307 Background: Human epidermal growth factor receptor 2 (HER2) plays a vital role in cell proliferation, apoptosis, and ultimately tumorigenesis. HER2 targeted treatment (HER2 tx) has improved survival for patients with HER2 amplified tumors; however, its role in HER2 nonamplified mutated GI cancers is not well established. The objective of this study is to determine trends in survival and response to HER2 tx for GI cancers based on tumor location and non-amplified HER2 mutation and HER2 amplification/overexpression status. Methods: The retrospective, multicenter analysis used the Flatiron Health Research Database, a deidentified electronic health record-derived database of real world (rw) patients with advanced GI cancers who received HER2 tx. Patients were categorized as HER2mut+ or HER2mut- based on the presence or absence of a definitely pathogenic or likely pathogenic HER2 non-amplification mutation. HER2Amp+ or HER2Amp- was defined by the presence or absence of HER2 amplification/overexpression (either HER2 gene amplification or 3+ by IHC and 2+/FISH+). Index date was defined by the start of HER2 tx. The cox proportion hazards models were used to compare rw overall survival (OS) and progression-free survival (PFS) across subgroups. Results: A total of 858 patients who received HER2 tx were included [esophagogastric (EGC) n=709 and non-EGC (colorectal, pancreatic, and HCC) n=149]. In HER2mut+ EGC, OS was longer among those with HER2Amp+ compared to HER2Amp-, with an OS of 17 and 7.1 months, respectively. Among HER2Amp+ EGC, there was no statistically significant difference in OS between HER2mut+ and HER2mut-. Similarly, there was no difference in PFS for HER2mut+/HER2Amp+ and HER2mut-/HER2Amp+ (7.4 vs. 7.4 mo). In the non-EGC cohort, there was a non-statistically significant trend towards worse OS and PFS with HER2mut+/HER2Amp+ compared to HER2mut-/HER2Amp+ (OS 5 vs. 16 mo; PFS 2.2 vs. 4.6 mo). Conclusions: Our study identifies a lack of benefit with HER2 tx in HER2mut+ EGC in the absence of HER2 amplification/overexpression. This highlights the importance of HER2 amplification rather than presence of a HER2 non-amplifying mutation for EGC. In EGC, coexisting HER2 mutation does not have a negative effect on HER2 tx efficacy in the setting of HER2 amplification; however, there may be a negative impact with coexisting HER2 mutations in the setting of HER2 amplified non-EGC. Further investigation is needed to determine if specific HER2 mutations alone can be targeted to induce a treatment response.

Supplementary Table 1. Statistical Assumption and Sample Size Estimates for MUCE Design

769 Background: Patients with metastatic pancreatic cancer and pathogenic germline BRCA mutations have improved survival with platinum-based chemotherapy and maintenance PARP inhibitor therapy. However, little is known about the impact of somatic BRCA mutations and BRCA Variants of Uncertain Significance (VUS) on treatment patterns and survival. Methods: The Flatiron Health Research Database contains de-identified patient-level data originating from over 280 cancer clinics in the United States. Using patient-level data from the Flatiron Health database collected between 2017 and 2024, we conducted multivariate analyses to evaluate the association of BRCA mutations with survival outcomes and treatment patterns in metastatic pancreatic cancer patients. A study protocol was developed prior to the implementation of the study. Results: Relative to benign variants, an overall survival benefit of germline and somatic pathogenic BRCA mutations was demonstrated. Multivariate Cox proportional analysis revealed a hazard ratio (HR) of 0.54 (95% CI: 0.41–0.70), p<0.001 for germline pathogenic and 0.61 (95% CI: 0.45–0.81), p<0.001 for somatic pathogenic mutations. Trend-level effects in germline variants of uncertain significance (VUS): HR = 0.74 (95% CI: 0.48–1.12), p=0.157, and somatic VUS: 0.82 (95% CI: 0.66–1.01), p=0.067 were also noted. Patients with germline pathogenic BRCA mutations were also more likely to receive first-line platinum therapy than their benign comparators. Multivariable OR: 2.12 (95% CI: 1.04 - 4.50), p=0.042. Other mutation categories, including VUS and somatic mutations, were not associated with different treatment patterns. Patients with germline pathogenic BRCA mutations had a longer time to next treatment (TTNT) than benign: TTNT = 6.35 months (95% CI: 4.5-7.82), p=0.003. Benign = 4.8 months (95% CI: 4.57-5.03). Somatic mutations did not have a significant relationship with treatment duration, but Germline VUS did show a trend-level effect: TTNT = 6.02 months (95% CI: 3.48-10.98), p=0.061. Patients with a germline pathogenic mutation were found to have significantly higher odds of receiving more systemic therapies: HR = 2.67 (95% CI: 1.43–4.93), p=0.002. Other categories were statistically indistinguishable from benign. Conclusions: Pathogenic BRCA mutations, both somatic and germline, are independently associated with enhanced survival outcomes in patients with metastatic pancreatic cancer. Treatment patterns do differ between groups. This data warrants further exploration but suggests patients with somatic BRCA mutations may benefit from the increased sensitivity to therapies as seen in patients with germline pathogenic BRCA alterations. Summary of key population and mutation demographics. Benign GermlinePathogenic GermlineVUS SomaticPathogenic SomaticVUS Total Total N (%) 2407 (86.0) 103 (3.7) 40 (1.4) 101 (3.6) 149 (5.3) 2800

