Personalized circulating tumor DNA (ctDNA) analysis in predicting residual disease after neoadjuvant chemoradiotherapy (nCRT) for esophageal squamous cell carcinoma (ESCC).

Abstract

e16062 Background: For ESCC patients who receive nCRT, high pathological complete response (pCR) rates could be achieved. A “Watch and Wait” strategy has been proposed for patients with clinical complete response (cCR) after nCRT. However, clinical assessments would inevitably misclassify non-pCR as cCR. We previously showed that ctDNA analysis with tumor-naïve fixed panel could improve the prediction of cCR. This study further investigates the value of personalized ctDNA analysis in predicting tumor response and residual disease after nCRT in ESCC. Methods: This is a side study of the prospective, diagnostic preSINO trial (NCT03937362). After completion of nCRT, patients underwent one or two clinical response evaluations (CREs) before planned surgery. CRE‐1 was 4–6 weeks after nCRT. CRE‐2 was 10–12 weeks after nCRT for patients without histological evidence of residual tumor at CRE-1. Individual tumors and matched white blood cells at baseline (B0) were whole-exome sequenced and somatic mutations were identified. Serial plasma samples were analyzed by brPROPHET tracking up to 50 somatic mutations at three timepoints: B0, CRE-1, CRE-2. This assay was developed to detect residual disease with a limit of detection of 0.004% and sample-level specificity of greater than 99% in the analytical validation. Residual disease was also detected and quantified through TNFP in the same samples for a head-to-head comparison. We correlated post-nCRT ctDNA status with pCR and major locoregional residual disease (MLRD) (>10% residual cancer or any residual nodal disease). Results: At this interim analysis, 55 ESCC patients (stage III, 87%) who received per-protocol treatments had sufficient sample to test ctDNA. The pCR rate was 27 % (15/55). 98% (2918/2966) of variants selected for patient-specific panel design were private to a specific patient, only 3% (92/2966) of variants were included in the fixed panel. After nCRT, ctDNA was detected by brPROPHET prior to nCRT in 100% of samples (n=55), significantly higher than that by TNFP assay (73%, 40/55). ctDNA was detected in 55% (30/55) of patients at CRE-1, 43% (10/23) at CRE-2. Importantly, the brPROPHET achieved higher sensitivity than TNFP in predicting non-pCR and MLRD owing to its ability to detect extremely low ctDNA fractions (to 1.14 x 10-5). Moreover, combining brPROPHET with CREs could further improve the sensitivity. Conclusions: Personalized ctDNA analysis can supplement the clinical assessment to decrease the false negative rate in identifying patients with residual disease after nCRT, and might become useful in the “Watch and Wait” strategy for the management of ESCC. Clinical trial information: NCT03937362 . [Table: see text]