Nature Cancer

Somatic mutations inactivating TET2 are among the most common drivers of clonal hematopoiesis (CH). While TET2 inactivation is associated with monocyte-derived inflammation and improved chimeric antigen-receptor-T cell function, its impact on immunotherapy response is unknown. In our mouse model, hematopoietic Tet2 mutation enhanced immune checkpoint blockade (ICB) response. Enhanced ICB response with Tet2 mutation required phagocytes, CD4 and CD8 T cells. Mechanistically, in Tet2-mutant tumor-infiltrating leukocytes (TILs), ICB preferentially induced anti-tumor states and restricted cell states linked to tumor progression. Tet2-mutant monocytes activated costimulatory programs, while Tet2-mutant T cells showed enhanced T cell memory signatures, lesser exhaustion and decreased regulatory phenotype. Our murine data was clinically relevant, since we found that melanomas from patients with TET2 driver mutation-CH (TET2-CH) showed enhanced immune infiltration, T cell activation, and T cell memory programs. In melanoma patients treated with ICB, TET2-CH was associated with 6-fold greater odds of clinical benefit. Collectively, our data establishes that hematopoietic Tet2 inactivation primes leukocytes for anti-tumor states associated with immunotherapy response and provides a potential biomarker for personalized therapy.

BackgroundProcarbazine-containing chemotherapy regimens associate with cytopenias and infertility, suggesting stem cell toxicity. Procarbazine in eBEACOPP (escalated dose bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, prednisolone) is increasingly replaced with dacarbazine (eBEACOPDac) to reduce toxicity, although limited genomic and clinical data support this substitution. MethodsTo assess mutagenic and clinical consequences of dacarbazine-procarbazine substitutions, we compared mutational landscapes in haematopoietic stem and progenitor cells (HSPCs) from patients treated with different Hodgkin regimens and children, sperm and bowel tissue from procarbazine-treated patients. We compared efficacy and toxicity data of a multicentre eBEACOPDac-treated patient cohort, with eBEACOPP clinical trial and real-world datasets. ResultseBEACOPP-treated patients exhibit a higher burden of point mutations, small insertions and deletions in HSPCs compared to eBEACOPDac and ABVD (doxorubicin, bleomycin, vinblastine, dacarbazine)-treated patients. Two novel mutational signatures, SBSA (SBS25-like) and SBSB were identified in HSPCs, neoplastic and normal colon from only procarbazine-treated patients. SBSB was also identified in germline DNA of three children conceived post-eBEACOPP and sperm of an eBEACOPP-treated male. The dacarbazine substitution did not appear to compromise efficacy; 3-year progression-free survival of 312 eBEACOPDac patients (93.3%; CI95=90.3-96.4%) mirrored that of 1945 HD18-trial eBEACOPP patients (93.3%; CI95=92.1-94.4%). eBEACOPDac-treated patients required fewer blood transfusions, demonstrated higher post-chemotherapy sperm concentrations, and experienced earlier resumption of menstrual periods. ConclusionsProcarbazine induces a higher mutational burden and novel mutational signatures in eBEACOPP-treated patients and their germline DNA raising concerns for hereditary consequences. However, replacing procarbazine with dacarbazine appears to mitigate gonadal and stem cell toxicity while maintaining comparable clinical efficacy.

Precision medicine holds great promise to improve outcomes in cancer, including haematological malignancies. However, there are few biomarkers that influence choice of chemotherapy in clinical practice. In particular, multiple myeloma requires an individualized approach as there exist several active therapies, but little agreement on how and when they should be used and combined. We have previously shown that a transcriptomic signature can identify specific bortezomib- and lenalidomide-sensitivity. However, gene expression signatures are challenging to implement clinically. We reasoned that signatures based on the presence or absence of gene mutations would be more tractable in the clinical setting, though examples of such signatures are rare. We performed whole exome sequencing as part of the CARDAMON trial, which employed carfilzomib-based therapy. We applied advanced machine learning approaches to discover mutational patterns predictive of treatment outcome. The resulting model accurately predicted progression-free survival (PFS) both in CARDAMON patients and in an external validation set of patients from the CoMMpass study who had received carfilzomib. The signature was specific for carfilzomib therapy and was strongly driven by genes on chromosome 1p36. Importantly, patients predicted to be carfilzomib-sensitive had a longer PFS when treated with carfilzomib/lenalidomide/dexamethasone than with bortezomib/carfilzomib/dexamethasone. However, in those predicted to be carfilzomib-insensitive, the latter therapy may have been capable of eradicating carfilzomib-resistant clones. We propose that the signature can be used to make rational therapeutic decisions and could be incorporated into future clinical trials.

