Tumor microbes are increasingly recognized for modulating tumor behavior and therapy responses. Intratumoral microbial burden (ITMB) analysis across cancers revealed regulation of immune pathways, and activated mast cells, mostly in colorectal (CRC) and gastric (STAD) cancers. High ITMB CRC leads to interferon regulation and is associated with improved outcomes in advanced disease. Single-cell sequencing revealed induction of interferon-related genes (IRGs) within microbes-containing human CRC. GI-luminal mismatch repair deficiency (MMRd) tumors had higher ITMB than proficient tumors (MMRp). In a rectal MMRd cohort with 100% remission after immune checkpoint blockade (ICB), tumor microbes and microbes-containing mast cells increased. In ICB-sensitive syngeneic murine MMRd tumor models, local tumor microbial depletion, impaired ICB efficacy while downregulating IFN signaling. Forced upregulation of IRGs in ADAR1-deficient cancer cells restored immunotherapy responses during microbial ablation. These data highlight dynamic interplay between ITMB, host defense, and immunogenicity which seems key to determine therapy responses.

Supplementary Figure 1: Number of cancer cases and deaths across indications in 2020, Millions

Abstract Background: Appendiceal adenocarcinomas (AA) are exceptionally rare malignancies that exhibit high rates of fatal peritoneal metastases with few effective treatment options. AAs have frequent MAPK signaling variants targetable with tyrosine kinase inhibitors. However, the efficacy of EGFR, BRAF, and KRAS inhibitors in AA is unknown. Methods: We performed a retrospective analysis of 1,505 patients with AA treated at Memorial Sloan Kettering (MSK) between 7/1993 and 11/2025. We curated clinicogenomic characteristics and treatment-specific outcomes for every patient who received EGFR-inhibitors[i], BRAF-V600Ei, or KRASi for metastatic AA. From the initial dataset, we also defined two separate propensity-matched comparison cohorts of patients treated with conventional non-MAPKi treatments to compare outcomes (1) in the second line (FOLFOX, FOLFIRI) or (2) treatment-refractory setting (TAS-102, regorafenib). Cohorts were generated using the nearest-neighbor method to balance patient sex, tumor histology, grade, KRAS and BRAF status between the main MAPKi cohort and each comparative cohort. Patient progression-free-survival (PFS) and overall survival (OS) after treatment initiation were assessed with the Kaplan-Meier curves with significance defined using log-rank and/or Restricted Mean Survival Time (RMST) methods when appropriate. Results: Forty-seven patients with metastatic AA were treated at MSK with EGFRi (n = 42), BRAF-V600Ei (n = 3), and KRASi (n = 2) primarily in the third line (66%, n=31/47). The cohort encompassed mucinous (55%), goblet cell colonic-type (40%) and colonic-type adenocarcinomas (4%) with predominantly poor differentiation (68%). EGFRi demonstrated limited efficacy with median PFS of 2.9 months and best tumor outcomes of progression (67%, n=28/42), stability (29%, 12/42), and regression (5%, 2/42) irrespective of treatment line, chemotherapy partner, and KRAS-mutation(mut) status. Two out of three patients treated with BRAF-V600Ei responded (67%), although responses were not durable (median PFS 3.5 months). One of the two patients treated with KRAS inhibitors exhibited exceptional ongoing disease control of 14.0 months while the other patient progressed within 2 months. Patients treated with EGFRi or BRAFi in the second line exhibited significantly worse 12-month PFS (p=0.03) and OS (p=0.003) compared to patients with propensity-matched clinicogenomic features treated with second-line FOLFIRI or FOLFOX. In a separate analysis of later-line therapy given to chemo-refractory patients, outcomes did not differ between MAPKi and regorafenib, but TAS-102 demonstrated the overall superior PFS (p=0.04). Conclusions: EGFR inhibitors are largely ineffective in metastatic appendiceal adenocarcinoma, whereas BRAF-V600E and KRAS inhibitors demonstrate early signs of clinical activity in a small subset of patients. Findings underscore the value of genomic profiling and future prospective trials for BRAF and KRAS inhibitors in AA. Citation Format: Somer Abdelfattah, Nayva Vemula, Tina Gowda, Efsevia Vakiani, Jinru Shia, Anne K. Koehne de Gonzalez, Henry Walch, Georgios Karagkounis, Nikolaus Schultz, Michael Berger, Rona Yaeger, Miteshkumar Patel, Garrett Nash, Andrea Cercek, Luis A . Diaz Jr, Michael B. Foote. Clinical benefit of MAPK inhibition in appendiceal adenocarcinoma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Breaking Barriers in the Fight against Rare Cancers; 2026 Jul 18-20; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86(14_Suppl):Abstract nr A045.

