Abstract:Well-differentiated grade 3 neuroendocrine tumors (NETs) have recently been described as a distinct category, and randomized data regarding efficacy of therapy for these patients are scarce. In the phase 3 CABINET trial, cabozantinib improved PFS compared with placebo in patients with advanced, previously treated, progressive extra-pancreatic NETs (epNETs) and pancreatic NETs (pNETs) of all grades. Here, we evaluate if these results remain consistent in a subgroup of patients with well-differentiated G3 NETs. Patients with locally advanced or metastatic epNETs or pNETs were randomized 2:1 in independent cohorts to receive cabozantinib 60 mg daily vs placebo. We analyzed outcomes of the subset of patients with G3 NETs (Ki-67 > 20%), combining patients in the pNET and epNET cohorts due to small sample sizes. Twenty-four patients had G3 NETs, 16 randomized to cabozantinib and 8 to placebo. Primary sites included pancreas (n = 12), GI tract (n = 7), unknown primary sites (n = 3), and lung/thymus (n = 2). Median PFS for patients with G3 NETs treated with cabozantinib was 7.9 vs 3 months with placebo (HR = 0.15, 95% CI: 0.04-0.57, 1-sided log-rank P = 0.0034). The confirmed overall radiographic response rate was 25% (4/16) with cabozantinib vs 0% (0/8) with placebo. Safety outcomes were consistent with published data for the trial as a whole. Subset analysis of the CABINET trial showed improved PFS associated with cabozantinib vs placebo for G3 NETs of pancreatic and extra-pancreatic origin. Despite limited numbers, these results suggest that cabozantinib can be an effective option for patients with advanced G3 NETs. ClinicalTrials.gov Identifier: NCT03375320.

3609 Background: Detection of circulating tumor DNA (ctDNA) is prognostic for recurrence after completion of curative-intent therapy for patients (pts) with colorectal cancer (CRC). Our MD Anderson INTERCEPT platform enrolls ctDNA(+) CRC pts onto matched clinical trials, with the goal of eradicating minimal residual disease (MRD) and curing more pts with CRC. To gain insights on how to identify CRC pts most likely to have true MRD, we retrospectively reviewed the role of non-specific radiographic abnormalities in forecasting CRC recurrence. Methods: For this retrospective review of CRC pts included in ctDNA intervention trials (Table) from 1/2019-12/2025, MRD at baseline was defined as (1) having no radiographically evident disease at time of study enrollment and (2) detection of ctDNA on an MRD assay (Signatera). ctDNA levels were quantified as mean tumor molecules (MTM) per mililiter (mL) plasma. Radiographic sites of disease progression on clinical trials were compared with pre-treatment imaging. Median recurrence-free survival (RFS) was estimated using Kaplan Meier and compared with Cox regression. Results: 54 pts with definitively treated CRC prior to subsequent ctDNA(+) were enrolled across 4 trials (Table); 45 pts (83%) had stage IV disease. No pts had radiographically evident CRC at trial start. Median ctDNA levels were 0.45 MTM/mL (IQR 0.12 – 2.4). After a median follow up of 6.4 months (IQR, 3.5-9.5), 42 (78%) pts had recurred after treatment on a ctDNA intervention trial. Median RFS was 8.1 months (95% CI 4.4 - 9.4). Sites of recurrence were liver (N=20, 48%), lung (N=13, 31%), lymph nodes (N=10, 24%), and peritoneum (N=3, 7%). Upon retrospective review, minute but discernible lesions were identified at the sites of recurrence in 25 (60%) of the pts who had a recurrence on MRD trials. At eligibility evaluation these were characterized as “benign” in 14 (56%) or as “post-procedural” in 11 (44%) of the cases. The presence of indistinguishable CRC deposits at the start of the trial was associated with inferior RFS (4.2 vs 9.3 months; HR 1.96 (95% CI 1.04-3.70; p = 0.04). Conclusions: Subclinical radiographic