Nature Cancer

Normal and oncogenic Ras proteins are functionally dependent on one or more lipid modifications1,2. Whereas K-Ras4b farnesylation is sufficient for stable association with the plasma membrane, farnesylated H-Ras, K-Ras4a, and N-Ras traffic to the Golgi where they must undergo palmitoylation before regulated translocation to cell membranes. N-Ras palmitoylation by the DHHC family of palmitoyl acyl transferases (PATs) and depalmitoylation by ABHD17 serine hydrolases is a dynamic process that is essential for the growth of acute myeloid leukemias (AMLs) harboring oncogenic NRAS mutations3-6. Here, we have tested whether co-targeting ABHD17 enzymes and Ras signal output would cooperatively inhibit the proliferation and survival of NRAS-mutant AMLs while sparing normal tissues that retain K-Ras4b function. We show that ABD778, a potent and selective ABHD17 inhibitor with in vivo activity, selectively reduces the growth of NRAS-mutant AML cells in vitro and is synergistic with the allosteric MEK inhibitor PD0325901 (PD901)7,8. Similarly, ABD778 and PD901 significantly extended the survival of recipient mice transplanted with three independent primary mouse AMLs harboring an oncogenic NrasG12Ddriver mutation. Resistant leukemias that emerged during continuous drug treatment acquired by-pass mutations that confer adaptive drug resistance and increase mitogen activated protein kinase (MAPK) signal output. ABD778 augmented the anti-leukemia activity of the pan-PI3 kinase inhibitor pictilisib9, the K/N-RasG12C inhibitor sotorasib10, and the FLT3 inhibitor gilteritinib11. Co-treatment with ABD778 and gilteritinib restored drug sensitivity in a patient-derived xenograft model of adaptive resistance to FLT3 inhibition. These data validate the palmitoylation cycle as a promising therapeutic target in AML and support exploring it in other NRAS-mutant cancers.

Chimeric antigen receptor (CAR)-T cell therapies demonstrate potent anti-tumor efficacy in hematologic malignancies, yet clinical outcomes remain unpredictable due to the bespoke nature of the treatment, which is manufactured from each patients own T-cells. While germline variants are known to influence response to immune checkpoint inhibitors1, 2, their role in CAR-T cell therapy is unknown. Here, we pair whole-genome germline sequencing of lymphoma patients from the ZUMA-13 and ZUMA-74 clinical trials of axicabtagene ciloleucel, along with correlative biomarkers and functional assays, to ascertain the impact of germline variants on CAR-T cell behavior. Hypothesizing shared mechanisms of the most common toxicities of CAR-T cells, namely cytokine release syndrome (CRS) with hemophagocytic lymphohistiocytosis (HLH)--a hyperinflammatory syndrome driven by T cell overactivation5, we first looked within 17 canonical HLH-associated genes6, and identified putative deleterious STXBP2 variants in 15% of ZUMA-1 patients with toxicity, which were absent in control subjects who did not experience high grade toxicity. Subjects with these variants had elevated baseline IFN-{gamma} and inflammatory cytokines, findings that were recapitulated in engineered STXBP2-deficient and STXBP2-variant-expressing primary CAR-T cells derived from healthy donors. However, STXBP2 variant enrichment was absent in ZUMA-7 for this toxicity phenotype, possibly reflecting differences in underlying disease burden and evolving clinical management between the trials. A more expansive genome-wide analysis revealed ADAMTSL3 (a negative regulator of TGF{beta}7) as the only gene nominally enriched for putative deleterious variants in both ZUMA-1 and ZUMA-7 among control subjects, suggesting a protective effect. Finally, we focused on associations between germline variants and CAR-T cell expansion after infusion, a more objective and granular continuous variable that is strongly associated with clinical response across most CAR-T products8. We found a strong association between PTPN22, a known negative regulator of T-cell activation 9-12 and an autoimmune risk gene13, 14, variant status and CAR-T cell expansion in both ZUMA-1 and ZUMA-7, with the patients having the highest level of CAR-T expansion across clinical trials harboring variants in the gene. Together, these findings demonstrate the first clear association between germline variants and the clinical behavior of engineered immune cell therapies, which has implications for cellular therapy design, monitoring, testing, clinical trial design, and patient care.

