Cancer Research

The study of tumor progression and metastasis in a physiologic setting commonly involves implantation of tumor cells into immunocompetent mice. For this purpose, tumor cells are labeled routinely with bioluminescent and/or fluorescent proteins prior to transplantation, despite the fact that these foreign proteins generate adaptive immune responses. We have described previously how incorporating centrally tolerized mouse strains into tumor and metastasis modeling can be used as a study design principle to properly control for artifactual immune responses against such reporters. In this issue of Cancer Research, Khan and colleagues applied a tolerized mouse strain—the Tol mouse—to overcome limitations associated with the use of xenoantigenic fluorescent and bioluminescent reporters in immunocompetent settings. The authors showed how antigenic responses against such proteins can confound interpretation of study results when measuring the efficacy of immunotherapy regimens. This study readily demonstrates the utility of employing centrally tolerized transgenic models to improve preclinical investigations of cancer metastasis and therapeutic resistance. See related article by Khan et al., p. 2165

Cancer drug resistance is multi-factorial, driven by heritable (epi)genetic changes but also phenotypic plasticity. Here, we dissected drivers of resistance by perturbing colorectal cancer patient-derived organoids longitudinally with drugs in sequence. Combined longitudinal lineage tracking, single cell multi-omics analysis, evolutionary modelling, and machine learning revealed that different targeted drugs select for distinct subclones, supporting rationally designed drug sequences. The cellular memory of drug resistance was encoded as a heritable epigenetic configuration from which multiple transcriptional programs could run, supporting a one-to-many (epi)genotype-to-phenotype map that explains how clonal expansions and plasticity manifest together. This epigenetic landscape may ensure drug resistant subclones can exhibit distinct phenotypes in changing environments while still preserving the cellular memory encoding for their selective advantage. Chemotherapy resistance was instead entirely driven by transient phenotypic plasticity rather than stable clonal selection. Inducing further chromosomal instability before drug application changed clonal evolution but not convergent transcriptional programs. Collectively, these data show how genetic and epigenetic alterations are selected to engender a "permissive epigenome” that enables phenotypic plasticity.

Targeted therapies and immunotherapy have improved treatment outcomes for many melanoma patients. However, patients whose melanomas harbor driver mutations in the neurofibromin 1 (NF1) tumor suppressor gene often lack effective targeted treatment options when their tumors do not respond to immunotherapy. In this study, we utilized patient-derived short-term cultures (STCs) and multiomics approaches to identify molecular features that could inform the development of therapies for patients with NF1 mutant melanoma. Differential gene expression analysis revealed that the epidermal growth factor receptor (EGFR) is highly expressed and active in NF1 mutant melanoma cells, where it hyper-activates ERK and AKT, leading to increased tumor cell proliferation, survival, and growth. In contrast, genetic or pharmacological inhibition of EGFR hindered cell proliferation and survival and suppressed tumor growth in patient-derived NF1 mutant melanoma models but not in NF1 wild-type models. These results reveal a connection between NF1 loss and increased EGFR expression that is critical for the survival and growth of NF1 mutant melanoma cells in patient-derived culture and xenograft models, irrespective of their BRAF and NRAS mutation status.

Normal tissues actively employ a phenomenon called cell competition to drive the elimination and replacement of less fit loser cells by fitter winner cells. This quality control mechanism promotes tissues health, by favouring the selective expansion of fitter cells. Indeed, through cell competition, many mutant cells are eliminated from tissues by fitter normal cells. However, some oncogenic mutations can turn cells into super-competitors that outcompete normal cells, promoting tumorigenic growth and metastasis. Several cellular stresses have been associated with the loser status such as oxidative stress, DNA damage responses, unfolded protein response and mitochondrial dysfunction. By affecting these pathways, metabolism and dietary choices can regulate cellular fitness and cell competition. However, how these pathways affect competitive interactions in vivo, during the early establishment of mutant clones, is relatively little understood. Recent work from Hemalatha and colleagues introduces real-time fluorescence ratio metric imaging of NAD(P)H and FAD, to investigate cellular redox status - live and over time, at single cell level - as cells compete in the mouse epidermis. Their work demonstrates that redox status changes dynamically during competition between cell carrying oncogenic mutations. It further shows that drugs that modulate mitochondrial metabolism and cellular redox are strong modulators of cell competition. The introduction of live redox imaging will prove a powerful tool to further dissect how metabolic states affect cell competition in normal physiology and in tumorigenesis.

