Background – There is limited data on the clinical utility of circulating tumor DNA (ctDNA) for treatment response monitoring (TRM) in patients with advanced solid tumors treated with Tyrosine Kinase Inhibitors (TKIs). Here we evaluate whether changes in ctDNA tumor fraction (TF) add value to actionable variant allele frequency (VAF) monitoring in predicting outcomes in a real-world (rw) cohort of advanced solid tumor patients treated with TKIs.
Methods – Tempus xM for TRM is a ctDNA assay that quantifies changes in ctDNA TF and classifies patients as molecular responders (MRs; ≥50% reduction in ctDNA TF between baseline and on-treatment time points) or molecular non-responders (nMRs). Deidentified patient records from the Tempus multimodal database were included if a patient had a baseline test ≤ 15 weeks prior to TKI start and an on-treatment test result 3-25 weeks post-TKI initiation. Actionable SNV/indels and matched TKIs were defined per ESMO guidelines. Differences in rw-overall survival (rwOS) were assessed using a log-rank test with follow-up times censored at 18 months.
Results – The evaluable cohort was 45 patients with 7 distinct solid tumors (49% NSCLC, 15% breast cancer). Thirty-one (69%) patients had actionable SNV/indels identified in one or both ctDNA test results (13 MRs, 18 nMRs). The majority of MRs (92%, n=12/13) had decreasing VAFs in the actionable variant, while half (n= 9/18) of nMRs had increasing VAFs. There was a significant difference in rwOS between MRs (no death events observed; median follow-up was 8.7 months) and nMRs (median rwOS: 14.4 months; p=0.004). For patients with decreasing VAFs (n=21), there was a significant difference in rwOS between MRs (no death events observed) and nMRS (median rwOS: 13 months, p=0.02).
Conclusions – This is a unique dataset that utilizes longitudinal tDNA TF as a TRM biomarker in advanced solid tumor patients treated with actionable TKIs. ctDNA TF monitoring added value in predicting rw outcomes beyond individual changes in actionable VAFs. These results should be further evaluated in prospective clinical datasets.
VIEW THE PUBLICATION
VIEW THE POSTER