A central problem in cancer immunotherapy with immune checkpoint blockade (ICB) is the development of resistance, which affects 50% of patients with metastatic melanoma1,2. T cell exhaustion, resulting from chronic antigen exposure in the tumour microenvironment, is a major driver of ICB resistance3. Here, we show that CD38, an ecto-enzyme involved in nicotinamide adenine dinucleotide (NAD+) catabolism, is highly expressed in exhausted CD8+ T cells in melanoma and is associated with ICB resistance. Tumour-derived CD38hiCD8+ T cells are dysfunctional, characterised by impaired proliferative capacity, effector function, and dysregulated mitochondrial bioenergetics. Genetic and pharmacological blockade of CD38 in murine and patient-derived organotypic tumour models (MDOTS/PDOTS) enhanced tumour immunity and overcame ICB resistance. Mechanistically, disrupting CD38 activity in T cells restored cellular NAD+ pools, improved mitochondrial function, increased proliferation, augmented effector function, and restored ICB sensitivity. Taken together, these data demonstrate a role for the CD38-NAD+ axis in promoting T cell exhaustion and ICB resistance, and establish the efficacy of CD38 directed therapeutic strategies to overcome ICB resistance using clinically relevant, patient-derived 3D tumour models.

The dynamic evolution of the immune tumor microenvironment (TME) during targeted therapy is a critical yet poorly understood determinant of treatment response and resistance. While most studies compare immune states before and after treatment, temporal immune changes during therapy remain largely uncharacterized, limiting development of effective combination strategies. Here, we investigated immune dynamics throughout targeted therapy using mouse melanoma models that recapitulate human therapeutic responses. Single-cell RNA sequencing (scRNA-seq) identified a previously unrecognized inflection point where the inflamed TME during tumor regression, characterized by robust NK cell infiltration, transitions to an immune-excluded state upon onset of drug-tolerant residual disease. We uncovered a unique macrophage subset (F4/80hiCCL5MHCIICD63) that orchestrates NK cell recruitment through CCR2/5 signaling during regression. Depletion of these macrophages using LysM-cre;iDTR mice significantly reduced NK cell infiltration. Specifically during residual disease, pharmacological inhibition of Ptpn22, a phosphatase that negatively regulates immune activation, reprogrammed macrophages, restored NK cell recruitment and enhanced therapeutic efficacy. Extending these findings to human cancer, longitudinal scRNA-seq analysis of melanoma and lung cancer patient samples revealed dynamic NK cell infiltration during targeted therapy, establishing a direct link between innate immune remodeling and treatment outcome. Unlike prior prognostic studies assessing immune states at single time points, our results provide mechanistic evidence of a temporal relationship between NK cell infiltration and therapeutic efficacy. Together, these findings position immune evolution as a driver of acquired resistance and identify macrophage-NK cell crosstalk as a therapeutically actionable axis to overcome immune exclusion and improve targeted therapy across multiple cancer types. One Sentence SummaryCancer therapy resistance emerges from a dynamic evolution of the tumor microenvironment, characterized by macrophage-driven NK cell infiltration during initial tumor regression, followed by exclusion of NK cells during residual disease, highlighting macrophage-NK cell interactions as a promising therapeutic target to improve clinical outcomes.

