. It also reprogrammed macrophages to an M1-like antitumor state. Moreover, the approach was generalizable to tumor cells with different oncogenic pathways, opening the door to precision oncology-based rationale combination therapies that have the potential to improve outcomes for patients with oncogene-driven lung cancers and likely other cancer types.

Data Valuation with Gradient Similarity.

High-quality data is crucial for accurate machine learning and actionable analytics, however, mislabeled or noisy data is a common problem in many domains. Distinguishing low- from high-quality data can be challenging, often requiring expert knowledge and considerable manual intervention. Data Valuation algorithms are a class of methods that seek to quantify the value of each sample in a dataset based on its contribution or importance to a given predictive task. These data values have shown an impressive ability to identify mislabeled observations, and filtering low-value data can boost machine learning performance. In this work, we present a simple alternative to existing methods, termed Data Valuation with Gradient Similarity (DVGS). This approach can be easily applied to any gradient descent learning algorithm, scales well to large datasets, and performs comparably or better than baseline valuation methods for tasks such as corrupted label discovery and noise quantification. We evaluate the DVGS method on tabular, image and RNA expression datasets to show the effectiveness of the method across domains. Our approach has the ability to rapidly and accurately identify low-quality data, which can reduce the need for expert knowledge and manual intervention in data cleaning tasks.

Androgen receptor activity inversely correlates with immune cell infiltration and immunotherapy response across multiple cancer lineages.

There is now increasing recognition of the important role of androgen receptor (AR) in modulating immune function. To gain a comprehensive understanding of the effects of AR activity on cancer immunity, we employed a computational approach to profile AR activity in 33 human tumor types using RNA-Seq datasets from The Cancer Genome Atlas. Our pan-cancer analysis revealed that the genes most negatively correlated with AR activity across cancers are involved in active immune system processes. Importantly, we observed a significant negative correlation between AR activity and IFNγ pathway activity at the pan-cancer level. Indeed, using a matched biopsy dataset from subjects with prostate cancer before and after AR-targeted treatment, we verified that inhibiting AR enriches immune cell abundances and is associated with higher IFNγ pathway activity. Furthermore, by analyzing immunotherapy datasets in multiple cancers, our results demonstrate that low AR activity was significantly associated with a favorable response to immunotherapy. Together, our data provide a comprehensive assessment of the relationship between AR signaling and tumor immunity.

C5aR1 inhibition reprograms tumor associated macrophages and reverses PARP inhibitor resistance in breast cancer.

Although Poly (ADP-ribose) polymerase (PARP) inhibitors (PARPi) have been approved in multiple diseases, including BRCA1/2 mutant breast cancer, responses are usually transient requiring the deployment of combination therapies for optimal efficacy. Here we thus explore mechanisms underlying sensitivity and resistance to PARPi using two intrinsically PARPi sensitive (T22) and resistant (T127) syngeneic murine breast cancer models in female mice. We demonstrate that tumor associated macrophages (TAM) potentially contribute to the differential sensitivity to PARPi. By single-cell RNA-sequencing, we identify a TAM_C3 cluster, expressing genes implicated in anti-inflammatory activity, that is enriched in PARPi resistant T127 tumors and markedly decreased by PARPi in T22 tumors. Rps19/C5aR1 signaling is selectively elevated in TAM_C3. C5aR1 inhibition or transferring C5aR1hi cells increases and decreases PARPi sensitivity, respectively. High C5aR1 levels in human breast cancers are associated with poor responses to immune checkpoint blockade. Thus, targeting C5aR1 may selectively deplete pro-tumoral macrophages and engender sensitivity to PARPi and potentially other therapies.

Proteogenomic insights into early-onset endometrioid endometrial carcinoma: predictors for fertility-sparing therapy response.

