Evaluating novel therapies and circulating tumor DNA (ctDNA) as a marker of response in curatively treated gastrointestinal cancers with microscopic residual disease.

TPS850 Background: ctDNA can identify the earliest sign of relapse (MRD) after curative therapies with specificities of 93-100%, positive predictive values over 95%, and median lead times of 8-9 months before radiographic relapse. ctDNA has identified a "new stage" of high-risk patients (pts) with no evidence of disease radiographically (rNED) and MRD who lack treatment (tx) strategies as, traditionally, they wait for radiographic progression of disease (POD) before starting therapies. Treating MRD is a promising strategy supported by observational studies where ctDNA clearance is linked with improved survival and on-tx kinetics predict drug efficacy . We designed a pilot investigator-initiated trial to 1) determine the feasibility of using ctDNA as a rapid signal for positive therapeutic activity and 2) obtain pilot data on the efficacy of a novel combination therapy (atezolizumab [A] + bevacizumab [B]) in eradicating MRD in previously curatively treated pts with GI cancers. Methods: We will enroll 20 pts with any-stage cancers divided over 4 cohorts (n=5 colorectal [CRC], n=5 hepatocellular/biliary tract carcinoma, n=5 upper GI, n=5 pancreatic adenocarcinoma) who have a positive Signatera ctDNA test and rNED any time after completing standard curative therapies. Pts are treated with A 1200 mg IV + B 15 mg/kg IV on day 1 of a 21-day cycle for up to 1 year with imaging every 12 weeks and serial ctDNA testing every 3 weeks. Co-primary endpoints are rates of pt enrollment in 12 months and rates of ctDNA responses at a 12-week landmark after tx initiation. Pts with ctDNA POD (ctDNA doubling on each of 2 sequential tests, any rate of rise in 3 sequential tests, or radiographic relapse) will stop while those with ctDNA complete response (CR; clearance on 2 sequential tests + ongoing rNED) or ctDNA partial/stable response (PR; not ctDNA CR or POD + ongoing rNED) will continue tx. Each cohort's Bayesian predictive probability of positive therapeutic activity will be calculated. If at least 60% of a cohort experiences ctDNA CR, it will be expanded by 5 pts. This scenario is equivalent to Simon's 2-stage Minimax design testing a null hypothesis of 20% CR vs. an alternative hypothesis of at least 60% CR at 5% alpha with 80% power. If at least 5/10 pts in an expanded cohort experience ctDNA CR, this will generate interest in developing a larger confirmatory trial to evaluate pt survival outcomes with A+B and ctDNA as a surrogate for survival. Secondary endpoints are toxicity and enrollment barriers. Exploratory endpoints are associations between ctDNA conversion and disease-free survival, tumor whole exome sequencing data, and peripheral blood immune profiles collected at baseline and on tx. Enrollment began in Q1 2023, paused in 7/2023 due to funding changes, and re-opened 8/2024 with 5 patients enrolled currently (n=4 CRC, n=1 biliary). Clinical trial information: NCT05482516 .