Peritoneal carcinomatosis is a common yet deadly manifestation of gastrointestinal cancers, with few effective treatments. To identify targetable determinants of peritoneal metastasis, we focused on appendiceal adenocarcinoma (AC), a gastrointestinal cancer that metastasizes almost exclusively to the peritoneum. Current treatments are extrapolated from colorectal cancer (CRC), yet AC has distinct genomic alterations, mucinous morphology and peritoneum restricted metastatic pattern. Further, no stable preclinical models of AC exist, limiting drug discovery and representing an unmet clinical need. We establish a first-in-class stable biobank of 16 long-term cultured AC patient-derived organoids (PDOs), including 3 matched, simultaneously resected primary AC-peritoneal carcinomatosis (AC-PC) pairs. By enriching for cancer cells, AC PDOs enable accurate genomic characterization relative to paucicellular AC tissue. We establish an organoid orthotopic intraperitoneal xenograft model that recapitulates diffuse peritoneal carcinomatosis and show that PC-organoids retain increased metastatic capacity, decreased growth factor dependency and sensitivity to standard of care chemotherapy relative to matched primary AC organoids. Single cell profiling of AC-PC pairs reveals dedifferentiation from mucinous differentiated states in primary AC into intestinal stem cell and fetal progenitor states in AC-PC, with upregulation of oncogenic signaling pathways. Through hypothesis-driven drug testing, we identify KRASMULTI-ON inhibitor RMC-7977 and Wnt-targeting tyrosine kinase inhibitor WNTinib as novel, clinically actionable strategies to target AC-PC more effectively.

Many cancer patients do not develop a durable response to the current standard of care immunotherapies despite substantial advances in targeting immune inhibitory receptors1-5. A potential compounding issue, which may serve as an unappreciated, dominant resistance mechanism, is an inherent systemic immune dysfunction that is often associated with advanced cancer6-12. Minimal response to inhibitory receptor (IR) blockade therapy and increased disease burden have been associated with peripheral CD8+ T cell dysfunction, characterized by suboptimal T cell proliferation and chronic expression of IRs (eg. Programmed Death 1 [PD1] and Lymphocyte Activation Gene 3 [LAG3])13, 14. Here, we demonstrate that up to a third of cancer patients express robust intracellular LAG3 (LAG3IC), but not surface LAG3 (LAG3SUR), in peripheral CD8+ T cells compared to CD4+ T cells and regulatory T cells (Tregs). LAG3IC is associated with: (i) expression of a LAG3IC-dominant IR module that includes PD1IC, NRP1IC, CD39IC, and TIGITIC; (ii) decreased CD8+ but not CD4+ T cell function that can be reversed by anti-LAG3 (and/or anti-PD1), despite limited constitutive surface IR expression; and (iii) poor disease prognosis. Systemic immune dysfunction is restricted to CD8+ T cells, including a high percentage of peripheral naive CD8+ T cells, indicating a TCR-independent mechanism that is driven by the cytokine IL6 and the chemokine IL8. Thus, the combination of an increased LAG3-dominant IR module and elevated systemic IL6 and/or IL8 may serve as predictive biomarkers and increase the possibility that cancer patients will benefit from therapeutic combinations targeting these systemic cytokines in the setting of PD1 and/or LAG3 blockade.

Oncogenic MYC promotes cancer cell proliferation, metabolism, and death, while also driving immunosuppression in the tumour microenvironment, complicating immune-based therapies. To counter MYC-driven immune evasion while leveraging MYC-dependent synthetic lethality (MYC-SL), we identified microtubule-targeting agents, including eribulin, as potent inducers of immunogenic cell death in MYChigh triple-negative breast cancer (TNBC). A screen of 528 oncology compounds using damage-associated molecular pattern (DAMP) reporters revealed that microtubule inhibitors induced key DAMPs, including HMGB1 secretion, calreticulin exposure, and double-stranded DNA release, leading to gasdermin-E associated cell death in MYChigh TNBCs. Immune cell co-culture assays showed immune activation, and patient-derived explant cultures confirmed pro-inflammatory cytokine responses. In vivo, cell-free media from eribulin-treated MYChigh murine TNBCs enhanced tumour protection in vaccination models compared to MYC-knockdown controls, linking MYC-dependent DAMP release to immunogenicity. These findings highlight a dual-function therapeutic strategy: agents that selectively induce MYC-dependent immunogenic cell death can provide both targeted cytotoxicity and local immune stimulation, thereby addressing a key limitation of conventional chemotherapeutics, offering a new approach for MYC-driven cancers.