Precision oncology matches tumors to targeted therapies based on the presence of actionable molecular alterations. However, most tumors lack actionable alterations, restricting treatment options to cytotoxic chemotherapies for which few data-driven prioritization strategies currently exist. Here, we report an integrated computational/experimental treatment selection approach applicable for both chemotherapies and targeted agents irrespective of actionable alterations. We generated functional drug response data on a large collection of patient-derived tumor models and used it to train ScreenDL, a novel deep learning-based cancer drug response prediction model. ScreenDL leverages the combination of tumor omic and functional drug screening data to predict the most efficacious treatments. We show that ScreenDL accurately predicts response to drugs with diverse mechanisms, outperforming existing methods and approved biomarkers. In our preclinical study, this approach achieved superior clinical benefit and objective response rates in breast cancer patient-derived xenografts, suggesting that testing ScreenDL in clinical trials may be warranted.

Radiotherapy is a pillar of cancer care and augments the response to immunotherapies. However, little is known regarding the relationships between the tumor immune ecosystem (TIES) and intrinsic radiosensitivity, and a pressing question in oncology is how to optimize radiotherapy to improve patient responses to immune therapies. To address this challenge, we profiled over 10,000 primary tumors for their metrics of radiosensitivity and immune cell infiltrate (ICI), and applied a new integrated in silico model that mimics the dynamic relationships between tumor growth, ICI flux and the response to radiation. We then validated this model with a separate cohort of 59 lung cancer patients treated with radiotherapy. These analyses explain radiation response based on its effect on the TIES and quantifies the likelihood that radiation can promote a shift to anti-tumor immunity. Dynamic modeling of the relationship between tumor radiosensitivity and the TIES may provide opportunity to personalize combined radiation and immunotherapy approaches.

Ionizing radiotherapy (RT) is a widely used palliative and curative treatment strategy for malignancies. In solid tumors, RT-induced double strand breaks lead to the accumulation of indels, and their repair by non-homologous end-joining has been linked to the ID8 mutational signature in resistant cells. However, the extent of RT-induced DNA damage in hematologic malignancies and its impact on their evolution and interplay with commonly used chemotherapies has not yet been explored. Here, we interrogated 580 whole genome sequencing (WGS) from patients with large B-cell lymphoma, multiple myeloma, and myeloid neoplasms and identified ID8 only in relapsed disease. Yet, it was detected after exposure to both RT and mutagenic chemotherapy (i.e., platinum). Using WGS of single-cell colonies derived from treated lymphoma cells, we revealed a dose-response relationship between RT and platinum and ID8. Finally, using ID8 as a genomic barcode we demonstrate that a single RT-resistant cell may seed systemic relapse.

Immune checkpoint blockade (ICB) has revolutionised cancer therapy, yet resistance -- both primary and acquired -- remains a significant obstacle, affecting the majority of patients. Here, we leveraged a large-scale, real-world clinicogenomic dataset to systematically explore the molecular underpinnings of ICB resistance in the post-progression setting. Analysing over 5,000 pan-cancer patients with clinical and pre-/post-treatment genomic and transcriptomic data, we identify distinct immunogenomic drivers of acquired vs. primary ICB resistance. Post-ICB progression, acquired resistance showed extended survival compared to primary resistance across all cancer types. The acquired resistance clinical phenotype was paralleled by a universally immune-inflamed, albeit dysfunctional, tumour microenvironment (TME) at the onset of acquired resistance, with sustained or ICB-induced inflammatory and interferon responses. We confirm previously described mechanisms of acquired resistance, including B2M loss-of-function (LoF) in non-small cell lung cancer (NSCLC), and identify novel potential mediators, including LoF of TGFBR2 in NSCLC, CYLD in head and neck cancer, and RUNX1 in triple negative breast cancer. Further supporting their involvement in resistance, these acquired ICB alterations associated with immune-escaped TMEs, characterised by active immunomodulatory oncogenic signalling, hyperproliferation and invasiveness, or altered tumour metabolism. These findings emphasise the heterogeneity of molecular drivers of acquired resistance to ICB within and across cancers, and highlight the potential for personalised therapeutic interventions post-progression to improve patient outcomes.