findings may confound true identification of pts with CRC MRD for clinical trials. Optimizing definitions of radiographically detectable disease is warranted in order to ensure homogenous populations of CRC that are the most likely to benefit from novel MRD therapies. Demographic characteristics across four trials. Number of Patients Percentage (%) Age at Enrollment < 30 1 2.4 30-39 3 7.3 40-49 8 19.5 50-59 12 29.3 60-69 14 34.1 70+ 3 7.3 Gender Male 34 63.0 Female 20 37.0 Race/Ethnicity White or Caucasian 34 63.0 Black or African American 5 9.3 Hispanic or Latino 11 20.4 Asian 1 1.9 Other 3 5.6 Stage at Enrollment I 0 0.0 II 0 0.0 III 9 16.7 IV 45 83.3 Trial Bintrafusp alfa 4 7 NK cells with cetuximab 15 27.8 Lifestyle interventions 16 29.6 Trifluridine/tipiracil 19 35.2

Supplementary Table 1: Details of each P/LP detected in the high-risk cohort along with the high-risk criteria

241 Background: Despite occurring in only 1-3% of metastatic colorectal cancer (mCRC) patients, brain metastases predict a strikingly poor median overall survival (mOS) of 2.6-7.4 months. There is limited data regarding the clinicopathological characterization of brain metastases (BMs) in mCRC. Although potential predictive patterns for brain metastases have been sparingly described, no guidelines exist for screening asymptomatic patients, limiting opportunities for earlier detection and intervention. A comprehensive understanding of the risk factors associated with the development of brain metastases in mCRC is essential to optimize detection and improve patient outcomes. Methods: Patients with mCRC and radiographically confirmed BMs were retrospectively identified utilizing the Palantir Foundry platform and review of patient electronic medical records. Between 2015 and 2024, data from 66 patients at MD Anderson Cancer Center (MDACC) and 37 from the Veneto Institute of Oncology were collected for data analysis. Clinicopathological data including gender, ethnicity, smoking history, tumor sidedness, and mutational status were assessed. Overall survival (OS) was calculated using the Kaplan-Meier method. Results: In the MDACC cohort, baseline characteristics of the 66 patients revealed a median age of 54.4 years at the time of metastatic disease diagnosis, with approximately 55% being male. The majority of patients were Caucasian (72.7%) and had no history of smoking (65.2%). Most had lung metastases (92.4%), a rectal primary (57.6%), retained mismatch repair protein expression (98.5%), and harbored mutations in KRAS at initial diagnosis (65.2%) and TP53 (66.7%). A minority of patients were BRAF mutated (7.6%) or HER2 amplified (7.6%). BMs were most often seen in patients with indolent disease, evidenced by a median of 30.3 months from metastatic disease diagnosis to detection of brain involvement. The median OS from the time of BMs detection was 10 months, while mOS from metastatic disease diagnosis was 49 months. Of the 37 patients from the Veneto Institute of Oncology, 54.1% were male with a mean age of 58 years, and the primary tumor location was either the descending colon (40.5%) or rectum (40.5%). Consistent with the MDACC cohort, 59.5% of patients harbored KRAS mutations and 83.8% had lung metastases. The mOS was 42.9 months from the time of metastatic disease diagnosis and 11.5 months from the detection of BMs. Conclusions: BMs in mCRC are associated with a poor prognosis and routine screening is not currently standard practice. In this preliminary analysis of two separate institutional patient cohorts, similar clinicopathological features appear to be associated with BM development and warrant further investigation. We plan to design a multivariate model to estimate lifetime risk of BMs at presentation, with the aim of identifying high-risk individuals and developing earlier detection strategies.

