Personalized tumor-informed circulating tumor DNA analysis in monitoring recurrence following resection of high-risk locally advanced stage gastrointestinal stromal tumor.

Abstract

11522 Background: Gastrointestinal stromal tumor (GIST) is the most common mesenchymal tumor of the gastrointestinal tract. Radical resection is the standard treatment of localized GIST, yet the 5-year recurrence rate for high-risk GIST is more than 50%. It remains unclear whether detecting post-surgery molecular residual disease (MRD) via circulating tumor DNA (ctDNA) can predict the recurrence of high-risk locally advanced GIST. Methods: Patients with high-risk locally advanced stage GIST who underwent R0 surgery were enrolled prospectively. Surgical tissue samples were collected. Blood samples were collected pre-surgery, within 1-month post-surgery, and every 3–6 months thereafter. Tumor-derived variants were identified by whole-exon sequencing of the surgical tissues. Up to 50 highly ranked variants with allele frequency 3.0% were selected for the personalized panel design, which was subsequently used to assess MRD status. Results: As date cutoff of December 2024, 44 eligible patients were enrolled, among whom 42 tissue samples, 41 pre- and 166 post-surgical blood samples were collected and analyzed, and the median follow-up time was 21 months. Tissue-based sequencing identified variants in KIT and PDGFRA in 37 (84.1%) and 4 (9.1%) patients, respectively. Positive MRD was detected in 56.1% (23/41) of all pre-surgical plasma samples. Pre-surgical MRD positivity was associated with tumor volume, mitoses and Ki-67 (P < 0.05). Landmark analysis within 1 month post-surgery showed that 4 patients (4/41, 9.8%) were positive for ctDNA. Patients with positive MRD at landmark showed marginally worse DFS compared with those with negative MRD (HR = 4.24, 95% CI = 0.81-22.26, P = 0.07).To date, 7 patients have been detected recurrence by CT scan. Among the 4 patients with both radiological and MRD positivity, longitudinal ctDNA detected recurrence with lead-time of 3 months compared with CT scan for 2 patients. The other 2 cases are simultaneously. Notably, these 4 patients didn’t receive regular adjuvant therapy. Furthermore, the longitudinal MRD positivity was associated with inferior DFS after adjusting sex, age, TNM stages and whether receiving adjuvant therapy in the multivariable cox regression (HR = 5.63, 95% CI = 1.09-28.99, P = 0.04). Additionally, 30 patients with consistent negative MRD during surveillance exerted significantly superior survival compared with patients whose MRD status converted to negative (n = 2), converted positive (n = 7) and remained consistently positive (n = 2) from landmark to longitudinal monitoring (P = 0.04). Conclusions: The present study suggests that personalized tumor-informed ctDNA has the potential to inform recurrence in high-risk locally advanced stage GIST patients, especially for patients who have not received regular adjuvant therapy. Clinical enrollment is still ongoing. Clinical trial information: NCT05408897 .