Quantitation of tumor-infiltrating lymphocytes

Supplemental Table 2: Patients with stage IV MMRd sporadic CRC from the main MSK cohort.

Microsatellite instability (MSI) is a predictive biomarker in several tumor types. However, many next-generation sequencing-based callers require matched normal samples, reference panels, or pretrained models, limiting their portability across assays and sequencing centers. We developed PROMIS (PROfiling of Microsatellite InStability), a tumor-only, reference-free pipeline that uses a discrete mixture model to characterize intrasample repeat-length distributions at predefined microsatellite loci. Locus-level classifications are then aggregated into a continuous MSI score. We benchmarked PROMIS in colorectal (CRC), endometrial (UCEC), and gastric (STAD) cancers from The Cancer Genome Atlas. PROMIS achieved an overall area under the receiver operating characteristic curve (AUC) of 0.995 and cohort-specific AUCs of 1.00 in CRC and stomach adenocarcinoma and 0.999 in uterine corpus endometrial carcinoma, comparable to established tools despite not using matched normals or pretrained models. Subsampling demonstrated robust performance with substantially fewer loci. In silico dilution showed progressively reduced MSI-microsatellite-stable discrimination, with the pooled AUC declining from 0.83 at 10% tumor fraction to 0.53 at 1%. At low tumor fractions, tumor-type-specific baseline microsatellite variability increasingly influenced PROMIS scores. Finally, in prostate and CRC cell-free DNA cohorts, including Illumina TSO500 data and an 18-gene panel, PROMIS yielded MSI scores concordant with orthogonal tissue- and panel-based classifications across the evaluated Illumina-based sequencing contexts. Accordingly, the present validation should be considered limited to Illumina-based sequencing platforms. PROMIS is intended to complement existing genomic profiling workflows by enabling MSI assessment from sequencing data already generated for broader molecular analyses. Prospective clinical validation remains necessary before clinical implementation.

Abstract Background: The standard treatment for locally-advanced rectal adenocarcinoma is trimodality therapy with neoadjuvant chemotherapy and chemoradiation, followed by surgery. A subset of these cases overexpress HER2; therapies that target HER2 signaling may improve therapeutic responses and organ preservation. Methods: We conducted a prospective investigator initiated single-arm study for patients with stage 2-3, HER2-overexpressed, RAS-wild-type, mismatch-repair-proficient rectal cancer. Patients received 6 weeks of induction HER2 targeted therapy with trastuzumab and tucatinib and then continued trastuzumab and tucatinib with the addition of 15 weeks of FOLFOX or CAPOX. Patients with a complete clinical response (cCR) transition to active surveillance without radiation or surgery. Patients with a non-cCR proceeded with SOC chemoradiation or surgery. The primary endpoint is cCR rate that exceeds 40% at completion of trastuzumab/tucatinib/chemotherapy. Overall response rate (ORR) is a key secondary endpoint. Results: Nine patients with stage II (n=2) and III (n=7) rectal adenocarcinoma with at least 2+ expression of HER2 by immunohistochemistry and Fluorescence in Situ Hybridization HER2-amplification were enrolled. Eight of nine patients completed 21 weeks of neoadjuvant therapy with combination HER2/chemotherapy. The study achieved an ORR of 78% (n=7/9) after 6 weeks of trastuzumab and tucatinib and 75% (n=6/8) after 15 additional weeks of combination HER2/chemotherapy. Four patients out of 8 exhibited a cCR after induction trastuzumab/tucatinib/chemotherapy and an additional patient achieved a cCR with subsequent radiation. All remain alive without metastatic recurrence after a median follow-up of 26 months (95% CI 15 mo - not reached). Only patients with 3+ HER2 expression achieved a cCR. Conclusion: Genomic HER2 amplification in early stage rectal adenocarcinoma predicts for complete clinical response to HER2 targeted therapy in combination with chemotherapy, which may guide non-operative management in this patient population. Citation Format: Michael B. Foote, Jinru Shia, Callahan Wilde, Marinela Capanu, Joanne Chou, Vetri Sudar Jayaprakasam, Marc Gollub, Miteshkumar Patel, Mina Ito, Leonard B. Saltz, Devika Rao, Neil Segal, Maliha Nusrat, Karuna Ganesh, Paul Romesser, Julio Garcia-Aguilar, Martin Weiser, Rona Yaeger, Luis A. Diaz, Andrea Cercek. Genomic HER2 amplification predicts for complete clinical response in a phase II study of induction tucatinib and trastuzumab combined with chemotherapy in locally advanced rectal adenocarcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(8_Suppl):Abstract nr CT011.