e15572 Background: The treatment landscape of metastatic colorectal cancer (mCRC) has evolved with identification of actionable biomarkers. However, despite inclusion of biomarker-directed therapies in NCCN guidelines, real-world uptake remains poorly defined. We evaluated predictors of receipt of biomarker-directed therapy in a large, real-world cohort of patients with BRAF V600E-mutated, KRAS G12C-mutated, and RAS/RAF wild-type (WT) mCRC. Methods: In this retrospective cohort study, we analyzed the Flatiron Health Research Database. Cohorts included patients with BRAF V600E-mutated, KRAS G12C-mutated, and RAS/RAF WT mCRC. The primary outcome was receipt of any NCCN-recommended biomarker-directed therapy at any line since FDA-approval. Univariable associations between demographic and clinical factors and therapy receipt were assessed within each cohort. Multivariable logistic regression estimated adjusted odds ratios (ORs) for treatment receipt, incorporating age, sex, baseline ECOG, primary tumor site, insurance, and stage. Results: A total of 12,226 patients were included. Among 1,336 patients with BRAF V600E, 26.2% (n = 350) received BRAF-targeted therapy. Treated patients were younger (median 66 vs 72 years, p < 0.001) and had better performance status. In multivariable analysis, increasing age (OR 0.96 per year, p < 0.001), ECOG 2 vs 0 (OR 0.60, p = 0.036), and ECOG 3 vs 0 (OR 0.19, p = 0.002) were independently associated with lower odds of therapy receipt. Race and tumor site were not associated with treatment receipt after adjustment. Among 93 patients with KRAS G12C, 10.8% (n = 10) received KRAS-targeted therapy. No significant predictors of receipt were identified, likely due to limited sample size. Among 10,797 patients with RAS/RAF WT mCRC, 44.5% (n = 4,803) received anti-EGFR targeted therapy. In multivariable analysis, older age (OR 0.98 per year, p < 0.001) and worse performance status (ECOG 2 vs 0: OR 0.79, p = 0.006; ECOG 3 vs 0: OR 0.70, p = 0.021) were associated with lower odds of treatment receipt. Male sex (OR 1.27, p < 0.001), rectal primary tumor site (OR 1.15, p = 0.008), Medicare insurance (OR 1.26, p = 0.003), and uninsured/self-pay status (OR 1.48, p < 0.001) were associated with higher odds of receiving targeted therapy, while race, ethnicity, and SES were not independent predictors. Conclusions: Receipt of guideline-recommended biomarker-directed therapy was suboptimal, particularly for BRAF V600E and KRAS G12C mutations. Older age and worse performance status consistently predicted lower treatment receipt, suggesting clinical fitness drives the gap between biomarker identification and therapy. Notably, nearly three-quarters of patients with BRAF V600E mutations did not receive targeted therapy. These findings reveal underutilization of precision oncology in mCRC and highlight the need for interventions to improve access to therapy for eligible patients.