Rhabdoid tumors (RT) are among the most aggressive pediatric malignancies, characterized by early onset, loss of SWI/SNF complex members (SMARCB1 or SMARCA4), and dismal outcomes despite multimodal therapy. Refractory and relapsing RT remain almost uniformly fatal, and targeted or immune-based approaches have yet to demonstrate clinical benefit. To explore novel therapeutic vulnerabilities, we systematically investigated the expression of clinically actionable surface proteins that could serve as targets for antibody-drug conjugates (ADCs), radiopharmaceutical therapy (RPT), or cellular immunotherapies. Based on large-scale transcriptomic analyses, we prioritized FAP, CXCR4, and IL13RA2 and performed comprehensive protein-level validation by immunohistochemistry in an unprecedented cohort of 60 rhabdoid tumors spanning all molecular subgroups (ATRT-TYR, ATRT-SHH, ATRT-MYC, and eMRT). Integrating these data with spatial and single-nucleus transcriptomic profiling, we identified subgroup- and cell-type-specific expression patterns, including heterogeneous FAP distribution between stromal and tumor compartments and a distinct IL13RA2-positive rhabdoid cell population with melanosomal and stem-like features. These findings define a set of biologically and clinically relevant surface targets in RT and provide a translational blueprint for rational ADC and RPT target discovery in pediatric cancer.

SARS-CoV-2 mRNA vaccination (COVID-19 vaccination) within 100 days of immune checkpoint inhibitor (ICI) treatment was reported to improve survival and prevent disease progression in patients with non-small cell lung cancer (NSCLC) and metastatic melanoma (Grippin et al., Nature, 2025). However, the clinical evidence, derived from real-world observational data, might suffer from methodological limitations, including immortal-time bias. These key limitations can be overcome by carefully designing a target trial emulation analysis. Using the data made publicly available by the authors, we emulated a target trial that would identify the causal effect of COVID-19 vaccination within 100 days of first ICI on overall survival and progression-free survival in patients with NSCLC and metastatic melanoma. In contrast to the original analysis, we found no evidence that COVID-19 vaccination improves survival outcomes in these populations. The original results likely reflect biases inherent to non-causal observational analyses. To clarify the true effect of COVID-19 vaccination in this setting, larger and suitably designed studies are needed.

Platinum-based chemotherapy remains a cornerstone of treatment for triple-negative breast cancer (TNBC), yet the molecular determinants governing platinum response remain poorly defined. By leveraging the randomized Phase II INFORM trial, which compared neoadjuvant cisplatin to anthracycline-based therapy in BRCA1/2-mutant breast cancer--we identified miR-362-3p as a specific regulator of cisplatin sensitivity. Higher plasma miR-362-3p expression was exclusively associated with favorable clinical outcome in the cisplatin arm, with no association observed in the AC arm, decoupling platinum-specific vulnerability from general chemotherapy response. We used gain- and loss-of-function TNBC models to establish that miR-362-3p functions as a potent sensitizer to cisplatin in vitro and in vivo. Integrated TCGA analysis and experimental validation identified BCLAF1, a key regulator of DNA damage response, as a direct repression target of miR-362-3p. We uncovered a novel role for the miR-362-3p/BCLAF1 axis in overcoming platinum resistance in TNBC.

