Background Doublet chemotherapy plus anti-epidermal growth factor receptor (EGFR) is a standard of care in left-sided, microsatellite stable RAS and BRAF wild-type metastatic colorectal cancer. Guidelines recommend treatment de-escalation after achieving disease control with induction. Maintenance and intermittent strategies were investigated but not directly compared.Methods We performed an individual patient data pooled analysis of 3 randomized phase II trials (PanaMa, Valentino, PRODIGE-28 TIME) focused on toxicity analysis. Only patients with left-sided, nonmicrosatellite instability high RAS and BRAF V600E wild type who started protocol-planned postinduction were included and stratified into 3 treatment groups: 5-fluorouracil and leucovorin plus anti-EGFR maintenance, anti-EGFR alone, or intermittent. Longitudinal toxicity data were collected and analyzed according to literature-based approach (toxicity over time), incorporating dimension of time into adverse event assessment and analyzing individual and groups of adverse events comparing treatment groups.Results Overall, 327 patients were included: 166, 109, and 52 patients received anti-EGFR plus 5-fluorouracil and leucovorin maintenance, anti-EGFR alone, and intermittent strategy, respectively. Mean adverse event grades for chemotherapy-related toxicity showed different longitudinal patterns. Mean grades in intermittent strategy were lower in early cycles and increased later, and higher values were reported for combination maintenance. Considering anti-EGFR-related skin toxicity, the mean adverse event grade was constantly lower for intermittent strategy compared with maintenance groups. Overall, grades 3 and 4 adverse events were more represented in maintenance groups vs intermittent strategy, although the predominant grade was 1 across cycles for all groups.Conclusions In our individual patient data analysis, indirectly comparing 3 clinical trials, intermittent strategy showed lower anti-EGFR skin-related toxicity vs maintenance. Shared decision making, considering patient and tumor features and treatment tolerability, may allow defining optimal de-intensification strategy.+*+9

3511 Background: Phase 3 CodeBreaK 300 (NCT05198934) explored two doses of sotorasib (soto) (240 mg and 960 mg) in combination with panitumumab (pani) 6 mg/kg versus investigator's choice (IC) and demonstrated superior progression-free survival (PFS) with soto 960 mg+pani versus IC in chemorefractory KRAS G12C-mutant mCRC. We evaluated whether early ctDNA clearance could serve as non-invasive biomarker of treatment (tx) response using one of the largest cohorts of mCRC patients (pts) treated with a KRAS G12C inhibitor plus an EGFR antibody and the only longitudinal ctDNA dataset from a KRAS G12C inhibitor trial in mCRC. Methods: ctDNA analyzed using the Guardant Infinity assay was quantified using KRAS G12C variant allele frequency (VAF) and methylation-based circulating tumor fraction (cTF) metrics. Clearance was defined at ≥50%, ≥80%, ≥90% and 100% reduction thresholds. Associations with objective response rate (ORR), PFS, and overall survival (OS) were assessed using logistic regression, Fisher-exact tests, Cox proportional hazards models, and Kaplan–Meier methods. Results: A total of 142 pts had evaluable baseline and Cycle 2 Day 1 ctDNA results. Early ctDNA clearance was more frequent in the soto arms than in the IC arm. This finding was consistent across ctDNA assessment methods, with high concordance between VAF and cTF. At ≥80% threshold, VAF clearance rates (95% CI) were 77.3% (62.2–88.5) and 65.9% (50.1–79.5) in the soto 960 mg and 240 mg arms (each N=44), respectively, compared with 15% (5.7–29.8) in the IC arm (N=40). Complete (100%) VAF clearance (95% CI) was observed in 45.5% (30.4–61.2), 31.8% (18.6–47.6), and 5.0% (0.6–16.9) of pts in the 960 mg, 240 mg, and IC arms, respectively. Reductions in ctDNA were correlated with radiographic tumor shrinkage. Lack of ctDNA clearance was associated with non-response, whereas pts achieving clearance were more likely to respond. Among pts treated with soto 960 mg who achieved ≥80% ctDNA clearance, the ORR (95% CI) by VAF was 35.3% (12/34; 19.7–53.5) and 32.3% (10/31; 16.7–51.4) by cTF. Decline in ctDNA was associated with improved PFS and OS independent of tx arm, confirming the role of ctDNA as a prognostic marker. Increasing clearance stringency did not enhance prognostic value. Conclusions: Early ctDNA clearance, assessed by KRAS G12C VAF or cTF, is prognostic for improved outcomes, supporting its utility as an early response biomarker. Clinical trial information: NCT05198934 . Survival (months) by VAF clearance (≥80%). Cohort VAF Clearance N mPFS (95% CI) mOS (95% CI) IC N 34 2.0 (1.9–4) 10.3 (6.6–14) IC Y 6 8.4 (3.7–NE) NE Soto 240 N 15 2.6 (1.8–3.8) 5.1 (3.2–7.5) Soto 240 Y 29 5.6 (3.8–8.5) 14.0 (8.4–NE) Soto 960 N 10 1.9 (1.6–3.8) 4.9 (1.6–7) Soto 960 Y 34 5.8 (5–7.5) NE (10.8–NE) m, median; NE, not evaluable; N, no; Y, yes.