PURPOSE:The addition of MEK inhibition to PD-L1 blockade improves progression-free survival (PFS) in patients with advanced biliary tract cancer (BTC). While MEK inhibitors may increase tumor cell immunogenicity, they can impair T-cell priming/effector function, limiting combination efficacy. We hypothesized that the addition of a CD27 agonist could restore T-cell function and enhance anti-tumor immunity in this combination. PATIENTS AND METHODS:We conducted a randomized, phase 2 trial evaluating atezolizumab (840mg IV days 1,15) in combination with the CD27 co-stimulatory mAb (CDX-1127/varlilumab [3mg/kg IV days 1, 15]), with/without the addition of a MEK inhibitor (cobimetinib [60mg oral daily days 1-21, off 22-28]) in unresectable BTC following at least 1 metastatic therapy. Overall response rate (ORR) and PFS were co-primary endpoints. Treatment-related changs in CD8+ tumor infiltrating lymphocytes (TIL) was the primary correlative outcome. RESULTS:The trial was closed early following interim preplanned ORR analysis. At closure, 57 patients had been enrolled (n=29[CAV],n=28[AV]. A majority (67%) had intrahepatic cholangiocarcinoma and 32% were immunotherapy-experienced. Both regimens were well-tolerated without new safety signals. Objective responses were rare (0%[CAV],3.8%[AV]). Median PFS was 2.40(CAV) and 1.84(AV) months (HR 0.67,95%CI[0.38,1.18]). Among immunotherapy-experienced patients, mPFS was 3.62(CAV) and 1.84(AV) months (HR 0.54,95%CI:[0.18-1.62]).  Treatment with CAV increased intratumoral CD8+ T-cell density compared with treatment with AV. CONCLUSIONS:The combinations of atezolizumab and varlilumab with/without cobimetinib were safe but neither meaningfully improved outcomes in BTC treated in the later lines. Correlative tissue studies validated preclinical work that MEK inhibition increases CD8+TILs.

BACKGROUND:Cysteine-cysteine chemokine receptors 2 (CCR2) and 5 (CCR5) contribute to immune suppression in tumor microenvironments. CCR2 and CCR5 antagonists have demonstrated antitumor activity in pancreatic ductal adenocarcinoma (PDAC) and colorectal cancer (CRC), respectively. This phase 1b/2, open-label study evaluated BMS-813160, a CCR2/5 dual antagonist, in combination with chemotherapy±nivolumab in advanced PDAC or metastatic CRC. METHODS:Part 1 included patients with metastatic untreated (first-line (1L)) PDAC, 1L CRC, or previously treated (second or third line (2/3L)) microsatellite stable (MSS) CRC. Patients received 2 weeks of BMS-813160 monotherapy (300 mg two times a day, 600 mg once daily, 300 mg once daily, or 150 mg once daily) and then BMS-813160+chemotherapy (gemcitabine+nab-paclitaxel (gem/nabP; 1L PDAC), 5-fluorouracil+leucovorin+irinotecan (FOLFIRI; 1L CRC)), or nivolumab (2/3L MSS CRC).Part 2 included patients with metastatic 1L PDAC or 2L CRC. Patients received BMS-813160 300 mg two times a day+gem/nabP±nivolumab (1L PDAC), BMS-813160 300 mg two times a day or 150 mg once daily+FOLFIRI (2L CRC), or chemotherapy alone. Primary endpoints were safety and pharmacodynamics (Part 1) and efficacy (Part 2). RESULTS:In Part 1, 22 of 75 (29%) and 54 of 72 (72%) patients experienced a treatment-related adverse event during monotherapy lead-in and overall, respectively. Two dose-limiting toxicities (rash and pericardial effusion with pericarditis, both grade 3) occurred. In Part 2, patients with 1L PDAC who received BMS-813160 300 mg two times a day+gem/nabP+nivolumab achieved an overall response rate (ORR) of 37% (13/35); the median duration of response (DOR) was 45 weeks (95% CI 26.1 to not evaluable). ORRs with BMS-813160 300 mg two times a day+gem/nabP and gem/nabP alone were 26% (9/35) and 28% (9/32), respectively; median DORs were 121 and 31 weeks, respectively. Progression-free survival rates at 24 weeks were 56% (BMS-813160 300 mg two times a day+gem/nabP+nivolumab), 56% (BMS-813160 300 mg two times a day+gem/nabP), and 50% (gem/nabP). ORRs in 2L CRC were 19% (6/32; BMS-813160 300 mg two times a day+FOLFIRI), 13% (4/32; BMS-813160 150 mg once daily+FOLFIRI), and 27% (7/26; FOLFIRI). CONCLUSIONS:In 1L PDAC, BMS-813160 300 two times a day+gem/nabP±nivolumab demonstrated durable antitumor response and was well tolerated. BMS-813160 combination regimens were tolerable in other cohorts, but clinical efficacy was not demonstrated. TRIAL REGISTRATION NUMBER:NCT03184870.

