Dimou, Anastasios

e20011 Background: The therapeutic landscape of lung cancer is rapidly evolving, presenting oncologists with the challenge of staying updated amidst an overwhelming influx of data. Clinical decision support (CDS) tools, including artificial intelligence (AI) and large language models (LLMs), may help bridge this gap. Evaluating the accuracy of LLMs in complex, real-world oncology scenarios is crucial to understanding their potential. Methods: Twenty-five de-identified lung cancer cases from the fellows’ clinic at Karmanos Cancer Institute, Detroit, MI, were analyzed. Two LLMs, GPT-4 (OpenAI) and Claude Opus (Anthropic), were assessed using advanced prompting techniques like persona-based and chain-of-thought prompting. Five board-certified lung cancer oncologists from NCI-designated centers evaluated LLM-generated responses based on accuracy, treatment recommendation comprehensiveness, and supportive care planning, using a 1–5 scale. Novel insights, the presence of fabricated information, and harmful recommendations were flagged as binary outcomes. Oncologists were blinded to the LLM source and actual treatment decisions. Results: Table 1 presents patient characteristics. GPT-4 achieved an average accuracy score of 4.2 (95% CI, 3.9–4.4), with 3.7 for comprehensiveness of medical/surgical treatment recommendations and 3.7 for supportive care planning. Six responses (32%) were flagged as potentially harmful, and two (8%) contained inaccuracies. Sixteen GPT-4 responses (64%) were rated trustworthy as a CDS tool. Claude Opus had an average accuracy score of 3.6 (95% CI, 3.1–4.1), scoring 3.6 for treatment recommendation comprehensiveness and 3.5 for supportive care planning. Nine responses (36%) were flagged for potential harm, and five (20%) included inaccuracies. Eleven Claude responses (44%) were deemed trustworthy. Significant differences were observed in accuracy (p=0.04) and trustworthiness (p=0.03) between models using McNemar's test. Other factors showed no statistical significance. Conclusions: GPT-4 outperformed Claude Opus in accuracy and trustworthiness, but both models demonstrated limitations, including harmful recommendations and inaccuracies. These findings highlight the need for improved LLM refinement before routine use as CDS tools in lung cancer treatment. Patient demographics and clinical characteristics. Category subcategory Number Median Age (range)- Yr 65 (26-78) Female 7 Male 18 Histology Adenocarcinoma 10 Squamous Cell Carcinoma (SCC) 7 Small Cell Carcinoma 6 Poorly Differentiated 2 Total 25 Stage NSCLC Stage 3 7 NSCLC Stage 4 13 Small Cell limited stage 3 Small Cell Extensive Stage 2

Abstract Recent studies have described the tumor microenvironment and the key transcription factors in untreated small cell lung cancer (SCLC), but less is known about the salient features in recurrent SCLC (rSCLC). This information could provide valuable insight about the effectiveness of second and later line therapies. In this study, we compared transcriptome profiles (by RNA-seq) of rSCLC against untreated primaries (pSCLC). rSCLC samples (n=23) were obtained from frozen biopsies in patients enrolled in our MC1923 (NCT04607954) phase 2 clinical trial whose disease progressed after chemoimmunotherapy. The pSCLC group (n=73) included publicly available data from fresh frozen primary tumors after surgical resections. Normalized expression data were used in immune deconvolution and gene set enrichment analysis (GSEA) in molecular signature database. Additional GSEA included gene sets describing ASCL1 (A), NEUROD1 (N), ASCL1 and NEUROD1 (AN), and POU2F3 (P) transcription factor activities which are used for subtyping of SCLC. Immune deconvolutions by multiple software algorithms suggested depletion of immune infiltrates in rSCLC compared with pSCLC including dendritic cells, B cells, CD8+ T cells, and M1 macrophages. Furthermore, we found decreased chemotaxis and increased apoptosis of dendritic cells, reduced TNF signaling, and reduced lymphocyte co-stimulation in rSCLC by GSEA. Additionally, in comparisons with pSCLC, rSCLC tumors had significantly lower enrichment scores for POU2F3 activity, representing the SCLC-P subtype (p < 0.007). Importantly, the SCLC-P was recently reported to be more immune inflamed than the other A, N, and AN subtypes. Altogether, our analyses provide strong evidence that the tumor microenvironment of rSCLC is significantly more immunosuppressed and with reduced immune cell infiltrates compared to pSCLC. Developing therapeutic strategies aimed at “warming” the tumor microenvironment and enhancing anti-tumor immune activity in ongoing multimodal therapies of recurrent SCLC will likely provide significant benefits to patients. Citation Format: Farhad Kosari, Mohamed Ali, Sarah H. Johnson, Alex Adjei, Julian R. Molina, Aaron S. Mansfield, Anastasios Dimou, Kaushal Parikh, Mohamed Shanshal, Katherine E. Smith, Ailsa G. Luce, Anna J. Schwecke, Janet Lensing, Ying-Chun Lo, John C. Cheville, George Vasmatzis, Konstantinos Leventakos. Small cell lung cancer (SCLC) at recurrence has a “colder” tumor microenvironment than primary SCLC according to gene expression profiling [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 6614.

