e20049 Background: Consolidation durvalumab after chemoradiation improves outcomes in patients with locally advanced NSCLC, but pneumonitis remains a clinically significant toxicity. The impact of taxane-based chemotherapy on pneumonitis risk and survival remains unclear. We evaluated this association in a real-world cohort of NSCLC patients treated with curative-intent chemoradiation and durvalumab. Methods: We conducted a single-center retrospective study of patients with NSCLC treated with concurrent chemoradiation followed by durvalumab, including unresectable and select oligometastatic cases treated with definitive intent between 2017-2025. Patients were grouped by receipt of taxane vs. non-taxane based chemotherapy. The primary endpoint was grade ≥2 pneumonitis. Multivariable analysis adjusted for treatment group, smoking status, nodal stage, delivered radiation dose, mean lung dose, and time from radiation completion to durvalumab initiation. Time to grade ≥2 pneumonitis was assessed using cumulative incidence methods, accounting for death as a competing risk. Overall survival (OS) was assessed with multivariable analysis adjusting for age, sex, smoking status, histology, delivered radiation dose, disease stage, and grade ≥2 pneumonitis. Results: A total of 289 patients were included (non-taxane n = 81; taxane n = 208). The median age was 65 years (range 39-87) in the non-taxane group and 70 years (42-86) in the taxane group. The majority of patients were male (58% in non-taxane vs 54% in taxane). Most patients were former smokers (62% vs 71%), with similar proportions of current smokers (23% in both groups). Histology differed between groups: squamous cell carcinoma was more common in the taxane group (45% vs 17%), whereas non-squamous histology was more frequent in the non-taxane group (64% vs 38%). Grade ≥2 pneumonitis occurred in 25% of non-taxane and 24% of taxane patients. On multivariable analysis, taxane use was not associated with pneumonitis risk (OR 1.00, 95% CI 0.54-1.90; p > 0.9). Being a former smoker (OR 2.20, 95% CI 1.07-4.91; p = 0.040) and increasing mean lung dose per 5 Gy (OR 1.74, 95% CI 1.23-2.55; p = 0.003) were independently associated with a higher risk of pneumonitis. Delivered radiation dose, nodal stage, and interval to durvalumab were not significant. Median OS for the cohort was 55 months (95% CI 34-76; p = 0.16), with no difference by treatment group. Taxane use was not significantly associated with OS on multivariable analysis (HR 1.31, 95% CI 0.83-2.08; p = 0.3). Conclusions: In this real-world cohort of patients treated with definitive chemoradiation and durvalumab, taxane-based chemotherapy was not associated with increased pneumonitis risk or worse survival. Mean lung dose and former smoking status were key predictors of pneumonitis, emphasizing the importance of radiation optimization and individualized risk assessment.

Background:KRASG12C inhibitors such as adagrasib and sotorasib have shown promise in treating KRASG12C- mutant non-small cell lung cancer (NSCLC), but treatment resistance remains a major challenge. In this study, we investigated the therapeutic potential of combining trastuzumab deruxtecan (T-DXd), an anti-human epidermal growth factor receptor 2 (HER2) antibody-drug conjugate (ADC), with sotorasib in KRASG12C -mutant NSCLC xenografts and evaluated HER2 expression in NSCLC patient samples. Methods:HER2 expression in 11 xenograft models and in 191 clinical and autopsy samples from 31 patients with advanced-stage NSCLC was assessed using breast cancer (BC) and gastroesophageal adenocarcinoma (GEA) HER2 immunohistochemistry (IHC) interpretation guidelines. Responses to sotorasib, T-DXd, and their combination were evaluated in therapy-naïve and sotorasib-relapsed (SR) xenografts. Results:The sotorasib-T-DXd combination induced deeper and more durable tumor regressions than either monotherapy, including in SR xenograft models. Enhanced activity was associated with sotorasib-induced adaptive HER2 upregulation, with stronger HER2 expression in SR tumors correlating with greater responses. In patient samples, HER2 expression was more frequent in KRAS-mutant NSCLC compared with tumors harboring other oncogenic drivers although the difference was not statistically significant. Discordance in HER2 IHC scoring was observed between BC and GEA interpretation guidelines. Conclusions:Sotorasib plus T-DXd produced durable regressions in KRAS G12C-mutant NSCLC xenografts, including SR tumors. Given the observed HER2 expression in a subset of patient tumors, this combination is promising and it is currently under investigation (NCT07012031). Our findings also highlight the limitations of applying current HER2 IHC interpretation guidelines to NSCLC, underscoring the need for optimized assays to guide patient treatment selection.