558 Background: Cholangiocarcinomas (CCAs) are rare cancers that are often advanced and incurable at diagnosis. Durvalumab (DURVA) + gemcitabine/cisplatin (GEM/CIS) is approved for first-line treatment for CCA. This study (NCT06501625) evaluates the safety/tolerability of ivosidenib (IVO) + DURVA/GEM/CIS to determine the recommended combination dose (RCD) (safety lead-in phase, SLI) and the preliminary activity of this combination (expansion phase). Here we present the SLI results. Methods: Patients (pts) with locally advanced unresectable or metastatic mIDH1 CCA with ECOG 0 or 1 and at least 1 measurable lesion (RECIST v1.1) were dosed in the SLI. Treatment with up to 1 cycle of DURVA/GEM/CIS was permitted prior to study treatment initiation. Pts received IVO 500 mg QD + DURVA 1500 mg IV infusion every 3 weeks for up to 8 cycles + GEM 1000 mg/m 2 IV and CIS 25 mg/m 2 IV on days 1 and 8 of each cycle, followed by IVO 500 mg QD and DURVA 1500 mg every 4 weeks of each cycle. DLTs were evaluated during the first cycle of study treatment. Results: Seven pts with locally advanced or metastatic mIDH1 CCA (based on local or central testing) were enrolled in the SLI: age range, 37-75 years, 3 (43%) female, and 3 (43%) received 1 cycle of DURVA/GEM/CIS prior to study treatment. Through the data cut off of July 10, 2025, pts received 1–5 cycles with 6 (86%) ongoing. All pts reported a treatment-emergent adverse event (AE); 4 pts (57%) required a dose interruption and 1 (14%) required a dose reduction. One pt (14%) discontinued study treatment due to an adverse event (the only DLT), which was an AE of drug-induced liver injury, with LFT and bilirubin elevation that resolved after discontinuation of the quadruplet. Six pts were DLT evaluable, and 1 was not DLT evaluable, due to receiving <75% of the planned dose of IVO during the first cycle. This pt experienced an adverse event of special interest (AESI) of grade 3 QT prolongation requiring IVO dose reduction, but with re-escalation to full dose IVO 500 mg + DURVA/GEM/CIS during the second cycle after optimization of electrolytes and resolution of QT prolongation within 7 days, not assessed as a DLT. No other pts experienced AESIs. All pts experienced an AE, and 6 (86%) had a grade ≥3 AE. Two pts (29%) had a serious AE — 1 was the DLT of drug-associated liver injury and 1 was a non-treatment related cholangitis that recovered with resumption of IVO at full. One pt had a partial response and 6 had stable disease at the time of the data cut-off. IVO 500 mg + DURVA/GEM/CIS was confirmed as the RCD and the expansion phase was initiated. Conclusions: IVO at 500 mg QD + DURVA /GEM/CIS demonstrated a safety profile similar to DURVA/GEM/CIS. This dose will be evaluated further during the expansion phase, which will enroll ~40 pts. Clinical trial information: NCT06501625 .