Although high clinical response rates are seen for immune checkpoint blockade (ICB) treatment of metastatic melanomas, both intrinsic and acquired ICB resistance remain considerable clinical challenges1. Combination ICB (anti-PD-1 + anti-CTLA-4) shows improved patient benefit2-5, but is associated with severe adverse events and exceedingly high cost. Therefore, there is a dire need to stratify individual patients for their likelihood of responding to either anti-PD-1 or anti-CTLA-4 monotherapy, or the combination. Since it is conceivable that ICB responses are influenced by both tumor cell-intrinsic and -extrinsic factors6-9, we hypothesized that a predictive genetic classifier ought to mirror both these features. In a panel of patient-derived melanoma xenografts10 (PDX), we noted that cells derived from the human tumor microenvironment (TME) that were co-grafted with the tumor cells were naturally replaced by murine cells after the first passage. Taking advantage of the XenofilteR11 algorithm we recently developed to deconvolute human from murine RNA sequence reads from PDX10, we obtained curated human melanoma tumor cell RNA reads. These expression signals were computationally subtracted from the total RNA profiles in bulk (tumor cell + TME) melanomas from patients. We thus derived one genetic signature that is purely tumor cell-intrinsic ("InTumor"), and one that comprises tumor cell-extrinsic RNA profiles ("ExTumor"). Here we report that the InTumor signature predicts patient response to anti-PD-1, but not anti-CTLA-4 treatment. This was validated in melanoma PDX and cell lines, which confirmed that InTumorLO tumors were effectively eliminated by adoptive cell transfer of T-Cell Receptor (TCR)-matched cytotoxic T cells, whereas InTumorHI melanomas were refractory and grew out as fast as tumors challenged with unmatched T cells. In contrast, the ExTumor signature predicts patient response to anti-CTLA-4 but not anti-PD-1. Most importantly, we used the InTumor and ExTumor signatures in conjunction to generate an ICB response quadrant, which predicts clinical benefit for five independent melanoma patient cohorts treated with either mono- or combination ICB. Specifically, these signatures enable identification of patients who have a much higher chance of responding to the combination treatment than to either monotherapy (p < 0.05), as well as patients who are likely to experience little benefit from receiving anti-CTLA-4 on top of anti-PD-1 (p < 0.05). These signatures may be clinically exploited to distinguish patients who need combined PD-1 + CTLA-4 blockade from those who are likely to benefit from either anti-CTLA-4 or anti-PD-1 monotherapy.

Antibody-drug conjugates (ADCs) targeting cell surface proteins TROP2 or HER2 are effective in metastatic breast cancer, but the precise clinical contribution of epitope expression is uncertain. We prospectively monitored circulating tumor cells (CTCs) in 33 patients receiving ADC therapies using quantitative imaging. The expression of TROP2 and HER2 are heterogeneous across single CTCs from untreated patients, comparable to matched tumor biopsies, and display poor association with clinical response. Within three weeks of treatment initiation, declining CTC numbers correlate with a durable response (TROP2: median time to progression 391 versus 97 days, HR 4.15, P=0.0046; HER2: 322 versus 66 days, HR 9.12, P=0.0002). Neither TROP2 nor HER2 expression is reduced at progression, compared to matched pretreatment CTCs, and switching ADC epitope while maintaining a similar payload shows poor efficacy. Thus, epitope downregulation is not a common driver of acquired resistance to TROP2 or HER2 ADCs, and second-line ADC therapies may benefit from distinct payloads. SIGNIFICANCEADCs target tumor-associated antigens, followed by internalization and release of drug payloads. However, clinical studies of epithelial-targeting ADCs show efficacy despite low tumor epitope expression. Our finding that epitope downregulation does not commonly accompany acquired resistance suggests alternative drivers of clinical efficacy and the need for testing non-cross-resistant payloads to overcome resistance.

Triple-negative breast cancer (TNBC) has high rates of recurrence despite chemotherapy and immune checkpoint blockade (ICB). Tumor-associated macrophages (TAMs) can either suppress or support anti-tumor immunity, but the mechanisms governing these states and therapeutic targets remain unclear. Here, integrating public scRNAseq datasets with TNBC cohorts, we identify a prognostic myeloid signature defined by CXCL9hi/C1Qlow TAM programs, associated with improved survival and increased lymphocyte activation pathways. Using immunocompetent p53-null syngeneic TNBC models spanning basal-like (2153L) and claudin-low (T12) subtypes, we show that immunomodulatory cyclophosphamide (CTX) reprograms hematopoiesis toward the monocytic lineage and induces an interferon (IFN) conditioned tumor milieu that supports CXCL9 monocyte-derived macrophages (Mo.Macs) in basal-like disease. Combining CTX with the next generation MERTK-selective inhibitor UNC2371 (MRX-2843) drives complete remissions in both models, but durable long-term responses occurred selectively in the basal-like subtype model. The expansion of antigen-presenting CXCL9 Mo.Macs and reduction of C1q phagocytic TAMs are observed in responding tumors. Mechanistically, MERTK inhibition relieves MAPK/SOCS1 mediated restraint of IFN signaling driving a positive feedback loop of IRF7/STAT1/IRF1 driven CXCL9 induction. Functionally, tumor control requires CXCL9-CXCR3 dependent CD4 T cell recruitment, accumulation of stem-like memory CD4 T cells, and germinal center like immune organization in tumor-draining lymph nodes. PD-1 blockade further increases durability, preventing recurrence in most treated basal-like tumors. Together, these findings define an IFN licensed, MERTK regulated myeloid checkpoint that can be therapeutically targeted to convert suppressive TNBC microenvironments into durable adaptive immunity, supporting clinical translation of CTX + MRX-2843 based combinations in basal-like TNBC. SignificanceSuppressive myeloid programing limits effective adaptive immune engagement in TNBC usually resulting in ICB treatment resistance and tumor recurrence. This study identifies a therapeutically actionable myeloid interferon checkpoint in which MERTK inhibition stabilizes CXCL9 monocyte-macrophage programming to promote CD4 T cell dependent immune memory and durable tumor control in basal-like TNBC.