Endometrial carcinoma remains a public health concern with a growing incidence, particularly in younger women. Preserving fertility is a crucial consideration in the management of early-onset endometrioid endometrial carcinoma (EEEC), particularly in patients under 40 who maintain both reproductive desire and capacity. To illuminate the molecular characteristics of EEEC, we undertook a large-scale multi-omics study of 215 patients with endometrial carcinoma, including 81 with EEEC. We reveal an unexpected association between exposome-related mutational signature and EEEC, characterized by specific CTNNB1 and SIGLEC10 hotspot mutations and disruption of downstream pathways. Interestingly, SIGLEC10Q144K mutation in EEECs resulted in aberrant SIGLEC-10 protein expression and promoted progestin resistance by interacting with estrogen receptor alpha. We also identified potential protein biomarkers for progestin response in fertility-sparing treatment for EEEC. Collectively, our study establishes a proteogenomic resource of EEECs, uncovering the interactions between exposome and genomic susceptibilities that contribute to the development of primary prevention and early detection strategies for EEECs.

MIM-CyCIF: masked imaging modeling for enhancing cyclic immunofluorescence (CyCIF) with panel reduction and imputation.

Cyclic Immunofluorescence (CyCIF) can quantify multiple biomarkers, but panel capacity is limited by technical challenges. We propose a computational panel reduction approach that can impute the information content from 25 markers using only 9 markers, learning co-expression and morphological patterns while concurrently increasing speed and panel content and decreasing cost. We demonstrate strong correlations in predictions and generalizability across breast and colorectal cancer, illustrating applicability of our approach to diverse tissue types.

Graph Structured Neural Networks for Perturbation Biology.

Computational modeling of perturbation biology identifies relationships between molecular elements and cellular response, and an accurate understanding of these systems will support the full realization of precision medicine. Traditional deep learning, while often accurate in predicting response, is unlikely to capture the true sequence of involved molecular interactions. Our work is motivated by two assumptions: 1) Methods that encourage mechanistic prediction logic are likely to be more trustworthy, and 2) problem-specific algorithms are likely to outperform generic algorithms. We present an alternative to Graph Neural Networks (GNNs) termed Graph Structured Neural Networks (GSNN), which uses cell signaling knowledge, encoded as a graph data structure, to add inductive biases to deep learning. We apply our method to perturbation biology using the LINCS L1000 dataset and literature-curated molecular interactions. We demonstrate that GSNNs outperform baseline algorithms in several prediction tasks, including 1) perturbed expression, 2) cell viability of drug combinations, and 3) disease-specific drug prioritization. We also present a method called GSNNExplainer to explain GSNN predictions in a biologically interpretable form. This work has broad application in basic biological research and pre-clincal drug repurposing. Further refinement of these methods may produce trustworthy models of drug response suitable for use as clinical decision aids. Availability and implementation: Our implementation of the GSNN method is available at https://github.com/nathanieljevans/GSNN. All data used in this work is publicly available.

Transcriptional and phenotypic heterogeneity underpinning venetoclax resistance in AML.

The venetoclax BCL2 inhibitor in combination with hypomethylating agents represents a cornerstone of induction therapy for older AML patients, unfit for intensive chemotherapy. Like other targeted therapies, venetoclax-based therapies suffer from innate and acquired resistance. While several mechanisms of resistance have been identified, the heterogeneity of resistance mechanism across patient populations is poorly understood. Here we utilized integrative analysis of transcriptomic and ex-vivo drug response data in AML patients to identify four transcriptionally distinct VEN resistant clusters (VR_C1-4), with distinct phenotypic, genetic and drug response patterns. VR_C1 was characterized by enrichment for differentiated monocytic- and cDC-like blasts, transcriptional activation of PI3K-AKT-mTOR signaling axis, and energy metabolism pathways. They showed sensitivity to mTOR and CDK inhibition. VR_C2 was enriched for NRAS mutations and associated with distinctive transcriptional suppression of HOX expression. VR_C3 was characterized by enrichment for TP53 mutations and higher infiltration by cytotoxic T cells. This cluster showed transcriptional expression of erythroid markers, suggesting tumor cells mimicking erythroid differentiation, activation of JAK-STAT signaling, and sensitivity to JAK inhibition, which in a subset of cases synergized with venetoclax. VR_C4 shared transcriptional similarities with venetoclax-sensitive patients, with modest over-expression of interferon signaling. They were also characterized by high rates of DNMT3A mutations. Finally, we projected venetoclax-resistance states onto single cells profiled from a patient who relapsed under venetoclax therapy capturing multiple resistance states in the tumor and shifts in their abundance under venetoclax selection, suggesting that single tumors may consist of cells mimicking multiple VR_Cs contributing to intra-tumor heterogeneity. Taken together, our results provide a strategy to evaluate inter- and intra-tumor heterogeneity of venetoclax resistance mechanisms and provide insights into approaches to navigate further management of patients who failed therapy with BCL2 inhibitors.

