Longitudinal Liquid Biopsy and Mathematical Modeling of Clonal Evolution Forecast Time to Treatment Failure in the PROSPECT-C Phase II Colorectal Cancer Clinical Trial

Abstract

Sequential profiling of plasma cell-free DNA (cfDNA) holds immense promise for early detection of patient progression. However, how to exploit the predictive power of cfDNA as a liquid biopsy in the clinic remains unclear. RAS pathway aberrations can be tracked in cfDNA to monitor resistance to anti-EGFR monoclonal antibodies in patients with metastatic colorectal cancer. In this prospective phase II clinical trial of single-agent cetuximab in wild-type patients, we combine genomic profiling of serial cfDNA and matched sequential tissue biopsies with imaging and mathematical modeling of cancer evolution. We show that a significant proportion of patients defined as wild-type based on diagnostic tissue analysis harbor aberrations in the RAS pathway in pretreatment cfDNA and, in fact, do not benefit from EGFR inhibition. We demonstrate that primary and acquired resistance to cetuximab are often of polyclonal nature, and these dynamics can be observed in tissue and plasma. Furthermore, evolutionary modeling combined with frequent serial sampling of cfDNA allows prediction of the expected time to treatment failure in individual patients. This study demonstrates how integrating frequently sampled longitudinal liquid biopsies with a mathematical framework of tumor evolution allows individualized quantitative forecasting of progression, providing novel opportunities for adaptive personalized therapies. Liquid biopsies capture spatial and temporal heterogeneity underpinning resistance to anti-EGFR monoclonal antibodies in colorectal cancer. Dense serial sampling is needed to predict the time to treatment failure and generate a window of opportunity for intervention. .

Longitudinal cancer evolution from single cells

“…On the one hand, in fact, the existing list of approaches that process single-cell data and extend phylogenetic methods by handling data-specific errors [13,14,15,16], are not suitable to handle multiple temporally ordered samples derived from the same tumor, and cannot be used to investigate the clonal prevalence variation in time. On the other hand, even though methods for longitudinal bulk sequencing data are starting to produce noteworthy results [17,18,19], they usually require complex computational strategies to deconvolve the signal coming from intermixed cell subpopulations. Furthermore, there is an ongoing debate whether multi-sample trees from bulk samples are indeed phylogenies or, conversely, if they might lead to erroneous evolutionary inferences [20].…”

Detection of mutations in circulating cell‐free DNA in relation to disease stage in colorectal cancer

“…Liquid biopsy includes the analysis of tumor‐derived biomarkers in any body fluid, such as plasma, urine, and cerebrospinal fluid. In particular, serial blood testing is proposed as a convenient real‐time tool to identify spatial and temporal heterogeneity predicting response or resistance to targeted agents …”

“…In particular, serial blood testing is proposed as a convenient real-time tool to identify spatial and temporal heterogeneity predicting response or resistance to targeted agents. 5 Circulating cfDNA is composed of small nucleic acid fragments liberated from cells by rupture, necrosis or apoptosis originating from normal and deceased cells. Thus, circulating tumor-derived DNA (ctDNA) is only identified via the detection of cancer-related mutations.…”

A mathematical model of ctDNA shedding predicts tumor detection size

“…For cancer relapse detection, we considered an aggressive lung tumor growing with = 1% per day (doubling time of 69 days) and assumed a sequencing panel that covers 20 tumor-specific mutations (Abbosh et al 2017;McDonald et al 2019;Newman et al 2016;Tie et al 2016;Reinert et al 2019;Khan et al 2018). Requiring that at least one of these 20 tumor-specific mutations needs to be called present in the plasma sample to infer that the tumor relapsed, we find an AUC (area under the curve) for the ROC (receiver operating characteristic) curve of 86% for tumors with 0.2 cm 3 (Fig.…”