The Prognostic Impact of Early ctDNA Kinetics in Metastatic Pancreatic Cancer Using the ctDNA-RECIST

Abstract

Purpose: Changes in ctDNA levels during systemic treatment may predict treatment efficacy in patients with metastatic pancreatic ductal adenocarcinoma (PDAC), but quantitative response criteria are not yet established. This study evaluates our recently proposed ctDNA-RECIST.

Experimental design: Blood samples were collected before treatment, before the second treatment cycle, and at the time of the first CT evaluation from 220 patients with metastatic PDAC receiving first-line palliative chemotherapy. Plasma ctDNA levels were measured using Droplet Digital PCR with HOXA9 methylation assays. ctDNA response was determined according to ctDNA-RECIST and correlated with overall survival (OS).

Results: ctDNA was positive before treatment in 71% of the patients and was related to OS [HR = 1.61; 95% confidence interval (CI), 1.19-2.19; P = 0.002]. Among ctDNA-positive patients, ctDNA maximal response (MR; n = 41) and ctDNA disease control (DC; n = 107) before the second treatment cycle had longer OS compared with ctDNA progressive disease (PD; n = 5; median OS: MR 11.9 months, DC 7.2 months, PD 3.6 months; P = 0.002). In Cox regression, ctDNA DC (HR = 1.55; 95% CI, 1.07-2.26; P = 0.021) and ctDNA PD (HR = 4.50; 95% CI, 1.74-11.6; P = 0.002) showed shorter OS compared with ctDNA MR. The same applied to ctDNA response at the time of the first CT evaluation assessed from the second treatment cycle (P < 0.001) and from treatment start (P < 0.001).

Conclusions: ctDNA-RECIST applied to liquid biopsies holds potential for early evaluation of treatment benefit in patients with metastatic PDAC, offering a novel, minimally invasive method to guide early clinical decision-making. Future studies should validate ctDNA-RECIST prospectively, preferably in randomized controlled trials.

Figure 1.

Study flow and ctDNA-RECIST schedule. Overview of sampling time points and patient flow throughout the study. The figure shows the number of patients included at each step, ctDNA positivity rates, and the number of patients evaluated for early and late ctDNA response. The positivity rate is based on the predefined cutoff of five positive droplets. A total of 153 patients with positive ctDNA before treatment had an available blood sample from time point B, and 128 patients had an available blood sample from time point C.

Figure 2.

Sankey plot of baseline ctDNA level quartiles and early ctDNA response. Flow of patients from quartiles of log-transformed baseline ctDNA levels (prior to treatment) to ctDNA-RECIST response categories at time point B. The width of the streams represents the number of patients in each transition.

Figure 3.

OS according to ctDNA response. A, OS according to early ctDNA response from prior to treatment (time point A) to prior to the second cycle of chemotherapy (time point B). In the Cox proportional hazards model, there was an HR = 1.55 (95% CI, 1.07–2.26; P = 0.021) for ctDNA DC compared with ctDNA MR and an HR = 4.50 (95% CI, 1.74–11.6; P = 0.002) for ctDNA PD compared with ctDNA MR. B, OS according to late ctDNA response from prior to treatment (time point A) to the time of the first CT evaluation (time point C). In the Cox proportional hazards model, there was an HR = 1.80 (95% CI, 1.21–2.66; P = 0.003) for ctDNA DC compared with ctDNA MR and an HR = 8.93 (95% CI, 3.32–24.01; P < 0.001) for ctDNA PD compared with ctDNA MR. C, OS according to ctDNA response from prior to the second treatment cycle (time point B) to the time of the first CT evaluation (time point C). In the Cox proportional hazards model, there was an HR = 2.06 (95% CI, 1.31–3.24; P = 0.002) for ctDNA DC compared with ctDNA MR and an HR = 11.14 (95% CI, 4.98–24.91; P < 0.001) for ctDNA PD compared with ctDNA MR.