Multiomic analysis of cervical squamous cell carcinoma identifies cellular ecosystems with biological and clinical relevance.

Cervical squamous cell carcinoma (CSCC) exhibits a limited response to immune-checkpoint blockade. Here we conducted a multiomic analysis encompassing single-cell RNA sequencing, spatial transcriptomics and spatial proteomics, combined with genetic and pharmacological perturbations to systematically develop a high-resolution and spatially resolved map of intratumoral expression heterogeneity in CSCC. Three tumor states (epithelial-cytokeratin, epithelial-immune (Epi-Imm) and epithelial senescence), recapitulating different stages of squamous differentiation, showed distinct tumor immune microenvironments. Bidirectional interactions between epithelial-cytokeratin malignant cells and immunosuppressive cancer-associated fibroblasts form an immune exclusionary microenvironment through transforming growth factor ß pathway signaling mediated by FABP5. In Epi-Imm tumors, malignant cells interact with natural killer and T cells through interferon signaling. Preliminary analysis of samples from a cervical cancer clinical trial ( NCT04516616 ) demonstrated neoadjuvant chemotherapy induces a state transition to Epi-Imm, which correlates with pathological complete remission following treatment with immune-checkpoint blockade. These findings deepen the understanding of cellular state diversity in CSCC.

Widespread BRCA1/2-independent homologous recombination defects are caused by alterations in RNA-binding proteins.

Defects in homologous recombination DNA repair (HRD) both predispose to cancer development and produce therapeutic vulnerabilities, making it critical to define the spectrum of genetic events that cause HRD. However, we found that mutations in BRCA1/2 and other canonical HR genes only identified 10%-20% of tumors that display genomic evidence of HRD. Using a networks-based approach, we discovered that over half of putative genes causing HRD originated outside of canonical DNA damage response genes, with a particular enrichment for RNA-binding protein (RBP)-encoding genes. These putative drivers of HRD were experimentally validated, cross-validated in an independent cohort, and enriched in cancer-associated genome-wide association study loci. Mechanistic studies indicate that some RBPs are recruited to sites of DNA damage to facilitate repair, whereas others control the expression of canonical HR genes. Overall, this study greatly expands the repertoire of known drivers of HRD, with implications for basic biology, genetic screening, and therapy stratification.

Accessible high-throughput single-cell whole-genome sequencing with paired chromatin accessibility.

Single-cell whole-genome sequencing (scWGS) enables the assessment of genome-level molecular differences between individual cells with particular relevance to genetically diverse systems like solid tumors. The application of scWGS was limited due to a dearth of accessible platforms capable of producing high-throughput profiles. We present a technique that leverages nucleosome disruption methodologies with the widely adopted 10× Genomics ATAC-seq workflow to produce scWGS profiles for high-throughput copy-number analysis without new equipment or custom reagents. We further demonstrate the use of commercially available indexed transposase complexes from ScaleBio for sample multiplexing, reducing the per-sample preparation costs. Finally, we demonstrate that sequential indexed tagmentation with an intervening nucleosome disruption step allows for the generation of both ATAC and WGS data from the same cell, producing comparable data to the unimodal assays. By exclusively utilizing accessible commercial reagents, we anticipate that these scWGS and scWGS+ATAC methods can be broadly adopted by the research community.

Efficacy of futibatinib, an irreversible fibroblast growth factor receptor inhibitor, in FGFR-altered breast cancer.