Precision oncology has made significant advances in the last few years, mainly by targeting actionable mutations in cancer driver genes. However, the proportion of patients whose tumors can be targeted therapeutically remains limited. Recent studies have begun to explore the benefit of analyzing tumor transcriptomics data to guide patient treatment, raising the need for new approaches for systematically accomplishing that. Here we show that computationally derived genetic interactions can successfully predict patient response. Assembling a broad repertoire of 32 datasets spanning more than 1,500 patients and including both tumor transcriptomics and response data, we predicted the response in 17 out of 21 targeted and 8 out of 11 checkpoint therapy datasets across 8 different cancer types with considerable accuracy, without ever training on these datasets. Analyzing the recently published multi-arm WINTHER trial, we show that the fraction of patients benefitting from transcriptomic-based treatments could potentially be markedly increased from 15% to about 85% by targeting synthetic lethal vulnerabilities in their tumors. In summary, this is the first computational approach to obtain considerable predictive performance across many different targeted and immunotherapy datasets, providing a promising new way for guiding cancer treatment based on the tumor transcriptomics of cancer patients.

Precision oncology aims to guide treatment decisions using biomarkers. While DNA-based panels are increasingly applied, RNA transcriptomics remain underused due to limited datasets and the absence of robust models. We assembled the largest transcriptomic resource for drug response prediction to date, spanning 69 cohorts, 3,729 patients, nine cancer types, and six frontline therapies: anti-PD-1/PD-L1 immune-checkpoint inhibitors, trastuzumab, bevacizumab, BRAF inhibitors, paclitaxel, and FAC/FEC (Fluorouracil-Adriamycin-Cyclophosphamide/Fluorouracil-Epirubicin-Cyclophosphamide) chemotherapy. We developed EXPRESSO (EXpression-Profile-RESponSe-Optimizer), a supervised machine-learning framework that predicts treatment response from pre-treatment transcriptomes by integrating drug targets and context-specific biomarkers. EXPRESSO achieves ROC-AUCs of 0.64-0.73 and odds ratios of 2.4-4.6 across therapies, outperforming 20 published transcriptomic signatures. Robustness analysis reveals that predictive performance plateaued for some therapies with increasing training cohorts but continued to improve for others. These findings suggest inherent limits of supervised brute-force learning for certain treatments, but additional data and deeper mechanistic modeling may further enhance transcriptomics-based predictors.

Pediatric solid tumors arise from diverse tissues during development and exhibit a wide range of molecular, cellular and genetic features. This diversity, combined with the low incidence of pediatric cancer makes it increasingly difficult to personalize therapy for individual patients based on the unique features of their tumors. Therefore, well-credentialed preclinical models that capture the diversity and heterogeneity of pediatric solid tumors are essential for identifying molecular targeted therapeutics for precision medicine. Here, we report 281 orthotopic patient derived xenografts (O-PDXs) from 224 patients representing 24 different types of pediatric solid tumors. We have performed genomic characterization of the O-PDXs and compared them to their corresponding patient tumors. To demonstrate the feasibility and utility of using such a diverse collection of O-PDXs in preclinical studies, we performed a preclinical pediatric precision medicine trial based on the NCI-COG Pediatric MATCH trial enrollment criteria. We also tested molecular targeted therapy for a novel oncogenic fusion recently reported in pediatric melanoma and precision drug delivery using nano-liposomal irinotecan. Our studies demonstrate the value of large, well-credentialed preclinical models for future precision medicine in pediatric oncology using single agents, drug combinations and novel drug formulations. Translational RelevanceThis study demonstrates the value of utilizing fully characterized preclinical models of pediatric solid tumors to evaluate the response to precision medicine approaches. Our results demonstrate the importance of performing comprehensive preclinical testing using multiple orthotopic patient derived xenografts to validate and prioritize vulnerabilities identified through genomic or integrated analyses which can be translated into clinical trials. Importantly, this study identified combinations using nano-liposomal irinotecan in a precision drug delivery approach that may benefit pediatric solid tumor patients. In addition, all models and their associated data are made freely available to the scientific community through the Childhood Solid Tumor Network.

