Citations

6073 Background: Laryngeal Squamous Cell Carcinoma (LaSCC) has varying outcomes based on the stage of cancer patients present with. Currently, a high proportion of patients are diagnosed with advanced-stage LaSCC complicating the treatment landscape and over-treating low risk patients. Risk stratification of LaSCC can help tailor treatment plans. Numerous studies have identified spatial architecture of tumor-infiltrating lymphocytes (TILs) as a prognostic biomarker in oral cavity and oropharyngeal SCC. In this work, we evaluate the prognostic value of an artificial intelligence (AI)-leveraged approach that characterizes the spatial architecture of TILs on digitized hematoxylin and eosin (H&E)-stained slides from patients with LaSCC. Methods: H&E slides from 192 patients with LaSCC were collected from Baylor Medical Center. This dataset was randomly divided into two equal cohorts, A and B. The slides were digitized as whole slide images at 40x magnification. The nuclei of all cells were automatically segmented using a deep-learning model (Hover-Net). Each nucleus was then classified as TIL or non-TIL based on morphological features. TILs and non-TILs were clustered based on proximity, and features related to the density and spatial distribution were extracted. The top features, determined by the least absolute shrinkage and selection operator, were used to train a Cox Proportional Hazards regression model that assigned a risk score for recurrence of cancer to each patient in cohort A. For validation, the model was applied to patients in cohort B. The 25 th percentile training risk score was used as a cutoff for classifying patients as high or low risk. The performance of the model in prognosticating loco-regional recurrence (LRR) was evaluated using survival analysis. Results: Patients in Cohort B identified as “high risk” by the model based on spatial organization of TILs had a significantly shorter survival time. Univariate survival analysis showed this model was prognostic for DFS with a hazard ratio of 2.57 (95% Confidence Interval: 1.12-5.89, p-value=0.048), meaning that patients classified as “high risk” are approximately 2.5 times more likely to develop LLR. Conclusions: We used computational pathology to characterize the architecture of TILs and develop a model to predict risk of LLR in LaSCC. With additional validation, this approach can be used to assist clinicians with making clinical decisions. Multivariate analysis. Variable Reference vs Comparison Pr(>|z|) HR (95% CI) Race Black vs Caucasian 0.93 0.92 (0.40 - 2.31) N N0 vs N+ 0.67 1.80 (0.12 - 26.06) T 1-2 vs 3-4 0.83 1.28 (0.13 - 12.38) Chemo Yes vs No 0.38 3.49 (0.21 - 56.90) Tobacco Yes vs No 0.26 2.08 (0.58 - 7.47) Alcohol Yes vs No 0.43 1.53 (0.54 - 4.35) TIL Architecture Risk High vs Low 0.03 * 0.26 (0.08 - 0.89)

6057 Background: The incidence rate of human papillomavirus associated oropharyngeal squamous cell carcinoma (HPV(+)OPSCC) has persistently increased over the past decade and is anticipated to rise further over the next decade. Cost effective, accurate, and less invasive screening options for HPV(+)OPSCC have remained stagnant leading to the interrogation of circulating tumor DNA in liquid biopsies such as plasma. We hypothesized that candidate methylated DNA markers (MDMs), previously described by our group, would be present in the plasma of patients with HPV(+)OPSCC and absent in controls. Methods: HPV(+)OPSCC cases were enrolled from a high-volume referral practice. Cancer-free control patients were identified from the surrounding 7-county catchment area and frequency matched to cases by age and alcohol use. cfDNA was extracted from 4 mL of plasma and bisulfite converted prior to Target Enrichment Long-probe Quantitative Amplified Signal assays for 15 MDMs. Strand counts for individual MDMs were normalized to a methylated genomic reference marker. An MDM score was derived via random forest modeling of the 15 MDMs. Area under the receiver operator characteristic curve (AUC) for discriminating cases from controls was evaluated for individual markers and for the MDM score. The sensitivity of the MDM score at 95% specificity was also evaluated. Results: The study consisted of 96 HPV(+)OPSCC and 100 cancer-free controls. The study population was predominately white (95%) with a median age of 61 yrs (IQR: 52-69 yrs). Eighty-nine percent of the cases were male with cancers affecting either the tonsil in 50 (52%) or base of the tongue in 46 (48%) patients; by design 50% of controls were male. Stage I:II:III:IV disease was present in 66%:22%:9%:3% of the cases. The median (IQR) AUC across the 15 MDMs was 0.83 (0.82-0.85) overall: 0.78 (0.77-0.83) for stage I disease and 0.92 (0.91-0.93) for stage II-IV disease. The MDM score for the combined panel of 15 MDMs achieved a cross-validated AUC of 0.93 (CI: 0.89-0.97) overall: 0.90 (CI: 0.84-0.96) for stage I disease and 0.98 (CI: 0.97-1.00) for stage II-IV disease. At 95% specificity, the cross-validated sensitivity of the MDM score for HPV(+)OPSCC was 80% (CI: 71-87%) overall: 73% (CI: 60-83%) for stage I disease and 94% (CI: 78-99%) for stage II-IV disease. Conclusions: We validated a set of previously discovered MDM markers in HPV(+)OPSCC in a robust case control cohort using plasma. The MDMs were detected in both early and late-stage disease. Additional prospective studies in larger intended use cohorts are needed to validate our results for clinical use.