Background Neoadjuvant treatment has become standard for patients with high-risk operable stage III melanoma, but the optimal regimen is unknown. Targeted therapy approaches yield high pathological response rates, while immunotherapy regimens show favorable recurrence-free survival (RFS). NeoACTIVATE was designed to address whether a neoadjuvant combination of both targeted therapy and immunotherapy might leverage the benefits of each. Methods We tested neoadjuvant treatment with 12 weeks of vemurafenib, cobimetinib, and atezolizumab for patients with BRAF-mutated (BRAFm) melanoma (cohort A) and cobimetinib and atezolizumab for patients with BRAF-wild-type (BRAFwt) melanoma (cohort B), regimens which we have shown generate a substantial major pathological response. After therapeutic lymph node dissection, patients received 24 weeks of adjuvant atezolizumab. Here, we report survival outcomes and their association with biomarkers assayed among the gut microbiome and peripheral blood immune subsets. Results With 49 months median follow-up, the median RFS was not reached for cohort A and was 40.8 months for cohort B. At 24 months after operation, 2 of 14 cohort A patients and 4 of 13 cohort B patients had experienced distant relapse. Key findings from correlative analyses included diversity, taxonomic and functional metagenomic gut microbiome signals associated with distant metastasis-free survival at 2 years. Notably, we observed a strong correlation between low microbial arginine biosynthesis (required for T-cell activation and effector function) and early distant recurrence (p=0.0005), which correlated with taxonomic differential abundance findings. Peripheral blood immune monitoring revealed increased double-positive (CD4+CD8+) T cells in patients with early recurrence. Conclusions Neoadjuvant treatment with cobimetinib and atezolizumab±vemurafenib was associated with a low rate of distant metastasis in patients with high-risk stage III melanoma. Freedom from early distant metastasis was highly associated with taxonomic differences in gut microbiome structure and with functional pathway alterations known to modulate T cell immunity. Identification of predictive biomarkers will permit optimization of neoadjuvant therapy regimens for individual patients. Trial registration number NCT03554083.

Abstract Introduction We describe the safety of sotorasib monotherapy in patients with KRAS G12C-mutated advanced non-small cell lung cancer (NSCLC) and discuss practical recommendations for managing key risks. Methods Incidence rates of treatment-related adverse events (TRAEs) were pooled from 4 clinical trials: CodeBreaK 100 (NCT03600883), CodeBreaK 101 (NCT04185883), CodeBreaK 105 (NCT04380753), and CodeBreaK 200 (NCT04303780) and graded according to CTCAE v5.0. Adverse events were deemed sotorasib-related per investigator causality assessment. Results In the pooled population (n = 549), TRAEs were reported in 388 (70.7%) patients (grade 1: 124 [22.6%]; grade 2: 117 [21.3%]; grade ≥ 3: 147 [26.8%]). Gastrointestinal and hepatic TRAEs, including diarrhea (171 [31.1%]), nausea (80 [14.6%]), elevated alanine aminotransferase (ALT; 68 [12.4%]), and elevated aspartate aminotransferase (AST; 67 [12.2%]) were the most common (≥10%). Dose interruption and dose reduction of sotorasib resulted in the resolution of >90% of diarrhea events; median time to resolution were 18.0 days and 22.0 days, respectively. Similar trends were observed for elevated ALT and AST events. Patients who stopped immunotherapy <3 months before initiating sotorasib had a higher incidence of treatment-related hepatotoxicity (80/240 [33.3%]) than those who stopped immunotherapy ≥3 months before initiating sotorasib (26/188 [13.8%]). Treatment-related pneumonitis/interstitial lung disease (ILD) and corrected QT (QTc) prolongation were observed in 9 (1.6%) and 4 (0.7%) patients, respectively. Two (0.4%) patients died with TRAEs, 1 with ILD whose ultimate cause of death was disease progression, and the other with an unknown cause. Conclusions Sotorasib has a well-characterized safety profile in patients with KRAS G12C-mutated advanced NSCLC, and key risks are manageable with dose modification.