Immune Checkpoint Blockade (ICB) has revolutionized cancer treatment, however mechanisms determining patient response remain poorly understood. Here we used machine learning to predict ICB response from germline and somatic biomarkers and interpreted the learned model to uncover putative mechanisms driving superior outcomes. Patients with higher T follicular helper infiltrates were robust to defects in the class-I Major Histocompatibility Complex (MHC-I). Further investigation uncovered different ICB responses in MHC-I versus MHC-II neoantigen reliant tumors across patients. Despite similar response rates, MHC-II reliant responses were associated with significantly longer durable clinical benefit (Discovery: Median OS=63.6 vs. 34.5 months P=0.0074; Validation: Median OS=37.5 vs. 33.1 months, P=0.040). Characteristics of the tumor immune microenvironment reflected MHC neoantigen reliance, and analysis of immune checkpoints revealed LAG3 as a potential target in MHC-II but not MHC-I reliant responses. This study highlights the value of interpretable machine learning models in elucidating the biological basis of therapy responses. Statement of SignificanceImmune checkpoint blockade works only in a fraction of patients for reasons that are still not fully understood. Our study reveals heterogeneity in the immune responses of ICB responders that correlates with characteristics of the neoantigen landscape. This heterogeneity is accompanied by differences in the duration of clinical benefit as well as by differences as to which immune checkpoint gene serves as a biomarker of ICB response. These findings suggest possible new strategies for improving ICB responses. HighlightsO_LIWe used machine learning to study ICB response across 708 patients from 8 studies across 3 tumor types (melanoma, RCC, and NSCLC). C_LIO_LICombining germline and somatic features improves prediction of ICB response C_LIO_LIInteractions between germline and somatic features reveal mechanisms contributing to ICB sensitivity. C_LIO_LIMHC-I vs. MHC-II reliance implicates LAG3 as a prognostic biomarker in the context of CD4 T cell driven responses. C_LIO_LIMHC-II neoantigen reliant responses provide superior durable clinical benefit in response to ICB. C_LI

BRAFV600E mutation confers a poor prognosis in metastatic colorectal cancer (CRC) despite combinatorial targeted therapies based on the latest understanding of signaling circuitry. To identify parallel resistance mechanisms induced by BRAF/MEK/EGFR co-targeting, we used a high throughput kinase activity mapping platform. We found that SRC kinases are systematically activated in BRAFV600E CRC following targeted inhibition of BRAF {+/-} EGFR, and that coordinated targeting of SRC with BRAF {+/-} EGFR increases efficacy in vitro and in vivo. SRC drives resistance to BRAF {+/-} anti-EGFR therapy independently of ERK signaling by inducing transcriptional reprogramming via beta-catenin (CTNNB1). The EGFR-independent compensatory activation of SRC kinases is mediated by an autocrine prostaglandin E2-loop that can be blocked with cyclooxygenase-2 (COX2) inhibitors. Co-targeting of COX2 with BRAF+EGFR promotes durable suppression of tumor growth in patient-derived tumor xenograft (PDX) models. COX2 inhibition represents a novel drug-repurposing strategy to overcome therapeutic resistance in BRAFV600E CRC.

Adding anti-PD1 antibodies to 5-FU/platinum chemotherapy improves survival in a subset of advanced gastroesophageal adenocarcinoma (GEA) patients. Beyond PD-L1 expression and mismatch repair status we have limited insight into molecular predictors of response or the relative contribution of PD-1 blockade. We conducted an investigator sponsored phase II trial (n = 47) sequentially adding pembrolizumab to standard 5-FU/platinum in previously untreated advanced GEA (ClinicalTrials.gov: NCT04249739). With an overall response rate of 67% the activity paralleled phase III chemoimmunotherapy trials. To understand on-treatment tumor and immune adaptations patients underwent serial biopsy of the primary tumor, including baseline, after one cycle of 5-FU/platinum, and after the addition of pembrolizumab. We leveraged transcriptional profiling from 358,067 cells to identify multicellular networks of malignant, stromal, and immune cells after chemotherapy and concurrent chemoimmunotherapy. The relative usage of pro-tumor and anti-tumor interaction hubs differed between fast and slow progressing patients. Chemotherapy induced early on-treatment formation of hubs centered on tumor-reactive T-cell and M1-oriented macrophage interactions with pro-inflammatory cytokines in slow progressors. Faster progression was characterized by increased MUC5A and MSLN containing programs in tumor cells and M2-oriented macrophages with immunosuppressive stromal interactions. After adding pembrolizumab we observed increased CD8 T-cell infiltration by scRNAseq and multiplex immunofluorescence and development of an immunity hub involving co-variation of the tumor-reactive CXCL13 program and epithelial interferon-stimulated gene programs enriched in slow progressors. Together this data provides prospective evidence of differential early on-treatment evolution of the gastric immune microenvironment and nominates candidate cellular interactions for clinical targeting.

