Interim PET Evaluation in Diffuse Large B-Cell Lymphoma Using Published Recommendations: Comparison of the Deauville 5-Point Scale and the ΔSUVmax Method

The value of interim 18F-FDG PET/CT (iPET)–guided treatment decisions in patients with diffuse large B-cell lymphoma (DLBCL) has been the subject of much debate. This investigation focuses on a comparison of the Deauville score and the change-in-SUVmax (ΔSUVmax) approach—2 methods to assess early metabolic response to standard chemotherapy in DLBCL. Methods: Of 609 DLBCL patients participating in the PET-Guided Therapy of Aggressive Non-Hodgkin Lymphomas trial, iPET scans of 596 patients originally evaluated using the ΔSUVmax method were available for post hoc assessment of the Deauville score. A commonly used definition of an unfavorable iPET result according to the Deauville score is an uptake greater than that of the liver, whereas an unfavorable iPET scan with regard to the ΔSUVmax approach is characterized as a relative reduction of the SUVmax between baseline and iPET staging of less than or equal to 66%. We investigated the 2 methods’ correlation and concordance by Spearman rank correlation coefficient and the agreement in classification, respectively. We further used Kaplan–Meier curves and Cox regression to assess differences in survival between patient subgroups defined by the prespecified cutoffs. Time-dependent receiver-operating-characteristic curve analysis provided information on the methods’ respective discrimination performance. Results: Deauville score and ΔSUVmax approach differed in their iPET-based prognosis. The ΔSUVmax approach outperformed the Deauville score in terms of discrimination performance—most likely because of a high number of false-positive decisions by the Deauville score. Cutoff-independent discrimination performance remained low for both methods, but cutoff-related analyses showed promising results. Both favored the ΔSUVmax approach, for example, for the segregation by iPET response, where the event-free survival hazard ratio was 3.14 (95% confidence interval, 2.22–4.46) for ΔSUVmax and 1.70 (95% confidence interval, 1.29–2.24) for the Deauville score. Conclusion: When considering treatment intensification, the currently used Deauville score cutoff of an uptake above that of the liver seems to be inappropriate and associated with potential harm for DLBCL patients. The ΔSUVmax criterion of a relative reduction in SUVmax of less than or equal to 66% should be considered as an alternative.