INTRODUCTION:5-fluorouracil (5-FU) ± targeted therapy is a standard of care in frail/elderly patients with an unresectable colorectal adenocarcinoma (CRC) in first-line setting. Panitumumab plus sotorasib combination (KRAS G12C inhibitor) are promising in advanced line in KRAS G12C-mutated CRC. Here we assess the safety and efficacy of 5-FU combination with panitumumab and sotorasib as first-line treatment in frail/elderly patients with unresectable KRAS G12C-mutated CRC. METHODS:In this ENGIC 01 - PRODIGE 107 - FFCD 2306 - COLOSOTO multicenter, open-label, prospective single-arm phase II trial, the main inclusion criteria are adult patients with unresectable locally advanced or metastatic KRAS G12C-mutated CRC, unfit for a doublet/triplet chemotherapy. All patients will receive 5-FU plus panitumumab and sotorasib in 2-week-cycles until progression or intolerance. The primary endpoint is 8-months progression-free survival (PFS). The secondary endpoints include median PFS, disease control rate, time to progression, overall survival, best objective response rate, duration of response, safety profile, quality of life and geriatric assessment. A 70% 8-months PFS is expected (H0 <50%), and 37 patients will need to be included. PERSPECTIVES:Treatment with 5-FU plus panitumumab and sotorasib could be a promising alternative to 5-FU ± targeted therapy in first-line setting in frail/elderly patients with unresectable KRAS G12C-mutated CRC.

3561 Background: The use of overall response rate (ORR) and progression-free survival (PFS) as surrogacy of overall survival (OS) has limitations in mCRC. Change in tumor load (ΔTL; tumor burden at progressive disease [PD] vs baseline) at treatment failure may reflect tumor aggressiveness and resistance biology. We evaluated the prognostic value of ΔTL in mCRC using IPD from twelve randomized first-line trials. Methods: IPD from ATEZOTRIBE, FIRE3, FIRE4, FIRE4.5, MACBETH, ML22011, MOMA, PANAMA, TRIBE, TRIBE2, TRIPLETE, and VALENTINO were pooled. ΔTL was defined as the ratio of the sum of longest target lesion diameters at PD vs baseline, per RECIST. Primary endpoint was OS; secondary endpoints included post-progression survival (PPS). Estimated marginal means were controlled for treatment type. Associations between ΔTL and outcomes were assessed using Cox models adjusted for prespecified baseline tumor burden and clinical covariates (ECOG, age, sex, sidedness, RAS/BRAF status, metastatic patterns, previous therapies). Incremental prognostic value beyond early tumor shrinkage (ETS) and depth of response (DpR) was assessed using likelihood ratio testing. Effect modification by treatment regimen was evaluated using ΔTL × treatment interaction terms for EGFR- vs VEGF-based therapy and for triplet vs doublet-chemotherapy across trials. Results: A total of 4852 patients evaluable for ΔTL were included. Mean ΔTL was 0.63 (SD 0.45) and significantly lower in patients receiving EGFR- vs VEGF-based therapy (0.58 vs 0.66, p<0.001) and triplet vs doublet chemotherapy (0.57 vs 0.67, p<0.001). Lower ΔTL (lower tumor load at PD vs baseline) was associated with better OS (HR 0.58 [95% CI 0.55–0.61], p<0.001) and PPS (HR 0.65 [0.61–0.69], p<0.001) after adjustment for baseline tumor burden and clinical covariates. For OS, ΔTL added prognostic information beyond ETS (likelihood ratio χ² = 282.13, p<0.001) and DpR (likelihood ratio χ² = 121.91, p<0.001). Lower ΔTL was associated with improved OS in patients receiving EGFR- (n = 1609; HR 0.62 [0.58–0.67], p<0.001) and VEGF-based therapy (n = 1916; HR 0.39 [0.35–0.43], p<0.001); doublet (n = 2177; HR 0.61 [0.57–0.65], p<0.001) and triplet (n = 1348; HR 0.34 [0.30–0.38], p<0.001) chemotherapy, with significant ΔTL × treatment interaction (VEGF vs EGFR: HR 0.50 [0.46–0.54], p<0.001; triplet vs doublet: HR 0.60 [0.54–0.66], p<0.001) after adjustment for treatment regimen, RAS/BRAF status and other clinical covariates. Conclusions: TL appears to be a strong, independent prognostic marker for OS and PPS in mCRC. ΔTL seems to vary by biologic and chemotherapy intensity, indicating it captures differential treatment effect/resistance not captured by ORR or PFS. Prospective validation as an early, treatment-sensitive prognostic endpoint is warranted.