Several alterations in fibroblast growth factor receptor (FGFR) genes have been found in breast cancer; however, they have not been well characterized as therapeutic targets. Futibatinib (TAS-120; Taiho) is a novel, selective, pan-FGFR inhibitor that inhibits FGFR1-4 at nanomolar concentrations. We sought to determine futibatinib's efficacy in breast cancer models. Nine breast cancer patient-derived xenografts (PDXs) with various FGFR1-4 alterations and expression levels were treated with futibatinib. Antitumor efficacy was evaluated by change in tumor volume and time to tumor doubling. Alterations indicating sensitization to futibatinib in vivo were further characterized in vitro. FGFR gene expression between patient tumors and matching PDXs was significantly correlated; however, overall PDXs had higher FGFR3-4 expression. Futibatinib inhibited tumor growth in 3 of 9 PDXs, with tumor stabilization in an FGFR2-amplified model and prolonged regression (> 110 days) in an FGFR2 Y375C mutant/amplified model. FGFR2 overexpression and, to a greater extent, FGFR2 Y375C expression in MCF10A cells enhanced cell growth and sensitivity to futibatinib. Per institutional and public databases, FGFR2 mutations and amplifications had a population frequency of 1.1%-2.6% and 1.5%-2.5%, respectively, in breast cancer patients. FGFR2 alterations in breast cancer may represent infrequent but highly promising targets for futibatinib.

Common and distinct patterns of acquired uniparental disomy and homozygous deletions between lung squamous cell carcinomas and lung adenocarcinoma.

Acquired uniparental disomy (aUPD) is a chromosomal alteration that can lead to homozygosity of existing aberrations. We used data from The Cancer Genome Atlas SNP-based arrays to identify distinct and common aUPD profiles in lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC). Moreover, we tested relevance of aUPD for homozygous deletion (HMD), overall survival (OS), and recurrence-free survival (RFS). Overall, we found significantly higher aUPD (q = 5.34E-09) in LUSC than in LUAD. A significant portion of HMD was associated with aUPD in LUSC (24.9%) and LUAD (19.7%). We identified segmental, whole-chromosome arm and whole-chromosome aUPD, in which whole 7p arm aUPD was restricted to LUSC, while whole-chromosome 3 aUPD was observed only in LUAD, and whole-chromosome 21 aUPD was common to both LUSC and LUAD. The most frequent aUPD and HMD were observed at CDKN2A/B region in both LUAD and LUSC. In LUAD, aUPD and HMD at CDKN2A/B region were associated with shorter OS (q < 0.021 and q < 0.005), and RFS (q < 0.005 and q < 0.005), while heterozygous deletion was not associated with OS and RFS. In contrast, no association was found between aUPD at CDKN2A/B region and survival in LUSC. In LUAD, CTLA expression was significantly lower in samples with aUPD at CDKN2A/B regions than in samples without copy number and allele-based changes. Immune infiltration correlated with aUPD or HMD at CDKN2A/B, gain at HLA class I region, and aUPD at whole-chromosome q-arm or whole chromosome in LUAD, but not in LUSC. Both LUSC and LUAD have common and distinct patterns of aUPD regions with differing frequencies of occurrence and associations with outcome.

Dual blockade of BRD4 and ATR/WEE1 pathways exploits ARID1A loss in clear cell ovarian cancer.

ARID1A, an epigenetic tumor suppressor, is the most common gene mutation in clear-cell ovarian cancers (CCOCs). CCOCs are often resistant to standard chemotherapy and lack effective therapies. We hypothesized that ARID1A loss would increase CCOC cell dependency on chromatin remodeling and DNA repair pathways for survival. We demonstrate that combining BRD4 inhibitor (BRD4i) with DNA damage response inhibitors (ATR or WEE1 inhibitors; e.g. BRD4i-ATRi) was synergistic at low doses leading to decreased survival, and colony formation in CCOC in an ARID1A dependent manner. BRD4i-ATRi caused significant tumor regression and increased overall survival in ARID1AMUT but not ARID1AWT patient-derived xenografts. Combination BRD4i-ATRi significantly increased γH2AX, and decreased RAD51 foci and BRCA1 expression, suggesting decreased ability to repair DNA double-strand-breaks (DSBs) by homologous-recombination in ARID1AMUT cells, and these effects were greater than monotherapies. These studies demonstrate BRD4i-ATRi is an effective treatment strategy that capitalizes on synthetic lethality with ARID1A loss in CCOC.

