Zhao, Yujie

TPS6126 Background: Oral cavity cancers (OCC) are a rare subtype of head and neck cancer (HNC) with poor outcomes and limited second-line treatment options in the recurrent or metastatic (R/M) setting. Treatment-related morbidity and repeated local recurrences in OCC result in significant impairment of function and quality of life. Rural OCC patients comprise a sizable portion (50%) of our practice and are usually of lower socioeconomic status with lack of access to specialist care. They also have multiple barriers to treatment access and ancillary services, translating to poorer clinical outcomes. There is a critical need for effective and accessible therapies for R/M OCC in rural patients. Triple oral metronomic chemotherapy (TOMC) with Methotrexate, Erlotinib and Celecoxib (MEC) suppresses angiogenesis and activates antitumor microenvironment in HNC. Studies using TOMC in R/M OCC outside the US show reasonable efficacy (6-month OS 52.9%), better safety and lower cost compared to standard second line options. We hypothesize that a decentralized clinical trial with TOMC in R/M OCC is feasible, safe and effective in addition to being economical and accessible for rural patients. Methods: MC240701 is a single institution, open label, decentralized, single arm pilot study of MEC in patients with R/M OCC. Eligible pts must be ≥18 years old, have pathologically confirmed R/M OCC not amenable to curative-intent therapy, ECOG 0-2 and adequate organ function. At least 1 measurable lesion by RECIST 1.1 OR non-measurable disease (evident mucosal lesions, CNS/bone disease, lesions not meeting RECIST criteria) is permitted. Patients must have received standard first line immunotherapy with or without chemotherapy OR should be unable/unwilling to receive first line treatment. Key exclusion criteria include patients unable to take pills by mouth, serious medical co-morbidity or autoimmune disease, immunocompromised patients and other active malignancy. Primary endpoint is to demonstrate feasibility of decentralized clinical trials, secondary outcomes include overall response rate, progression free and overall survival and toxicity. Additional goals are to evaluate patient impact of decentralized treatment through surveys and interviews After initial screening and evaluation, patients are remotely consented and will receive study drugs by mail. All patients will receive the same treatment of Methotrexate 9mg/m 2 PO weekly, Erlotinib 150 mg PO daily, and Celecoxib 200 mg PO BID, each cycle will be 28 days. Treatment will continue until disease progression or treatment tolerance, study duration is for 2 years. All subsequent clinical and laboratory monitoring and response assessments (every 3 cycles) will be remotely performed or have an option for local testing. Enrollment began July 2025 and 4 out of planned 25 patients have been accrued. FDA IND:174540. Clinical trial information: NCT06997068 .

e18149 Background: Laryngeal neuroendocrine carcinoma (LNEC) is rare and biologically heterogeneous. The independent prognostic roles of tumor differentiation and stage remain incompletely defined. Methods: Patients diagnosed with LNEC (2000–2022) were identified from the Surveillance, Epidemiology, and End Results (SEER) database. Overall survival (OS) and cancer-specific survival (CSS) were estimated using a Kaplan–Meier analysis. Tumor differentiation (Diff) was harmonized across SEER grade fields: legacy Grade Recode (through 2017) was used when available; otherwise, Grade Clinical/Pathological (2018+) codes (1=well, 2=moderate, 3=poor, 4=undifferentiated) defined Diff. Multivariable Cox regression adjusted for stage, Diff, age, and sex. Cases with missing survival time, stage, or Diff were excluded (n=97). Results: The cohort included 157 patients (median age 63 years; 63% male). Diff was well in 4%, moderate in 29%, poor in 34%, and undifferentiated in 12% of patients. At diagnosis, 22% of patients had localized, 27% had regional, and 28% had distant disease. Treatments included surgery in 31%, chemotherapy in 42%, and adjuvant radiation in 13% of patients. Median OS differed significantly by Diff, ranging from 52.5 months (mo) for well, 90 mo for moderate, 12 mo for poor, and 17 mo for undifferentiated (p<0.001). Five-year OS was 50%, 58%, 23%, and 37%, respectively, with a similar gradient for CSS (p=0.001). Survival remained differentiation-associated within surgical (p=0.004) and chemotherapeutic subgroups (p<0.001). Median OS by stage was 96 mo for localized disease, 30 mo for regional nodes only, 8.5–7.5 mo for regional direct extension ± nodes, and 11 mo for distant disease. On multivariable analysis, stage and differentiation—driven primarily by poorly differentiated tumors—were independently associated with OS. Conclusions: Both tumor differentiation and stage independently predicted survival in LNEC, indicating that tumor biology and anatomic extent each contribute to prognosis. Outcomes remain suboptimal in poorly differentiated and advanced-stage LNEC, underscoring the need for treatment optimization and clinical trial development. Variable HR (95% CI) p Differentiation (Diff) Moderate vs Well 1.76 (0.49–6.31) 0.39 Poor vs Well 6.11 (1.63–22.93) 0.007 Undifferentiated vs Well 2.28 (0.60–8.76) 0.23 Stage Nodes only vs Localized 2.21 (1.14–4.38) 0.023 Direct extension only vs Localized 14.04 (2.44–80.97) 0.003 Direct + nodes vs Localized 7.49 (3.19–17.63) <0.001 Distant vs Localized 3.34 (1.63–6.83) 0.001 Male vs Female 2.28 (1.34–3.89) 0.002