e20538 Background: Programmed death-ligand 1 (PD-L1) immunohistochemistry (IHC) is commonly used to guide immunotherapy selection in advanced non–small cell lung cancer (NSCLC); however, its ability to consistently reflect tumor immune biology and treatment benefit remains limited. RNA-based immune gene expression profiling provides a complementary approach to characterize the tumor immune microenvironment using formalin-fixed paraffin-embedded specimens. We evaluated associations between RNA-based immune gene expression and duration of immunotherapy (IO) in advanced NSCLC. Methods: This retrospective biomarker study included patients with stage IV, non– EGFR , non– ALK NSCLC treated with immune checkpoint inhibitors. Tumor RNA expression profiling was performed using a multiplex, amplification-free gene expression assay quantifying 204 genes. Gene expression was analyzed as categorical variables (high expression defined as log 2 ≥1) and as continuous measures. The primary endpoint was duration of IO, defined as time from initiation of IO until progression of disease as documented by treating physician, initiation of alternative therapy, or death. Overall survival (OS) was evaluated as an exploratory secondary endpoint. Associations were assessed using univariate duration analyses, Kaplan–Meier methods, and Cox proportional hazards models. All tests were two-sided, and p<0.05 was considered significant. Analyses were exploratory and hypothesis-generating. Results: Sixty-one patients were included. Median duration of IO was 108 days (range, 7–2006). PD-L1 IHC categories (<1%, 1–49%, ≥50%) were not associated with IO duration (log-rank p=0.538). In contrast, multiple RNA-based tumor immune gene expression markers were significantly associated with duration of IO in Cox models (Table). Higher dichotomized (categorical) expressions of CDK6 , ERBB3 , MDM2 , PCSK9 , and STK11 were associated with shorter IO duration, whereas STK11 mutation status was not associated with IO duration. Of these genes, high CDK6 and PCSK9 expressions were also associated with worse OS. Conclusions: RNA expression profiling identified multiple biomarkers associated with IO duration and, for select genes, OS that were not captured by PD-L1 IHC or DNA-based mutation status. These findings suggest RNA-based assays may provide complementary biologic correlates of IO benefit in advanced NSCLC and warrant prospective validation. RNA Biomarker IO duration Hazard Ratio HR (95% CI) p Value OS HR p Value CDK6 3.25 (1.24-8.50) 0.017 28.86 (7.31-114.02) <0.001 ERBB3 2.19 (1.12-4.27) 0.022 0.70 (0.27-1.83) 0.767 MDM2 2.10 (1.08-4.08) 0.030 1.78 (0.79-4.01) 0.166 PCSK9 2.67 (1.16-6.13 0.021 3.64 (1.44-9.16) 0.006 STK11 2.57 (1.08 -6.12) 0.033 1.16 (0.40-3.33) 0.787

Abstract Background: Harmonizing patient longitudinal data is critical to uncovering variables and events that can influence outcomes or molecular data, yet existing tools have significant limitations in integrating multilayered time-series data, particularly in linking treatment events with survival outcomes. Due to their observational nature, real-world data (RWD) can be comprehensive and heterogeneous, posing a challenge when visualizing and interpreting the data. We developed ShinyEvents, an open-source tool and application to facilitate interaction and exploration of longitudinal data, which we demonstrate in the application of the AACR Project GENIE a global consortium that pools real-world cancer genomic and clinical data to advance precision oncology. Methods: ShinyEvents is a web-based framework that allows users to upload longitudinal data and generate interactive patient timelines to view clinical events and perform cohort-level analyses through treatment clustering and endpoint assignment. The tool provides informative cohort visualizations, such as a Sankey diagram of the treatment line, swimmer diagrams of the clinical course and treatment duration, as well as heatmaps to view unsupervised clustering on patient treatments. Our tool can infer real-world progression-free survival (rwPFS) based on user-defined endpoints and perform Kaplan-Meier and Cox proportional hazards regression analysis. We incorporated the AACR Project GENIE data on non-small cell lung cancer (NSCLC) and colorectal cancer (CRC) into a dedicated wed instance to visualize and interact with the data. The application is publicly accessible at the following link: https://shawlab-moffitt.shinyapps.io/ShinyEvents_AACR_GENIE/. Conclusions: ShinyEvents provides a unified framework integrating longitudinal real-world data with survival analytics to facilitate transparent and reproducible collaboration between clinicians and data scientists. Based on the GENIE data, the tool is able to provide dynamic longitudinal visualization on the patient treatment regimen and relate back to the molecular data by identifying complexities surrounding sample collection in relation to treatment regimens. This standardized approach to RWD analysis will facilitate additional collaboration across the global GENIE network. Citation Format: Alyssa Obermayer, Joshua Davis, Roger Li, Rodrigo Rodrigues Pessoa, Brandon J. Manley, G. Daniel Grass, Aik Choon Tan, Dung-Tsa Chen, Timothy Shaw. ShinyEvents: Harmonizing real-world longitudinal data for clinical insights and survival analytics [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 9.