3615 Background: During nonoperative management (NOM) for rectal cancer after a clinical complete response (cCR) or near-CR (nCR), 25–30% develop local regrowth and 5–10% develop distant metastasis, highlighting the need for improved risk stratification and surveillance. Circulating tumor DNA (ctDNA) has emerged as a prognostic biomarker, but its utility in NOM decision making remains undefined. Methods: We retrospectively studied 110 patients with stages I-III, microsatellite stable rectal adenocarcinoma achieving cCR/nCR after neoadjuvant therapy (2020-2024) managed with NOM and tumor-informed ctDNA testing in the INTERCEPT program (Signatera Exome). We assessed longitudinal ctDNA status during NOM and the first post-treatment ctDNA in relation to local regrowth and/or distant metastasis (Kaplan–Meier/log-rank; Fisher’s exact). We also evaluated per-sample accuracy for events occurring within ±90 days of each blood draw. Results: Over a median follow-up of 25 months, 23 (20.9%) patients developed local regrowth and 12 (10.9%) developed distant metastases (Table 1). Patients with ever positive longitudinal ctDNA had an associated worse 2-year regrowth-free survival (41.7% vs 83.9%, P=0.0002) and metastasis-free survival (41.7% vs 94.9%, P<0.0001) than those with persistently negative ctDNA. Among patients with an evaluable first post-treatment ctDNA result (within 180 days post treatment; n=72), those with a positive result had an associated lower regrowth-free survival (P = 0.0006) and metastasis-free survival (P < 0.0001). In per-sample analysis (n=669), ctDNA showed low sensitivity and high specificity for local regrowth (41.4% and 94.3%) and higher sensitivity and specificity for distant metastasis (73.8% and 97.4%). Twenty-two of 23 patients with local regrowth underwent salvage surgery; ctDNA positivity at local regrowth was associated with more advanced pathological T stage (66.7% of ypT3–4 vs 15.4% of ypT0–2, P=0.01). Conclusions: During NOM for rectal cancer, ctDNA may be used to identify a small subgroup at high risk of local regrowth and/or distant metastasis. However, many local regrowth occurs despite persistently negative ctDNA, consistent with limited sensitivity. Negative ctDNA results should therefore not prompt de-escalation of endoscopic and radiologic surveillance when a NOM strategy is used. Overall rates of local regrowth and distant metastasis. Local regrowth (n=23) a P Distant metastasis (n=12) P Longitudinal ctDNA <.0001 <.0001 Persistently negative (n=95) 14/95 (14.7%) 3/95 (3.2%) Ever positive (n=15) 9/15 (60.0%) 9/15 (60.0%) First post-treatment ctDNA b 0.005 <.0001 Negative (n=67) 15/67 (22.4%) 5/67 (7.5%) Positive (n=5) 4/5 (80.0%) 4/5 (80.0%) a Two had synchronous and 5 had metachronous distant metastasis. b Within 6 months post treatment, n = 72.

Surgical and systemic management of neuroendocrine liver metastases (NELM) has evolved over the past decades. This study aimed to evaluate temporal changes in prognosis and identify factors associated with improved survival after hepatic cytoreduction for NELM. Patients who underwent hepatic cytoreduction for NELM during 1998–2023 were retrospectively analyzed and divided into earlier period (EP, 1998–2008) and later period (LP, 2009–2023) cohorts. Clinicopathological features, perioperative outcomes, and survival were compared. Covariate-adjusted survival curves and propensity score matching were used to compare overall survival (OS) between major hepatectomy and liver-parenchyma-sparing cytoreduction (LPSC). The study included 293 patients, 132 in the EP cohort and 161 in the LP cohort. The LP cohort had a lower rate of major hepatectomy (32.9