Genomic alterations converging on persistent activation of the PI3K/mTOR pathway represent one of the most frequently altered signaling circuitries in cancer. However, the clinical efficacy of mTOR inhibitors (mTORi) has been limited. Here, we took advantage of the widespread activation of PI3K/mTOR signaling in head and neck squamous cell carcinoma (HNSCC) and the promising effects of mTORi in HNSCC experimental models and recent clinical trials, to gain a mechanistic understanding of the anti-tumoral activity of mTORi. A genome-wide CRISPR screen revealed that treatment with mTORi promotes the autophagic degradation of ferritin (ferritinophagy), consequently increasing free intracellular iron, inducing lipid peroxidation, and ultimately driving cancer cell demise by ferroptosis. These findings provide a rationale for synergistic combinations repurposing approved drugs that disable cellular ferroptotic defense mechanisms. Together, this study provides a molecular framework underlying the anti-tumor activity of mTORi in HNSCC, thereby revealing multimodal precision therapies for HNSCC and many human malignancies displaying overactive PI3K/mTOR signaling.

Lung adenocarcinoma (LUAD) is a heterogeneous disease with substantial genomic differences between individuals of Chinese and European ancestry. Deciphering the timing of driver mutations may lead to insights into tumor evolution that can inform diagnostic and therapeutic approaches for LUAD. Here, we conducted whole-genome sequencing on LUAD samples from 251 patients with Chinese ancestry to reconstruct the evolutionary trajectories of somatic alterations, especially those across the non-coding regions. Tobacco-related mutations preferentially occurred early and plateaued at 28 packs of cigarettes per year. Well-known driver mutations (e.g., EGFR, TP53, and RB1) also occurred at the early stage, displaying ancestry heterogeneity among smokers. In contrast to exogenous mutagens, endogenous mutagen-related alterations (APOBEC) occurred late. The 3’UTR was the most frequently altered non-coding element in LUAD, with recurrent disrupting mutations in the 3’UTR of SFTPB and SFTPA1. Unlike other cancer types, TERT promoter mutations were observed specifically among female LUAD patients. Clustered mutations (e.g., doublet-base substitutions, multi-base substitutions, and kataegis) influenced LUAD evolution and were overrepresented in driver genes. These findings provide insights into the dynamic nature of genomic alterations during lung tumorigenesis.

p53 is a transcription factor that functions as a tumor suppressor and the active unit of which is a tetramer. Nearly all cancers inactivate the p53 pathway, primarily through missense mutations in TP53. Most mutant p53 proteins lose their function and often exhibit increased protein stability. In addition to this loss of function, mutant p53 can drive oncogenicity through a dominant-negative effect by forming mixed tetramers with wild-type (WT) p53 to inhibit its activity. The study in this issue of Cancer Research by Klemm and colleagues provides definitive evidence for the mechanism by which mutant p53 inactivates WT p53 function. The p53R248Q mutant has a longer half-life than WT p53, resulting in an approximate ratio of 3 or 4 mutant molecules to every WT molecule. This imbalance facilitates the dominant-negative effect, which can be overcome either by exogenously increasing WT p53 levels or by selectively degrading mutant p53. In experiments whereby mutant p53 was degraded using a degron-tagged construct combined with iberdomide as a molecular glue, remarkable therapeutic efficacy was observed when this approach was used in combination with an MDM2 inhibitor. This work paves the way for therapeutic strategies that aim to degrade mutant p53 proteins in cases in which a WT TP53 allele is retained. See related article by Klemm et al., p. 1978

The arginine methyltransferase CARM1 is amplified/overexpressed in a variety of cancers, including triple-negative breast cancer (TNBC), spurring the interest of developing CARM1 inhibitors (CARM1i). Here, we discovered that CARM1i treatment leads to elevated CARM1 levels as well as activation of AKT, which could result in long-term treatment resistance in breast cancer cells. CARM1 methylated MAP2K4 at arginine 58 in TNBC, and methylated-MAP2K4 localized to the nucleus and potentiated the proliferation- and metastasis-promoting function of MAP2K4 via activation of JNK signaling. Inhibition of MAP2K4 by CARM1i led to AKT activation, which was abrogated by treatment with a PI3K inhibitor (PI3Ki). Combining CARM1i and PI3Ki elicited synergistic anti-cancer effects in TNBC cell lines, organoids, and patient-derived xenografts. This study demonstrates that, through inhibiting MAP2K4 methylation, CARM1i abrogates the feedback loop of MAP2K4 and PI3K signaling, supporting treatment with CARM1i as a therapeutic approach to improve the sensitivity of TNBC to PI3Ki.