e16326 Background: Well- to moderately differentiated gastrointestinal and pancreatic neuroendocrine tumors (NETs) show substantial heterogeneity in clinical outcomes. While tumor-related characteristics are established prognostic factors, the impact of patient-level comorbidities, metabolic factors, and sociodemographic characteristics on overall survival (OS) remains incompletely described. We evaluated OS across clinical, metabolic, renal, and demographic subgroups in a real- world cohort of patients with well- to moderately differentiated NETs. Methods: We conducted a retrospective cohort study using a harmonized clinical dataset spanning diagnosis, physical assessment, laboratory, and demographic domains. The cohort included adult patients with well- to moderately differentiated NETs of pancreatic and gastrointestinal origin; poorly differentiated tumors were excluded. Baseline characteristics, including body mass index (BMI, categorized as < 25 vs. ≥25 kg/m²), smoking status (ever vs. never), age at primary diagnosis, and race/ethnicity, were extracted. Advanced kidney disease (AKD) was defined as serum creatinine > 1.5 mg/dL on ≥2 occasions. OS was estimated with the Kaplan-Meier method, and differences were assessed with the log-rank test. Results: The cohort included 132 patients (Oncology Research Information Exchange Network, ORIEN). Overall survival was significantly worse in patients aged ≥50 years, with a shorter median OS than in patients younger than 50 years. For patients who were 50 or older at primary diagnosis, the median overall survival was 14.46 years, whereas for those younger than 50, the median OS was not reached(p = 0.046). There was no significant difference in median OS based on other clinical factors, including BMI, smoking status, advanced kidney disease, and race/ethnicity. Conclusions: In this real-world analysis of patients with GEP-NETs, age was the dominant predictor of overall survival.Traditional metabolic and renal risk factors did not demonstrate significant prognostic value in this cohort, suggesting that tumor biology or age-related background mortality may outweigh these comorbidities in this setting. Future studies utilizing larger, multi-institutional datasets are warranted to perform multivariate adjustments and isolate disease-specific survival outcomes.