The costs of cancer therapies are rising rapidly across the globe, with novel therapies like targeted treatment and immunotherapies as major contributors, but their effectiveness can be low or uncertain due to limited post market surveillance. Reliable biomarkers to identify patients highly unlikely to respond to cancer therapies represent an increasingly important clinical and societal need, as they could prevent unnecessary treatments, reduce side effects, and alleviate pressure on healthcare systems. Here, we developed a robust statistical framework and applied it to whole-genome and transcriptome sequencing data of cancer patients (n = 2,596) with advanced disease. Our approach systematically identified known and potentially novel genomic and transcriptomic biomarkers of non-response, such as immune evasion driver events in skin melanoma patients treated with anti-PD-1 checkpoint inhibitors, and KRASG12 mutations in metastatic colorectal cancer patients treated with different chemotherapy regimens. Despite the identification of these promising non-response signals, an analytical power analysis revealed that for most treatments and/or cancer types the cohort sizes are underpowered. Our results underscore the promises and the urgent need for expanding response-annotated real-world comprehensive genomics datasets to enable robust biomarker identification and validation.

Somatic mutation profiling is central to cancer diagnosis and treatment selection. However, most studies focus on individual actionable mutations, overlooking the broader mutational context that shapes tumor evolution and treatment response. Here, we introduce OncoBERT, a language model that learns contextual representations of somatic mutations from large-scale clinical sequencing data spanning >210,000 patients, 113 cancer types and 20 institutions. OncoBERT uncovers robust patient-specific mutational subtypes across diverse cohorts and targeted sequencing panels, revealing clinically meaningful mutation patterns that are associated with differential response to chemotherapy, targeted therapies, and immunotherapy. Importantly, integrating OncoBERTs contextual representations with clinically approved biomarkers of immunotherapy response, such as tumor mutational burden (TMB) and microsatellite instability (MSI), significantly improved prediction of clinical benefit. By further incorporating matched tumor transcriptomic profiles, we linked OncoBERT-defined mutational subtypes to distinct cancer hallmark programs and tumor microenvironment states. Together, OncoBERT provides a scalable framework for deciphering somatic mutational landscapes, enabling improved patient stratification and advancing precision oncology.

The histopathological phenotype of tumors reflects the underlying genetic makeup. Deep learning can predict genetic alterations from tissue morphology, but it is unclear how well these predictions generalize to external datasets. Here, we present a deep learning pipeline based on self-supervised feature extraction which achieves a robust predictability of genetic alterations in two large multicentric datasets of seven tumor types.

Metastasis drives mortality and morbidity in cancer. While some patients develop broad metastatic disease across multiple organs, others exhibit organ-specific spread. To identify mechanisms underlying metastatic organotropism, we analyzed clinico-genomic data from over 7,000 patients with metastatic cutaneous melanoma in three independent cohorts (one primary discovery and two validation cohorts including a nationwide electronic health record-derived deidentified database), leveraging machine learning approaches to clinical data. We found that female sex and increased tumor mutational burden associate with decreased metastatic potential, while older age associates with increased lung and adrenal metastases. Using unsupervised analyses, patients clustered into five metastatic patterns: a "highly metastatic" cluster characterized by involvement of many organs, a "low metastatic" cluster characterized by few metastatic sites (mostly lymph node metastases), and three additional clusters each characterized by metastasis to specific sites (brain, lung, liver). Mutations in B2M and PTEN associated with increased overall metastatic potential. PTEN mutations were also associated with brain metastases but were enriched only in the "highly metastatic" cluster and not the brain-specific cluster. Mutations in GNAQ or GNA11 (GNA) associated with increased liver metastasis. To validate this association, we tested and demonstrated liver tropism in two GNA-mutant genetically engineered cutaneous melanoma mouse models of metastasis. Overall, our study elucidates distinct phenotypes of metastasis in patients with melanoma and identifies novel clinical and genomic associations that illuminate the drivers of clinical metastatic organotropism.