The 5-year overall survival rate for pancreatic cancer remains ~13%, underscoring the urgent need for improved treatment strategies. TGFβ is a promising target due to its significant involvement in the desmoplasia, immune suppression, and chemoresistance characteristic of pancreatic cancer. Over 300 clinical trials targeting TGFβ have been conducted in unselected patient cohorts; however, none of the therapies have gained FDA approval. Nevertheless, TGFβ blockade may hold promise for a subset of cancers with non-functional TGFβ signaling. Greater than 25% of pancreatic cancers carry mutations in SMAD4, a key component of canonical TGFβ signaling. In this study, we investigated the potential for stratifying patients based on SMAD4 mutational status to identify tumors susceptible to TGFβ inhibition. Analysis of SMAD4 expression in human pancreatic tumors revealed that SMAD4 mutation or loss is associated with worse disease-free survival. Intriguingly, intratumoral SMAD4 expression displayed heterogeneity among human pancreatic cancer samples. SMAD4 deficient genetically engineered mouse models and orthotopic SMAD4 knockout tumor models exhibited reduced survival, increased metastasis, and alterations in the tumor microenvironment compared to SMAD4 wildtype controls, consistent with gene and protein expression changes in the absence of functional SMAD4. Importantly, treating mice bearing SMAD4 deficient tumors with a blocking TGFβ antibody reduced tumor weight and improved survival. These findings suggest that genomic stratification by TGFβ axis alterations, such as SMAD4 mutations, may be a promising approach to identifying patients likely to benefit from a TGFβ inhibitor.

Solid cancers are complex ‘ecosystems’ comprised of diverse cell types and extracellular molecules, where heterotypic interactions significantly influence disease etiology and therapeutic response. However, our current understanding of tumor microenvironments (TMEs) remains incomplete, hindering the development and implementation of novel TME-targeted drugs. To maximize cancer therapeutic development we require a systems-level understanding of the malignant, stromal, and immune states that define the tumor and determine treatment response. In their recent study, Liu and colleagues took a new approach to resolving the complexity of stromal heterogeneity. They leveraged extensive single-cell spatial multi-omics datasets across various cancer types and platforms to identify four conserved spatial Cancer-Associated Fibroblast (CAF) subtypes, classified by their spatial organization and cellular neighborhoods. Their work expands upon prior efforts to develop a CAF taxonomy, which primarily relied on single-cell RNA-Sequencing (scRNA-Seq) and yielded a multitude of classification systems. This study advances our understanding of CAF biology by establishing a link between spatial context and CAF identity across diverse tumor types. Departing from recent single-cell transcriptomics studies that employed a marker-based approach for sub-state identification, Liu and colleagues conducted de novo discovery of CAF subtypes using spatial neighborhood information alone. By positioning spatial organization as the defining axis of CAF heterogeneity, this research redefines CAF classification and provides a new framework for exploring the rules governing tumor ecosystems and developing novel ecosystem-based therapeutic strategies.

The gut microbiome has emerged as a key regulator of response to cancer immunotherapy. However, a better understanding of the underlying mechanisms by which the microbiome influences immunotherapy is needed to identify strategies to optimize outcomes. To this end, we developed a mathematical model to obtain insights into the effect of the microbiome on the immune system and immunotherapy response. This model was based on i) gut microbiome data derived from preclinical studies, ii) mathematical modeling of the antitumor immune response, iii) association analysis of microbiome profiles with model-predicted immune profiles, and iv) statistical models that correlate model parameters with the microbiome. The model was used to investigate the complexity of murine and human studies on microbiome modulation. Comparison of model predictions with experimental observation of tumor response in the training and test datasets supported the hypothesis that two model parameters, the activation and killing rate constants of immune cells, are the most influential in tumor progression and are potentially affected by microbiome composition. Evaluation of the associations between the gut microbiome and immune profile indicated that the components and structure of the gut microbiome affect the activation and killing rate of adaptive and innate immune cells. Overall, this study contributes to a deeper understanding of microbiome-cancer interactions and offers a framework for understanding how microbiome interactions influence cancer treatment outcomes.

