ELANTABLY, DINA

e21607 Background: The association between timing of palliative care referral and survival outcomes in patients with melanoma brain metastases (MBM) treated with immune checkpoint inhibitors remains poorly characterized. Methods: We conducted a retrospective cohort study of patients with MBM treated with combination ipilimumab and nivolumab at Mayo 3 sites from 2016 to 2024. Palliative care referral timing relative to MBM diagnosis was categorized as early (≤56 days), late ( > 56 days), or none. Patients without referral documentation were excluded from referral-based analyses. Overall survival (OS) was estimated using Kaplan–Meier methods and compared with log-rank tests. Multivariable Cox proportional hazards models were used to identify factors independently associated with OS, with stepwise selection (entry p < 0.25; stay p < 0.15). Results: Among 147 patients, 139 had referral information: 33 (23.7%) early, 56 (40.3%) late, and 50 (36.0%) no referral. Median OS differed significantly by referral timing (log-rank p < 0.0001): 2.8 months (early), 7.9 months (late), and 47.8 months (no referral). Among patients with ECOG 2–3, a higher proportion had early or late referral (each 42.9%) compared with no referral (14.3%), whereas patients with ECOG 0–1 were more evenly distributed across referral groups (early 21.6%, late 40.0%, none 38.4%). In multivariable analysis of 88 patients with complete covariate data, both early (HR 4.54; p = 0.0016) and late referral (HR 2.86; p = 0.011) were independently associated with worse OS compared with no referral. Progressive disease versus objective response (HR 4.45; p = 0.0002), hemorrhagic metastases (HR 2.79; p = 0.0015), male sex (HR 2.07; p = 0.044), and increasing age (HR 1.02 per year; p = 0.032) were also associated with inferior OS. Treatment with ipilimumab 1 mg/kg plus nivolumab 3 mg/kg showed a trend toward improved OS compared with ipilimumab 3 mg/kg plus nivolumab 1 mg/kg (HR 0.48; p = 0.055). Conclusions: Timing of palliative care referral was strongly associated with overall survival in patients with MBM treated with ipilimumab and nivolumab. Although patients with ECOG 2–3 were more frequently referred early or late, ECOG category overall did not clearly differentiate referral timing groups, suggesting referral may reflect unmeasured clinical factors beyond baseline performance status. Prospective studies incorporating more granular functional and symptom measures are needed to better define the role of palliative care integration in this population. Multivariable cox model for overall survival. Variable Comparison HR P value Referral timing Early vs None 4.54 0.0016 Late vs None 2.86 0.011 Best overall response PD vs ORR 4.45 0.0002 SD vs ORR 2.16 0.098 Hemorrhagic metastases Yes vs No 2.79 0.0015 Sex Male vs Female 2.07 0.044 Age Per year increase 1.02 0.032 Regimen Ipi1+Nivo3 vs Ipi3+Nivo1 0.48 0.055

e20595 Background: Circadian regulation of immune function has been proposed as a modifier of immune checkpoint inhibitor (ICI) efficacy, with retrospective studies suggesting improved outcomes with earlier infusion timing. Whether these findings translate to routine oncology practice, where ICIs are administered across multiple lines, frequently combined with chemotherapy, and subject to treatment interruptions, remains uncertain. We evaluated the association between ICI infusion timing and overall survival (OS) in a real-world cohort of patients with non-small cell lung cancer (NSCLC). Methods: A retrospective cohort study included adults with NSCLC who received at least one line of ICI therapy at a community oncology center between January 2018 and December 2024. Infusion timing was defined at the patient level as morning-only ( < 1300 hours), evening-only (≥1300 hours), or mixed timing, reflecting receipt of infusions exclusively in different time-of-day categories across treatment lines. OS was measured from diagnosis to death or last follow-up. Kaplan-Meier and Cox proportional hazards models were used. Multivariable analyses adjusted for stage, ECOG performance status, body mass index, and premature ICI discontinuation. Line of therapy was excluded due to collinearity with mixed-timing classification and risk of immortal time bias. Results: A total of 127 patients were included (morning-only n = 77; mixed timing n = 36; evening-only n = 14). Staging data were available for 99.2%, and 94 deaths occurred. Median OS was 495 days for morning-only, 850 days for mixed timing, and 388 days for evening-only patients. Kaplan-Meier analysis showed no statistically significant OS differences across groups. In sensitivity analysis excluding patients with mixed infusion timing, morning-only and evening-only groups demonstrated overlapping survival curves, with no evidence of a time-of-day–specific survival advantage. In univariate analysis, mixed timing was associated with a borderline reduction in mortality compared with morning-only infusions (HR 0.63, 95% CI 0.39-1.00), while evening-only timing was not (HR 1.63, 95% CI 0.85-3.16). In multivariate analysis restricted to stage III-IV disease, infusion timing was not independently associated with OS. ECOG performance status, premature ICI discontinuation and advanced staging were the strongest predictors of mortality. Conclusions: In this real-world NSCLC cohort, ICI infusion timing was not independently associated with overall survival after adjustment for patient fitness and treatment continuity. These findings suggest that in pragmatic settings, clinical factors may outweigh subtle chronotherapy effects reported in more selected populations.