Abstract Purpose: This study aims to assess whether early-onset, grade 1 (G1) and low-grade (LG) treatment-related adverse events (TrAEs) can serve as predictive markers for favorable survival outcomes in advanced non-small cell lung cancer (NSCLC) across different treatment modalities. Methods: We analyzed data from 577 NSCLC patients across 11 cohorts treated at Moffitt Cancer Center: 5 immunotherapies (n=383), 3 targeted therapies (n=88), and 3 chemotherapies (n=106). Data for analysis used AE data derived from Common Terminology Criteria for Adverse Events (CTCAE v4-5), treatment response including comparison of responder (complete response (CR) and partial response (PR)) versus non-responder (stable disease (SD) and progressive disease (PD)) and comparison of disease control (DC: CR/PR/SD) versus PD using Wilcoxon two-sample test, progression-free survival (PFS) and overall survival (OS) using Kaplan-Meier survival curve with log-rank test. Our analytic approach leveraged multiple AE parameters to develop a set of innovative AE biomarkers. Early-onset G1/LG TrAEs were defined as those occurring within 30 days of treatment initiation. Results: Early-onset AE analysis across all treatment types revealed that (a) Immunotherapy had lower frequency of G1 and LG TrAEs compared to chemotherapy and targeted therapy; (b) higher frequency of G1 and LG TrAEs were associated with better treatment response in immunotherapy (responder vs non-responder with p=0.047 (G1) and 0.069 (LG); DC vs PD with p=0.005 (G1) and 0.018 (LG)), but no significant results in chemotherapy and targeted therapy; (c) For patients who did not encounter HG non-treatment related AEs (non-TrAEs), if they frequently experienced G1 and LG TrAEs, their survival outcomes tended to be better compared to the ones with less or no AE experiences in immunotherapy (median PFS: 5.5 vs 3.5 months with p=0.03 for G1 and 5.5 vs 3.3 months with p=0.015 for LG; median OS: 18.2 vs 12.2 months with p=0.008 for G1 and 16.1 vs 12.2 months with p=0.04 for LG). Please note that non-TrAEs represent a strong surrogate for compromised baseline health status. Without proper adjustment, this factor may confound the observed association between early-onset G1 and LG TrAEs and survival outcomes. For in chemotherapy and targeted therapy, both G1 and LG TrAEs did not show significant survival association. Conclusion: G1 and LG TrAEs within 30 days of therapy initiation were associated with better treatment response and improved survival in advanced NSCLC, especially in immunotherapy-treated patients. These findings support the use of early-onset G1/LG TrAE profiles as potential predictive biomarkers. Citation Format: Dung-Tsa Chen, Andreas N. Saltos, Zachary Thompson, Junmin Whiting, Sebastian Viracacha, Timothy I. Shaw, Ignacio I. Wistuba, Jhanelle E. Gray. Early-onset low-grade adverse events as predictive biomarkers in advanced NSCL: A multi-treatment cohort analysis [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 3734.