Diffuse midline gliomas (DMGs) are devastating brain tumors that occur primarily in children. The salient feature of these tumors is the presence of a H3K27M mutation (K27M), associated with the worst prognosis. Development of effective strategies for treating K27M+ DMG is desperately needed to help improve patient outcomes. Here, we identified the cell surface antigen CD99 as notably expressed in DMGs, particularly in K27M+ DMGs. The increased expression of CD99 in K27M+ DMGs was a result of the onco-histone K27M mutation. In K27M+ DMG cells, CD99 inactivation impaired tumor growth by inducing cell differentiation. The development of a therapeutic anti-CD99 chimeric antibody, 10D1, with a membrane-proximal binding epitope enabled the evaluation of the antitumor efficacy of targeting CD99 in preclinical models of K27M+ DMG. 10D1 suppressed DMG growth in vitro and in vivo by inducing apoptosis. When combined with radiation treatment, 10D1 exhibited improved antitumor efficacy and prolonged xenograft survival. Together, these findings provide a strong justification for the clinical development of 10D1 as a therapy for targeting CD99 to treat DMGs.

Small cell carcinoma of the ovary, hypercalcemic type (SCCOHT) is a rare, deadly form of ovarian cancer that uniformly harbors mutations in SMARCA4, a member of the SWI/SNF chromatin remodeling complex. SWI/SNF impacts RNA splicing, and dysregulation of splicing can generate immunogenic tumor antigens. Here, we explored the relationship between SMARCA4 loss and RNA splicing dysregulation. SCCOHT primary tumors harbored tumor-associated outlier splicing events compared to normal tissues. Many of the tumor events were retained introns encoding novel peptides predicted to bind to MHC-I complexes. Immune cells were observed in primary SCCOHT tumors, suggesting a potentially immune reactive tumor microenvironment. Mutations in several SWI/SNF subunits were associated with higher rates of outlier retained introns across tumor types in TCGA data. Interestingly, RNA sequencing of isogenic SCCOHT cell lines demonstrated a role for SMARCA4 in intron retention. Distinct protein-protein interactions between splicing factors identified in SCCOHT cell lines supported a role for SMARCA4 in splicing regulation. Further, SWI/SNF localized to genes which were differentially spliced. Mass spectrometry analyses confirmed expression of some of these novel peptides and a subset of these are predicted to bind to MHC-I complexes. A pool of these novel peptides derived from retained introns in SCCOHT triggered proliferation and expression of TNFa and INFb in primary human T cells. Together, these data suggest that SMARCA4 loss in SCCOHT leads to intron retention. Furthermore, T cell activation by novel peptides encoded by these tumor-specific splicing events suggests intron retention could be a source of tumor-associated antigens in SCCOHT.

The RAS G12V mutation is among the most common oncogenic drivers in pancreatic, colorectal, and non-small cell lung cancers. Within a tumor cell, RAS G12V exists predominantly in the active, GTP-bound (“RAS(ON)”) state leading to excessive downstream oncogenic signaling. The intrinsic GTP hydrolysis rate of RAS G12V is about 12-fold lower than that of KRAS G12C, biasing the cellular RAS G12V pool to the ON state and emphasizing the importance of targeting RAS G12V(ON) for maximal suppression of this oncogenic driver. Additionally, achieving selectivity for RAS G12V over wild-type RAS has, to date, been extremely challenging using conventional small molecules because the Val-12 residue is neither amenable to covalent inhibition nor to formation of polar, noncovalent interactions.The investigational agent RMC-5127 is a potent, orally bioavailable, RAS (ON) G12V-selective, noncovalent tri-complex inhibitor. In KRAS G12V mutant cancer cells, RMC-5127 forms a tri-complex between KRAS G12V(ON) and cyclophilin A (CypA), driving near-immediate disruption of RAS effector binding through steric occlusion and extinction of KRAS G12V(ON) signaling.RMC-5127 suppressed ERK phosphorylation and cell growth in multiple human KRAS G12V-addicted cancer cell lines in vitro, and a single dose induced dose-dependent, deep, and durable suppression of RAS pathway activation in vivo in subcutaneous xenograft models of KRAS G12V mutant cancers in mice. Across a panel of preclinical PDAC and NSCLC models harboring KRAS G12V, RMC-5127 monotherapy induced tumor regressions in the majority of models and was well-tolerated. In addition, dose-dependent exposure of RMC-5127 was observed in the whole brain of naïve mice, indicating the compound is brain penetrant, and RMC-5127 exhibited profound antitumor activity in relevant intracranial KRAS G12V mutant tumor xenograft models, with regressions observed at well-tolerated doses. Citation Format Anne Edwards. Discovery of RMC-5127, an oral, RAS (ON) G12V-selective, noncovalent, tri-complex inhibitor [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_2):Abstract nr ND06.