1114 Background: While treatment and management of TNBC has improved, there is a need for novel prognostic biomarkers to better inform outcomes and guide therapeutic options. ILF2 is a poorly characterized protein with pleiotropic functions that is highly expressed in TNBC. Here we evaluated the associations of ILF2 with 1) genomic and transcriptomic data, 2) tumor microenvironment (TME), and 3) clinical outcomes in TNBC. Methods: 15,544 breast cancer (BC) samples, including 3,038 TNBC, were tested by NGS (592, NextSeq; WES, NovaSeq) and WTS (NovaSeq; Caris Life Sciences, Phoenix, AZ). ILF2 -high (H) and ILF2 -low(L) TNBC were defined by respective quartiles. Immune cell fractions were estimated by WTS deconvolution (Quantiseq). Real world overall survival (OS) was obtained from insurance claims and calculated from tissue collection to last contact using Kaplan-Meier estimates. Statistical significance was determined by chi-square, Fisher’s exact, and Mann-Whitney U test with p-values adjustments (q < .05). Results: ILF2 expression (median Log2(TPM+1) was higher (all q < .05) in key subgroups: ductal compared to lobular carcinoma (6.4 vs 6.0); primary compared to metastatic BC (6.4 vs 6.3); African American compared to White (6.4 vs 6.3); basal compared to luminal A, luminal B, HER2 PAM50 subtypes (6.9 vs 5.8, 6.3, 6.3); and TNBC compared to HR+HER2+, HR-HER2+, HR+HER2- subtypes (6.7 vs 6.3, 6.4, 6.2). Biopsied tissues from primary TNBC (pTNBC) and metastatic TNBC (mTNBC) patients were stratified into ILF2- H and ILF2- L groups. In both mTNBC and pTNBC, ILF2- H groups had 1) higher percentage of young patients (age < 50) (pTNBC: 35.5% vs 19.8%; mTNBC: 28.1% vs 17.3%; all q < .05); 2) higher mutation frequency of TP53 (pTNBC: 94.5% vs 79.6%; mTNBC: 92.4% vs 74.4%), but lower frequencies for PIK3CA ( pTNBC: 5.1% vs 23.4% , mTNBC: 8.8% vs 27.4%), CDH1 ( pTNBC: 0.8% vs 6.1%; mTNBC: 2.8% vs 12.2%; all q < .05); 3) higher infiltration of NK cells (pTNBC: 3% vs 2.6%; mTNBC: 2.8% vs 2.6%), but lower infiltration of M2 Mφ (pTNBC: 2.5% vs 3.3%; mTNBC: 2.6% vs 3.2%) and Tregs (pTNBC: 1.5% vs 1.9%; mTNBC: 1.4% vs 1.7%; all q < .05); 4) higher expression levels of immune checkpoint ( CD274, PDCD1LG2, CTLA4, LAG3, HAVCR2, FOXP3, IDO1, CD276 , FC: 1.2-3.1; all q < .05), stem cell genes (C D44, NANOG, POU5F1, KLF4, ALDH1A1 , FC: 1.4-2.4; all q < .05), and drug efflux genes ( ABCC3, ABCC11, ABCC2, ABCB1, ABCG2, ABCC1 , FC: 1.1-4.5; all q < .05) compared to ILF2- L group. In pTNBC, ILF2- H had significantly shorter OS vs ILF2- L group (22.3 vs 28.9 months, HR 1.2 [95% CI 1-1.5], p = .03), but no significant differences were observed between mTNBC ILF2 groups (HR 1.1 [95% CI 0.93-1.3], p = .2). Conclusions: ILF2 -H TNBC patients showed differential genomic and transcriptomic alterations that relate to therapy resistance, immune suppressive TME, and shorter OS. Further studies are warranted to validate the effects of ILF2 upregulation on therapeutic efficacy.