TPS610 Background: Even with the advance of systemic therapy in gastrointestinal (GI) cancers over the past decade, the therapeutic options are still limited for patients (pts) with colorectal cancer (CRC) and hepatocellular carcinoma (HCC) who are refractory to those standard therapies. Nectin-4 is overexpressed in about 70% of CRC and HCC cases and correlates with worse outcomes; and preclinical study suggested that Nectin-4 expression causes 5-fluorouracil (5-FU) resistance in CRC. Based on these findings, the authors suggest Nectin-4 may be a novel therapeutic target for CRC and HCC. Enfortumab vedotin (EV) is an antibody-drug conjugate (ADC) designed to specifically target Nectin-4 expressing cancer cells. In a landmark phase III trial, EV demonstrated improved overall survival (OS) compared to investigator’s chosen chemotherapy in patients with refractory metastatic urothelial cancer. In this study, we hypothesize that targeting Nectin-4 with EV may have clinical benefits in advanced or metastatic HCC or CRC refractory to current frontline standard therapies. Methods: This study is a multi-indication, open-label, single-treatment arm, two parallel-cohort phase II study of EV in adult participants with advanced or metastatic CRC or HCC who have been previously treated with one or more lines of systemic therapy. A two-stage design using Bayesian predictive probability is used to evaluate overall response rate (ORR). In each cohort, the first stage will enroll 10 participants for futility analysis. If at least one responder is observed, the trial will enroll 10 additional participants for a total of 20. The trial will be considered promising to warrant future study if there are at least 4 responders out of the 20 participants. Estimated sample size is 20 for each cohort. EV at a dose of 1.25 mg/kg will be administered intravenously (IV) on days 1, 8, and 15 of each 28-day cycle. Key inclusion criteria include measurable disease, ECOG 0-1; CRC pts must have received fluoropyrimidine, oxaliplatin and irinotecan with or without biologic agents; HCC pts must have received first-line immunotherapy combination or a multikinase inhibitor. Uncontrolled brain metastases, grade ≥ 2 neuropathy or impaired organ function that poses risk are key exclusion criteria. Primary endpoint is ORR and secondary endpoints include safety, duration of response, progression-free survival and OS. The trial is in progress and registered on clinicaltrials.gov under NCT06553885. Funding and enfortumab vedotin drug support provided from Astellas Pharma Global Development, Inc./Pfizer, Inc. through the Investigator Sponsored Research program. Clinical trial information: NCT06553885 .

Background and Aims: Accurate pre-operative identification of high-risk intraductal papillary mucinous neoplasms (IPMNs) remains a major clinical challenge, particularly for branch-duct (BD) lesions where guideline-based criteria incompletely capture biologic aggressiveness. We investigated whether tumoral mucin expression identifies high-risk IPMN pathology (i.e., high-grade dysplasia or invasive carcinoma) and whether computed tomography (CT)-derived radiomic features can serve as non-invasive biomarkers to enhance pre-operative risk assessment beyond international consensus guidelines (ICG) criteria. Methods: Multiplex immunofluorescence quantified MUC1, MUC2, MUC5AC, and MUC6 expression in tissue microarrays from 101 surgically resected IPMNs classified as low-risk (low-grade dysplasia) or high-risk (high-grade dysplasia or invasive carcinoma). Associations were evaluated using Wilcoxon rank-sum tests, and their discriminatory capability evaluated using receiver operating characteristic curves. For mucins predictive of high-risk pathology, a CT-based ‘radiomic’ signature was developed. Incremental value beyond ICG criteria was evaluated using discrimination metrics and decision curve analysis. Results: Reduced MUC6 expression was significantly associated with high-risk pathology (p = 0.001) and had the highest discriminatory performance (AUC = 0.72). A CT-derived radiomic signature predictive of low MUC6 expression achieved an AUC of 0.75 and, when integrated with ICG high-risk stigmata (HRS), demonstrated improved discrimination and favorable decision-curve characteristics compared with HRS alone, including among BD-IPMNs. Conclusions: Loss of tumoral MUC6 expression is associated with high-risk IPMN pathology and may be approximated using CT-derived radiomic features, supporting the feasibility of non-invasive molecular phenotyping. These findings suggest that integration of molecular and imaging biomarkers with guideline-based criteria may enhance pre-operative IPMN risk stratification; however, prospective external validation in broader surveillance populations and multi-institutional cohorts is warranted prior to clinical implementation.

