Failure rates of mutation-based ctDNA assays in lung cancer treatment monitoring.

Abstract

e18752 Background: Despite the wide adoption of noninvasive liquid biopsy approaches for cancer genotyping, little is known about the longitudinal performance of these assays for monitoring therapeutic responses in patients with advanced disease. Here, we report results of a pooled analysis that collects existing evidence regarding the failure rates of circulating tumor DNA (ctDNA) assays designed for therapeutic monitoring in patients with advanced or metastatic lung cancer. Methods: We searched PubMed for original studies published between February 2012 and January 2022, following these inclusion criteria: (1) Adults older than 18 years old with advanced or metastatic lung cancer treated with chemotherapy, targeted therapy, or immunotherapy; (2) Tumor-informed or plasma-only circulating tumor DNA assays based on droplet digital PCR (ddPCR) or targeted deep sequencing technologies for the detection of tumor-specific allele fractions; (3) Measurement of circulating tumor DNA levels at sequential timepoints. We defined failures as a lack of detectable mutant allele fraction (MAF) in patients with metastatic lung cancer in samples collected before treatment initiation. Two reviewers independently screened studies, extracted data, assessed the risk of bias, and evaluated the quality of evidence. Divergences were resolved by a third independent reviewer reaching a consensus. Study quality assessment covered: (1) representativeness of patient population, (2) risk of funding bias, (3) description of ctDNA analysis and outcomes, and (4) statistical quality and interpretation. Results: From 142 studies retrieved by an automated PubMed query, we extracted data from 42 studies which fulfilled the eligibility criteria - 11 registered trials, 31 prospective/retrospective cohorts. MAF levels in the circulation were measured using PCR-based assays (23/42, 55%); the remainder used NGS-based approaches (19/42, 45%). The median number of targeted genes was 1 (range 1-6) vs 73 (range 36-214), respectively. The median number of samples analyzed per study was 79 (range 3-334). On average, 31% of samples collected before treatment did not have a detectable MAF in any of the target genes analyzed, despite the high disease burden typically experienced in patients with metastatic lung cancer. Only 57% of PCR-based and 78% NGS-based studies succeeded in providing molecular response assessment metrics. We identified 10/42 studies with critical quality assessment concerns and 18 studies had funding provided by a company with investments in the commercial cfDNA assay. Conclusions: Available liquid biopsy assays that report tumor-specific mutations have a high rate of failure for measuring disease burden and response outcomes during therapy. Approaches that could be reliably used for monitoring patients with advanced cancers undergoing treatment are needed to improve patient care.