End-of-treatment and serial PET imaging in primary mediastinal B-cell lymphoma following dose-adjusted EPOCH-R: a paradigm shift in clinical decision making

Dose-adjusted-EPOCH-R obviates the need for radiotherapy in most patients with primary mediastinal B-cell lymphoma. End-of-treatment PET, however, does not accurately identify patients at risk of treatment failure, thereby confounding clinical decision making. To define the role of PET in primary mediastinal B-cell lymphoma following dose-adjusted-EPOCH-R, we extended enrollment and follow up on our published phase II trial and independent series. Ninety-three patients received dose-adjusted-EPOCH-R without radiotherapy. End-of-treatment PET was performed in 80 patients, of whom 57 received 144 serial scans. One nuclear medicine physician from each institution blindly reviewed all scans from their respective institution. End-of-treatment PET was negative (Deauville 1-3) in 55 (69%) patients with one treatment failure (8-year event-free and overall survival of 96.0% and 97.7%). Among 25 (31%) patients with a positive (Deauville 4-5) end-of-treatment PET, there were 5 (20%) treatment failures (8-year event-free and overall survival of 71.1% and 84.3%). Linear regression analysis of serial scans showed a significant decrease in SUVmax in positive end-of-treatment PET non-progressors compared to an increase in treatment failures. Among 6 treatment failures, the median end-of-treatment SUVmax was 15.4 (range, 1.9-21.3), and 4 achieved long-term remission with salvage therapy. Virtually all patients with a negative end-of-treatment PET following dose-adjusted-EPOCH-R achieved durable remissions and should not receive radiotherapy. Among patients with a positive end-of-treatment PET, only 5/25 (20%) had treatment-failure. Serial PET imaging distinguished end-of-treatment PET positive patients without treatment failure, thereby reducing unnecessary radiotherapy by 80%, and should be considered in all patients with an initial positive PET following dose-adjusted-EPOCH-R (clinicaltrials.gov identifier 00001337).