AbstractThe stromal compartment of many solid tumors plays a critical role in shaping an immunosuppressive microenvironment that limits the effectiveness of immune-based therapies1. Among stromal constituents, cancer-associated fibroblasts (CAFs) have emerged as key regulators of antitumor immunity2-5. Here, we identify a distinct subset of CAFs in both murine and human stroma-rich cancers that secrete osteoprotegerin (OPG)-- a soluble decoy receptor that neutralizes receptor activator of nuclear factor kappa-B ligand (RANKL) and TNF-related apoptosis-inducing ligand (TRAIL), both of which are involved in T cell function. In vitro, OPG directly impairs CD8 T cell-mediated killing of target cells. In murine models of pancreatic and breast cancer, antibody-mediated blockade of OPG promotes robust immune infiltration into the tumor microenvironment, leading to significant tumor regression. Stromal profiling revealed that OPG blockade induces a shift in CAF cells--reducing immunosuppressive OPG fibroblasts while expanding interferon-responsive fibroblasts, thus recalibrating the tumor stroma toward a pro-immunogenic landscape. These findings uncover a previously unrecognized mechanism of stromal immune suppression and highlight OPG as a stromal immune checkpoint controlling CD8 T cell infiltration. Targeting OPG may offer a novel therapeutic strategy to convert immunologically "cold" tumors into T cell-infiltrated, tumor microenvironment.

Targeting metabolism has long been a theory for cancer therapy, but clinical development has been limited by toxicities, compound availability, overall efficacy, and patient specificity1. CPI-613, a lipoic acid analogue that interferes with enzymes involved in mitochondrial metabolism, has demonstrated clinical activity in lethal malignancies including relapsed or therapy refractory Acute Myeloid Leukemias (AMLs)2,3 and Phase III trials are ongoing1. Using metabolomics, we investigated blood and bone marrow samples from a cohort of 29 relapsed or refractory AML patients involved in Phase I and II studies undergoing CPI-613 treatment (NCT01768897, NCT02484391) including 13 that achieved a complete response. We show that CPI-613 treatment in patients induced defined alterations related to the tricarboxylic acid (TCA) cycle and associated redox, anabolic and catabolic metabolism. These findings are consistent with targeting of several ketoacid dehydrogenase (KADH) enzymes that use lipoic acid as a cofactor and are related to mitochondrial metabolism. The alterations were observed systemically but were more pronounced within the leukemic bone marrow microenvironment consistent with its mechanistic target. Machine learning revealed that metabolic status and changes associated with mitochondrial metabolism were predictive of treatment response, indicating that mechanism-based metabolite biomarkers to a targeted metabolic cancer therapy may be feasible. Finally, we confirm using isotope tracing and flux analysis that these effects are due to disruptions to substrate utilization into the mitochondria. Our findings provide evidence that a tolerated, anti-cancer therapeutic can act by targeting mitochondrial metabolism in humans.