140 Background: MRD is a state where circulating tumor DNA (ctDNA) is detected without visible radiologic disease after curative intent procedures. As this is associated with a high risk of recurrence, several clinical trials are ongoing for CRC in this setting with the goal of eradicating the disease before overt metastases establish. We aim to provide an overview of this field of research. Methods: Clinicaltrials.gov was screened for colorectal, colon, and rectal cancer trials containing the keywords minimal residual disease or MRD. Information on geographical region, phase, tumor stage and site, requirements on ctDNA, radiological no evidence of disease (NED) and prior adjuvant therapy, duration and type of therapy, and endpoints was collected. Results: A total of 41 MRD trials (USA 46%, Asia 24%, Europe 24%, multinational 5%) were identified, of which 23 are recruiting, 8 active not recruiting, 8 not yet recruiting, and 2 have an unknown status. Among interventional trials (n = 39), 37% are phase 3, 39% phase 2, and 20% phase 1, with 54% having a randomized design. Localized-only disease is included in 49%, localized or stage IV in 34%, stage IV-only in 10%, and not defined in 7%. CRC is enrolled in 23 trials, colon only in 13, and multiple disease types in 5. All except one trial have specific requirements for ctDNA positivity, with timing criteria being mentioned in 31 trials, of which 16 define a specific amount of time from resection or adjuvant therapy (range 1-12 weeks). ctDNA negative arms are present in 10 trials and all but 5 specifically require radiological NED. Post-adjuvant therapies for MRD are studied in 25 trials, ctDNA-informed adjuvant therapy in 13, and either of these in 2. Chemotherapies are studied in 23 trials, with some evaluating multiple regimens (FOLFOXIRI in 7, FOLFOX / CAPOX in 9, FOLFIRI in 5, TAS-102 ± other in 4, TEMIRI in 2, and capecitabine / 5-FU in 2), with durations from 3 to 6 months and number of patients varying from 25 to 4812. The primary endpoint is disease-free survival (DFS), recurrence-free survival (RFS), or time to recurrence in 12 trials (52%), ctDNA clearance in 5 (22%), and other in 6 (26%). Novel approaches are studied in 20 trials (checkpoint inhibitors ± other in 9, vaccines in 3, immune-modulation in 2, NK cells in 2, CAR-T cells in 1, ADCs in 1, and targeted therapy based on dMMR, HER2 , and BRAF -V600E in 4). Compared with chemotherapy trials, treatment duration is more variable (single infusion to 12 months) and sample size smaller (range 10-327), with ctDNA clearance (n = 8, 40%) and safety (n = 3, 15%) more often being the primary endpoint and DFS/RFS (n = 7, 35%) and other (n = 2, 10%) being used less frequently. Conclusions: The landscape of MRD trials in CRC reflects longstanding questions on optimal adjuvant therapy delivery. Heterogeneous eligibility criteria, increasing use of innovative therapeutics, and novel endpoints such as ctDNA clearance, warrant a consensus on MRD trial design.

Supplementary Table 2: Variety of genes tested in the unselected cohort (83–86)

Abstract Background: The CEDAR study (NCT03916510) showed that combining our first-generation oncolytic immunotherapy with CRT in LARC significantly improved the response rate (RR) by magnetic resonance imaging (MRI) compared to standard CRT alone, warranting further study. The ongoing FORTRESS study (NCT06459869) aims to demonstrate improved RR for our next-generation oncolytic immunotherapy NG-350A, in combination with CRT in pMMR LARC, compared to contemporary CRT outcomes. Methods: Stage II/III LARC patients (pts) with risk factors for recurrence (cT3a-d, N0-2, MRF+ or EMVI+) received NG-350A at 1x e12 on Day (D) 1 and 3x e12 virus particles on D3 & 5 in week (W)1, 5 & 9. All pts received pelvic long-course CRT (capecitabine 825mg/m2 orally twice a day) to 50Gy (option for 4Gy boost) during W2 to 6. The active study period was 12 Ws; the study allowed for total neoadjuvant therapy. Efficacy was assessed in week 12 using MRI Tumor Regression Grade, endoscopy and digital rectal examination. Circulating tumor DNA (ctDNA) was measured throughout the active study period and follow up using the NeXT Personal™ Dx assay (Personalis, City, CA). The primary objective is the proportion of pts achieving a response [(near) complete clinical response or (n)cCR] to NG-350A in combination with CRT. Key secondary objectives include safety and tolerability of NG-350A in combination with CRT and the effect of study treatment on ctDNA as a molecular marker of tumor burden and response. Results: Ten pts have been enrolled