ctDNA analysis for personalization of consolidation immunotherapy in localized non-small cell lung cancer.

2547 Background: Detection of molecular residual disease via circulating tumor DNA (ctDNA) analysis after chemoradiation (CRT) in localized non-small cell lung cancer (NSCLC) predicts risk of relapse. We explored the hypotheses that (1) patients with undetectable ctDNA after CRT may not require consolidation immunotherapy (CI) and (2) ctDNA analysis could monitor the effectiveness of CI in patients with residual ctDNA after CRT. Methods: We applied CAPP-Seq ctDNA analysis to 88 plasma and matched leukocyte samples collected pre-CRT, post-CRT but pre-CI, and mid-CI in 22 patients with Stage IIB-IIIB NSCLC treated with CRT followed by CI. Identification of patient-specific tumor variants was performed using tumor tissue or pretreatment plasma, and ctDNA was quantified using a tumor mutation-informed bioinformatic strategy. Freedom from progression (FFP) defined radiographically by RECIST 1.1 criteria was compared in patients with ctDNA detected or not detected at pre-CI and mid-CI landmarks. Results: Median follow up from the start of CRT was 11 months. ctDNA detection was associated with inferior rates of FFP when compared to patients with ctDNA not detected both pre-CI (12-month 33% vs. 76%, P = 0.015, HR 7.51, 95% CI 1.47-38.24) and mid-CI (12-month 0% vs. 86%, P < 0.0001, HR 123.3, 95% CI 16.21-937.8). In patients with undetectable ctDNA after CRT, FFP was similar to a historical cohort of patients with undetectable ctDNA after CRT alone (12-month 88% vs. 87%, P = 0.56, HR 0.55, 95% CI 0.07-4.18), suggesting that such patients may not benefit from CI. All patients with detectable ctDNA pre-CI in whom ctDNA increased mid-CI developed progressive disease. Finally, in 2 patients with ctDNA detected after CRT, CI led to elimination of ctDNA at the mid-CI timepoint. One of these patients developed an isolated local recurrence 22 months after CRT and the other patient is currently disease free at 11 months, suggesting clinical benefit from CI. Conclusions: Our results suggest that ctDNA analysis may allow personalization and response monitoring of CI following CRT for NSCLC. Validation in more patients followed by prospective testing in clinical trials will be required to establish clinical utility of such an approach.