Prognostic evaluation value of (18)F-FDG PET-CT in Hodgkin's lymphoma after treatment

Abstract

Objective: To investigate the prognostic value of (18)F-fluorodeoxygen-D-glucose-positron emission tomography /computerized tomography ((18)F-FDG-PET-CT) in Hodgkin's lymphoma (HL) at the end of first-line treatment (PET-end), by comparing the ratio of maximum standardized uptake value (SUV(max)) of lesion and liver SUV (rLL), SUV(max) reduction between baseline PET (PET-0) and PET-end (ΔSUV(max)), and Deauville 5-point scale (5-PS). Methods: Patients with HL newly treated in our hospital from August 2006 to December 2015 were retrospectively analyzed. All the patients enrolled in the study underwent post-treatment FDG PET-CT. The rLL and ΔSUV(max) were calculated, and all the cases were scored using Deauville 5-PS. The receiver operating characteristic (ROC) approach was applied to identify the optimal cut-point value, and survival curves according to different PET-CT assessment methods were estimated using the Kaplan-Meier analysis. The prognostic efficacy of different PET-CT assessment methods was compared, and DeLong test was used to verify it. Kaplan-Meier method and multivariate analysis using the Cox proportional hazards model were performed to analyze the potential independent risk factors. Results: There were 5 patients progressed within a 3-year follow-up. In the three PET-CT assessment methods, the predictive value of rLL and Deauville 5-PS were significant effective. ROC analysis for rLL as a progression predictor showed an optimal cut-point of 1.29. Deauville 5-PS=4 and rLL=1.29 showed the best prognostic accuracy. The sensitivity of rLL and Deauville 5-PS were both 80.0%, and the specificity of each was 98.0% and 93.7%, respectively. The positive predictive value (PPV) and negative predictive value (NPV) of rLL were 66.7% and 98.7%, while the PPV and NPV of 5-PS were 44.4% and 98.7%. The 3-years progression-free survival (PFS) rates of rLL≥1.29 group and rLL<1.29 group were 33.3% and 98.7%, with significant difference (P<0.001). The 3-years PFS rates of post-treatment Deauville 5-PS<4 group and Deauville 5-PS≥4 group were 98.7% and 55.6%, with significant difference (P<0.001). The prognostic evaluation efficacy of rLL was positively correlated with that of Deauville 5-PS (r=0.75, P<0.05). Area under curves (AUC) of rLL and Deauville 5-PS were 0.93 (95%CI: 0.825-1.000) and 0.91 (95%CI: 0.757-1.000), respectively. DeLong test showed the significant difference between the two methods (P<0.05). The univariate analysis results showed that clinical baseline stage, post-treatment rLL and Deauville 5-PS were associated with the prognoses of HL patients (P<0.05). The multivariate analysis results showed that post-treatment rLL and Deauville 5-PS were independent prognostic factors of HL (P<0.05). Conclusions: The rLL and Deauville 5-PS are potential prognostic factors for HL response assessment. The new semi-quantitative method rLL has methodological advantages over visual analysis, and it is a good supplement for Deauville 5-PS. rLL can improve prognostic evaluation accuracy of PET-CT and is useful to early identify patients with HL at a high risk of relapsing after first-line treatment.