Analysis of Plasma Cell-Free DNA by Ultradeep Sequencing in Patients With Stages I to III Colorectal Cancer

Abstract

Importance: Novel sensitive methods for detection and monitoring of residual disease can improve postoperative risk stratification with implications for patient selection for adjuvant chemotherapy (ACT), ACT duration, intensity of radiologic surveillance, and, ultimately, outcome for patients with colorectal cancer (CRC).

Objective: To investigate the association of circulating tumor DNA (ctDNA) with recurrence using longitudinal data from ultradeep sequencing of plasma cell-free DNA in patients with CRC before and after surgery, during and after ACT, and during surveillance.

Design, setting, and participants: In this prospective, multicenter cohort study, ctDNA was quantified in the preoperative and postoperative settings of stages I to III CRC by personalized multiplex, polymerase chain reaction-based, next-generation sequencing. The study enrolled 130 patients at the surgical departments of Aarhus University Hospital, Randers Hospital, and Herning Hospital in Denmark from May 1, 2014, to January 31, 2017. Plasma samples (n = 829) were collected before surgery, postoperatively at day 30, and every third month for up to 3 years.

Main outcomes and measures: Outcomes were ctDNA measurement, clinical recurrence, and recurrence-free survival.

Results: A total of 130 patients with stages I to III CRC (mean [SD] age, 67.9 [10.1] years; 74 [56.9%] male) were enrolled in the study; 5 patients discontinued participation, leaving 125 patients for analysis. Preoperatively, ctDNA was detectable in 108 of 122 patients (88.5%). After definitive treatment, longitudinal ctDNA analysis identified 14 of 16 relapses (87.5%). At postoperative day 30, ctDNA-positive patients were 7 times more likely to relapse than ctDNA-negative patients (hazard ratio [HR], 7.2; 95% CI, 2.7-19.0; P < .001). Similarly, shortly after ACT ctDNA-positive patients were 17 times (HR, 17.5; 95% CI, 5.4-56.5; P < .001) more likely to relapse. All 7 patients who were ctDNA positive after ACT experienced relapse. Monitoring during and after ACT indicated that 3 of the 10 ctDNA-positive patients (30.0%) were cleared by ACT. During surveillance after definitive therapy, ctDNA-positive patients were more than 40 times more likely to experience disease recurrence than ctDNA-negative patients (HR, 43.5; 95% CI, 9.8-193.5 P < .001). In all multivariate analyses, ctDNA status was independently associated with relapse after adjusting for known clinicopathologic risk factors. Serial ctDNA analyses revealed disease recurrence up to 16.5 months ahead of standard-of-care radiologic imaging (mean, 8.7 months; range, 0.8-16.5 months). Actionable mutations were identified in 81.8% of the ctDNA-positive relapse samples.

Conclusions and relevance: Circulating tumor DNA analysis can potentially change the postoperative management of CRC by enabling risk stratification, ACT monitoring, and early relapse detection.

Conflict of interest statement

Conflict of Interest Disclosures: Drs Sharma, Salari, Wu, Tin, Shchegrova, Assaf, Balcioglu, Hafez, Goel, Rabinowitz, Billings, Swenerton, Aleshin, Lin, and Zimmermann, Messrs Sethi, Srinivasan, Olson, and Dashner, and Ms Navarro, reported receiving support from Natera Inc outside the submitted work. Dr Billings reported receiving support from Trovagene, OmniSeq, MissionBio, and Metastat outside the submitted work. Dr Zimmermann has a pending patent for Provisional. Dr Lindbjerg Andersen reported receiving grants from Novo Nordisk Foundation, Danish Council for Strategic Research, Danish Council for Independent Research, and Danish Cancer Society during the conduct of the study. No other disclosures were reported.

Figures

Figure 1.. Patient Enrollment, Sample Collection, and Definition of the Patient Subgroups Used to Address the Defined Clinical Questions

ACT indicates adjuvant chemotherapy; CEA, carcinoembryonic antigen; CRC, colorectal cancer; CT, computed tomography; and ctDNA, circulating tumor DNA.

Figure 2.. Preoperative and Postoperative Circulating Tumor DNA (ctDNA) Monitoring in Patients With Colorectal Cancer (CRC)

A, Kaplan-Meier estimates of recurrence-free survival (RFS) for 94 patients with stages I to III CRC stratified by postoperative day 30 ctDNA status. The 3 censored ctDNA-positive patients were all treated with adjuvant chemotherapy (ACT) and were likely cured by this treatment (see patients 33, 62, and 130 in B). B, Recurrence rate and longitudinal ctDNA status in ctDNA-positive patients receiving ACT. C, Kaplan-Meier estimates of RFS for 58 ACT-treated patients, stratified by ctDNA status at first post-ACT visit. D, Kaplan-Meier estimates of RFS for 75 patients with longitudinal samples, stratified by longitudinal post–definitive-treatment ctDNA status. A patient was classified as testing positive if 1 or more plasma samples after definitive treatment was ctDNA positive. The Kaplan-Meier plots were halted when the proportion of patients in follow-up was less than 10%. Shaded areas in the Kaplan-Meier plots indicate 95% CIs. HR indicates hazard ratio.

Figure 3.. Association of Circulating Tumor DNA (ctDNA) Analysis With Early Detection of Relapse and Detection of Clinical Actionable Mutations

A, Comparison of time to relapse by ctDNA and standard-of-care computed tomography (CT). The mean time from surgery to relapse detection was 5.5 months (range, 0.4-17.7 months) for ctDNA and 14.2 months (range, 5.9-31.1 months for CT). Dashed lines indicate mean time in months of recurrence based on CT and ctDNA. B, For all patients with relapsing disease, the ctDNA levels in plasma increased over time from ctDNA detection to radiologic response. Early time points before and during adjuvant chemotherapy were omitted. Each colored curve represents data from a different patient. C, Fraction of recurrence in ctDNA-positive patients with actionable mutations detected in plasma. D, The actionable variants occurred with variant allele frequencies (VAFs) similar to the nonactionable variants. Association between the mean ctDNA VAF and the VAF of the actionable mutations is shown.