1107 Background: Chronic stress-mediated β2-adrenergic receptor (β2-AR) signaling promotes tumor growth via immunosuppression in the tumor microenvironment (TME) in preclinical models. Blockade of β2-AR has shown higher survival benefit in patients with TNBC in observational studies compared to other breast cancer (BC) subtypes. However, the molecular and immunological features associated with ADRB2 (gene for β2-AR) gene expression in TNBC are unknown, prompting this investigation. Methods: 3,038 TNBC samples were analyzed via NGS (592-gene panel, NextSeq; WES/WTS, NovaSeq; Caris Life Sciences, Phoenix, AZ). Immune cell fractions were calculated by deconvolution of WTS: Quantiseq. TNBC ADRB2 -high(H) and ADRB2 -low(L) RNA expression were classified as above or below the 50 th percentile, respectively. Real-world overall survival (OS) was obtained from insurance claims and calculated from tissue collection to last contact using Kaplan-Meier estimates. Statistical significance was assessed using chi-square and Mann-Whitney U tests with multiple comparison adjustments (q < 0.05). Results: ADRB2 gene expression was lowest in TNBC (median (TPM: 1.6) compared to N = 453 HR-HER2+ (1.9), N = 629 HR+HER2+ (2.0) and N = 4,918 HR+HER2- (2.2) BC (all q < 0.05). African American or Black patients (N = 670) had lower expression of ADRB2 compared to European American or White (N = 1,412) TNBC patients (1.3 vs 1.7, q < 0.05). ADRB2 -H TNBC had higher mutation frequency of PIK3CA (21% vs 15.4%), CDH1 (7% vs 3.5%), NF1 (8% vs 4%), AKT1 (3.5% vs 2.1%), but lower frequency of TP53 (81.6% vs 87.5%), NOTCH1 (2.5% vs 4.5%) and NOTCH3 (4.4% vs 11.7%) compared to ADRB2 -L, all q < 0.05. ADRB2 -H had greater PD-L1 (22C3) positivity (39.1% vs 30.2%, q < 0.05), higher % of B cells (4.5 vs 3.4), M1 Mφ (3.4 vs 2.8), M2 Mφ (3.9 vs 2.2), Tregs (2.2 vs 1.3), NK cells (3.1 vs 2.6), DC (3.1 vs 2.9), CD8 + T cells (0.9 vs 0.2), all q < 0.05. ADRB2 -H TNBC had higher T-cell inflamed score (95 vs -80), IFN γ score (-0.23 vs -0.37), MAPK activation score (-0.46 vs -1.7), all q < 0.05; and higher expression of immune checkpoint genes ( CD274, PDCD1, PDCD1LG2, CTLA4, LAG3, HAVCR2, FOXP3, IDO1, TNFSF14, TIGIT, BTLA, CEACAM1, CD47, CD274 ; fold change: 1.6-3.7, all q < 0.05). ADRB2 -H tumors had higher expression of genes related to inflammatory response, IFNγ response, IL6-JAK-STAT3 signaling (normalized enrichment score (NES): 1.9 – 2.1), while ADRB2 -L had enrichment of MYC targets V1, MYC targets V2, E2F targets and G2M checkpoint (NES: 2.5 – 4.2), all FDR < 0.01. ADRB2 -H TNBC had better OS (mOS: 23.6 vs 18.6 months; HR 0.81, 95% CI 0.73-0.89, p < 0.0001) compared to ADRB2 -L. Conclusions: High ADRB2 expression in TNBC is associated with better survival and an immune enriched TME, elevated immune checkpoints and other targetable vulnerabilities. Future studies are needed to investigate ADRB2 as a potential stress biomarker and therapeutic target.