Minimal residual disease (MRD) monitoring in multiple myeloma (MM) relies on invasive bone marrow (BM) biopsies, which often yield insufficient tumor material. We performed whole genome sequencing of cell-free DNA from 163 plasma samples from 51 patients to develop a non-invasive MRD classifier. BM WGS identified a median of 2,502 clonal mutations, enabling cfDNA tracking at levels comparable to BM-based MRD testing. The cfDNA classifier achieved a mean AUC of 0.86 against multiparameter flow cytometry and targeted immunoglobulin sequencing (clonoSEQ), and MRD negativity after one year of maintenance was strongly associated with two-year relapse-free survival probability (Hazard Ratio = 24), with cfDNA changes preceding clinical progression by a median of 12.6 months. To establish a BM-agnostic mode, a plasma-only classifier trained on baseline cfDNA established a mean AUC of 0.79 compared to BM clinical testing and stratified relapse risk (HR = 4.18), enabling MRD detection in patients with suboptimal BM samples. Longitudinal cfDNA profiles detected subclonal evolution in half of profiled patients. Cell-free DNA whole genome sequencing provides a sensitive, scalable, and clinically informative platform for non-invasive MRD monitoring in MM. Statement of Translational SignificanceCurrently, myeloma MRD assays require invasive, painful bone marrow sampling and fail to account for spatial heterogeneity. High-depth, cell-free whole genome sequencing tracks thousands of personalized mutations in blood, identifying molecular relapse a median of 12.6-19 months before clinical progression and showing a stronger association with progression-free survival than standard bone-marrow tests (hazard ratio = 24 for BM-informed mutation lists). This scalable, comprehensive approach enables dynamic multiple myeloma monitoring and risk-adapted treatment. Summary ParagraphMinimal residual disease (MRD) monitoring is one of the strongest predictors of progression-free and overall survival in multiple myeloma (MM); as such, clinicians currently rely on invasive bone-marrow sampling that patients tolerate poorly for serial assessment. We show that longitudinal whole-genome sequencing of plasma cell-free DNA, guided by personalized mutation catalogs from diagnostic bone marrow, detects residual tumor DNA at ultra-low levels. This approach tracked disease dynamics across serial samples, achieved strong concordance with standard clinical MRD assays, accurately predicted progression-free survival, and detected subclonal evolution driving relapse. While developed initially using bone marrow reference, we found that cell-free DNA collected at baseline provided comparable mutational information to serve as a reference for subsequent MRD blood draws. This plasma-only classifier requires no marrow input, identified persistent disease when diagnostic marrow was insufficient for referencing, stratified relapse risk, and detected rising MRD probabilities a median of 19 months before progression. Together, these results establish cell-free DNA whole genome sequencing as a minimally invasive platform for comprehensive genomic surveillance in MM, reducing dependence on painful repeat marrow biopsies and enabling earlier, risk-adapted intervention. More broadly, this strategy can extend to diverse cancers to guide personalized therapy and identify relapse earlier.

The abnormal tumor vasculature can present a barrier to the infiltration of anti-tumor immune cells, which impairs immune surveillance and response to immunotherapy. Here, we show that targeting the epigenetic factor DNA methyltransferase 1 in endothelial cells (ECs) reduces angiogenesis while imparting profound changes to the tumor immune microenvironment (TIME), including increased proportions of CD4+ memory T-cells and NK cells. Depleting CD4+ T-cells, or blocking lymphocyte egress from the lymph nodes with FTY720, rescues tumor growth in mice with conditional deletion of Dnmt1 in ECs (Dnmt1iECKO) and dramatically shortens overall survival, whereas NK cells are dispensable. Tumors implanted in Dnmt1iECKO mice show reduced vascular branching, elevated expression of Vcam1, increased vessel-associated T-cells, and a shift in vascular specification including increased proportions of immune-permissive post-capillary venules (PCVs) and interferon-stimulated ECs (IFN-ECs). Deleting Dnmt1 in EC cultures strikingly potentiates responses to combinations of IFN{gamma} and TNF and, notably, up-regulates important T-cell co-stimulatory molecules for memory CD4+ T-cells, including Icosl, Cd40, and Tnfsf4. Finally, immune checkpoint blockade (ICB) administered to Dnmt1iECKO mice with experimental melanoma lung metastasis reduces tumor burden, with some mice showing tumor eradication. Our findings identify endothelial Dnmt1 as a key regulator of vascular-mediated anti-tumor immunity, providing a rationale for integrating epigenetic modulation of the vasculature with cancer immunotherapy regimens.