MYC-driven medulloblastomas (MBs) represent the most aggressive and deadly subgroup of MB, the most common malignant pediatric brain tumor. Direct targeting of MYC itself remains an unmet clinical need, therefore focusing on vulnerabilities driven by MYC may be a viable option for novel therapeutic interventions. Using whole-genome CRISPR screening, we identified the de novo pyrimidine synthesis enzyme CTP synthase (CTPS1) as a strong dependency in MYC-driven MB. CTPS1 is the final and rate-limiting step in the de novo pyrimidine synthesis pathway. Targeted inhibition of CTPS1 leads to decreased tumor cell proliferation and markedly reduces MYC expression in G3 MB models. Mechanistically, we demonstrate that single agent CTPS1 inhibition activates the replication stress signaling pathway mediated by ATM-CHK2 and ATR-CHK1. Blockade of CHK1 kinase activity increases sensitivity to CTPS1 inhibition and significantly impedes heterotopic MB tumor growth. CTPS1 enzymatic activity requires the amino acid glutamine, therefore we inhibited CTPS1 using the glutamine antagonists, JHU083 and JHU395. These compounds are prodrugs of 6-diazo-5-oxo-L-norleucine (DON) which were developed to exhibit better tumor targeting and enhanced blood-brain barrier penetrability. Combining JHU083 and CHK1 inhibition demonstrates potent synergy against patient-derived MB xenografts in vivo. Our findings strongly suggest that combining de novo pyrimidine synthesis and ATR-CHK1 inhibitors is a promising treatment for MYC-driven MBs. Key PointsO_LICTPS1 is a unique vulnerability in MYC-driven medulloblastoma C_LIO_LICTPS1 inhibition activates the ATR-CHK1 replication stress response pathway for cell survival C_LIO_LIBlockade of CTPS1 enzymatic activity synergizes with CHK1 inhibition in vitro and in vivo C_LI Importance of the StudyMYC hyperactivation in tumors drives multiple anabolic processes which contribute to tumor proliferation and aggressiveness in patients. We show that targeting de novo pyrimidine synthesis (via CTPS1) limits tumor growth and targets MYC itself through a feedback mechanism. CTPS1 inhibition potently combines with CHK1 blockade and enhances disease control in both heterotopic and orthotopic models of medulloblastoma (MB). Our results support the clinical evaluation of combined CTPS1 and CHK1 inhibition in patients with MYC-driven MB.

Anti-PD-1/PD-L1 agents have transformed the treatment landscape of advanced non-small cell lung cancer (NSCLC). While our understanding of the biology underlying immune checkpoint blockade in NSCLC is still incomplete, studies to date have established predictive roles for PD-L1 tumor expression and tumor mutational burden (TMB). To expand our understanding of the molecular features underlying response to checkpoint inhibitors in NSCLC, we describe here the first joint analysis of the Stand Up 2 Cancer - Mark Foundation (SU2C-MARK) Cohort, a resource of whole exome and/or RNA sequencing from 393 patients with NSCLC treated with anti-PD-(L)1 therapy, along with matched clinical response annotation. We identify a number of associations between molecular features and outcome, including: 1) favorable (e.g., ATM altered), and unfavorable (e.g., TERT amplified) genomic subgroups, 2) distinct immune infiltration signatures associated with wound healing (unfavorable) and immune activation (favorable), and 3) a novel de-differentiated tumor-intrinsic subtype characterized by expression of endodermal lineage genes, immune activation, and enhanced response rate. Taken together, results from this cohort extend our understanding of NSCLC-specific predictors, providing a rich set of molecular and immunologic hypotheses with which to further our understanding of the biology of checkpoint blockade in NSCLC.

The co-occurrence of germline and somatic oncogenic alterations is frequently observed in breast cancer, but their combined biologic and clinical significance has not been evaluated. To assess the role of germline-somatic interactions on outcomes in routine practice, we developed an integrated clinicogenomic pipeline to analyze the genomes of over 4,500 patients with breast cancer. We find that germline (g)BRCA2-associated tumors are enriched for RB1 loss-of-function mutations and manifest poor outcomes on standard-of-care, front-line CDK4/6 inhibitor (CDK4/6i) combinations. Amongst these tumors, gBRCA2-related homologous recombination deficiency (HRD) as well as baseline RB1 LOH status promote acquisition of RB1 loss-of- function mutations under the selective pressure of CDK4/6i, causing therapy resistance. These findings suggest an alternative therapeutic strategy using sequential targeting of HRD in gBRCA- associated breast cancers through PARP inhibitors prior to CDK4/6i therapy to intercept deleterious RB1-loss trajectories and thus suppress the emergence of CDK4/6 inhibitor resistance. More broadly, our findings demonstrate how germline-somatic driven genomic configurations shape response to systemic therapy and can be exploited therapeutically as part of biomarker-directed clinical strategies.