TPS264 Background: B7-homolog 3 protein (B7-H3), an immune checkpoint protein, is overexpressed in many types of solid tumors but has limited expression in normal tissues. B7-H3 expression is correlated with tumor progression, metastasis, and poor clinical outcomes across various malignancies. GSK’227 (HS-20093) is a novel antibody-drug conjugate composed of a fully human anti–B7-H3 monoclonal antibody linked to a topoisomerase I inhibitor via a protease-cleavable linker. GSK’227 has shown acceptable safety and promising antitumor activity in an Asian patient (pt) population with advanced solid tumors (NCT05276609; NCT05830123). This study (NCT06885034) will first assess GSK’227 as monotherapy for efficacy, safety, tolerability, PK, and immunogenicity and subsequently in combination therapy, in pts with previously treated advanced, unresectable, gastrointestinal (GI) solid tumors, in a global population. Methods: This Phase 1b/2, open-label, multicenter study will assess early efficacy and safety signals of GSK’227 in pts with advanced, unresectable GI solid tumors. The study will include Cohort A, evaluating two doses (1:1 randomization) of GSK’227 administered 3-weekly (Q3W) in pts with advanced, unresectable colorectal cancer (CRC, sub-cohort CRC-A). The study will also assess two additional doses (1:1 randomization) administered 2-weekly (Q2W) (sub-cohort CRC-B). In parallel, Cohort B will assess a single Q3W dose of GSK’227 in pts with advanced, unresectable pancreatic ductal adenocarcinoma (PDAC). The study has two parts per cohort. Part 1 involves signal seeking (and dose optimization for CRC only) and an optional extension, in which GSK’227 is evaluated as monotherapy. Subsequently, Part 2 (expansion) may evaluate combination treatments with GSK’227. The optional extension component and the expansion part will be guided by the totality of emerging data . Eligible adults must have histologically confirmed advanced, unresectable CRC or PDAC, 1–2 lines of prior treatment (only 1 for PDAC), and ECOG PS of 0–1. Part 1’s primary endpoint is confirmed ORR. Secondary endpoints include unconfirmed ORR, DoR, PFS, safety/tolerability, PK, immunogenicity, and pt-reported AEs/tolerability. Efficacy will be assessed per RECIST v1.1, with imaging every 6 weeks (±7 days) from date of randomization (sub-cohorts CRC-A and CRC-B) or first dose (cohort B) for the first 48 weeks, and then every 12 weeks (±7 days) thereafter. Safety follow-up will be assessed at 30 (±3), 60 (±7), and 90 (±7) days after the last dose. Safety, tolerability, and efficacy analyses will be summarized descriptively; efficacy analyses will also include point estimates with 2-sided 95% CIs. Clinical trial information: NCT06885034 .

e15552 Background: The derived neutrophil-to-lymphocyte ratio (dNLR) is an established prognostic marker in solid tumors including metastatic colorectal cancer (mCRC), but its predictive value remains unclear. The dNLR was evaluated as a prognostic biomarker and as a potential predictor of treatment outcome during fluorouracil/folinic acid (FU/FA) ± panitumumab (pmab) maintenance therapy in patients with RAS wild-type mCRC enrolled in the PANAMA trial (NCT01991873). Methods: Patients with available baseline blood counts prior to induction therapy were grouped according to dNLR using a predefined cut-off of 2.2. Progression-free survival (PFS) and overall survival (OS) from initiation of maintenance therapy, as well as PFS of reinduction therapy, were estimated using the Kaplan–Meier method and compared by log-rank testing and Cox proportional hazards regression. Multivariable Cox models adjusted for confounders were used to assess independent prognostic effects. Predictive effects were evaluated using interaction tests within Cox proportional hazards models for maintenance and reinduction. Results: Of 241 patients included into the full analysis set with available data, n = 140 had baseline dNLR ≤2.2 and n = 101 dNLR > 2.2. While median PFS during maintenance therapy did not differ according to baseline dNLR (dNLR ≤2.2 vs. > 2.2: 10.1 vs. 9.7 months; log-rank p = 0.25), OS was significantly longer in patients with dNLR ≤2.2 (30.8 vs 22.7 months; log-rank p < 0.001). Baseline dNLR remained independently associated with OS in multivariable Cox regression (HR 1.56, 95% CI 1.15–2.12; p = 0.004). Despite numerical prolongation of PFS and OS during maintenance therapy by the addition of pmab to FU/FA, no significant interaction between baseline dNLR and maintenance treatment arms was observed (PFS (maintenance): dNLR ≤2.2 = 8.8 vs 5.6 months; dNLR > 2.2 = 9.7 vs 5.8 months; interaction p = 0.72; OS: dNLR ≤2.2 = 33.7 vs 28.2 months; dNLR > 2.2 = 26.1 vs 20.1 months; interaction p = 0.294). By contrast, PFS after treatment reinduction was significantly shorter after FU/FA + pmab compared with FU/FA maintenance in patients with dNLR ≤2.2, whereas no difference according to prior maintenance treatment was observed in patients with dNLR > 2.2 (dNLR ≤2.2 = 2.6 vs 7.4 months; dNLR > 2.2 = 5.8 vs 6.2 months; interaction p = 0.036). Conclusions: Baseline dNLR is an independent prognostic biomarker for OS in patients with RAS WT mCRC treated within the PANAMA trial. Potential information might be derived for the optimal choice of maintenance and reinduction strategies.