Tumor infiltrating CD8 + T cells (TILs) progress to a state of terminal exhaustion (Ttex) which have impaired functionality and are nonrenewable. However their precursors (Tpex) are renewable and can generate efficient effector cells. We started from the observation that melanoma patients undergoing therapy with checkpoint inhibitors show increased survival when their T cells have low BCL11B mRNA. In line with this, ablation of Bcl11b in CD8 + TILs conferred a superior anti-tumor response in murine melanoma and ovarian cancer models. Bcl11b KO TILs failed to progress to the Ttex state and retained elevated stemness. Bcl11b exerted its role by repressing expression of essential transcription factors (TF) controlling stemness, and conversely by promoting expression of exhaustion-associated TFs and inhibitory receptor genes, through complex epigenetic control. In addition, Bcl11b KO CD8 + T cells showed increased Ag-specific cytolytic activity and elevated Gzmb and Prf1 proteins, but no increase in their mRNAs, however presented higher expression of genes with role in translation. Furthermore, CRISPR-CAS9-mediated deletion of BCL11B in human TILs from a patient with poor response to adoptive cell therapy with autologous TILs, improved their cytolytic activity and promoted expression of the stemness-associated TF TCF1, underlying its potential therapeutic use. HIGHLIGHTS:Adoptive transfer of Bcl11b KO CD8 + TILs surpasses WT in tumor burden reduction Bcl11b ablation reprograms TILs and impairs the progression to Ttex state Bcl11b KO CD8 + T cells have elevated cytotoxicity and kill only Ag-MHCI targets BCL11B deletion in nonresponder ACT-TIL improves cytolytic activity and elevates TCF1. GRAPHICAL ABSTRACT:

842 Background: Few studies have evaluated the role of early-onset adverse events (AEs) on treatment outcomes for patients with GI cancer. This study aims to develop a pan-GI cancer AE profile to facilitate modernization of treatment strategies. Methods: We evaluated 10 study cohorts from Moffitt Cancer Center: 3 in biliary (n=115), 2 in colorectal (n=89), 2 in pancreatic r (n=44), 1 in gastric (n=23), 1 in hepatocellular (n=14), and 1 in pan GI cancers (n=16), for a total of 301 patients. Treatment included 1 cohort with immunotherapy (IO), 2 with targeted therapy (TT), 2 with combination of TT and IO, and 5 with chemotherapy (Chemo)+TT. Data for analysis used CTCAE version 4 AE data, treatment response, progression-free survival (PFS), and overall survival (OS). Our analytic approach leveraged multiple AE parameters to develop a set of innovative AE biomarkers. Early-onset AEs were defined as events that occurred between the first day and day 30 after the first treatment. Results: In the 10 aggregated GI cohorts (n=301), patients experiencing with higher frequency of grade 1 (G1) early-onset treatment related (Tr) AEs tended to improve PFS and OS (p<0.05). Other TrAEs also showed a non-significant trend toward hazard ratio (HR) <1, suggesting a potential but inconclusive benefit for PFS and OS. In contrast, all non-TrAEs had a HR>1, indicating poor survival with significant associations for grade 2 (G2) and grade > 2 (G>2) in OS. Similar findings were observed across the 9 pooled TT cohorts (n=247). For the 3 merged IO cohorts (n=143), patients presenting G1 early-onset TrAEs exhibited improved OS compared to those without AEs (p=0.02). Most grade 1-2 (G1/2) TrAEs showed HR<1 in PFS and OS. Conversely, all non-TrAEs had HR>1 with significance achieved in G2 and higher grade for OS. The 5 combined chemo cohorts (n=107) showed a nonsignificant HR <1 in PFS and OS for most G1/2 TrAEs; conversely, all non-TrAEs were associated with HR>1 with significant effects in G2 or higher grade for OS. Conclusions: This study presents compelling evidence supporting the clinical relevance of early-onset AEs in predicting pan-GI cancer patient outcomes. Specifically, G1 and G1/2 early-onset TrAEs showed robust results as potential indicators of improved OS or PFS in the entire combined cohorts and in the subset analysis for TT and IO. Even the cohorts with Chemo had a nonsignificant HR<1. Timely recognition of treatment response or disease progression is essential for optimizing clinical decision-making. The pan-GI cancer early-onset AEs may serve this function, supporting more precise treatment strategies and improved patient outcomes.