Immune checkpoint blockade (ICB) has become a standard of care in the treatment of metastatic melanoma (MM). Although ICB is particularly successful in some MM patients, more than half do not obtain a durable benefit. Biomarkers that predict response are urgently needed and overcoming intrinsic resistance is key to improving the success of ICB therapy. Using single cell RNA sequencing, we characterized the immune landscape of pre- and early on-treatment biopsies taken from a cohort of MM patients (n>20) exposed to ICB therapy. Our analysis identified >20 immune cell types and confirmed previously described associations between the abundance of various CD8 T cell populations and ICB outcome. Unexpectedly, we found that lack of response was associated with an increased occurrence of a granulysin-expressing (GNLY+) natural killer (NK) cell population. This observation was replicated in other MM cohorts and in a breast cancer cohort in which paired biopsies were also collected pre and early-on ICB therapy. Spatial proteomics revealed that whereas NK cells colocalized with CD8 T cells within the tumour bed in responding lesions, these cells accumulated at the tumour margin in non-responding lesions. Strikingly, depletion of NK cells in an NRAS-driven melanoma mouse model, which exhibits an immune-excluded phenotype and is refractory to ICB, promoted massive immune cell infiltration and tumour clearance upon anti-PD1 exposure. These data highlight a differential immune cell topography between early on-treatment responding and nonresponding MM lesions, which could be exploited to develop a robust stratification biomarker, and unravel an unexpected contribution of NK cells in primary resistance to ICB.

Tumor metastasis, the main cause of death in cancer patients, requires outgrowth of tumor cells after their dissemination and residence in microscopic niches. Nutrient sufficiency is a determinant of such outgrowth1. Fatty acids (FA) can be metabolized by cancer cells for their energetic and anabolic needs but impair the cytotoxicity of T cells in the tumor microenvironment (TME)2, 3, thereby supporting metastatic progression. However, despite the important role of FA in metastatic outgrowth, the regulation of intratumoral FA is poorly understood. In this report, we show that tumor endothelium actively promotes tumor growth and restricts anti-tumor cytolysis by transferring FA into developing metastatic tumors. This process uses transendothelial fatty acid transport via endosome cargo trafficking in a mechanism that requires mTORC1 activity. Thus, tumor burden was significantly reduced upon endothelial-specific targeted deletion of Raptor, a unique component of the mTORC1 complex (RptorECKO). In vivo trafficking of a fluorescent palmitic acid analog to tumor cells and T cells was reduced in RptorECKO lung metastatic tumors, which correlated with improved markers of T cell cytotoxicity. Combination of anti-PD1 with RAD001/everolimus, at a low dose that selectively inhibits mTORC1 in endothelial cells4, impaired FA uptake in T cells and reduced metastatic disease, corresponding to improved anti-tumor immunity. These findings describe a novel mechanism of transendothelial fatty acid transfer into the TME during metastatic outgrowth and highlight a target for future development of therapeutic strategies.

Compared to immortalized cell lines, patient-derived organoids and other ex vivo models have been shown to better recapitulate patient responses to therapy. High cost and technical complexity have prevented the creation of pan-cancer ex vivo datasets, limiting comprehensive analyses and predictive modeling for ex vivo drug response. We present the Pan-PreClinical (PPC) project: a drug screen atlas of 2.1M experiments across 1,982 ex vivo samples and 3,100 drugs spanning 134 cancer indications tested across 26 studies. We develop a contrastive Bayesian model to harmonize across studies, identifying 303 tissue-specific drug sensitivities and demonstrating drug sensitivities are predictive of clinically-relevant molecular profiles. Integrating established cell line databases reveals systematic biases across 55 cancer subtypes, with cell line screens favoring drugs targeting highly proliferative cells and undervaluing cell-cell communication targets. We leverage PPC to establish an ex vivo foundation model and computational platform for scalable ex vivo cancer biology and predictive oncology.