Predicting diffuse large B-cell lymphoma outcomes with lesionto- liver maximum standardized uptake value for interim-treatment and end-of-treatment positron emission tomography-computed tomography

Abstract

Background: 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography-computed tomography (PET-CT) has been used in response evaluation systems for malignant lymphomas and is an important tool for determining efficacy and prognosis. The Deauville 5-point scale (D-5PS) is an 18F-FDG PET-CT imageinterpretation protocol for patients with lymphoma. Nevertheless, a number of limitations in visual image interpretation, such as interobserver disagreement and the increase of false-positive results, suggests that new parameters are needed. In this study, we aimed to evaluate the prognostic values of interim-treatment (I-) and end-of-treatment (EOT) PET-CT by comparing D-5PS to the semiquantitative lesion-to-liver maximum standardized uptake value ratio (RLL). Methods: A total of 90 patients with diffuse large B-cell lymphoma (DLBCL) (45 I-PET and 45 EOTPET) were analyzed, and the RLL was calculated. Patients were additionally evaluated using the D-5PS system. We determined the optimal cutoff value of RLL using receiver operating characteristic (ROC) analysis. Kaplan-Meier survival analysis was used to compare the outcome predictions, while multivariate Cox regression analysis was used to identify the predictive factors. Results: Among the patients examined, 41 (20 I-PET and 21 EOT-PET) experienced progression, and 49 (25 I-PET, 24 EOT-PET) did not. The optimal cutoff values of the RLL for predicting disease progression were 1.37 for I-PET (sensitivity 75%, specificity 88%) and 2.03 for EOT-PET (sensitivity 45.5%, specificity 100%), while the cutoffs of the D-5PS were scores 4 for I-PET (sensitivity 80%, specificity 72%) and 5 for EOT-PET (sensitivity 40.9%, specificity 100%). The prognostic efficacy was higher for the RLL at interim than for the D-5PS [area under the curve (AUC) =0.848 vs. 0.741]. The EOT prognostic efficacy of both evaluation methods was essentially equivalent (AUC =0.785 vs. 0.725). Univariate and multivariate analyses showed that RLL and D-5PS were independent factors affecting DLBCL outcomes for both interim and EOT assessment. Conclusions: RLL and D-5PS have independent predictive values for the interim and EOT evaluation of outcomes in patients with DLBCL. The RLL has better interim predictive ability than does D-5PS and can optimize D-5PS interpretation, thus improving interim outcome prediction.