8598 Background: Immune checkpoint inhibitors (ICI) have become a cornerstone of treatment for lung cancer. Glucagon-like peptide-1 receptor agonists (GLP-1 RA) have been associated with improved outcomes in cancer patients in prior studies; however, their impact on outcomes among lung cancer patients receiving immunotherapy remains limited. Methods: We conducted a retrospective cohort study using the TriNetX real-world data platform. Adult patients with lung cancer treated with ICIs, including PD-1, PD-L1, or CTLA-4 inhibitors, were included. Patients were stratified based on exposure to GLP-1 RA. Propensity score matching (1:1) was performed to balance baseline demographics and comorbidities, including type 2 diabetes mellitus and body mass index. Primary outcomes included all-cause mortality, intensive care unit (ICU) admission, and hospitalization. Secondary outcomes included immune-related pneumonitis, colitis/enteritis, cachexia, and major adverse cardiovascular events (MACE). Results: After matching, 2,013 patients were included in each group, with a median follow-up of 372 days in the GLP-1 RA group and 383 days in the non-GLP-1 RA group. Mean age was 66.8 and 67.2 years, respectively, and approximately 50% of patients were male. Most patients were White (77.2% vs 71.8%), followed by Black (14.1% vs 14.0%) and Asian (2.1% vs 1.8%). GLP-1 RA use was associated with a significantly lower all-cause mortality rate compared with non-use (35.7% vs. 53.4%; p < 0.0001), with a hazard ratio of 0.65 (95% CI, 0.59–0.71). Significant reductions were also observed in ICU admissions (22.8% vs 29.0%; p = 0.0001), hospitalizations (64.1% vs 71.8%; p = 0.0001), pneumonitis (5.2% vs 6.8%; p = 0.0457), and MACE (19.9% vs 25.3%; p = 0.0026). No significant differences were observed in colitis/enteritis (9.3% vs 9.4%; p = 0.9704) or cachexia (8.0% vs 9.8%; p = 0.0650). In a subgroup analysis of lung cancer patients not receiving ICI, GLP-1 RA use was similarly associated with lower mortality, ICU admission, hospitalization, and MACE, without differences in immune-related toxicities. Conclusions: GLP-1 RA use among lung cancer patients was associated with improved survival, reduced healthcare utilization, and fewer selected adverse events in patients receiving ICI therapy. These associations likely reflect established metabolic and cardiovascular benefits of GLP-1 RA and may also involve antitumor mechanisms suggested in prior studies. Residual confounding related to healthcare access and socioeconomic factors cannot be fully excluded and should be considered when interpreting these findings. Prospective studies are warranted to validate these observations and clarify underlying biological mechanisms.