Regulatory T (Treg) cell abundance in the tumor microenvironment is associated with poor responses to immunotherapies, such as nivolumab. The transcription factors Helios (IKZF2) and Eos (IKZF4) are abundantly expressed in Treg cells and contribute to the immunosuppressive phenotype associated with this T-cell subset. We and others have postulated that degradation of Helios and Eos could re-polarize Treg cells towards a less suppressive, more inflammatory phenotype that would result in an enhanced antitumor effector T cell response. This presentation will describe the discovery and characterization of BMS-986449, a Cereblon E3 Ligase Modulatory Drug (CELMoD™) molecular glue degrader of Helios and Eos. BMS-986449 is a potent and orally bioavailable degrader of IKZF2/4 with minimal impact on IKZF1/3 levels in regulatory T-cells as assessed by global proteomics. BMS-986449 affects in vitro reprogramming of Treg cells to have a more effector-like phenotype that results in promising in vivo efficacy in syngeneic tumor models. In cynomolgus monkeys, oral administration of BMS-986449 (0.3 mg/kg, QD) was well-tolerated and maintained ≥80% degradation of Helios in circulating Treg cells over 24h. Given the positive outcomes from these pre-clinical studies and an acceptable safety profile, BMS-986449 was advanced into Phase I/II clinical trials for patients with advanced solid tumors. Citation Format Aaron Balog. The discovery of BMS-986449, a highly potent and selective degrader of the transcription factors Helios (IKZF2) and Eos (IKZF4) for the treatment of solid tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_2):Abstract nr ND05.

Relapsed/refractory (R/R) AML is associated with poor outcomes. Preclinical studies have shown that SENTI-202, a first-in-class CAR NK cell therapy, selectively kills AML blasts and leukemia stem cells (LSCs) while sparing hematopoietic stem & progenitor cells (HSPCs) via its Logic Gated CAR design (Kaveri, 2024). Here, we report interim clinical trial data in patients with R/R AML who received SENTI-202. Interim correlative results are submitted as a separate abstract. SENTI-202-101 is an ongoing multicenter Phase 1 dose finding study in adult patients (pt) with R/R hematologic malignancies including AML (NCT06325748). 2 SENTI-202 Dose Levels (DLs): 1e9 & 1.5e9 CAR NK cells/dose & 2 Schedules: 3 or 5 doses/ 28-day Cycle (I on Days 0, 7, 14; & II on Days 0, 3, 7, 10, 14, respectively), following lymphodepletion (LD) with fludarabine and cytarabine are being evaluated. Adverse events (AEs) including dose limiting toxicities (DLT) are graded per CTCAEv5 or ASTCT criteria. Efficacy is assessed by investigator per ELN 2022 and measurable residual disease (MRD) is assessed locally. Pharmacokinetics (PK) is assessed at multiple timepoints by ddPCR detection of transgene in peripheral blood (PB) and cytometry by time of flight (CyTOF) is conducted on serial bone marrow (BM) samples. As of 6 Jan 2025, 6 R/R AML pts with a median age 63.5 years (range 26, 72) were treated across both DLs in Schedule I with median 4.5 (3, 6) SENTI-202 doses. At baseline, the median time from diagnosis was 0.95yr (0.5, 6.2), with a median of 1.5 (1, 3) prior lines & median BM blasts 41.82% (8.9, 92.5). SENTI-202 was well tolerated with no DLTs. Grade (G) 3-4 AEs in > 1 pt were febrile neutropenia and decreased platelet count (3 pts each), all unrelated to SENTI-202 except 1 pt with decreased platelets; and all related to LD. SENTI-202 related AEs in > 1 pt were G1-2 pyrexia reported as CRS (3 pts). There were no G5 AEs or SENTI-202 related SAEs. 4 of 6 response evaluable pts achieved composite complete remission (cCR) across both DLs (3 CRs, 1 CRh currently receiving a 2nd cycle). 3/3 CRs were MRD-. All responses are ongoing at time of data cut with longest follow up of 5+ months & 2 pts receiving post treatment BM transplant. SENTI-202 transgene was detected in PB up to Day 15 in all 5 patients with PK data. Preliminary CyTOF analyses of BM showed a decrease in LSCs & maintenance/expansion of HSPCs in responders. Interim clinical data from ongoing Phase 1 trial reveals a well-tolerated safety profile for SENTI-202 administered after LD. 4/6 patients achieved cCR with all 3 CR patients being MRD- as assessed locally and responses ongoing. Preliminary PK profile is consistent with allogeneic NK cell therapies and CyTOF analyses reveal LSC killing & HSPC protection consistent with SENTI-202 Logic Gated Gene Circuit design. Dose finding is ongoing & additional data from pts in Schedule II will be reported. Citation Format Stephen A. Strickland, Eghtedar, Nosha Farhadfar, Ashish R. Bajel, Farhad Ravandi, Rochelle Emery, Brian Garrison, Muharrem Muftuoglu, Michael Andreeff, Kanya Rajangam, Gary Schiller. First-in-human, multicenter study of SENTI-202, a CD33/FLT3 selective off-the-shelf logic gated CAR NK cell therapy in hematologic malignancies including AML: Clinical data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_2):Abstract nr CT014.