BACKGROUND:Sotorasib 960 mg plus panitumumab (soto960+pani) was investigated in chemorefractory KRAS G12C-mutated metastatic colorectal cancer (mCRC) in the phase 3 CodeBreaK 300 and phase 1b CodeBreaK 101 studies. In CodeBreaK 300, soto960+pani significantly improved progression-free survival (PFS) versus investigator's choice therapy (trifluridine/tipiracil [T/T] or regorafenib). The phase 3 SUNLIGHT study evaluated T/T plus bevacizumab (T/T+bev) in patients with unselected refractory mCRC and found longer survival times with T/T+bev than T/T monotherapy. Matching-adjusted indirect treatment comparisons (MAIC) were performed to compare the efficacy and safety of soto960+pani with new standard-of-care T/T+bev treatment. MATERIALS AND METHODS:Clinical outcomes and adverse events (AEs) from CodeBreaK 300 and 101 (for soto960+pani) were compared with those from SUNLIGHT (for T/T+bev). By reweighting individual patient-level data from the pooled CodeBreaK studies, differences in baseline characteristics were adjusted. Odds ratios (ORs) were estimated for objective response rates; hazard ratios (HRs) were used for PFS and overall survival (OS). RESULTS:From a pool of 93 patients, the effective sample size of soto960+pani with matched characteristics was 29 patients. Soto960+pani increased the likelihood of treatment response, with an adjusted OR of 5.7 (95% CI, 2.6-12.8) versus T/T+bev. HR for PFS was 0.77 (95% CI, 0.47-1.25); HR for OS was 0.44 (95% CI, 0.22-0.87), suggesting a survival benefit favoring soto960+pani. Grade ≥ 3 AEs occurred in 58% and 72% of soto960+pani-treated and T/T+bev-treated patients, respectively. CONCLUSION:In this MAIC analysis, soto960+pani demonstrated statistically significant improvement in response rates and OS in patients with chemorefractory KRAS G12C-mutated mCRC.