Melanoma is the deadliest form of skin cancer. The median age at diagnosis is 66. While most patients are treated with immunotherapy, the use of targeted therapy is a valid alternative for patients whose tumors harbor a BRAF or c-KIT driver mutation. These agents, while effective, come with a variety of side effects which limit their use, especially in older patients. We sought to assess the efficacy and toxicity of these agents in older melanoma patients. Melanoma patients over 65 treated with BRAF/MEK or c-KIT inhibitors were retrospectively identified, and their data were analyzed for treatment efficacy and toxicity. All data were compared using the Chi-square test for categorical comparisons and the Kruskal–Wallis method for median comparisons. One hundred and sixteen patients were identified. One hundred and six patients were treated with BRAF/MEK inhibitors. The assessed response rate (RR) was 83% and was comparable across different subgroups, including advanced line patients and those with a more aggressive disease. The median progression free survival (PFS) was 7.9 months, and the median overall survival (OS) was 15.7 months. Twenty-seven percent experienced grade 3–4 toxicity leading to a 24% treatment discontinuation rate. Another 10 patients were treated with the c-KIT inhibitor imatinib, for whom the assessed RR was 55%. The median PFS was 4.3 months, and the median OS was 22.6 months. Forty percent needed dose reductions, yet none had to stop treatment due to adverse effects. The use of targeted therapy in older patients is effective yet challenging due to toxicity. Deploying mitigation strategies can help maximizing their usefulness.

e20588 Background: MET exon 14 skipping mutations are seen in about 3% of patients (pts) with NSCLC, with 2 FDA approved METi for pts with stage IV NSCLC harboring it. Besides the original clinical trials, there is a paucity of real-world data about their outcomes, and adverse events (AE). Our aim was to compare and analyze the utilization of capmatinib (cap) and tepotinib (tep) in this population. Methods: We performed a retrospective chart review of the electronic medical records of all pts with stage IV NSCLC treated with METi between the date of FDA-accelerated approval of these drugs, 5/6/20 for cap and 2/3/21 for tep, and 12/1/23. All pts were treated at the Mayo Clinic Health System. We collected data about baseline characteristics, diagnosis, AE, and outcomes. Chi-squared test was used to compare the frequency of AE, while Kaplan-Meier method was used to calculate median progression free survival (mPFS). Results: A total of 49 pts were treated with METi. Cap was used in 45 pts, 14 received tep, and 10 pts received both, with tep following cap. In the cap group, median age of diagnosis was 74 years (y), 42.22% were male, and the median number of previous lines of therapy (mPLT) was 0. In the tep group, median age of diagnosis was 72 y, 21.43% were male, and the mPLT was 2. Median ECOG at the time of diagnosis was 1 in both groups. There was no statistically significant difference between AE in both groups. Peripheral edema occurred in 75.56% pts who received cap, and in 64.29% with tep (p = 0.41). Pneumonitis happened in 4.44% with cap, and 7.14% with tep (p = 0.69). Creatinine elevation was seen in 37.78% with cap, while 42.86% with tep (p = 0.73), grade 3 elevation was seen in 5.88% and 16.67% respectively. Abnormality in liver function tests occurred in 35.56% of pts in the cap group, and 28.57% in the tep group (p = 0.63), grade 3/4 abnormality occurred in 31.25% and 0% respectively. Finally, pancreatic enzymes elevation was seen in 33.33% of patients in each group (p = 0.67), with none being grade 3/4. Cap was discontinued in 20% of pts due to AE, while it happened in 28.57% of pts with tep. Dose reduction happened in 64.44% of pts who received cap, and 42.86% who received tep. The mPFS was 12.03 months (m) in the cap group, and 8.94 m in the tep group, which was not statistically significant (p = 0.18). Similarly, the mPFS in those treated with tep following cap was 8.94 m. Conclusions: We performed the first real-word comparison between cap and tep. The results demonstrated similar outcomes with both drugs, and no significant differences in AE, suggesting these agents can be employed equivalently. Importantly, rates of therapy discontinuation and dose reduction were high with both, indicating the importance of closely monitoring these pts. The retrospective nature and small sample size are initial limitations that can be overcome with future multi-centric prospective studies.