to date, and enrollment is ongoing. Treatments have been well-tolerated, and no serious adverse events (SAEs) or new safety signals related to NG-350A have been detected. Related AE are limited to the effects of systemic viremia; to date, no AEs attributable to the CD40 agonist transgene encoded by NG-350A have been observed. Transient aPTT prolongation has been observed in 4 pts. Four (all stage IIIB) out of 5 evaluable pts (80%) to date achieved a ncCR at W12 and all pts continued to consolidation chemotherapy immediately after the active study period. ctDNA clearance (incl. 1 transient) was observed at or before W12 in all pts. Results will be updated with W12 response data from all pts. Conclusions: In this small pt cohort, NG-350A plus long-course CRT thus far has achieved a high RR. Early efficacy and safety data suggest the combination may reach/exceed the RR level seen in CEDAR (not screened for MMR status). Preliminary ctDNA kinetics data assessed by the ultrasensitive tumor-educated whole genome sequencing assay are consistent with the encouraging RR and may enable non-operative management after CRT, as implemented in CEDAR, with the potential for patients to avoid additional chemotherapy or surgical interventions. Citation Format: Eric D. Miller, Pannaga Malalur, Sonal S. Noticewala, Arvind Dasari, Sheela Rao, Andry Santoso, Maria Hawkins, Douglas Brand, Rui Ru Ji, Korinna Pilz, Oliver Rosen. Phase 1b trial of NG-350A, a CD40 agonist antibody expressing adenoviral vector, in combination with chemoradiotherapy (CRT), in patients with mismatch repair-proficient (pMMR) locally advanced rectal cancer (LARC): Initial results from the FORTRESS study [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(8_Suppl):Abstract nr CT155.

TPS3634 Background: Currently, there are no biomarkers validated prospectively in randomized studies for resected colon cancer (CC) to determine need for adjuvant chemotherapy (AC). However, circulating tumor DNA (ctDNA) represents a highly specific and sensitive approach (especially with serial monitoring) for identifying minimal/molecular residual disease (MRD) post-surgery in CC patients (pts), and may outperform traditional clinical and pathological features in prognosticating risk for recurrence. CC pts who do not have detectable ctDNA (ctDNA-) are at a much lower risk of recurrence and may be spared the toxicities associated with AC. Furthermore, for CC pts with detectable ctDNA (ctDNA+) who are at a very high risk of recurrence, the optimal AC regimen has not been established. We hypothesize that for pts whose CC has been resected, ctDNA status may be used to risk-stratify for making decisions about AC. Methods: In this prospective phase II/III trial, up to 1,912 pts with resected stage III A, B (all pts) and stage II, IIIC (ctDNA+ only) CC will be enrolled. Based on the post-operative ctDNA status using personalized and tumor-informed assay (SignateraTM, bespoke assay), those who are ctDNA- (Cohort A) will be randomized to immediate AC with fluoropyrimidine (FP) + oxaliplatin (Ox) for 3-6 mos per established guidelines vs. serial ctDNA monitoring. Patients who are ctDNA+ post-operatively or with serial monitoring (Cohort B) will be randomized to FP+Ox vs. more intensive AC with addition of irinotecan (I) for 6 mos. The primary endpoints for Cohort A are time to ctDNA+ status (phase II) and disease-free survival (DFS) (phase III) in the immediate vs. delayed AC arms. The primary endpoint for Cohort B is DFS in the FP+Ox vs FP+Ox+I arms for both phase II and phase III portions of the trial. Secondary endpoints include prevalence of detectable ctDNA post-operatively, time-to-event outcomes (overall survival and time to recurrence) by ctDNA status, and the assessment of compliance to adjuvant therapy. Biospecimens including archival tumor tissue, as well as post-operative plus serial matched/normal blood samples, will be collected for exploratory correlative research. Active enrollment across the NCTN started in June 2022. NCT#: NCT05174169. Support: U10-CA-180868, -180822; UG1CA-189867; Natera, Inc. Clinical trial information: NCT05174169 .