594 Background: The incidence of hepatocellular carcinoma (HCC) diagnoses is increasing worldwide with an estimated 1 million new cases of liver cancer in the year 2025. HCC is commonly diagnosed through imaging and clinical evaluation alone. Many patients with HCC never undergo biopsy or complete molecular testing. As a result, molecular profiling of large numbers of HCC patients has not been reported. This impairs our ability to identify mutations that may be treated with approved targeted therapies or could be targeted in future clinical trials. Methods: Utilizing the Flatiron Health-Foundation Medicine Clinico-Genomic Database (CGDB), we collected tissue and circulating tumor DNA-derived somatic testing results from 447 patients with HCC. We evaluated the frequency at which alterations were discovered, excluding alterations present in less than 1% of patients. Univariate analysis using Fischer’s exact testing was then undertaken to evaluate if alterations were present in different proportions of patients with distinct HCC risk factors. The number of alterations associated with an FDA approved targeted therapy in another cancer type was determined. Alterations currently being investigated with therapeutic agents in clinical trials enrolling HCC patients was also determined using clinicaltrials.gov as a reference. Results: Molecular data from 447 HCC patients was analyzed. Fifty six alterations were present in over 1% of individuals. Genes altered in at least 3% of patients included TERT (26.5%), CTNNB1 (19.7%), TP53 (18.8%), MYC (13%), ARID1A (5.6%), CDKN2A (5.4%), RB1 (5.1%), CCND1 (4.9%), RAD21 (4.9%), LYN(4.5%), DNMT3A (4%), FGF19 (4%), PTEN (4%), FGF3 (3.6%), MLL2 (3.4%), TSC2 (3.4%), FGF4 (3.1%), and NTRK1 (3.1%). A higher proportion of patients with hepatitis C were found to have alterations in TERT (p = 0.019) but less likely to have an RB1 alteration (p = .003). Patients with obesity had a higher proportion of CCND1 mutations than non-obese patients (p= .035). FDA approved targeted therapies in other cancer types are available for 10 of the 56 alterations present in over 1% of patients and there are clinical trials evaluating targeted therapies that are actively recruiting patients with HCC for 5 of the alterations including CTNNB1, FGF19, VEGFA, ERBB2, and APC. Conclusions: There are a significant number of genetic alterations present in hepatocellular carcinoma patients that may be targetable with approved therapies available in other tumor types or clinical trial treatments. Additional research assessing the efficacy of existing targeted treatments and trial therapies in HCC patients is greatly needed.

669 Background: The incidence of neuroendocrine tumors (NETs) is approximately 7 per 100,000 persons and rising. By location, tumors in the gastroenteropancreatic (GEP) region, particularly midgut NETs, are most common. The Oncology Research Information Exchange Network (ORIEN) database contains complementary clinical, genomic, and transcriptomic profiling, providing opportunities to identify novel associations between molecular features and clinical outcomes in GEP-NETs. Methods: Survival analyses were performed using the log-rank testing, and clinical features were evaluated using Wilcoxon and chi-squared tests. Mutational analyses utilized sample-level enrichments from whole exome sequencing data, and statistical tests were performed using the one-sided Fisher Exact test. Transcriptomic analyses utilized a student’s t-test. We reviewed 240 samples from 226 patients, with 124 of pancreatic origin, 90 small bowel, 16 gastric, and 10 colorectal. All specimens were well-differentiated; further grading information was unavailable on the ORIEN platform. A p-value <0.05 was considered significant. Results: Patients with metastatic disease were significantly younger at diagnosis (58 vs. 61 years, p=0.0242) and had significantly higher tumor mutational burden (TMB) (0.48 vs. 0.32 mut/Mb, p=0.001). Among the 100 most common alterations across the entire cohort, BPTF (p=0.03) and PRKCA (p=0.007) mutations were more prevalent in pancreatic NETs than in other primary sites. The most frequently mutated genes identified included TTN (27.7%), MUC16 (25.1%), CCDC168 (22.1%), KIR3DL1 (20%), TGIF2LX (20%), MUC17 (19.1%). TTN mutation was significantly associated with higher TMB (p<0.0001). The most common copy number alteration was MUC3a amplification (7q22.1) in 61.5% of samples. This amplification was associated with more prolonged overall survival (OS) with a significance of p=0.0501. Patients with NETs harboring a mutation in CSNK2A2 (p=0.0149), WDR74 (p=0.0386), or NCMAP (p=0.0248) experienced significantly shorter OS compared to the cohort as a whole. Conclusions: We report the first clinical, genomic, and transcriptomic analysis of ORIEN GEP-NET cases. These findings create multiple avenues for further investigation and reinforce the value of multi-institutional consortia such as ORIEN in deepening our knowledge of well-differentiated NETs. Most prevalent mutations in GEP-NETs from ORIEN database. Mutation Prevalence (Percentage of Samples) TTN 27.7% MUC16 25.1% CCDC168 22.1% KIR3DL1 20.0% TGIF2LX 20.0% MUC17 19.1%