Background Patients with early-stage solid tumors who harbor minimal residual disease (MRD) following surgical resection as detected by circulating tumor DNA (ctDNA), have a high risk of recurrence. It remains uncertain whether therapeutic intervention of MRD in this context reduces cancer recurrence. We evaluated the efficacy of PD-1 blockade in resected early-stage mismatch repair deficient (MMRd) tumors with ctDNA-detected MRD, given high sensitivity of MMRd tumors to immune checkpoint blockade. Methods We conducted a prospective clinical study assessing PD-1 blockade with pembrolizumab in patients with resected MMRd solid tumors treated with surgery and standard of care adjuvant therapy. Treatment with PD-1 blockade was guided by ctDNA status, measured 6-10 weeks after curative intent surgery and standard of care adjuvant therapy. Study arms included, Arm A (Interventional) comprised of ctDNA-positive patients treated with pembrolizumab for 6 months, and Arm B (Observational) comprised of ctDNA-negative patients. The primary endpoint was ctDNA clearance at 6 months in 40% of study patients. Recurrence-free survival (RFS) and overall survival (OS) were included as exploratory endpoints. Results Three-hundred and three patients with resected MMRd solid tumors were screened and ctDNA status was measured in 174 of these patients in 16 different tumor types. Twenty-two patients (12.6%) were ctDNA positive and enrolled in Arm A (Interventional). Of these 22, 13 ctDNA-positive patients received six months of PD-1 blockade and 9 did not because of radiographically detected disease recurrence prior to initiation of therapy. Fifty-two ctDNA-negative patients were enrolled in Arm B (Observational) and observed. The median follow-up time for all patients was 32.1 months. The study met the primary endpoint; 85% (11/13) of patients in Arm A (Interventional) achieved ctDNA clearance at 6 months, the median RFS was not reached, and the recurrence rate was 38% (5/13). In those that were ctDNA positive that did not receive PD-1 blockade the median RFS was 0.8 months (95% CI 0.3-1.3) and the recurrence rate was 100% (9/9). In Arm B (Observational) the median RFS was not reached, and the recurrence rate was 5.9% (9/152). Overall survival at 2 years was 92% (95%CI: 79-100%) in ctDNA positive patients who received PD-1 blockade and 78% (95%CI: 55%-100%) in patients that did not receive PD-1 blockade. In ctDNA negative patients, the OS at 2 years was 98% (95%CI: 96%-100% ). Conclusion Adjuvant PD-1 blockade directed by ctDNA status post-resection early-stage MMRd malignancies may provide clinical benefit in a agnostic of tumor type. Citation Format Yelena Yuriy Janjigian, Michael B. Foote, Melissa Lumish, Marinela Capanu, Joanne Chou, Julio Garcia-Aguilar, Martin Weiser, Jason Konner, Vivian Strong, Elizabeth Jewell, Jennifer Mueller, Steven Maron, Ping Gu, Rona Yaeger, Sree Chalasani, Andrea Cercek, Jia Li, Maliha Nusrat, Ryan Sugarman, David Jones, Daniela Molena, David Solit, Brian Loomis, Marc Ladanyi, Michael Berger, Luis A. Diaz. Circulating tumor DNA status to direct adjuvant immunotherapy for mismatch repair deficient tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_2):Abstract nr CT002.

