A multi-trial analysis published in Clinical Cancer Research has reinforced the potential of circulating tumor DNA (ctDNA) as a predictive biomarker for treatment response in advanced non–small cell lung cancer (aNSCLC). The study, led by Friends of Cancer Research, examined ctDNA dynamics in 940 patients across eight clinical trials involving tyrosine kinase inhibitors (TKIs) for oncogene-driven aNSCLC.
The researchers used droplet digital PCR (ddPCR) technology to detect ctDNA from patient samples. Patients whose ctDNA was detected at baseline but cleared by 10 weeks after treatment initiation experienced significantly improved overall survival (OS) and progression-free survival (PFS) compared to those with persistent ctDNA detection.
These associations held even after adjusting for variants such as age, sex, smoking status, and performance status. Of note, ctDNA detection at 10 weeks appeared to be a more robust predictor of outcomes than early radiographic response via the Response Evaluation Criteria in Solid Tumors (RECIST), which did not show a statistically significant correlation with OS or PFS.
Furthermore, in patients with stable disease at early RECIST assessment, ctDNA clearance was strongly associated with better OS (HR = 4.15, P < .001). However, adding RECIST response to ctDNA categories did not enhance predictive accuracy.
The study supports the growing body of evidence positioning ctDNA as a minimally invasive, early endpoint for monitoring therapeutic efficacy. Co-author Gary Pestano, Chief Development Officer at Biodesix, says, “With solid tumors in particular, we see a blood-based surrogate as having multiple benefits because of the non-invasive nature of sample collection, added to our ability to conduct longitudinal analyses.”
Beyond oncology, ctDNA has broad utility as a biomarker in areas such as infectious disease and transplantation, predicts Pestano. “Working in consortia such as this one with Friends of Cancer Research serves to harmonize our collective efforts in helping evolve standards that support regulatory endpoints, for evaluating a complex but critical diagnostic application such as residual disease monitoring,” he says.