Abstract Tumor resident microbes are a well-recognized component of the tumor microenvironment. Microbial subcellular location across tumors along with their functionality remains to be determined. Bulk microbial profiling techniques lack subcellular and spatial resolution and ultimately cannot distinguish between microbial signals or live microbial presence. To address these limitations, we performed orthogonal methods for functional microbial-host characterization. We first developed advanced quantitative fluorescent imaging methodologies that allows visualization of microbial cellular compartmentalization across three different tumor types (total n=30). Using this methodology, we performed spatial microbial transcriptomics at the regional and single cell levels to determine microbial distribution and to interrogate microbial regulation of tumor cell signaling in human pancreatic tumors (n=55). To confirm presence of viable microbes, we performed multiplexed culturomics of patient tumors and normal adjacent tissue specimens (n=80), followed by Whole Genomic Sequencing (WGS) analysis. We tested the effect of the isolated clinical strains on tumor cell signaling pathways with in vitro co-culture assays, and upon genetic fluorescent labelling we defined their role on in vivo tumor growth in murine models. These experiments confirmed their role in promoting tumors, driving resistance to therapeutics and modulation of host signaling mechanisms. Overall, our results identified several pathways under microbial regulation within cancer cells that can drive immune evasion through impaired antigen presentation. In summary, using multiple complimentary novel methodologies we characterize the microbial niche of tumors (MiNT) that uncover microbial regulation of host cell signaling and patient outcomes. Microbial modulatory approaches may be needed to reverse resistance to therapies in pancreatic cancer. Citation Format: Vidhi Chandra, Le Li, Seyda Baydogan, Fuduan Peng, Thais Bartelli, Haoyue Liu, Fernando Jimenez-Arancon, David Romanin, Javier A. Gomez, Steven Maron, Erick M. Riquelme, Mark Hurd, Anirban Maitra, Luis A. Diaz, Ismet Sahin, Adriana Paulucci-Holthauzen, Jared K. Burks, Huamin Wang, Jay Kolls, James R. White, Linghua Wang, Michael P. Kim, Florencia McAllister. Functional interrogation of pancreatic cancer resident microbes reveals their role in host modulation [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 4908.

Genomic alterations detected in plasma cfDNA and tumor tissue

PURPOSE:Microsatellite stable (MSS) colorectal cancers, in contrast to microsatellite instability-high colorectal cancers, have few mutations and are insensitive to immune checkpoint blockade (ICB). Colorectal cancers treated with targeted agents often acquire a high number of genomic alterations at progression. We asked whether targeted therapy could be used to generate a high tumor mutational burden (TMB) in MSS colorectal cancer and sensitize these tumors to ICB. EXPERIMENTAL DESIGN:In patients with MSS metastatic colorectal cancer treated with targeted therapy, we evaluated baseline and progression TMB and response to ICB for patients whose tumors developed high TMB. We determined types of alterations, mutational signatures, neoantigenicity, and clonality associated with emergent genomic alterations in cases of acquired high TMB. RESULTS:Among 26 cases, nine acquired high TMB at progression. Three of these patients received ICB but none had a response. In the TMB-high cases, we found no induction of tumor-infiltrating lymphocytes or PD-L1 expression. Acquired genomic alterations consisted predominantly of single-nucleotide variants, were enriched for single base substitution 17a/b mutational signature, and did not enhance predicted MHC class I binding. TMB was higher in plasma, driven by highly subclonal acquired alterations, compared with tissue samples, which harbored few resistance alterations. CONCLUSIONS:A substantial number of MSS colorectal cancers acquire high TMB following targeted therapy. However, this change is not associated with sensitization to ICB. The high TMB is due to subclonal alterations unique to individual disease sites that are inadequate to elicit a robust antitumor immune response. See related commentary by Parseghian and Eluri, p. 999.