6123 Background: Pemetrexed (PTX) is safe and tolerable with responses reported in pts with R/M SGC. Given enhanced responses with PTX and pembrolizumab (PMB) for lung cancer, we hypothesized that PTX and PMB will have activity for SGC. Herein we present the efficacy results in pts with adenoid cystic carcinoma (ACC). Methods: MC200708 is a single arm phase II study of PTX and PMB in pts with R/M SGC (NCT04895735) with 2 cohorts: ACC (cohort A) and non-ACC (cohort B). Key eligibility criteria: ≥18 years, ECOG 0-1, measurable disease. Prior ICI and/or PTX was allowed. Key exclusion criteria: serious comorbidities, autoimmune disease and brain metastases. Simon’s 2-stage design was used for each cohort. Primary endpoint was overall response rate (ORR) for Cohort A but was amended to clinical benefit rate (CBR=SD+PR+CR) due to prolonged SD in several pts. Secondary endpoints: progression free survival (PFS), overall survival (OS), and toxicity. Exploratory analyses in Cohort A1 (post-amendment pts) included PSMA PET imaging and circulating PSMA extravesicles (EVs) as a biomarker. All pts received PTX 500 mg/m2 IV + PMB 200 mg IV q3 weeks until progression or treatment intolerance. Imaging was q3 cycles. Results: 20 pts (11 Cohort A + 9 Cohort A1) were enrolled from August 2021-Feb 2025. All 20 patients were eligible and received ≥ 1 cycle of treatment. Median age was 60.5 yrs, 50% male, performance status 0 (80%) or 1 (20%). 11 pts had no prior therapies (55.0%). The remaining pts had 1 (35%) or 2 (10%) prior lines of therapy. 75% of pts had no prior ICI. In Cohort A, the ORR was 0% (0 of 11). Cohort A1 met criteria for success based on CBR with 1 PR (12.5%) and 5 SD (62.5%) in the first 8 pts. For the whole cohort of 20 patients, the ORR was 5% (95% CI: 0.1-24.9) and the CBR rate was 75% (95% CI: 51-91) with 1 PR and 14 pts with SD. The duration of response for the 1 PR patient was 12.4 months. The median duration of response for the 15 pts with SD was 7.5 (2.0-29.7) months. 3 pts (15%) had PD and 2 pts (10%) went off prior to response assessment. Median cycles of treatment was 4 (2-30). In the 17 pts who discontinued, treatment was stopped in 9 (53%) for PD, 4 (23.5%) due to adverse events (AE), and 4 (23.5%) due to physician/patient preference. With a median follow-up of 14.5 months (1.5-39.2), the 1-year PFS rate is 52.1% (32.6 – 83.2%) and the OS rate is 82.1% (95% CI: 65.6 – 100%). The overall grade 3+ AE rate regardless of attribution was 80% (16/20). The most common grade 3 AEs were lymphocyte count decrease (40%), hypertension (20%), fatigue (10%), and pneumonitis (10%). 2 pts had a grade 4 AE (lymphocyte count decrease and thrombotic thrombocytopenic purpura). There were no treatment-related deaths. Analysis of correlative studies for Cohort A1 with PSMA imaging and EV biomarker ongoing. Conclusions: PTX and PMB has modest activity in pts with R/M ACC with CBR of 75% and prolonged benefit for some patients. Clinical trial information: NCT04895735 .

1615 Background: Immune checkpoint inhibitors (ICIs) have revolutionized lung cancer treatment but cause adverse events, including immune-related colitis (ir-colitis). Accurate identification of ir-colitis is critical for improving patient outcomes and clinical research. However, current methods relying on ICD codes have substantial limitations: non-specific categories that fail to capture the cases or distinguish cases from infectious or ischemic etiologies, as well as incomplete coding that leads to missed cases. Large language models (LLMs), with advanced natural language understanding capabilities, offer a scalable approach for automated phenotyping using unstructured clinical notes. We aimed to develop and evaluate an LLM-based approach for identifying ir-colitis in patients with lung cancer receiving ICIs and compare it with ICD code based method. Methods: We identified 5,278 lung cancer patients with ICI administration at Mayo Clinic (2012-2025). To investigate ir-colitis occurrence within two years after ICI initiation, patients were stratified into three ICD-based cohorts: (1) IR-Colitis, with clinician-assigned ir-colitis-specific diagnosis codes; (2) Non-IR Colitis, with colitis diagnosis codes excluding ir-colitis; and (3) No-Colitis, absent of any colitis-related codes. We used three open-source LLMs to process clinical notes : Llama4, MedGemma, and Qwen. Prompts were designed to extract both gold standard evidence (explicit ir-colitis terminology, such as "immune-mediated colitis" and "checkpoint inhibitor colitis") and supportive evidence (such as pathology and lab results). Results: Llama4 demonstrated superior accuracy among three LLMs in manual evaluation and was thus selected for all cohort analyses. Among ICD-based IR-Colitis patients (n = 339), 82.0% had confirmatory evidence while 18% lacked evidence, suggesting potential false-positive coding. In the Non-IR Colitis cohort (n = 297), LLM-based extraction identified 43.8% patients in fact had ir-colitis, demonstrating substantial misclassification. In the No-colitis cohort (n = 4,642, random sample of 996 for analysis), the LLM identified 10.7% with documented ir-colitis, revealing substantial missing by administrative coding. Conclusions: LLM-based phenotyping demonstrates superior sensitivity for identifying ir-colitis compared to ICD codes alone, successfully uncovering cases missed by administrative coding. This scalable approach enables more comprehensive case identification for immunotherapy safety research and clinical surveillance. ICD-Based Cohorts LLM-Identified ir-Colitis Gold Standard Evidence Supportive Evidence Any Evidence ir-Colitis (n = 339) 277/339 (81.7%) 275/339 (81.1%) 278/339 (82.0%) Non-ir-Colitis (n = 297) 129/297 (43.4%) 126/297 (42.4%) 130/297 (43.8%) No Colitis (n = 996) 106/996 (10.6%) 104/996 (10.4%) 107/996 (10.7%)