Personalized tumor-informed circulating tumor DNA as predictor of progression risk after long-term responses to immunotherapy in advanced non-small-cell lung cancer.

2547 Background: Immune checkpoint inhibitors (ICIs) have remarkably improved survival in advanced non-small-cell lung cancer (NSCLC), with about 30%～40% of patients achieving long-term responses. However, biomarkers for predicting progression remain undefined. Circulating tumor DNA (ctDNA) has demonstrated its ability to predict recurrence in resected NSCLC, but its potential to forecast progression following prolonged responses to ICIs requires investigation. Methods: CR1STAL study is a multicenter, prospective cohort study investigating ctDNA surveillance to monitor progression risk in advanced NSCLC treated with first-line ICIs (NCT05198154). Patients with advanced NSCLC with long-term responses, defined as a PFS of about 1 year, were enrolled. Peripheral blood samples were collected alongside radiographic evaluations. ctDNA was detected using a personalized tumor-informed assay. Somatic variants were identified using a targeting 1,021 genes, followed by the design of individualized target-capture. ctDNA-positive was defined as the detection of ctDNA at any time during surveillance. The primary endpoint was PFS, defined as the time from enrollment until progression or death. Secondary endpoints included OS and ORR. Exploratory endpoints included the association between ctDNA features and survival, and comparison to other biomarkers. Results: We analyzed 199 sample from 42 NSCLC patients. The median age was 60.5 years with 88.1% male, and 64.3% at stage IV. The median number of sample collections was 4, with a median follow-up time of 24.7 months. ctDNA was detected in 54.8% of patients (23/42), with 82.7% of patients (19/23) showing ctDNA-positive occurring within 2 years of ICIs treatment. A total of 23 PFS events were observed. The ctDNA-positive group showed significantly worse PFS compared to the negative group (HR: 7.65, p < 0.001), with a positive predictive value of 90.0% and a specificity of 88.2%. Additionally, ctDNA-positive provided a median lead time of 6.6 months prior to radiological progression. ctDNA-positive significantly associated with poorer OS (HR: 68.42, p = 0.003) and lower ORR (60.9% vs 89.5%, p = 0.036). 18 exhibited clonal mutations. Compared to the ctDNA-negative group, the patients with clone had significantly worse PFS (HR: 9.38, p < 0.001) than those with subclone (HR: 4.16, p = 0.063). The ctDNA positivity rate was 84.6% in cases of local progression, 80.0% in distant metastases with brain exhibiting lower positivity rates. Additionally, peripheral CEA showed inferior predictive value for PFS (HR: 1.76, p = 0.303) than ctDNA. Conclusions: ctDNA has emerged as a promising biomarker for predicting progression risk of ICIs in advanced NSCLC patients with long-term responses. ctDNA surveillance enables earlier detection of progression and supports treatment adjustments through adaptive therapy. Clinical trial information: NCT05198154 .