Introduction: Local treatments are routinely used in addition to systemic therapy for treatment of oligometastatic lung cancers. However, there are significant challenges in identifying patients with oligometastatic NSCLC that are likely to benefit from local consolidative radiotherapy. Negative results from NRG-LU002 further highlight limitations in accurate patient identification. Our prior work demonstrated that higher pre-radiotherapy (RT) circulating tumor DNA (ctDNA) maximum variant allele fraction (VAF) was associated with worse OS (HR 5.7, p=0.004) and PFS (HR 4.7, p=0.02) in multivariate Cox regression analysis. Here we hypothesize that applying a multimodal algorithm-based cell-free DNA analytic approach (ctDNA-derived tumor fraction estimation or CTFE) to determine pre-RT ctDNA tumor fraction (ctDNA TF) can more robustly stratify patients who stand to benefit from consolidative RT for oligometastatic NSCLC.

Methods: Patients with oligometastatic NSCLC (metastasis in 1-5 organ systems) who underwent liquid biopsy ctDNA analysis using the 105-gene, hybrid-capture, NGS-based Tempus xF assay at any time prior to receiving RT were included in this study. ctDNA TF was estimated (CTFE) using an ensemble algorithm that incorporates pathogenic variant allele frequencies, copy number information, and germline information from plasma cell-free DNA. Limit of blank (LOB) was determined using 20 healthy subjects using the classical non-parametric rank-based approach. Overall survival (OS) and progression-free survival (PFS) were calculated from time of RT to minimize guarantee-time bias.

Results: Out of a real-world cohort of 1,487 patients with oligometastatic NSCLC, 309 patients who underwent RT were included in the final analysis. Median and mean ctDNA TF were 0.007 and 0.053, (range 0-0.522), respectively. Using LOB TF (0.0009) to determine ctDNA TF detectable versus undetectable, 221 patients were ctDNA TF detectable and 88 patients were ctDNA TF undetectable prior to RT. Median OS from time of RT was 25 months for ctDNA TF undetectable compared to 16.8 for ctDNA TF detectable patients (p=0.007). Median PFS was 8.7 months for ctDNA TF undetectable versus 5.2 months for ctDNA TF detectable patients (p=0.01). In multivariate Cox proportional hazards regression analysis, we observed that age, histology, and gender were not associated with OS or PFS. However, ctDNA TF was independently associated with worse OS (HR 23.0, 95% CI 4.038-109.0, p=0.0002) and worse PFS (HR 45.2, 95% CI 7.776-218.4, p<0.0001).

Conclusions: In oligometastatic NSCLC, pre-RT liquid biopsy-based ctDNA TF is a stronger predictor of survival (HR 23.0) than ctDNA maximum VAF (HR 5.7) and can be successfully utilized to risk-stratify patients who are likely to benefit from consolidative RT. Lack of ctDNA detection prior to RT based on CTFE analysis, likely reflective of minimal micrometastatic disease burden, was associated with significantly improved PFS and OS in patients who underwent RT.

VIEW THE PUBLICATION (MA03.05)