Plasma-Based Measurements of Tumor Heterogeneity Correlate with Clinical Outcomes in Metastatic Colorectal Cancer

SIMPLE SUMMARY: Molecular characterization of circulating tumor DNA (ctDNA) can offer a window into tumor genetic heterogeneity, especially in metastatic cancers where different lesions may harbor different mutations. The presence of multiple tumor clones may be reflected by dispersed variant allele frequencies of mutations detected in a single ctDNA sample. We hypothesized that the degree of dispersion of somatic mutations detected with a targeted next-generation sequencing assay may correlate with clinical outcomes in metastatic colorectal cancer. We found that patients with high ctDNA-based tumor heterogeneity after first-line bevacizumab and chemotherapy had shorter progression-free survival and worse objective response. Plasma-based measurements of tumor heterogeneity may have prognostic value in various cancer types and should be further explored for assessing treatment response and other clinical applications. ABSTRACT: Sequencing circulating tumor DNA (ctDNA) from liquid biopsies may better assess tumor heterogeneity than limited sampling of tumor tissue. Here, we explore ctDNA-based heterogeneity and its correlation with treatment outcome in STEAM, which assessed efficacy and safety of concurrent and sequential FOLFOXIRI-bevacizumab (BEV) vs. FOLFOX-BEV for first-line treatment of metastatic colorectal cancer. We sequenced 146 pre-induction and 89 post-induction patient plasmas with a 198-kilobase capture-based assay, and applied Mutant-Allele Tumor Heterogeneity (MATH), a traditionally tissue-based calculation of allele frequency distribution, on somatic mutations detected in plasma. Higher levels of MATH, particularly in the post-induction sample, were associated with shorter progression-free survival (PFS). Patients with high MATH vs. low MATH in post-induction plasma had shorter PFS (7.2 vs. 11.7 months; hazard ratio, 3.23; 95% confidence interval, 1.85–5.63; log-rank p < 0.0001). These results suggest ctDNA-based tumor heterogeneity may have potential prognostic value in metastatic cancers.