Background Cancer cachexia, a metabolic disorder marked by severe muscle loss and common among certain types of cancers, significantly impacts prognosis and quality of life in patients. A core feature of cachexia is skeletal muscle mass reduction, which can be efficiently monitored by assessing the skeletal muscle area (SMA) in computed tomography (CT) scans routinely performed in cancer care. However, manual annotation of SMA on CT scans is labor-intensive and time-consuming, while existing automated tools often suffer from variability in accuracy and lack full automation, limiting their clinical utility. To address these challenges, we developed a reliable and fully automated pipeline to deliver consistent, accurate measurements. By integrating SMA and skeletal muscle index (SMI) with clinical data, our approach facilitates multimodal survival analysis and cachexia prediction, generating insights that empower clinicians to enhance patient care. Method We trained two deep learning (DL) models (nnU-Net 2D) using annotated CT images of the mid and end-third lumbar (L3) level from 50 gastroesophageal and 15 pancreatic cancer patients. Models were trained using 5-fold cross-validation to ensure robustness. To ensure reliability at inference, uncertainty estimation methods, model ensembling, dropout, and post-hoc calibration, were employed. Metrics such as variance, entropy, and coefficient of variation quantified model uncertainty, and a threshold-based approach flagged high-error SMA estimations for expert review. Our pipeline processes axial CT series, identifies L3 slices, annotates skeletal muscle, generates uncertainty maps, and SMA/SMI estimates. We combined the radiology derived SMA/SMI with clinical features to form multimodal data, which was used for survival analysis and prediction of cachexia at the time of cancer diagnosis using a multi-layer perceptron (MLP) model. Results On the gastroesophageal dataset, the DL model achieved an average Dice score of 97.80% ± 0.93% for skeletal muscle segmentation. Across the pancreatic, colorectal, and ovarian datasets, the DL model's SMA estimates differed with a median of 2.48% compared to manual expert segmentation. Uncertainty metrics demonstrated strong correlations with SMA estimation differences, with correlation coefficients of 0.83 (variance), 0.76 (entropy), and 0.73 (coefficient of variation). The multimodal MLP model for cachexia prediction achieved 70% accuracy and an F1 score of 76.92%. The multimodal survival analysis showed significant improvements in concordance index compared to using clinical data alone, with increases of 5.6% on the pancreatic, 5.9% on the colorectal, and 1.5% on the ovarian dataset. Conclusion Our automated, uncertainty-aware tool offers a robust and reliable solution for monitoring skeletal muscle changes, enabling diagnosis and intervention in managing cancer cachexia. Citation Format Sabeen Ahmed, Nathan Parker, Margaret Park, Daniel Jeong, Lauren Peres, Evan W. Davis, Jennifer B. Permuth, Erin Siegel, Matthew B. Schabath, Yasin Yilmaz, Ghulam Rasool. Reliable skeletal muscle area quantification and clinical data integration for predicting cancer cachexia [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 1143.

Background Breast cancer incidence is rising in women aged 20-49 years, yet data on trends in mortality for this age group are limited. Mortality varies by several factors, underscoring the need for comprehensive analyses. We, therefore, investigated trends in breast cancer incidence-based mortality (IBM) by race and molecular subtypes. Methods We performed a population-based study of women aged 20-49 years diagnosed with breast cancer between 2010 and 2020 using SEER 17-registry data. We applied IBM method to evaluate mortality trends by molecular subtype and race/ethnicity. We used Joinpoint regression models to identify changes in IBM trends, using annual percent changes (APC). Additionally, we evaluated breast cancer survival rates by race/ethnicity. Results We analyzed data on 11,661 breast cancer deaths among women aged 20-49 years. Overall, IBM declined from 9.70/100,000 in 2010 to 1.47/100,000 in 2020. The declines occurred in all subtypes and racial/ethnic groups, with some differences in the points of inflexion. IBM for luminal A decreased consistently from 2010, with more marked decline from 2017 (APC, -32.88; 95%CI, -55.17 to -21.30). TNBC followed a similar pattern, with marked decline in 2018 (APC, -32.82; 95%CI, -41.47 to -17.79). Non-Hispanic Black (NHB) women had the highest IBM in 2010 (16.56/100,000) and 2020 (3.41/100,000) compared with Non-Hispanic White (NHW) women (9.18/100,000 in 2010 and 1.16/100,000 in 2020). Declines in IBM in NHB women become pronounced from 2016 (APC, -24.15; 95% CI -34.36 to -17.92), narrowing the gap with other racial groups. Significant declines in IBM for non-Hispanic Asian/Pacific Islander (API) women started earlier from 2013, compared to 2017 for NHW, non-Hispanic American Indian/Alaska Native, and Hispanic women. Survival analyses showed that NHB women had the worst survival outcomes, while NHW and non-Hispanic API had the best. Conclusion IBM among US women aged 20-49 years with breast cancer declined significantly from 2010 to 2020, with variations by molecular subtype and race/ethnicity. The rapid decline in IBM after 2016-2018 likely reflects advancements in breast cancer management, particularly in novel treatments. Although NHB women still have the highest IBM, the differences have narrowed in recent years. Efforts should continue to focus on further enhancing access to early screening for high-risk women and effective treatment to fully mitigate racial disparities in mortality. Citation Format Yueshiu Lyu, Suleepon Uttamapinan, Adetunji T. Toriola. Trends in breast cancer incidence-based mortality among US women aged 20-49 by molecular subtypes and race [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 6207.

