Deep sequencing of circulating tumor DNA detects molecular residual disease and predicts recurrence in gastric cancer

Abstract

Identifying locoregional gastric cancer patients who are at high risk for relapse after resection could facilitate early intervention. By detecting molecular residual disease (MRD), circulating tumor DNA (ctDNA) has been shown to predict post-operative relapse in several cancers. Here, we aim to evaluate MRD detection by ctDNA and its association with clinical outcome in resected gastric cancer. This prospective cohort study enrolled 46 patients with stage I-III gastric cancer that underwent resection with curative intent. Sixty resected tumor samples and 296 plasma samples were obtained for targeted deep sequencing and longitudinal ctDNA profiling. ctDNA detection was correlated with clinicopathologic features and post-operative disease-free (DFS) and overall survival (OS). ctDNA was detected in 45% of treatment-naïve plasma samples. Primary tumor extent (T stage) was independently associated with pre-operative ctDNA positivity (p = 0.006). All patients with detectable ctDNA in the immediate post-operative period eventually experienced recurrence. ctDNA positivity at any time during longitudinal post-operative follow-up was associated with worse DFS and OS (HR = 14.78, 95%CI, 7.991-61.29, p < 0.0001 and HR = 7.664, 95% CI, 2.916-21.06, p = 0.002, respectively), and preceded radiographic recurrence by a median of 6 months. In locoregional gastric cancer patients treated with curative intent, these results indicate that ctDNA-detected MRD identifies patients at high risk for recurrence and can facilitate novel treatment intensification studies in the adjuvant setting to improve survival.

Conflict of interest statement

Y.H.G., Y.F.G., Y.Y.Z., L.Y.J., Y.S.C., L.Y., X.Y., and X.F.X. are current employees of Geneplus-Beijing. V.K.L reports grants and personal fees from Takeda, honoraria from BMS, grants from Guardant Health, grants from Adaptimmune. J.Z. served as a consultant for AstraZeneca and Geneplus-Beijing and received honoraria and funding from Bristol-Myers Squibb, Merck, Roche, OrigiMed, Innovent, and Adaptive Biotechnologies Inc.

Figures

Fig. 1. Clinical determinates of ctDNA detection in gastric cancer.

Fractions of cases with ctDNA detected were shown in gastric cancer groups with different AJCC/UICC stage (a), early/advanced stage (b), tumor site (e) and histologic Lauren classification (f). Differences were assessed using Fisher’s Exact test, and p value were shown when less than 0.05. c tumor volume of cases with ctDNA detected (ctDNA+) or not (ctDNA-) were shown. The line indicates median with interquartile range. d Correlation with maximum VAF of cell-free DNA mutations and tumor volume were shown, the line indicates best fit values and 95% confidence intervals of linear regression. g Multivariable analysis results were shown.

Fig. 2. Patient survival is associated with ctDNA detection results.

Kaplan–Meier curves for disease-free survival data in relation to ctDNA detection in Plasma obtained after surgery before initiation of adjuvant treatment (a) and after completion of adjuvant treatment (c). b, d show Kaplan–Meier curves of overall survival data in relation to ctDNA detection in the same samples as disease-free survival curves. ctDNA+: ctDNA was detected; ctDNA−: ctDNA was not detected. The number of patients in each group and the log-rank p value are shown.

Fig. 3. Mutation tracking at serial time points predict patient survival.

Kaplan–Meier curves of disease-free survival (a) and overall survival (b) associated with detection of ctDNA in any post-operative plasma samples in patients with gastric cancer resected. Numbers of patients and log-rank p value are shown. c Detailed survival data of patients are shown, with gray bars indicating disease-free survival and orange bars indicating survival after recurrence. ctDNA+: ctDNA was detected; ctDNA−: ctDNA was not detected.

Fig. 4. Potential application of ctDNA detection in post-surgery surveillance for patients with gastric cancer.

Detailed radiology results, ctDNA and CEA changes for patient P002 and P041 are shown. Dynamic changes of ctDNA (red line) and CEA (blue line), adjuvant chemotherapy with SOX regimen (gray box) and S-1 (yellow box) are shown at bottom panels. Representative radiology images are shown, including abdominal and pelvic CT scan, chest CT scan, PET-CT, and bone scan.