3575 Background: The randomized FIRE-4 study evaluated the effect of cetuximab rechallenge versus investigator’s choice in the 3 rd -line treatment of patients (pts) with RAS-wildtype (RAS-wt) metastatic colorectal cancer (mCRC). The present analysis focuses on the time between the end of initial anti-EGFR-based treatment and start of 3 rd -line therapy (TIS-3) as a potential predictor of response. Methods: In FIRE-4, all pts received induction therapy with FOLFIRI plus cetuximab. After 1 st PD, an anti-EGFR-free “window therapy” was recommended. At diagnosis of 2 nd PD, pts who had responded to induction therapy were re-evaluated for RAS status and, after exclusion of RAS mutations, were offered randomization to either rechallenge with cetuximab plus chemotherapy or investigator’s choice. Overall survival in 3 rd -line (OS-3) was evaluated as a primary endpoint. Results: Of the 87 pts entering 3 rd -line treatment in FIRE-4, 45 received (FOLF)IRI plus cetuximab (rechallenge arm), and 42 received investigator’s choice (standard arm). Numerically superior outcome parameters regarding ORR-3 (OR 2.70), PFS-3 (HR 0.87) and OS-3 (HR 0.86) were observed in the rechallenge arm, without, however, reaching the level of statistical significance. To evaluate the relevance of TIS-3, we focused on the median (13.8 months) and the first quartile (9.0 months). Based on these cut-off values, treatment effects were evaluated in pts with longer and shorter TIS-3 intervals (see Table). Conclusions: Longer TIS-3 intervals were associated with longer survival times in 3 rd -line treatment. This observation was particularly true for pts receiving cetuximab rechallenge, but also for those receiving investigator’s choice. Clinical trial information: NCT02934529 . Treatment arm N Interval PFS (mo) HR OS (mo) HR FOLF(IRI) + Cetuximab 21 > 13.8 mo 7.4 0.475P=0.019 20.0 0.639P=0.187 24 < 13.8 mo 4.1 12.6 34 > 9.0 mo 5.9 0.680P=0.274 19.6 0.601P=0.173 11 < 9.0 mo 4.0 10.8 Investigator’s choice 23 > 13.8 mo 5.6 0.772P=0.415 16.2 0.516P=0.055 19 < 13.8 mo 4.4 12.2 32 > 9.0 mo 5.5 0.690P=0.313 16.9 0.458P=0.042 10 < 9.0 mo 4.5 7.8

Abstract Background Standard of care for non-metastatic colon cancer is surgery followed by stage-guided adjuvant therapy and/or structured follow-up. Upfront surgery in resectable colon cancer is irrespective of local T/N stage, whereas adjuvant systemic therapy is recommended according to pathological staging. The role of neoadjuvant chemotherapy remains unclear. Whereas perioperative systemic therapy in colon cancer seems to be safe and may lead to pathologic downstaging, evidence on improved survival and quality of life remains scarce. The PROTECTOR / FIRE‑10 trial aims to generate evidence that perioperative systemic therapy improves survival without compromising quality of life in locally advanced, mismatch-repair proficient colon cancer patients. Methods and design Open-label, randomized, controlled, multicenter, phase III study with two parallel arms. Patients with locally advanced colon or upper rectal cancer staged cT3-4 and/or cN+ are randomized in a 2:1 fashion (favoring preoperative therapy) to investigate the efficacy, patient reported quality of life, and safety of preoperative therapy followed by surgery (Arm A) versus direct surgery followed by non-study specific stage-guided adjuvant therapy (Arm B). Stratification during randomization will be performed according to the following parameters: Fit for mFOLFOXIRI vs. mFOLFOX/CAPOX vs. 80%-mFOLFOX/CAPOX, ECOG 0 vs. ECOG 1-2, and left-sided primary vs. right-sided primary tumor. Only patients with confirmed mismatch-repair proficient and/or microsatellite stable tumor can be included. Preoperative treatment in Arm A is performed for a maximum of 6 biweekly cycles of FOLFOX/FOLFOXIRI or for a maximum of 4 triweekly cycles CAPOX (i.e., appr. 12 weeks). Patients in both arms should undergo quality-controlled surgery of the primary tumor, performed as complete mesocolic excision. Patients will be followed up with regard to relapse, survival and if applicable subsequent anti-cancer treatments until death or for at least 5 years after randomization, whichever date is earlier. Discussion The PROTECTOR / FIRE‑10 trial compares preoperative systemic therapy to upfront surgery (with stage-guided adjuvant therapy) in patients with locally advanced colon cancer. Trial registration This study is registered with clinicaltrials.gov (NCT06899477) and EudraCT (2023-508076-11-00).