1106 Background: TNBC, although an aggressive breast cancer (BC) subtype, is highly immunogenic and the only BC subtype where the ICI pembrolizumab is approved. However, predictive biomarkers for pembrolizumab benefit are limited. The chemokines CXCL9 and CXCL10 attract CD8 + T cells into the tumor microenvironment (TME) and are associated with chemotherapy benefit, but little is known about their role in prediciting pembrolizumab benefit in TNBC. We investigated the association of CXCL9 , CXCL10 and their cognate receptor CXCR3 with TME and ICI efficacy. Methods: 3,038 TNBC samples were analyzed via NGS (592-gene panel, NextSeq; WES/WTS, NovaSeq; Caris Life Sciences, Phoenix, AZ). Tumor mutational burden (TMB) totaled somatic mutations per tumor (high > 10 mt/MB). Immune cell fractions were estimated using WTS deconvolution (Quantiseq). CXCL9 / CXCL10 / CXCR3 -high (H) and -low (L) tumors were classified by RNA expression above or below the 50th percentile. Real-world overall survival (OS) was derived from insurance claims and calculated from start of pembrolizumab to last contact using Kaplan-Meier. Statistical significance was assessed using chi-square and Mann-Whitney U with multiple comparison adjustments (q < .05). Results: TNBC expressed higher levels of CXCL9 and CXCL10 (median (TPM): 5.3 and 14.7) compared to N = 1,082 HER2+ (4.8 and 10.3, p < 0.05) and N = 4,918 HR+HER2- (2.7 and 7, q < .05) BC. CXCR3 expression was higher in TNBC compared to HR+HER2- (1.9 vs 1.7, q < .05), but no difference when compared to HER2+ (1.9 vs 2, q = 0.97) BC. CXCL9 / CXCL10 / CXCR3 -H TNBC had higher median OS post pembrolizumab [ CXCL9 -H vs -L: 26.5 vs 15.7 months (mo), HR: 0.65 (95% CI 0.5-0.84); CXCL10 -H vs -L: 26.0 vs 20.6 mo, HR 0.74 (0.57-0.95); CXCR3 -H vs -L: 32.6 vs 18.3 mo, HR 0.68 (0.52-0.88), all p < 0.05 ]. CXCL9 -H, CXCL10 -H and CXCR3 -H had higher PD-L1 positivity (22C3), TMB high, higher T cell inflamed score, TP53 mutations, elevated B and CD8 + T cells infiltration, but not neutrophils, and higher expression of immune checkpoint genes (Table). Conclusions: High CXCL9 / CXCL10 / CXCR3 expression is associated with longer survival in patients with TNBC post pembrolizumab, and characterized by an immune-enriched TME. Further investigation is needed to evaluate this chemokine axis in TNBC and its potential as a therapeutic target to enhance ICI efficacy. TME characteristics by CXCL9, CXCL10 and CXCR3 expression. CXCL9 CXCL10 CXCR3 High Low q-value High Low q-value High Low q-value PD-L1 % 54 14 <.05 54 14.7 <.05 49 19 <.05 TMB high % 14.3 8 <.05 12.6 9.7 <.05 12.4 9.8 <.05 B cell (median %) 4.4 3.5 <.05 4.3 3.6 <.05 5 3.4 <.05 CD8+T cell (median %) 1.2 0 <.05 1 0 <.05 1.3 0 <.05 Neutrophil (median %) 4.2 4.5 <.05 4.2 4.4 <.05 4.2 4.3 0.2 T cell inflamed score 100 -84 <.05 98 -84 <.05 108 -100 <.05 CTLA4 (median TPM) 3.2 0.7 <.05 3 0.8 <.05 3.4 0.7 <.05 LAG3 (median TPM) 6.6 2.1 <.05 6.9 2 <.05 6.8 2 <.05 TP53 mutation % 88 81 <.05 90.5 78.6 <.05 86 83 0.08

e21598 Background: While large language models (LLMs) like GPT-4 (OpenAI) and Claude Opus (Anthropic) perform well on general medical exams, limited data exist on their application in oncology, particularly in multiple-choice question (MCQ) scenarios. This study benchmarks the accuracy and assesses the clinical reasoning of LLMs on cutaneous oncology MCQs from the American Society of Clinical Oncology (ASCO) question bank. Methods: Skin oncology MCQs from the ASCO question bank were tested using GPT-4 (OpenAI) and Claude Opus (Anthropic). Questions were entered into OpenAI and Anthropic API playgrounds without additional prompting, with identical settings for both models (Temperature = 0, Token = Max). Each question was tested three times per model to assess response consistency and precision. Initial responses were collected, and chain-of-thought (COT) prompting was applied to encourage stepwise reasoning mimicking an oncologist’s thought process. The percentage of revised answers after COT prompting was analyzed. Accuracy before and after COT prompting was compared. Incorrect responses were reviewed by board-certified oncologists specializing in skin cancer at National Cancer Institute (NCI)-designated centers. Each oncologist scored