761 Background: A novel targeted therapy designed to restore wild-type function to the p53 Y220C mutant protein has demonstrated promising activity in an early-phase trial (NCT04585750). While the TP53 Y220C mutation is found in < 1% of all solid malignancies, its prevalence is enriched in certain cancers including PDAC. Here we report results of comprehensive genomic profiling (CGP) in a cohort of patients with advanced PDAC, with a focus on TP53 Y220C mutations. Methods: A total of 27,377 cases of clinically advanced PDAC underwent hybrid capture-based CGP to assess all classes of genomic alterations (GAs). Cases were sequenced to a mean coverage depth of 650X. Microsatellite instability (MSI) status and tumor mutation burden (TMB) were determined using sequencing data; PD-L1 expression was measured by immunohistochemistry (Dako 22C3, tumor proportion score [TPS]). Results: TP53 Y220C alterations were identified in 488 (1.8%) PDAC patients (Y220C+). In comparison to PDAC patients lacking the TP53 Y220C alteration (Y220C-), Y220C+ patients were of similar age, but were more likely to be female (52.9% vs 46.8%; p = 0.0008) and harbor KRAS mutations (96.7% vs 92.7%; p = 0.007). Meanwhile, alterations in ATM were more frequent in the Y220C- group (4.0% vs 1.4%; p = 0.0008). Frequencies of other GAs were similar between Y220C+ and Y220C- groups, including alterations in the homologous recombination defect-associated genes BRCA1/2 and RAD21 , as well as ERBB2 , BRAF , MTAP and PIK3CA . MSI-high status and elevated TMB (≥ 10 mutations/Mb) were extremely uncommon in both groups. PD-L1 expression was similar in the Y220C+ and Y220C- groups, with positive PD-L1 expression (TPS ≥ 1) observed in 31.5% and 34.9% of patients, respectively. Conclusions: The potentially targetable TP53 Y220C mutation was identified in close to 2% of clinically advanced PDAC cases. Those with TP53 Y220C mutations were more likely to be female and harbor KRAS mutations, and less likely to harbor ATM mutations. However, TP53 Y220C mutations were not associated with other distinct genomic characteristics. These findings are consistent with previous reports of interactions between TP53 and KRAS alterations in PDAC. Given the emergence of drugs aimed at restoring p53 function in cases with the Y220C mutation, further study of this alteration in PDAC is warranted. Age, number of genomic alterations (GA) per tumor, and alteration frequency of select commonly altered genes based on TP53 Y220C status. TP53 Y220C+ (n = 488) TP53 Y220C- (n = 26,889) P Value Median Age, years (range) 66 (36-89+) 66 (21-89+) NS GA/tumor 5.2 5.0 NS KRAS 96.7% 92.7% 0.007 CDKN2A 60.1% 56.6% NS CDKN2B 28.6% 29.5% NS MTAP 24.4% 23.8% NS ARID1A 7.8% 8.7% NS ATM 1.4% 4.0% 0.0008 BRCA2 2.3% 3.1% NS NS = non-significant.