Abstract Background: MMRd tumors respond exceptionally to immunotherapy, and pharmacologic MMR inactivation has the potential to improve immunosensitivity of MMR-proficient tumors. We previously showed that temozolomide plus cisplatin (TMZ+CDDP) induces MMRd through MSH2 silencing. However, clinical translation revealed limited efficacy, potentially due to delayed MMR inactivation preventing sufficient tumor mutational burden (TMB) and microsatellite instability (MSI) accumulation on treatment. We hypothesized that compounds enabling rapid MMR inactivation could overcome this limitation. Methods: Preclinical models treated with TMZ+CDDP were analyzed longitudinally for TMB and MSI. Clinical trial data (NCT04457284) combining TMZ+CDDP with nivolumab in metastatic colorectal cancer patients were assessed using serial ctDNA profiling. We performed computational screening of open-source databases to identify compounds inducing rapid MLH1 or MSH2 downregulation (≤3 days) across cancer cell lines. Candidates were studied using CT26 cells transfected with an out-of-frame luciferase-microsatellite reporter, where MSI-induced frameshift mutations restore luciferase expression, enabling real-time MSI monitoring. These repurposed compounds underwent MMR expression analyses. Results: In vivo, TMZ+CDDP induced MSH2 loss only after 4 weeks (W) of treatment, with MMRd genotype recapitulated only after 8W. Clinical trial analysis of 16 evaluable patients revealed that only 5 patients (31%) developed a MMRd-like genotype with gains in TMB and MSI at a median of 8W, which associated with improved survival. Critically, patients failing to increase TMB and/or MSI developed aneuploid gains in MMR genes on treatment suggesting a compensatory resistance mechanism to mutagenesis. These findings indicate that delayed MMR inactivation with TMZ+CDDP limits clinical efficacy. To address this limitation, our drug screen identified 6 compounds inhibiting significantly Msh2 and/or Mlh1 after short course treatment. Three compounds were confirmed to generate sustained bioluminescence increases within 1-2W compared to 4-8W with TMZ+CDDP. Two compounds were showed to completely abolish Msh2 expression by 1 or 2W, while the third reduced expression by 70%. The three other agents showed only transient Msh2 or Mlh1 downregulation without change in bioluminescence. These novel compounds demonstrate substantially accelerated MMR inactivation kinetics compared to TMZ+CDDP. Conclusions: While TMZ+CDDP can induce a MMRd genotype in patients, delayed MMR inactivation and compensatory MMR gene amplification limit therapeutic efficacy. Repurposed compounds enabling rapid MSH2 inhibition may prevent adaptive resistance mechanisms and improve clinical responses when combined with immunotherapy, warranting further preclinical/clinical development. Citation Format: Benoit Rousseau, Miteshkumar Patel, Karthik Rangavajhula, Lin Zhang, David Mieles, James R. White, Oliver Artz, Shrey Patel, Somer Abdelfattah, Neil Segal, Luis A. Diaz. Targeted drug screening identifies novel compounds enabling accelerated mismatch repair deficiency (MMRd) for immunotherapy sensitization [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 241.

Distribution of single-cell KRAS G12C VAF in clinical samples

PURPOSE:Mismatch repair-deficient (MMRd) colorectal cancers are classified based on MMR protein loss and BRAFV600E mutations. BRAF wild-type sporadic MMRd tumors exhibit a diverse landscape of alternative oncogenes, including gene fusions, with unclear biological and clinical significance. We evaluated mutually exclusive subtypes of sporadic MMRd tumors defined by oncogenic mitogen-activated protein kinase (MAPK) variants and gene fusions to determine the relationship among predominant genomic driver, MMR deficiency mechanism, and clinical outcomes. EXPERIMENTAL DESIGN:We assessed 6,789 patients with colorectal cancer sequenced by MSK-IMPACT to identify 518 patients with sporadic MMRd colorectal cancer. We defined mutually exclusive oncogenic alteration subtypes and then assessed differences in allele-specific MMR-inactivating events, co-occurring oncogenic variants, and patient outcomes. We validated findings in an Italian cohort (n = 69). RESULTS:We identify 4 sporadic MMRd colorectal cancer subtypes: (i) oncogenic fusion-positive, (ii) RAS-mutant (mut), (iii) BRAFV600E-mut, and (iv) MAPK/fusion driver-negative. These mutually exclusive subtypes were associated with conserved molecular lineages of MMR gene inactivation and WNT signaling variants. Oncogenic fusions were disproportionately prevalent in non-Caucasians, among nonsmokers, and in the transverse colon, compared with subtypes that were enriched in smokers (BRAF-mut) and male, younger patients (RAS-mut and MAPK/fusion-driverless). Oncogene-defined molecular lineages were strong predictors of patient outcomes and response to immunotherapy and tyrosine kinase inhibition for metastatic disease. Fusion-positive patients demonstrated improved survival compared with BRAF-mut cancers and benefited from immunotherapy and fusion inhibitors. MAPK/fusion driver-negative tumors were aneuploid, responded poorly to immunotherapy, and were sensitive to EGFR blockade. CONCLUSIONS:Overall, MAPK and fusion oncogenic drivers distinguish MMRd colorectal cancer molecular lineages that inform molecular and clinical phenotypes.