the responses on clarity, clinical reasoning, bias, and confidence in the models’ utility in clinical practice. Qualitative feedback was collected via free-text comments and analyzed descriptively. Results: Of 66 skin oncology MCQs, Claude correctly answered 55 questions (83% accuracy, 95% CI: 72%–91%), while ChatGPT correctly answered 53 (80% accuracy, 95% CI: 68%–89%). Both models performed comparably, with overlapping confidence intervals. A chi-square test showed no significant difference between their performances (p = 0.65). After COT prompting, both tools remained at their respective accuracy level. Thematic analysis of incorrect answers revealed that LLM responses frequently deviated from clinical guidelines and failed to reflect evidence-based oncology practice, often neglecting key trial data like DREAMSeq and frontline therapeutic strategies. The most frequent errors were Treatment Planning Errors, where recommendations misaligned with NCCN guidelines, and Interpretation Errors, involving misinterpretation of clinical trials or patient data. Anchoring bias emerged as a recurring issue, with models fixating on details (e.g., BRAF mutation status) at the expense of broader reasoning. Conclusions: This study demonstrates that while large language models like GPT-4 and Claude Opus show promise in the field of oncology, they still face significant challenges in clinical reasoning and adherence to evidence-based practices. Continued refinement and training on specialized medical data are essential before these models can be reliably integrated into clinical workflows.

e21573 Background: In melanoma, biomarkers such as lactate dehydrogenase (LDH), circulating tumor DNA (ctDNA), and the neutrophil-lymphocyte ratio (NLR) have been linked to disease progression. This study aims to compare survival outcomes in melanoma patients based on high versus low levels of these biomarkers, assessing their impact on survival within their respective groups. By evaluating these biomarkers, we hope to identify potential prognostic indicators that could guide treatment decisions and improve personalized care in melanoma management. Methods: We conducted a retrospective survival analysis of 43 melanoma patients, excluding 16 due to incomplete data at Dignity Health Cancer Institute in Phoenix, Arizona. Kaplan-Meier survival analysis was used to compare survival between groups with high vs low levels of biomarkers. Patients were stratified by median values for LDH, NLR, and the presence of detectable ctDNA. Patients included in the analysis were treated systemically with pembrolizumab, encorafenib/binimetinib, nivolumab, or ipilimumab/nivolumab. Survival curves were generated, and statistical significance was assessed using log-rank tests. Data was sourced from our electronic medical record and Signatera™ assay to obtain the primary endpoints. Specimens were standardized in accordance with Natera’s™ specifications. All analyses were executed in GraphPad Prism software. Results: When stratifying by ctDNA, 18 patients had undetectable ctDNA, while 9 had detectable ctDNA with an average level of 1330 (SD = 2067). The median survival for the high ctDNA group was 220 days, whereas the low ctDNA group had significantly longer survival, with only two deaths observed (p = 0.006) within the time period. For LDH, using a cutoff of 204 (IQR = 90.5), patients with high LDH had a median survival of 220 days, while those with low LDH had a significantly longer survival (p = 0.004). When comparing NLR with a cutoff at 2.94 (IQR = 3.7), no significant difference in survival was found between high and low NLR groups (p = 0.72). Conclusions: Our findings show that patients with higher levels of ctDNA and LDH have significantly shorter survival compared to those with low levels of these biomarkers. While NLR did not show a significant survival difference, the study suggests that higher ctDNA and LDH could be important markers for identifying patients at greater risk of poor outcomes. These results underscore the potential value of monitoring these biomarkers to assess disease progression and survival in melanoma patients. Further studies with larger cohorts are needed to validate these findings and refine melanoma prognostic models.