BACKGROUND:Doublet chemotherapy plus anti-EGFR is a standard of care in left-sided, microsatellite stable, RAS/BRAF wt (wt) mCRC. Guidelines recommend treatment de-escalation after achieving disease control with induction. Maintenance and intermittent (stop&go) strategies were investigated but not directly compared. METHODS:We performed an individual patient data pooled analysis of three randomized phase II trials (PanaMa, Valentino, PRODIGE-28 TIME), focused on toxicity analysis. Only patients with left-sided, non-MSI-H, RAS/BRAF V600E wt who started protocol-planned post-induction were included and stratified into three treatment groups: 5-FU/LV plus anti-EGFR maintenance, anti-EGFR alone, stop&go. Longitudinal toxicity data were collected and analyzed according to literature-based approach (Toxicity over Time; ToxT), incorporating dimension of time into adverse event (AE) assessment and analyzing individual and groups of AEs comparing treatment groups. RESULTS:Overall, 327 patients were included: 166, 109 and 52 patients received anti-EGFR plus 5-FU/LV maintenance, anti-EGFR alone and stop&go, respectively. Mean AE grades for chemotherapy-related toxicity showed different longitudinal patterns. Mean Grades in stop&go were lower in early cycles and increased later, higher values were reported for combination maintenance. Considering anti-EGFR-related skin toxicity, the mean AE grade was constantly lower for stop&go compared to maintenance groups. Overall, grade 3/4 AEs were more represented in maintenance groups versus stop&go, although the predominant grade was 1 across cycles for all groups. CONCLUSIONS:In our IPD analysis, indirectly comparing three clinical trials, stop&go showed lower anti-EGFR skin-related toxicity versus maintenance. Shared decision-making, considering patient and tumor features and treatment tolerability, may allow defining optimal de-intensification strategy.

3588 Background: The combination of TAS-102 and bevacizumab is standard in the treatment of advanced colorectal cancer (aCRC). Ramucirumab (Ram) is a mAb that blocks VEGFR2. Ram in combination with Tas-102 did not improve OS compared to Tas-102 alone in the ITT population of the RAMTAS trial. Subgroups such as female patients (pts) and pts with left-sided tumors appeared to benefit from the combination (DOI:10.1016/j.annonc.2024.08.2264). Here we aimed at identifying angiogenesis-related, soluble, predictive biomarkers defining a RAMTAS subpopulation benefitting from the addition of Ram to Tas-102. Methods: Serum samples were available from 194 of 428 pts from the RAMTAS study (96 pts randomized to TAS-102, 98 pts to TAS-102 plus Ram). Serum levels of 85 angiogenesis-related proteins obtained prior to treatment with Tas-102 +/- Ram (baseline, BL) and at cycle 2 day 1 prior to the start of the next cycle (treatment, TR) were quantified via multiplex-ELISA by Myriad-RBM. Multivariate Cox regression models were developed within the combination arm using stepwise selection and ridge regularization. Model performance was assessed via C-index and validated using 10 x 5-fold cross-validation (CV) as well as full quality control for multicollinearity and proportional hazard assumptions. To distinguish predictive from general prognostic effects, the identified signatures were applied to the monotherapy control arm. Results: A TR model incorporating just 4 serum proteins demonstrated robust discrimination in the combination arm (Training C-index: 0.712; CV C-index: 0.708 ± 0.063). This model successfully stratified patients into a “Ram benefit” and a “no Ram benefit” group with a corresponding median PFS of 7.0 mo and 1.9 mo, respectively (HR 6.97, p < 0.001). A Baseline (BL) model yielded a C-index of 0.699, also utilizing just 4 serum proteins. Application of both models to the TAS-102 mono arm resulted in poor discrimination (C-index ~0.60) indicating that the signatures are specific predictors for the combination. In a comparative survival analysis, "Ram-benefit" TR model patients achieved a significantly longer PFS with the combination compared to monotherapy (7.00 vs. 2.14 mo; HR 0.558, p=0.0078). Conversely, "no-Ram-benefit" patients derived no significant benefit from the addition of Ram (PFS 1.91 vs. 1.91 mo; HR 0.946, p=0.4204). Conclusions: Oligo-marker serum signatures, when incorporating early on-treatment dynamics after one cycle of Tas-102/Ram, may effectively predict a survival benefit for aCRC-pts receiving TAS-102 in combination with Ram. We propose that a two-stage “Screen & Monitor” (BL+TR) strategy will help to identify primary refractory patients who gain no benefit from the addition of Ram. Such an approach could maximize therapeutic utility while reducing potential toxicity. Clinical trial information: NCT03520946 .