Supplementary Table 3: Stratified analysis of the clinical factors in the unselected cohort (MEN1 and DNA repair genes)

126 Background: NNT/NNH analyses can estimate the magnitude of benefit vs harm of therapeutic agents and can inform clinical decision making. We quantified the benefit/risk profile of fruquintinib (F; a highly selective, oral inhibitor of all 3 VEGFRs) vs other later-line mCRC monotherapies. Methods: Six phase 3 randomized (2:1), placebo (P)-controlled trials (F: FRESCO, FRESCO-2; regorafenib [rego]: CONCUR, CORRECT; TAS-102: TERRA, RECOURSE) were identified. NNT (average number of pts needing treatment to prevent 1 additional death/progression event) for overall survival (OS) and progression-free survival (PFS) was calculated as the inverse of difference in survival function between treatment and P. NNH (average number of pts needing treatment for 1 additional pt to have an adverse event; AE) for treatment-emergent AEs (TEAEs) was calculated as the inverse of absolute risk increase relative to P. Low NNTs and high NNHs suggest a more favorable treatment profile. Results: F had the lowest NNT (most favorable) for 6-month OS (FRESCO-2) and PFS (FRESCO) vs rego and TAS-102 (Table). F had the highest NNH (most favorable) for grade ≥3 TEAEs, palmar-plantar erythrodysesthesia (PPE; FRESCO-2), and fatigue (FRESCO). TAS-102 (TERRA) had the highest NNH for diarrhea, followed by F (FRESCO-2). Consistent with mechanism of action of anti-VEGFRs, F and rego had lower NNHs for hypertension (HTN) vs TAS-102 (TERRA). F (FRESCO) had the lowest NNH for discontinuation due to TEAEs. Conclusions: In this cross-trial comparison, F had the lowest NNT for OS (FRESCO-2) and PFS (FRESCO/FRESCO-2), indicating survival benefit. F had the highest NNHs for grade ≥3 TEAEs, PPE (FRESCO-2), and fatigue (FRESCO), indicating a low risk profile. The analysis highlights fruquintinib as a favorable treatment for pts with mCRC. Clinical trial information: NCT02314819 , NCT04322539 , NCT01584830 , NCT01103323 , NCT01955837 , NCT01607957 . NNT/NNH (95% CI) for mCRC treatments.* F Rego TAS-102 FRESCO NCT02314819 N=416 FRESCO-2 NCT04322539 N=691 CONCUR NCT01584830 N=204 CORRECT NCT01103323 N=760 TERRA NCT01955837 N=406 RECOURSE NCT01607957 N=800 6-month OS 6.5(5.5–8.1) 5.3(4.8–5.9) 5.9(NA) 9.8(NA) 14.1(NA) 7.2(NA) 6-month PFS 4.2(3.7–4.8) 4.4(3.9–5.1) 5.0(NA) 9.1(NA) 7.1(NA) 7.2(NA) Grade ≥3 TEAEs 2.4 (2.0–3.1) 8.1 (5.0–22.5) 2.6 (2.0–3.7) 2.5 (2.2–3.0) 2.8 (2.3–3.6) 5.9 (4.1–10.2) TEAE leading to discontinuation 10.8 (6.7–28.5) NA † (18.0–NA) 12.5 (6.2–NA) 17.5(11.3–39.2) 250.0 (15.3–NA) 50.0 (22.9–NA) PPE 2.2 (1.9–2.5) 6.0 (4.8–8.0) 1.4 (1.3–1.7) 2.6 (2.2–3.0) NA (NA) NA (46.0–NA) HTN 2.4 (2.0–3.0) 3.6 (3.0–4.5) 5.3 (3.6–9.5) 4.5 (3.7–5.8) 14.3 (10.0–25.2) NA (NA) Diarrhea 5.1 (3.8–7.4) 7.3 (5.2–12.3) 6.2 (4.3–11.4) 3.8 (3.2–4.8) 7.9 (5.7–13.0) 5.0 (3.9–6.9) Fatigue 32.5 (10.3–NA) 25.8 (10.1–NA) 10 (5.3–80.3) 5.3 (3.8–8.4) 7.4 (5.0–13.9) 8.3 (5.4–18.1) *All trials except TERRA included best supportive care in both arms. † Lower rate of pts discontinued F vs P due to TEAEs, thus NNH was not applicable (NA).