e21600 Background: Uveal melanoma (UM), the most common intraocular malignancy in the U.S., metastasizes in 50% of patients, frequently to the liver (~90%) with poor prognosis (median overall survival [mOS] 1 year). Metastatic UM (mUM) literature is limited (due to its rarity/complexity) and management remains challenging, as standard therapies—liver-directed approaches (e.g. transarterial chemoembolization [TACE]) and systemic therapy with tebentafusp (limited to HLA-A*02:01-positivity)—offer limited efficacy, with no durable responses or significant survival benefits. Interestingly, Delcath’s recent FOCUS Phase 3 trial demonstrated the efficacy of the melphalan/Hepatic Delivery System for unresectable mUM, with an objective response rate (ORR) of 36.3%, 1-year OS of 80%, and median progression-free survival (PFS) of 9 months. Here, we present a large retrospective study of liver-predominant mUM treated with immunoembolization (IE) at a tertiary academic center. This study provides valuable data to the underreported mUM literature. We also compare our findings with Delcath’s trial to contextualize results and identify clinical factors influencing treatment response and prognosis. Methods: We retrospectively reviewed charts of mUM patients treated with IE (age ≥18 years, 2010–2023). Data collected included demographics, clinical/pathology/molecular details, and management/outcomes. Subgroup analysis on prior systemic therapy use was performed (similar to Delcath). The primary outcome was OS at 12 and 24 months. Results: 43 patients were included (100% White, 62.8% female, median age 62). Most had choroidal type (95.3%), hepatic-only metastases (90.7%), and were treatment-naïve before IE (72.1%). Across 249 IE procedures (5.8 per patient, mean duration 9.1 months), we observed median PFS and OS of 9.2 and 23 months, respectively - comparable to Delcath’s trial (9 and 20.5 months). Despite higher disease progression in our study (62.8% vs. 25.3%) and lower best clinical response (27.9% vs. 73.7%), 1-year (67% vs. 80%) and 2-year OS rates (49% vs. 43%) were similar. Subgroup analysis revealed better outcomes in previously treated patients (n = 12) compared to treatment-naïve patients (n = 31), with improved median PFS (29.9 vs. 7.2 months) and 1- and 2-year OS (75% vs. 65% and 58.3% vs. 41.9%, respectively). This contrasts with Delcath’s findings, where prior therapy did not significantly affect PFS or OS. Median PFS was comparable in treatment-naïve groups (7.2 vs. 9 months) but markedly improved in our previously treated group (29.9 vs. 9.2 months). Conclusions: We highlight IE’s comparable efficacy to Delcath’s melphalan/Hepatic Delivery System regarding mUM PFS and OS, with added benefits for previously treated patients. While our findings advance understanding of mUM management, more prospective studies are needed to improve therapeutic strategies/patient outcomes.

2562 Background: Pancreatic ductal adenocarcinoma (PDAC) is highly lethal cancer with limited immunotherapeutic options. Pelareorep (pela) is an intravenously delivered unmodified reovirus containing a double stranded RNA genome that has been studied as an immunotherapeutic in multiple cancers including breast, anal, colorectal and pancreatic. We previously reported high tumor response rates in first-line metastatic PDAC patients treated with pela combined with gemcitabine, nab-paclitaxel and atezolizumab. We report here the immunologic effects of pela in a cohort of first-line metastatic PDAC subjects treated with pela plus chemotherapy and atezolizumab and the correlation of these effects with tumor response. Methods: To examine its effects on pancreatic cancer, a phase 1/2 Simon 2-stage platform study (GOBLET) was performed that included patients with first-line locally advanced/metastatic unresectable PDAC. Anti-reovirus T cell activity was assessed by interferon gamma secretion (ELISPOT). Changes in the expression of plasma proteins were analyzed by Olink Response panels. T cell receptor sequencing (TCR-seq; ImmunoSEQ Assay, Adaptive Biotechnologies) was performed on tissue collected prior to the start of therapy and on blood from baseline through 3 treatment cycles to identify TILs expansion. Tumor responses were scored according to the modified RECIST v1.1 criteria. Results: Increases in anti-reovirus T cell activation as determined by ELISPOT after cycle 3 of therapy were observed in 6/8 subjects. Three subjects with maximum responses (> 300 spots) showed >30% decreases in tumor volume. Significant changes in plasma proteins as determined by Olink included PD-L1, CXCL9, CXCL10, CXCL11 and IFN-γ. Pre-treatment tumor tissue was used to identify the TIL clonal populations prior to therapy. Increased TIL clones in the blood pre-treatment was associated with tumor responses. Pela treatment increased the expansion of pre-existing and new TIL clones in the blood after one cycle of treatment. Sustained increases in pre-existing TIL clonal populations in the blood through cycle 3 of therapy were observed in subjects who exhibited reductions in tumor volume. Conclusions: These findings, while preliminary, demonstrate that pela induces not only anti-reovirus T cells but also activates innate and adaptive anti-tumor immunity in PDAC subjects treated with chemotherapy and atezolizumab. Tumor responses are associated with both the presence of TIL clones in the blood prior to treatment and the expansion of pre-existing TILs in the blood on treatment. These findings provide additional insights into the immunologic mechanisms by which pela-based therapy may provide clinical benefit in patients with metastatic PDAC.