The spectrum of MAPK-ERK pathway genomic alterations in gynecologic malignancies: Opportunities for novel therapeutic approaches.

OBJECTIVE: To investigate the incidence of MAPK/ERK pathway genomic alterations among patients with gynecologic malignancies. METHODS: We accessed the American Association of Cancer Research Genomics Evidence of Neoplasia Information Exchange publicly available dataset (v13.0). Patients with malignant tumors of the ovary, uterus, and cervix were identified. Following stratification by tumor site and histology, we examined the prevalence of MAPK/ERK pathway gene alterations (somatic mutation, and/or structural chromosome alterations). We included the following RAS-MAPK pathway genes known to be implicated in the dysregulation of the pathway; KRAS, NRAS, BRAF, HRAS, MAP2K1, RAF1, PTPN11, NF1, and ARAF. Data from the OncoKB database, as provided by cBioPortal, were utilized to determine pathogenic gene alterations. RESULTS: We identified a total of 10,233 patients with gynecologic malignancies; 48.2% (n = 4937) with ovarian, 45.2% (n = 4621) with uterine and 6.6% (n = 675) with cervical cancer respectively. The overall incidence of MAPK pathway gene alterations was 21%; the most commonly altered gene was KRAS (13%), followed by NF1 (7%), NRAS (1.3%), and BRAF (1.2%). The highest incidence was observed among patients with mucinous ovarian (71%), low-grade serous ovarian (48%), endometrioid ovarian (37%), and endometrioid endometrial carcinoma (34%). CONCLUSIONS: Approximately 1 in 5 patients with a gynecologic tumor harbor a MAPK/ERK pathway genomic alteration. Novel treatment strategies capitalizing on these alterations are warranted.

MIM-CyCIF: Masked Imaging Modeling for Enhancing Cyclic Immunofluorescence (CyCIF) with Panel Reduction and Imputation.

CyCIF quantifies multiple biomarkers, but panel capacity is compromised by technical challenges including tissue loss. We propose a computational panel reduction, inferring surrogate CyCIF data from a subset of biomarkers. Our model reconstructs the information content from 25 markers using only 9 markers, learning co-expression and morphological patterns. We demonstrate strong correlations in predictions and generalizability across breast and colorectal cancer tissue microarrays, illustrating broader applicability to diverse tissue types.

Identifying homologous recombination deficiency in breast cancer: genomic instability score distributions differ among breast cancer subtypes.

PURPOSE: A 3-biomarker homologous recombination deficiency (HRD) score is a key component of a currently FDA-approved companion diagnostic assay to identify HRD in patients with ovarian cancer using a threshold score of ≥ 42, though recent studies have explored the utility of a lower threshold (GIS ≥ 33). The present study evaluated whether the ovarian cancer thresholds may also be appropriate for major breast cancer subtypes by comparing the genomic instability score (GIS) distributions of BRCA1/2-deficient estrogen receptor-positive breast cancer (ER + BC) and triple-negative breast cancer (TNBC) to the GIS distribution of BRCA1/2-deficient ovarian cancer. METHODS: Ovarian cancer and breast cancer (ER + BC and TNBC) tumors from ten study cohorts were sequenced to identify pathogenic BRCA1/2 mutations, and GIS was calculated using a previously described algorithm. Pathologic complete response (pCR) to platinum therapy was evaluated in a subset of TNBC samples. For TNBC, a threshold was set and threshold validity was assessed relative to clinical outcomes. RESULTS: A total of 560 ovarian cancer, 805 ER + BC, and 443 TNBC tumors were included. Compared to ovarian cancer, the GIS distribution of BRCA1/2-deficient samples was shifted lower for ER + BC (p = 0.015), but not TNBC (p = 0.35). In the subset of TNBC samples, univariable logistic regression models revealed that GIS status using thresholds of ≥ 42 and ≥ 33 were significant predictors of response to platinum therapy. CONCLUSIONS: This study demonstrated that the GIS thresholds used for ovarian cancer may also be appropriate for TNBC, but not ER + BC. GIS thresholds in TNBC were validated using clinical response data to platinum therapy.