BACKGROUND:Clinical benefit from later lines of therapy in metastatic colorectal cancer (mCRC) is limited. Patient selection for treatment is crucial for optimal risk-benefit evaluation. Codon-specific KRAS mutations have been implicated as predictive biomarkers for efficacy of trifluridine/tipiracil (TAS-102). However, their predictive impact for TAS-102 plus Bevacizumab (TAS-Bev), the new standard of care in mCRC, is unknown. METHODS:This large patient-based, real-world, retrospective cohort study of mCRC patients who received TAS-102 alone or in combination with bevacizumab between January 1, 2020, and March 1, 2023. A sensitivity analysis was performed in 2 independent cohorts from MD Anderson Cancer Center, Houston, TX, and Tempus AI, Inc., Chicago, IL, prior to polling the data together. RESULTS:A total of 946 mCRC patients were evaluated; KRAS was mutated in 57.8% cases, 39.4% involving G12 codon, and 9.8% G13. We found no association of KRAS-G12 mutations with overall survival (hazard ratio [HR] = 1.1, 95% confidence interval [CI] = 0.90 to 1.25; P = .441) on TAS-102 contradicting the reported predictive impact of these biomarkers. Moreover, we reported inferior overall survival for KRAS-G13 mutant patients (HR = 1.3, 95% CI = 1.02 to 1.69; P = .045). On TAS-Bev, no association was found between KRAS-G12 (HR = 0.8, 95% CI = 0.59 to 1.11; P = .188) or KRAS-G13 (HR = 1.66, 95% CI = 0.99 to 2.79; P = .072) mutations and overall survival. CONCLUSIONS:No significant associations between KRAS-G12 and overall survival were found for TAS-102 or TAS-Bev. Further research is needed to assess the impact of these mutations on combined TAS-Bev therapy prior to any clinical application.

199 Background: Fruquintinib is a highly selective, oral inhibitor of all 3 vascular endothelial growth factor receptors. The phase 3 FRESCO-2 (NCT04322539) study demonstrated a significant survival benefit with fruquintinib + best supportive care (BSC) vs placebo + BSC as third or later line of therapy for patients (pts) with metastatic colorectal cancer (mCRC). Based on FRESCO-2, fruquintinib is approved, including in the US and EU, for previously treated mCRC, irrespective of biomarker status. This analysis aimed to evaluate baseline (BL) circulating tumor DNA (ctDNA) and mutation profiles, as well as post-treatment (tx) changes in ctDNA and their potential correlation with clinical outcomes. Methods: Plasma samples collected at BL (n=26) and Cycle 3 Day 1 (C3D1) post-tx (n=7) from pts in FRESCO-2 were analyzed using the Guardant Infinity tissue-free ctDNA detection and quantification platform. BL assessments included genomic profiling, tumor mutational burden (TMB), and tumor fraction (TF) measured by methylation. For pts in the fruquintinib arm with paired BL and post-tx (C3D1) samples, genomic and methylation molecular response (MR) scores were calculated using Guardant’s validated MR algorithm. Associations with overall survival (OS) and progression-free survival (PFS) were evaluated. Results: All plasma samples tested had detectable ctDNA levels at BL and 52% of pts were identified as TMB-high (≥20 mutations/Mb). The most frequently mutated genes included APC , KRAS , and TP53 , across both tx arms. BL TF was comparable between pts in the fruquintinib (n=17) and placebo (n=9) arms, and was observed to be higher in TMB-high pts. Overall, baseline TF levels above the cohort median were associated with shorter OS, regardless of tx arm, although the sample numbers were low (n=26). No significant associations were observed between BL mutation status of TP53 , KRAS , or PIK3CA and OS/PFS. Of the 6 pts in the fruquintinib arm who achieved a best overall response of stable disease, 5 demonstrated a reduction in ctDNA levels with a concordant decrease in both genomic and methylation MR scores at C3D1 compared to BL. Of these 5 pts, 3 met the criteria for molecular responders, defined as ≥50% decrease in ctDNA. Conclusions: This preliminary analysis suggests that ctDNA-based biomarkers, including TMB and methylation TF, may offer prognostic insights in pts with mCRC and support MR evaluation. The concordance between genomic and methylation MR and the observed associations with clinical outcomes underscores the utility of methylation-based ctDNA assessments as a potential tool for predicting the therapeutic benefit of fruquintinib. Due to limited sample numbers, this analysis warrants further investigation in larger cohorts to better understand the relationship between BL ctDNA parameters, early changes following fruquintinib tx, and clinical outcomes. Clinical trial information: NCT04322539 .