8071 Background: Data from the PACIFIC study demonstrated a 5-yr median progression free survival (PFS) and overall survival (OS) of 16.9 and 47.5 months respectively for patients with stage 3A and stage 3B NSCLC demonstrating a high risk of relapse. The 8th ed. of the IASLC staging project reports that the 60-month OS for stage 3A, B and C disease is 41%, 24% and 12% respectively, indicating that improved treatment options are needed for this patient population. Induction chemo-IO (ID chemo-IO) followed by concurrent chemo-radiation (CRT) has not been fully investigated due to concerns regarding the toxicity profile, but this may representant an important unexplored avenue for research. We aim to describe the outcomes of patients who received ID chemo-IO followed by CRT with or without maintenance IO. Methods: We conducted a retrospective study across all Mayo Clinic sites. Patients were included if they were diagnosed with Stage 3 NSCLC deemed ineligible for resection and/or requiring raditation over a large-field ] by a multidisciplinary team with plans for ID chemo-IO prior to CRT and maintenance IO. A total of 927 cases of stage 3 NSCLC patients were screened and those receiving ID chemo-IO followed by CRT were extracted (n = 36). Clinical endpoints are PFS, OS, overall response rate (ORR) and treatment related adverse events (TrAE) defined using CTCAE 5.0. Results: The median age was 67 years, 52.8% were female, all patients were past smokers and 97.2% identified as white. The tumor histologies were adenocarcinoma (66.67%), squamous (27.8%), mucinous adenocarcinoma (2.8%) and sarcomatoid (2.8%). Majority were Stage 3B (47.2%) followed by 3C (38.9%) and 3A (13.9%). N2 and N3 disease was noted in 36.1% and 55.6% of patients respectively. PD-L1 by IHC was > 1-49% (47.2%), > 50% (22.2%) and < 1% (19.4%). Patients received ID chemo-IO with pembrolizumab (83.3%), nivolumab (11.1%) and atezolizumab (2.78%). Average number of ID chemo-IO cycles were 3.67 cycles. The ORR to ID chemo-IO was 86.11%. The median PFS was 23.67 mths and the median OS was 32.07 months. About 94% of patients received CRT afterwards and 66.67% received maintenance IO with pembrolizumab (53.17%) or durvalumab (45.83%). Any grade TrAE occured in 58.33% of patients. Pneumonitis was the most common event (38.9%). Grade 3 -4 pneumonitis was noted in 28.5% (4/11) patients. Other toxicities included hepatitis (8.3%), colitis (11.1%) and myasthenia gravis (2.8%) and 22% of these TrAE’s were grade 3 or 4. Conclusions: Our results indicate that ID chemo-IO shows promise as a therapeutic approach for patients with unresectable stage III NSCLC, especially for patients deemed ineligible or high risk for upfront radiation. Further investigation is needed through a randomized clinical trial to systematically explore and validate the safety and efficacy of ID chemo-IO.

e23270 Background: Melanoma is the most lethal form of skin cancer which have seen drastic improvement in patient survival over the last decade. BRAF is the most common driver mutation in melanoma, seen in ≈50% of all patients (≈30% of patients > 65). The use of BRAF & MEK inhibitors yields high response rates, though it is limited by toxicity and development of acquired resistance. Acknowledging these limitations, we aimed at assessing the efficacy and toxicity of these agents in melanoma patients > 65. Methods: We identified electronic medical records of all melanoma patients > 65 that were treated with BRAF & MEK inhibitors at our institution over the years 2017-2023. Data was retrospectively collected including demographics, disease characteristics, treatment given, response patterns and toxicity. Results: 106 patients were identified. The median age was 71 (65-89), 55% were males. The median follow up was 28.9 months. 32 (30%) patients were treated at first line, 58 (55%) at second line and 16 (15%) at third line. 88 (83%) evaluable patients received treatment for unresectable/metastatic disease of which 84% had an objective response (60% partial response, 24% complete response). Response rate (RR) was 82%, 85% and 87% for first, second and third line respectively. RR was 86% for patients with an increased serum lactate dehydrogenase. 33% of patients had active central nervous system (CNS) metastases at treatment initiation, for whom the systemic RR was 80% and the CNS RR was 39%. The median progression free survival (PFS) was 7.9 months, with a PFS of 6, 8.3 and 13.4 months for first, second and third line respectively. The median PFS was 5.7 months for patients with active CNS disease. The median overall survival (OS) from treatment initiation was 15.7 months (12.5, 14.5, 31.9 for first, second & third line respectively). The median OS for patients with CNS disease was 9.8 months. 16 patients were treated in the adjuvant setting, of which 6 (37%) finished treatment as planned, 3 stopped due to disease progression (19%) and 6 stopped due to toxicity (37%). 88% of patients experienced adverse events (AEs) of any grade, with 27% having grade 3-4 toxicity. There were no grade 5 toxicities reported. 36% of patients experienced one AE, while 52% experienced two or more. 21% of patients stopped treatment due to toxicity, 28% needed a dose reduction and 20% changed to different BRAF/MEK inhibitors. 10% of patients needed systemic corticosteroids to treat AEs. The most common toxicities seen were fever (30%), fatigue (30%), skin rash (22%), elevated liver function tests (20%), arthralgia/myalgia (15%), nausea (14%) and diarrhea (11%). 7% had decrease in cardiac ejection fraction with one grade 3 leading to treatment discontinuation. Conclusions: Treatment with BRAF & MEK inhibitors is challenging in older patients, with a high prevalence of toxicity leading to treatment modifications and discontinuation. Efficacy seems to be the same as in the general population.

