Importance: Longitudinal circulating tumor DNA (ctDNA) has been shown to predict response and survival in patients with metastatic melanoma treated with anti–programmed cell death 1 (PD-1) antibodies. Pseudoprogression, defined as radiologic finding of disease progression prior to response, has been a challenge to clinicians.
Objective: To establish whether ctDNA at baseline and up to week 12 of treatment can differentiate between the radiologic findings of pseudoprogression and true progression in patients with metastatic melanoma.
Design, setting and participants: This explorative biomarker study examined circulating BRAF and NRAS mutations in a cohort of 125 patients with melanoma receiving PD-1 antibodies alone or in combination with ipilimumab between July 3, 2014, and May 24, 2016. Pseudoprogression was defined retrospectively as radiologic progression not confirmed as progressive disease at the next radiologic assessment. Plasma samples of ctDNA at baseline and while receiving treatment were taken for analysis prospectively over the first 12 weeks of treatment. Favorable ctDNA profile (undetectable ctDNA at baseline or detectable ctDNA at baseline followed by >10-fold decrease) and unfavorable ctDNA profile (detectable ctDNA at baseline that remained stable or increased) were correlated with response and prognosis.
Main outcomes and measures: Early differentiation of pseudoprogression from true progression using longitudinal ctDNA profile.
Results: According to guidelines by Response Evaluation Criteria in Solid Tumors (RECIST), progressive disease occurred in 29 of the 125 patients (23.2%). Of the 29 patients, 17 (59%) were 65 years or younger, 18 (62%) were men, 9 (31%) had pseudoprogression, and 20 (69%) had true progression. Of the 9 patients (7%) with confirmed pseudoprogression, all patients had a favorable ctDNA profile. At a median follow-up of 110 weeks, 7 of 9 patients (78%) were alive. All but 2 patients with true progression had an unfavorable ctDNA profile. Sensitivity of ctDNA for predicting pseudoprogression was 90% (95% CI, 68%-99%) and specificity was 100% (95% CI, 60%-100%). The 1-year survival for patients with RECIST-defined progressive disease and favorable ctDNA was 82% vs 39% for unfavorable ctDNA (hazard ratio [HR], 4.8; 95% CI, 1.6-14.3; P = .02). Overall survival was longer in patients with a partial response (54 of 125 patients [43%]) compared with patients with progressive disease and a favorable ctDNA profile (11 of 125 patients [9%]; HR, 0.09; 95% CI, 0.01-0.80; P < .01).
Conclusions and relevance: The results demonstrate that ctDNA profiles can accurately differentiate pseudoprogression from true progression of disease in patients with melanoma treated with PD-1 antibodies. Results of this blood test performed at regular intervals during systemic treatment reflect tumor biology and have potential as a powerful biomarker to predict long-term response and survival.