Background:Disparities in cancer outcomes persist between racial, ethnic, and socioeconomic groups. One potential cause is lack of appropriate representation in dose-finding clinical trials. We investigated the extent of disparities in phase I clinical trials and recent changes in the setting of institutional efforts to mitigate disparities, legislative interventions, FDA guidance for sponsors and the COVID-19 pandemic. Methods:We performed a retrospective review of patients enrolled in phase I clinical trials at the University of Colorado Cancer Center in 2018-2019 and 2022-2023. We collected demographics, area deprivation index (ADI), tumor type and other clinical variables. Differences between cohorts were evaluated with t-tests, chi-Square test, or Fisher exact test. Progression-free survival (PFS) and overall survival (OS) were calculated using the Kaplan-Meier method. Hazard ratios (HR), confidence intervals (CI) and p-values were derived using the Cox-proportional hazards method. Results:A total of 361 patients were included (209 and 152 in the 2018-2019 and 2022-2023 cohorts, respectively). The population consisted of 85.0% White, 3.3% Asian, 1.4% Black, 0.3% Native Hawaiian or Pacific Islander and no American Indian/Alaskan Native (AIAN) patients by race, and 9.1% Hispanic by ethnicity. The most common tumor type was colorectal cancer (18.3%). Compared to 2018-2019, we observed increases in non-English speakers from 1.9% (4/209) to 6.6% (10/152) (p = 0.028) and in translated informed consent forms (ICFs) from 1.4% (3/209) to 5.9% (9/152) (p = 0.033) in 2022-2023. There were no significant changes in race, ethnicity, insurance, or tumor type, although there was a moderate increase in Hispanic patients from 8.1% to 10.5%. There were no differences in clinical outcomes by race, ethnicity, or ADI scores in the overall study population. However, in the most common cancer type, colorectal cancer, higher ADI scores were associated with decreased median PFS and OS. Conclusion:The interventions resulted in an increase in accrual of non-English speaking patients, however, there was not yet a significant change in overall race and ethnicity. Our study confirms poorer outcomes for patients with higher ADI scores. Further research is warranted to understand disparities in clinical trial accrual, and intervention is needed to improve outcomes for disadvantaged patients.

Supplementary Table 4. Treatment-related adverse events by CTCAE v5.0, grade 1-4 occurring in at least 10% of patients and all grade 5, at the patient level.

Supplementary Figure S6. Survival outcomes of the trial.

Supplementary Figure S10. Serial peripheral blood mass cytometry by patient.

Supplementary Table 1. Flow cytometric Abs and Reagents

Supplementary Figure S2. Human and murine immune system in peripheral lymphatic organs and tumor of HIS-BRGS mice bearing CRC307P PDX.

Depth and duration of response by subject. A, Waterfall plot of best percent change in aggregate size of target lesions. B, Swimmer plot of duration of response. As of the data cutoff, subject 36-002 remained on study. In (A), subject 30-002 is not included due to a 35% decrease in target lesions but unequivocal progression in nontarget lesions. PI, principal investigator.

PURPOSE:Rising rates of early-onset colon cancer (EOCC) present challenges in deciding how to optimally treat patients. Although standard of care for stage II CC is surgical resection, adding chemotherapy for high-risk disease, evidence suggests treatment selection may differ by age. We investigated whether adjuvant chemotherapy (AC) administration rates differ between patients with early- and later-onset stage II CC. METHODS:Data originated from the nationwide Flatiron Health electronic health record (EHR)-derived deidentified database spanning January 1, 2003, to August 1, 2021. Adults with stage II CC were grouped as age 18-49 years (EOCC) and those age 50 years or older (later-onset colon cancer [LOCC]). Demographics, Eastern Cooperative Oncology Group score, tumor stage and site, and chemotherapy were included. Primary outcomes included rates of AC administration by age and ethnicity; secondary outcomes included overall survival (OS) and time to metastatic disease (TTMD). Univariate and multivariable logistic regression models evaluated relationships between chemotherapy administration, age, and ethnicity, adjusting for significant covariates. RESULTS:One thousand sixty-five patients were included. Median age of patients with EOCC was 45.0 years versus 69.0 years for patients with LOCC. Adjusted multivariate analysis showed patients with EOCC received AC significantly more often than patients with LOCC. Non-Hispanic patients received AC at significantly lower rates than Hispanic patients in both cohorts. Subanalysis of stage IIA patients showed that patients with EOCC were more likely to receive AC than patients with LOCC. No significant differences in OS or TTMD were observed by age regardless of AC administration in stage II overall; however, patients with stage IIA EOCC receiving AC had significantly longer TTMD than those not receiving AC. CONCLUSION:AC was given preferentially in stage II EOCC, even in stage IIA, despite deviation from guidelines. This may expose low-risk patients to unnecessary toxicities and suggests bias toward treating younger patients more aggressively, despite unclear evidence for better outcomes.

Treatment-emergent adverse events by Common Terminology Criteria for Adverse Events v5.0, grade 1–4 occurring in at least 10% of patients and all grade 5, at the patient level.