Background: A unique group of non-small cell lung cancer (NSCLC) is driven by alterations in HER2. Early studies, mostly from advanced NSCLC, suggest that NSCLC with HER2 alterations confers a poor prognosis. However, modern studies are scant. Methods: We performed an analysis of a United States-based database on stage I-III NSCLC patients diagnosed during 2019-2024. Cases with complete data on EGFR, ALK, and HER 2 alterations were included. Study cohorts were divided into: EGFR or ALK alterations (EGFR/ALK group), HER2 alterations (HER2 group), or negative for these alterations (triple-negative group). Outcomes of interest were survival. Results: Analyses were performed on data from 3486 patients: 515 patients (15%) in the EGFR/ALK group, 173 patients (5%) in the HER2 group, and 2798 patients (80%) in the triple-negative group. Median disease-free survival (DFS) was 41.3 months, 26.6 months, and 21.2 months, respectively. Median overall survival (OS) was 62.5 months, 63.7 months, and 40.1 months, respectively. Multivariable analysis showed that DFS and OS were significantly worse among the triple-negative group than HER2 group: adjusted HRs 1.44 (95% CI: 1.08-1.90, p = 0.01) and 1.94 (95% CI: 1.25-3.01, p = 0.003), respectively. In the subgroup of patients with HER2 alterations, no significant difference in DFS or OS was found among patients with HER2 mutation, HER2 amplification, or HER2 overexpression in this exploratory analysis. Conclusions: When classified by the status of EGFR, ALK and HER2, early-stage NSCLC patients with HER2 alterations had significantly better DFS and OS than those with triple-negative biomarkers.

Longitudinal data analysis of the patient’s treatment course is critical to uncovering variables that influence outcomes. However, existing tools have significant limitations in integrating multilayered time-series data, particularly in linking treatment events with survival outcomes. Here, we developed ShinyEvents, a web-based framework for complex longitudinal data analysis. ShinyEvents allows users to upload data and generate interactive timelines of clinical events, enabling cohort-level analyses such as treatment clustering and endpoint assignment. It also provides informative cohort visualizations, such as a Sankey diagram of the treatment line and a Swimmer diagram of the clinical course. Finally, our tool can infer real-world progression-free survival (rwPFS) based on user-defined endpoints and perform Kaplan-Meier and Cox proportional hazards regression analysis. With these features, the tool can then associate treatment lines with clinical outcomes. As a case study, we analyzed Moffitt patients with muscle-invasive bladder cancer treated with neoadjuvant chemotherapy followed by surgery. Patients treated with cisplatin and gemcitabine exhibited more favorable rwPFS and overall survival, which is consistent with prior reports. Altogether, ShinyEvents provides a unified framework for integrating longitudinal real-world data with survival analytics, fostering transparent and reproducible collaboration between clinicians and data scientists. A live demo is available at https://shawlab-moffitt.shinyapps.io/shinyevents/.

Adjusted mean change in total skeletal muscle area (tSMA) and psoas skeletal muscle area (pSMA) by clinically relevant covariables in subgroups by edema status.

BACKGROUND:Muscle loss influences pancreatic ductal adenocarcinoma (PDAC) outcomes, but treatment-related edema may cause overestimation of total skeletal muscle area (tSMA), confounding our understanding of muscle changes. However, no studies have quantified the impact of edema on tSMA and psoas skeletal muscle area (pSMA) changes. Thus, we sought to (i) assess the impact of edema on tSMA and pSMA changes between diagnosis and follow-up and (ii) explore the utility of pSMA as a clinically relevant measure of muscle and muscle loss among patients with PDAC. METHODS:Body composition was measured using CT scans at diagnosis and follow-up from 95 patients enrolled in the Florida Pancreas Collaborative cohort study. Edema was assessed by opacifications in subcutaneous fat, and tSMA and pSMA changes were expressed as percent change between diagnosis and follow-up. We used multivariable generalized linear models to estimate mean tSMA and pSMA changes overall and by edema status. Spearman correlation was used to measure interrelationships of tSMA and pSMA. RESULTS:tSMA increased between diagnosis and follow-up (Δ = 0.66) but only in patients with edema (Δ = 3.35) whereas non-edematous patients lost tSMA (Δ = -2.03). Conversely, pSMA decreased regardless of edema status. Furthermore, tSMA and pSMA were strongly correlated overall (r = 0.75) and in non-edematous patients (r = 0.83). CONCLUSIONS:Edema inflated estimates of tSMA at follow-up in patients with PDAC, but pSMA was impervious to edema and may represent a suitable proxy for tSMA. IMPACT:pSMA is a reliable measure of muscle and muscle loss and should be considered in future studies assessing muscle loss in patients with PDAC.