9558 Background: Ipilimumab 3 mg/kg plus nivolumab 1 mg/kg (Ipi3+Nivo1) is the standard immunotherapy regimen for metastatic melanoma with brain metastases (MBM) but is associated with significant immune-related toxicity, particularly neurologic toxicity. Ipilimumab 1 mg/kg plus nivolumab 3 mg/kg (Ipi1+Nivo3) has demonstrated improved tolerability in extracranial melanoma, though comparative data in MBM are limited. We compared efficacy and toxicity of these dosing strategies in a multisite real-world cohort. Methods: We retrospectively identified patients with MBM treated with ipilimumab plus nivolumab across three Mayo Clinic sites between 2016 and 2024. Patients were classified by induction dosing as Ipi1+Nivo3 or Ipi3+Nivo1. Overall survival (OS) and progression-free survival (PFS) were estimated using the Kaplan–Meier method and compared using log-rank tests. Best overall response (BOR) and toxicity outcomes were summarized descriptively. Results: Among 147 patients (Ipi1+Nivo3 n=32; Ipi3+Nivo1 n=115), 144 were evaluable for survival. Median age was higher in the Ipi1+Nivo3 group (69.5 vs 66.0 years); males predominated in both groups (71.9% vs 61.7%), and the cohort was predominantly White (>96%). Most patients had ECOG 0–1 in both groups (84.4% in Ipi1+ Nivo3 and 90.4% in Ipi3+Nivo1). BRAF mutation status was balanced between regimens, with approximately half of patients harboring BRAF-mutant disease (51.6% vs 53.2%). Median OS was 18.5 months (95% CI, 5.7–54.7) with Ipi1+Nivo3 versus 10.6 months (95% CI, 6.5–19.1) with Ipi3+Nivo1 (log-rank p=0.25). Two-year OS was 45.9% and 35.0%, and five-year OS was 24.5% and 19.2%, respectively. Median PFS was 7.3 months (95% CI, 1.8–10.9) versus 2.8 months (95% CI, 2.1–4.4) (log-rank p=0.16). BOR was available for 138 patients; objective response rates were 35.5% and 31.8%, respectively. 3–5 adverse events occurred in 24.1% of Ipi1+Nivo3 versus 30.5% of Ipi3+Nivo1 patients. Hospitalizations were less frequent (16.7% vs 36.1%), and neurologic adverse events occurred only with Ipi3+Nivo1 (11.9% vs 0%). Conclusions: In this multisite real-world MBM cohort, Ipi1+Nivo3 demonstrated comparable efficacy, numerically longer survival, and a more favorable toxicity profile compared with standard Ipi3+Nivo1. Notably, no neurologic toxicity was observed with the Ipi1+Nivo3 regimen. These findings support further prospective evaluation of lower dose ipilimumab strategies. Response and survival outcomes. Outcome Ipi1 + Nivo3 Ipi3 + Nivo1 Complete response (CR) 2 / 31 (6.5%) 8 / 107 (7.5%) Partial response (PR) 9 / 31 (29.0%) 26 / 107 (24.3%) Objective response rate (ORR = CR+PR) 11 / 31 (35.5%) 34 / 107 (31.8%) Disease control rate (DCR = CR+PR+SD) 18 / 31 (58.1%) 51 / 107 (47.7%) Median PFS, months (95% CI) 7.3 (1.8–10.9) 2.8 (2.1–4.4) 2-year PFS 27.6% 22.3% Median OS, months (95% CI) 18.5 (5.7–54.7) 10.6 (6.5–19.1) 2-year OS 45.9% 35.0% 5-year OS 24.5% 19.2%

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%)

Background: In the last decade, a surge of advancements in and approvals of immunotherapies has revolutionized the treatment of non-small-cell lung cancer (NSCLC). Even as an increasing number of immune checkpoint inhibitors and antibody–drug conjugates gained FDA approval for the remarkable response and durable remission they imparted to some patients, persistent challenges of primary and acquired resistance, imperfect predictive biomarkers, and toxicity remain. Methods: This review identifies and summarizes the most significant emerging clinical data being reported for immunotherapies in NSCLC. Results: We found that the most recent clinical trials are centered on antibody–drug conjugates, bispecific and multispecific antibodies, cellular therapies, and immunocytokines, while other ongoing trials continue to explore new monoclonal antibodies as well as a variety of immunotherapy modalities. Conclusions: Immunotherapy in NSCLC remains an area of rapid advancement, with the potential for further improvements in safety and efficacy as monotherapy and in multimodal approaches.