Abstract Background: Neuropilin 2 (NRP2) is a co-receptor that enhances signaling of growth factor and cytokine receptors and is expressed in tumor and microenvironment (TME) cells in melanoma and other tumors. Neuropilin 2 expression in tumor cells is associated with more aggressive tumors. Expression in tumor associated macrophages enhances efferocytosis and immunologically silent tumor cell death. Methods: Whole excisional lymph node biopsies were obtained from 10 treatment naïve patients with metastatic cutaneous melanoma who subsequently received anti-PD1 treatment. A single 5mm formalin-fixed paraffin embedded tissue section was used to assess a panel of 39 analytes by cyclic multiplex immunofluorescence (MxIF). Fields of view (~1mm2, n=288) were selected from pathologist-annotated regions of the tumor core, tumor-immune interface and adjacent lymphoid tissue. Pixel-based single cell segmentation and a supervised classifier approach was applied to resolve 12 distinct tumor, stromal and immune cell phenotypes and functional states (NRP2 positive or negative) in 1.5 million cells. Results: Tumor recurrence was observed in 6 patients (relapsed), whereas 4 patients were recurrence-free (non-relapsed) during follow up. Among the different cell types, NRP2 expression was more frequent in macrophages (44%), tumor cells (40%), dendritic cells (40%), followed by Tregs (31%) and T helper cells (29%), and less frequent in neutrophils (24%), cytotoxic T cells (23%) and B cells (15%). In colocalization permutation analysis, NRP2 expressing macrophages clustered away from tumor cells in non-relapsed patients while a random spatial distribution of NRP2 expressing macrophages with respect to tumor cells was observed in relapsed patients. In comparing the odds of tumor cell interactions with NRP2-expressing macrophages to those with NRP2 non-expressing macrophages, we found in the tumor:stromal interface (10% tumor cells) that cells from both non-relapsed and relapsed patients were more likely to interact with NRP2 non-expressing macrophages (OR = 0.73, 95% CI = [0.71, 0.75] and OR = 0.56, 95% CI = [0.55, 0.58] for non-relapsed and relapsed patients, respectively). However, in the tumor core (90% tumor cells), tumor cells from relapsed patients had significantly higher odds of interacting with NRP2 expressing macrophages, (OR = 1.15, 95% CI = [1.12, 1.18]), and this was not seen for non-relapsed patients (OR = 0.68, 95% CI = [0.66, 0.71]). In FOVs dominated by tumor cells (75% tumor), tumor cells from relapsed patients had 4.97 times the odds of expressing NRP2 compared to tumor cells from non-relapsed patients (95% CI = [2.23, 11.09], p = 0.0001). Conclusions: These spatial analytics suggest complex interactions among NRP2 expression in tumor and TME of melanoma determining risk for relapse with anti-PD1 treatment. Citation Format: Anastasios Dimou, Raymond M. Moore, Caitlin Ward, Alexey Leontovich, Ruifeng Guo, Chen Wang, Shankar Suman, Betty Dicke, Jill M. Schimke, Noah A. Stueven, Chathu A. Atherton, Wendy Nevala, Svetomir N. Markovic. Spatial analysis of neuropilin 2 expression in the microenvironment of melanoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2905.