Supplementary Table 1: Analysis of basic clinical and data regarding each cohort.

Intraductal papillary mucinous neoplasms (IPMNs) are common precursors to invasive pancreatic ductal adenocarcinoma (PDAC), with the risk of progression varying by IPMN type and anatomic location. Despite some IPMNs having a high risk of progression, there are limited non-surgical options for precision prevention in patients with IPMNs. One of the main challenges is that existing radiologic and molecular markers are insufficient for reliably assessing the risk of progression, mostly due to a lack of validated intervention targets. Currently, cancer prevention for patients with IPMN is centered on surgery or surveillance using a risk-based strategy; the clinical ability to stratify risk of cancer progression of individual IPMN tumors is poor and essentially no effective non-surgical interventions exist. Thus, the objective of this work is to apply statistical features extraction and use latent features derived from sequencing data as input to machine learning (ML) models to identify unexplored markers driving the progression of IPMNs to invasive PDAC. Data consisting of formalin-fixed, paraffin-embedded tissue cores sampled from 34 unique patients with the following pathological diagnoses: 8 non-IPMN-derived PDAC, 7 IPMN-derived PDAC, 7 high-grade IPMNs, and 12 low-grade IPMNs, were analyzed using the Moffitt STAR 2.0 Cancer Mutation and Molecular Biomarker Profiling panel. This STAR 2.0 next generation sequencing method performed using the TruSight Oncology 500 panel from Illumina, Inc., is designed to interpret sequence information for over 500 somatically altered genes. The analyses of sequencing data incorporated statistical and ML analyses of the mutational profiles of patients’ genomes followed by integration of trinucleotide sequence-derived mutational features. By applying non-negative matrix factorization to DNA trinucleotide motif mutational data, we identified 4 distinct mutational signatures. These signatures exhibited varying degrees of similarity to the Single Base Substitution Signatures from COSMIC (Sanger Institute). However, the contribution of these signatures to the mutational profile in each sample is more complex, as samples may carry a combination of signatures rather than a single, defining signature. Preliminary results from the discriminative ML models showed high performance in predicting type of malignancy (multiclass area under the curve > 0.8), using the mutational counts and each sample-to-signature contribution. We demonstrate that insights extracted from mutational profiles have the potential to enhance the interpretation of mutational patterns and improve the stratification of IPMNs. Further analyses are needed to fully understand the complex interplay of mutational processes across the samples, beyond the initial identification of signatures. We aim to uncover key gene patterns, focusing on codon mutations and their mapping to protein alterations, to better understand the molecular mechanisms driving the progression of IPMNs and identify potential therapeutic targets for early intervention. Aleksandra Karolak, Evan W. Davis, Mouktik Isukapalli, Rohit Veligeti, Margaret A. Park, Jamie K. Teer, Daniel K. Jeong, Kun Jiang, Dung-Tsa Chen, Jennifer B. Permuth, Ghulam Rasool. Mutational profiling and machine learning for risk stratification and biomarker identification in intraductal papillary mucinous neoplasms progressing to pancreatic cancer [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Artificial Intelligence and Machine Learning; 2025 Jul 10-12; Montreal, QC, Canada. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(13_Suppl):Abstract nr A043.