Abstract About 3–10% of patients with acute ischemic stroke (AIS) have active cancer. Malignancy-associated hypercoagulability (MAH) is an established cause of cancer-associated stroke. Nonbacterial thrombotic endocarditis (NBTE) is a severe manifestation of MAH. However, no prospective study has examined its prevalence and anticoagulation management in AIS. We conducted a prospective observational study at a tertiary center including patients with AIS or transient ischemic attack (TIA) and active malignancy. All patients underwent transthoracic echocardiography, followed by transesophageal echocardiography when indicated. Among 3,491 screened patients, 16 with active cancer and AIS/TIA were included. Eleven (68.8%) showed embolic patterns suggestive of MAH, and 6 (37.5%) had NBTE. NBTE patients numerically more frequently had multiterritory embolic infarctions (100% vs. 60.0%, p = 0.102) and prior ischemic stroke (66.7% vs. 10.0%, p = 0.028). 4/6 were on direct oral anticoagulants (DOACs), while none were on low-molecularweight heparin (LMWH). Under LMWH, vegetations resolved or regressed in 3/4, whereas under DOACs, progression or recurrent embolism occurred in several patients. Twelve-month mortality was high in both MAH (90.9%) and NBTE (83.3%) groups. NBTE was frequently identified in this prospective cohort of patients with cancer-associated stroke and consistently associated with multi-territory embolic infarction. Recognition may enable earlier diagnosis. Larger studies are needed to define optimal anticoagulation strategies.

3550 Background: Primary tumor sidedness is nowadays a major driver for the choice of biologics in the first line of pMMR/MSS RAS and BRAF wt mCRC patients (pts). It has been suggested that it might be replaced by a broader molecular profiling, but the evidence provided by individual trials is uncertain for the adoption of different panels and underrepresentation of gene-altered cases. We recently showed that HER2+ status does not predict resistance to first-line doublets + anti-EGFRs. Methods: We collected individual patient data from 7 CTs in first-line mCRC: TRIBE2, TRIPLETE, VALENTINO, CAPRI-2, FIRE-3, PARADIGM and CALGB/SWOG80405. pMMR/MSS RAS and BRAF wt and HER2− cases treated with doublets + anti-EGFRs/bev with available NGS data were included. Hyperselected tumors were those without MET+ status, ALK/ROS1/NTRKs/RET fusions, and HER2/PIK3CAexon20/PTEN/AKT1 mutations. Propensity score adjustment was used to assess clinical outcomes (aHR and aOR) according to first-line doublets + anti-EGFRs/bev. Results: Out of 1198 included pts, 990 (83%) had left-sided and 208 (17%) right-sided tumors. Anti-EGFR based regimens were associated with numerically longer OS (aHR: 0.86) in left-sided tumors, but not in right-sided ones (aHR: 0.98), with an insignificant p for interaction [int] = 0.60). Hyperselected and gene-altered cases were 1077 (91%) and 121 (9%), respectively. The anti-EGFR benefit in OS was magnified in hyperselected pts (aHR: 0.83), with an aHR of 1.16 in gene-altered ones (p int =0.11, adjusted also for tumor sidedness). As reported in the table, there was no interaction between the effect of biologics and primary tumor location either in the hyperselected (p int =0.86) or in the gene-altered group (p int =0.41). Conclusions: Hyperselection outperforms tumor sidedness as a predictive marker of benefit from doublets + anti-EGFR/bev in pMMR/MSS RAS/BRAF wt and HER2− mCRC. The OS benefit from anti-EGFR-based regimens is restricted to pts with hyperselected tumors, independently of primary tumor location. Therefore, in hyperselected mCRC, doublets + anti-EGFRs remain the standard for left-sided pts and may deserve consideration in right-sided ones. Hyperselected Gene-altered Left Right P int Left Right P int Anti-EGFRs N=537 Bev N=365 Anti-EGFRs N=87 Bev N=88 Anti-EGFRs N=57 Bev N=31 Anti-EGFRs N=19 Bev N=14 ORR (%) 81 67 72 65 74 52 58 57 aOR [95% CI] 2.29 [1.67 – 3.14] 1.44 [0.75 – 2.77] 0.24 3.48 [1.29 – 9.97] 1.02 [0.25 – 4.20] 0.23 mPFS* 13.2 12.9 11.1 10.2 11.2 11.3 8.0 8.8 aHR [95% CI] 1.02 [0.88 – 1.19] 1.01 [0.73 – 1.39] 0.73 1.06 [0.63 – 1.77] 1.23 [0.57 – 2.66] 0.75 mOS* 38.8 35.5 34.6 29.2 30.2 27.4 20.2 32.1 aHR [95% CI] 0.84 [0.71 – 0.98] 0.85 [0.61 – 1.17] 0.86 1.05 [0.63 – 1.75] 1.46 [0.66 – 3.22] 0.41 *Months.