Background Neoadjuvant and adjuvant immune checkpoint inhibitors added to SOC (surgery + postoperative radiotherapy [PORT] ± concurrent chemotherapy) yielded promising efficacy results in participants (pts) with LA HNSCC in early phase studies. The randomized, open-label, phase 3 KEYNOTE-689 study (NCT03765918) evaluates neoadjuvant and adjuvant pembrolizumab + SOC vs SOC in this population. Methods Adults with newly diagnosed resectable LA HNSCC (larynx/hypopharynx/oral cavity stage III/IVA; oropharyngeal stage III/IVA p16− or stage III T4 N0-2 p16+) were randomized 1:1 to 2 cycles neoadjuvant and 3 cycles concurrent (during PORT) and 12 cycles adjuvant pembrolizumab 200 mg IV Q3W + SOC vs SOC. SOC included surgery for all pts + PORT 60 Gy in 30 fractions for low-risk, PORT 66 Gy in 33 fractions + 3 cycles concurrent cisplatin 100 mg/m2 Q3W for high-risk, and PORT 70 Gy in 35 fractions + cisplatin for gross residual disease. The primary endpoint is event-free survival (EFS) per RECIST 1.1 by blinded independent central review. Key secondary endpoints are major pathological response (mPR; ≤10% invasive SCC) by blinded independent pathologist review and overall survival (OS). Efficacy endpoints are sequentially assessed in 3 populations: pts with tumors with PD-L1 combined positive score (CPS) ≥10, CPS ≥1, and all pts. Treatment-related adverse events (TRAEs) are graded per CTCAE v4.03. Results From December 2018 to October 2023, 363 pts were randomized to pembrolizumab + SOC and 351 to SOC. As of 25 July 2024 (first interim analysis), median follow-up was 38.3 months (range, 9.0-66.5). Baseline demographics were balanced between arms. The CPS ≥10 population included 234 pts in the pembrolizumab + SOC arm and 231 in the SOC arm; the CPS ≥1 population included 347 and 335 pts, respectively. EFS (CPS ≥10: median 59.7 vs 26.9 months, HR 0.66, 95% CI 0.49-0.88, P=.00217; CPS ≥1: 59.7 vs 29.6 months, HR 0.70, 95% CI 0.55-0.89, P=.00140; all pts: 51.8 vs 30.4 months, HR 0.73, 95% CI 0.58-0.92, P=.00411) and mPR rate difference (CPS ≥10: 13.7%, 95% CI 9.7-18.7, P<.00001; CPS ≥1: 9.8%, 95% CI 7.0-13.3, P<.00001; all pts: 9.3%, 95% CI 6.7-12.8, P<.00001) analyses were statistically significant with pembrolizumab + SOC vs SOC in all prespecified populations. Additional follow-up for OS is ongoing. Grade ≥3 TRAE frequency was similar (44.6% with pembrolizumab + SOC vs 42.9% with SOC); 4 and 1 deaths occurred due to TRAE, respectively. Immune-mediated AEs occurred in 43.2% of pts with pembrolizumab + SOC, most commonly hypothyroidism (24.7%). Conclusions Adding neoadjuvant and adjuvant pembrolizumab to SOC significantly improved EFS and mPR rate difference in pts with resectable LA HNSCC independent of CPS. The safety profile of pembrolizumab was consistent with expectations. Citation Format Ravindra Uppaluri, Robert I. Haddad, Yungan Tao, Christophe Le Tourneau, Nancy Y. Lee, William Westra, Rebecca Chernock, Makoto Tahara, Kevin Harrington, Arkadiy L. Klochikhin, Irene Braña, Gustavo Vasconcelos Alves, Brett G.M. Hughes, Marc Oliva, Iane Pinto Figueiredo Lima, Tsutomu Ueda, Tomasz Rutkowski, Ursula Schroeder, Paul-Stefan Mauz, Thorsten Fuereder, Simon Laban, Nobuhiko Oridate, Aron Popovtzer, Nicolas Mach, Yevhen Korobko, Diogo Alpuim Costa, Anupama Hooda-Nehra, Cristina P. Rodriguez, R. Bryan Bell, Cole Manschot, Kimberly Benjamin, Burak Gumuscu, Douglas Adkins. Neoadjuvant and adjuvant pembrolizumab plus standard of care (SOC) in resectable locally advanced head and neck squamous cell carcinoma (LA HNSCC): Phase 3 KEYNOTE-689 study [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_2):Abstract nr CT001.