Although depression is reported to be higher among patients with pancreatic cancer than in the general population, research on depression and stress levels in this population is limited. To address this gap, we investigated the prevalence of self-reported depression and lifetime stressor exposure in a cohort of treatment-naïve patients with pancreatic ductal adenocarcinoma (PDAC) or other types of pancreatic tumors such as pancreatic neuroendocrine tumors and intraductal papillary mucinous neoplasms who received care at one of 15 institutions participating in the multi-institutional study Florida Pancreas Collaborative. Depression severity was assessed using the Edmonton Symptom Assessment System-revised (ESAS-r) and the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC), and acute and chronic stressor exposure was assessed with the Stress and Adversity Inventory (STRAIN). PDAC patients reported higher average depression symptom severity at the time of diagnosis and after 6 months compared to non-PDAC patients (p = 0.027 and p = 0.063, respectively). On the other hand, non-PDAC patients experienced a higher mean number and severity of lifetime stressors (p = 0.021 and p = 0.039, respectively) than PDAC patients. Across the sample, greater stressor exposure (measured by stressor count, severity, and event type) was associated with higher odds of clinically significant depressive symptoms. We also observed that chronic stressors were significantly associated with lower odds of advanced disease (OR = 0.896, p = 0.002). Among PDAC patients who completed both STRAIN and ESAS-r (n = 52), greater severity of acute life events was associated with a significant increase in ESAS-r depression scores between baseline and 6-month follow-up (p = 0.015). These findings highlight distinct patterns of depression and stress across pancreatic tumor types and reveal a robust association between lifetime stress exposure and depressive symptoms. Together, they underscore the need for systematic screening and integrated psychosocial support for patients with pancreatic cancer.

PURPOSE:Newer targeted therapy (NTT), addressing alterations in BRAF, MET, NTRK, ROS1, and RET has become a common therapeutic option for non-small cell lung cancer (NSCLC). To date, only RET inhibitor has been shown in a phase III study to confer a survival advantage when used in the frontline setting. This study investigates timing of NTT and its impact on survival using a large, population-base data set. METHODS:We searched a nationwide, electronic health record-derived, deidentified database for patients with advanced NSCLC treated with NTT between May 2014 and March 2023. Time to treatment initiation (TTI) was calculated from the diagnosis of advanced NSCLC. Landmark analytic technique was used to address immortal time bias. RESULTS:Among 857 patients analyzed, the median TTI was 3.8 months. By month 2 or month 3 after diagnosis, patients who already initiated NTT had significantly better survival than those who had not initiated NTT at those time points: Hazard ratio (HR), 0.65 (95% CI, 0.53 to 0.81; P < .001) and HR, 0.69 (95% CI, 0.55 to 0.85; P = .001), respectively. A multivariate analysis indicated that delayed TTI was an independent prognostic factor of decreased survival, along with impaired performance status and squamous cell carcinoma histology. Subgroup analyses excluding RET inhibitor still demonstrated a statistically significant survival advantage in favor of earlier NTT initiation. CONCLUSION:Among advanced NSCLC patients undergoing NTT, survival was significantly better among those who began treatment within 2 or 3 months after diagnosis than those whose treatment was delayed.

Sensitivity analysis scatterplots of tSMA and pSMA.

Breast cancer is a leading cause of cancer-related mortality among women worldwide and is known to have higher mortality among women with African ancestry. Herein, we describe the creation and characterization of a multiethnic breast cancer tissue microarray (ME-BrTMA) representing tumors from non-Hispanic White (n = 41), non-Hispanic Black (NHB; n = 45), and Hispanic patients from Puerto Rico (n = 36) and Florida (n = 52). This ME-BrTMA comprises five blocks with a total of 610 cores: 371 breast cancer tumor cores, 93 breast stromal cores, 96 normal breast tissue cores, 30 non-breast cancer tumor cores, and 20 cores representing normal tissues. Initial characterization of the ME-BrTMA includes standard IHC staining of well-characterized clinical biomarkers, including the estrogen hormone receptors and progesterone hormone receptors, HER2, and Ki-67, interpreted by the coauthoring pathologist (Marilin Rosa). The IHC results indicated good but imperfect alignment with clinical diagnoses. Cores from breast cancer tumors from the NHB cohort most frequently scored negative for estrogen receptor (63%, P < 0.005) and progesterone receptor (80%, P < 0.005) and most frequently have high expression of the Ki-67 proliferation marker (38%, P < 0.05). Prediction Analysis of Microarray 50 (PAM50) analysis using RNA from secondary patient blocks showed that the NHB group also most frequently scored in the basal-like category (61%, P < 0.05). Taken together, the initial characterization of the ME-BrTMA suggests that it may serve as a representative resource to understand the underlying biology of breast cancer and its relationship to patient outcomes. SIGNIFICANCE:The ME-BrTMA described herein provides a resource that may serve as a tool to understand the underlying biology of breast cancer.

Sensitivity analysis changes in tSMA and pSMA bar charts.