10051 Background: CBL0137 is a novel agent that targets the FAcilitates Chromatin Transcription complex (FACT), a histone chaperone that regulates chromatin remodeling during transcription, replication, and DNA repair. CBL0137 has been shown to have anti-tumor activity in multiple preclinical models of pediatric cancer. We report the phase 1 trial of CBL0137 in children with relapsed or refractory solid tumors including central nervous system (CNS) tumors or lymphoma (NCT04870944). Methods: Patients (age 12 months to 21 years) with relapsed/refractory solid tumors, central nervous system tumors or lymphoma were eligible. In the dose escalation phase, a rolling-six design was used to evaluate CBL0137 administered intravenously (IV) once per week on Days 1 and 8 of a 21-day cycle. The starting dose was 400 mg/m 2 , with escalation to the adult phase 2 dose (RP2D) of 540 mg/m 2 . The maximum tolerated dose (MTD) was determined based on cycle 1 dose limiting toxicity (DLT) using Common Toxicity Criteria for Adverse Events (v5). Pharmacokinetics (PK) and cytokine analyses were performed during cycle 1. Seven additional patients ( < 18 years old) were accrued to a PK expansion cohort at the RP2D. Results: Sixteen patients were enrolled; 14 were evaluable for DLT assessment (12 at 400mg/m 2 [6 in dose escalation and 6 in PK expansion] and 2 at 540mg/m 2 ). The median (range) age was 10 (4-20) years. Diagnoses included high grade glioma (n = 6), osteosarcoma (n = 5), ependymoma (n = 2), neuroblastoma, hepatoblastoma and Burkitt’s lymphoma (1 each). One of 12 evaluable patients treated at 400 mg/m 2 experienced a DLT: Grade 3 photosensitivity. At 540 mg/m 2 , the two evaluable patients both had fever and dose limiting Grade 3 hypotension. Non-dose limiting Grade 3-4 toxicities included anemia, lymphopenia, neutropenia, thrombocytopenia, hypokalemia, anorexia, photosensitivity and fever. For both dose levels, a total of 12 patients were reported to have at least one episode of Grades 1-3 fever and/or Grade 1 cytokine release syndrome (CRS). PK parameters (mean±SD) for CBL0137 (400 mg/m 2 ) were T max = 0.7 ± 0.5 h, C max = 1310 ± 417 ng/mL, and AUC 0-24h = 15300 ± 6790 h·ng/mL. Given the unexpected toxicities of fever, hypotension, and CRS, cytokine samples were collected in patients enrolled in the PK expansion cohort; all had Grade 1 CRS and significantly elevated levels of both IL-10 (adj p = 0.001) and IL-1RA (adj p < 0.001) at 24 hours vs pre-infusion. Conclusions: The RP2D and MTD of CBL0137 in children and adolescents with solid or CNS tumors is 400mg/m 2 IV weekly on Days 1 and 8 of 21-day cycles. CBL0137 leads to immune activation, with cytokine elevation and the possibility of CRS, an unexpected toxicity not previously reported in adults. A Phase 2 cohort in patients with Diffuse Midline Glioma is ongoing and includes further assessment of immune activation. Clinical trial information: NCT04870944 .