4078 Background: ICI achieves deep and durable responses in patients with dMMR and microsatellite instability-high (MSI-H) EGC, creating an opportunity for non-operative management (NOM) of localized disease. However, limited data exists regarding long-term outcomes with NOM and factors associated with ICI failure, which we evaluated in this retrospective cohort study. Methods: We identified patients with EGC managed at MSK from 2007-2025 whose tumors were dMMR by IHC or MSI-H by next generation sequencing and analyzed patient/tumor characteristics and clinical outcomes in those who received curative-intent therapy (systemic therapy and/or surgery; ICI was first introduced in 2018). Clinical complete response (cCR) was defined as the absence of disease by available clinical, radiographic, and endoscopic assessment. Results: A total of 211 patients with localized dMMR/MSI-H EGC were identified, with median follow-up of 48 months (IQR 24-77). Primary tumor sites were stomach (71%), GE junction (17%), and esophagus (12%). Among 122 patients who underwent germline testing, 15 (12%) had Lynch syndrome. Initial treatment was upfront surgery in 85 patients (40%), chemotherapy (chemo) in 68 (32%), ICI alone in 44 (21%), and chemo/ICI in 14 (7%). Among 58 patients treated with ICI +/- chemo, 27 (47%) achieved a cCR. 6 patients underwent surgery despite cCR, with pathologic complete response (pCR) rate of 4/6 (67%); pCR rate in those without cCR was 3/16 (19%), yielding an overall pCR rate of 32%, higher than the pCR rate observed with chemo alone (5/48 [10%]). 21 patients elected for NOM after cCR; among those with ≥1 year of follow-up, 16 of 17 (94%) remain alive and surgery-free at 1 year. cCR rates were similar between those receiving ICI alone (45%) and chemo/ICI (50%). Notably, 2/2 (100%) patients treated with dual ICI therapy achieved cCR. Lower cCR rates were observed in patients with shorter time on ICI therapy (5/17 [29%] with ≤3 months of ICI vs 14/30 [47%] with > 3 and < 6 months vs 8/11 [73%] with ≥6 months) and in those with clinical nodal involvement (6/17 [35%] vs 14/19 [74%] in N0); 3 patients who underwent surgery had pCR at tumor but residual nodal disease. cCR rate was also numerically lower in the 19% of patients with MMR heterogeneity, defined by focal loss of MMR proteins on IHC or discordant MMR status between tissue specimens (4/11 [36%] vs 23/47 [49%] in non-heterogeneous cases). Conclusions: In this MSK cohort, ICI induced deep responses in nearly half of patients with localized dMMR/MSI-H EGC. Patients achieving cCR who pursued NOM demonstrated excellent outcomes, with most remaining surgery-free at 1 year, supporting the viability of NOM after careful multidisciplinary review.

Supplementary Figure 2: Number of cancer cases and deaths across geography in 2020, Millions

Supplementary Table 1: Authors and affiliations

PURPOSE:Mutational data from multiple solid and liquid biospecimens of a single patient are often integrated to track cancer evolution. However, there is no accepted framework to resolve if individual samples from the same individual share variants due to common identity versus coincidence. EXPERIMENTAL DESIGN:Utilizing 8,000 patient tumors from The Cancer Genome Atlas across 33 cancer types, we estimated the background rates of co-occurrence of mutations between discrete pairs of samples across cancers and by cancer type. We developed a mutational profile similarity (MPS) score that uses a large background database to produce confidence estimates that two tumors share a unique, related molecular profile. The MPS algorithm was applied to randomly paired tumor profiles, including patients who underwent repeat solid tumor biopsies sequenced with Memorial Sloan Kettering-IMPACT (n = 53,113). We also evaluated the MPS in sample pairs from single patients with multiple cancers (n = 2,012), as well as patients with plasma and solid tumor variant profiles (n = 884 patients). RESULTS:In unrelated tumors, nucleotide-specific variants are shared in 1.3% (cancer-type agnostic) and in 10% to 13% (cancer-type specific) of cases. The MPS method contextualized shared variants to specify whether patients had a single cancer versus multiple distinct cancers. When multiple tumors were compared from the same patient and an initial clinicopathologic diagnosis was discordant with molecular findings, the MPS anticipated future diagnosis changes in 28% of examined cases. CONCLUSIONS:The use of a novel shared variant framework can provide information to clarify the molecular relationship between compared biospecimens with minimal required input.