189 Background: Aspirin use is established as adjuvant therapy for PIK3CA mutant (mt) stage II-III colorectal cancer, but its role in mCRC remains unclear, with conflicting findings after resection of liver metastases. We evaluated overall survival (OS) in mCRC patients by PI3K status and aspirin use. Methods: Institutional datasets were used to acquire the data. PI3K alterations were classified as PIK3CA mt or other ( PTEN and PIK3R1 ). Aspirin use was defined as ≥30 days of use at any time after diagnosis. To mitigate immortal time bias, a landmark analysis at 12 months was performed for patients who initiated aspirin within 335 days of diagnosis. OS was assessed using Kaplan–Meier and Cox regression adjusted for age, ECOG, primary tumor location, number of metastatic sites, aspirin use, PIK3CA , and RAS/ BRAF . The interaction between PI3K/ PIK3CA and aspirin use were evaluated with the Wald test. Results: Of 1,529 patients (137 aspirin users, 1392 non-users; median age 55 years; 60% male), 484 (32%) had PIK3CA mt (exon 9, 44%; exon 20, 16%; other, 31%), and 103 (7%) other PI3K mutations. Among aspirin users, OS was similar for PI3Kmt vs wildtype (wt) (39.2 vs 38.9 months; HR 0.79; 95% CI 0.46-1.40; p = 0.403) and for PIK3CA mt vs wt (38.1 vs 39.2 months; HR 0.97; 95% CI 0.55-1.70; p = 0.914). In patients not using aspirin, OS was worse for PI3Kmt vs wt (47.3 vs 58.5 months; HR 1.20; 95% CI 1.00-1.40; p = 0.022) and for PIK3CA mt vs wt (45.7 vs 59.1 months; HR 1.30; 95% CI 1.10-1.50; p = 0.001). In the multivariable analysis, age ≥55 years, poor ECOG PS, multiple metastatic sites, and RAS mutations remained significant adverse prognostic factors, while PIK3CA mt and aspirin use were prognostic only in the univariable model (Table). No significant interaction was observed between aspirin use and PI3K ( p = 0.199) or PIK3CA ( p = 0.498). Conclusions: PIK3CA and PI3K pathway mutations are associated with worse prognosis in mCRC, particularly among patients not receiving aspirin. This effect was not observed in aspirin users with no significant interaction detected. These findings are hypothesis-generating, and further investigation is needed to clarify the role of aspirin in PI3K/ PIK3CA mt mCRC. Variables Univariable HR(95% CI) p Multivariable HR(95% CI) p Age (≥55 vs <55) 1.40 (1.20–1.60) <0.001 1.31 (1.13–1.50) <0.001 ECOG (1 vs 0) 1.40 (1.20–1.60) <0.001 1.29 (1.10–1.50) 0.001 ECOG (≥2 vs 0) 2.10 (1.70–2.60) <0.001 1.93 (1.54–2.40) <0.001 Primary location (Right vs Left) 1.37 (1.18–1.60) <0.001 1.16 (0.98–1.50) 0.008 Number of metastasis (≥2 vs 1) 1.40 (1.19-1.70) <0.001 1.35 (1.14–1.60) <0.001 Aspirin use (Yes vs No) 1.30 (1.00-1.60) 0.032 1.27 (1.00–1.60) 0.054 PIK3CA ( mt vs wt) 1.20 (1.10–1.40) 0.003 1.08 (0.93–1.30) 0.331 RAS (mt vs RAS/ BRAF wt) 1.30 (1.15–1.50) <0.001 1.24 (1.06–1.40) 0.006 BRAFV600E ( mt vs RAS/ BRAF wt) 1.40 (0.99-2.10) 0.057 1.22 (0.84–1.80) 0.291