Assessment of Tumor Burden in Lymphoma Patients with Deauville Score 4 Disease on Post Therapy FDG PET

Introduction: Lymphoma tumor FDG uptake is typically measured by SUVmax and the study is reported following the Lugano criteria with 5-point Deauville score (DS). Residual lesions with a DS 4 on post therapy scan are considered partial metabolic response and may be indicative of residual lymphoma. Tumor burden by metabolic tumor volume (MTV) and total lesion glycolysis (TLG) has been shown prognostic of survival, but the diagnostic value is uncertain. Our goal was to assess the feasibility and robustness of tumor segmentation in patients with DS 4 residual lesions, which is the most challenging disease group, and to determine the diagnostic value of MTV and TLG for malignant residual lymphoma.

Methods: This is a retrospective study consisted of consecutive patients with aggressive lymphoma referred for FDG PET/CT from 2017 to 2022. Cases with DS 4 on end of therapy scans were identified by searching FDG PET reports and confirmed on images. Patient level classification to benign or malignant lymphoma was based on biopsy and/or imaging follow-ups. Patients with follow-up images showing improvement after therapy, mixed DS 4 and DS 5, and interim scans in Hodgkin’s lymphoma were excluded. SUVmean of the liver was measured in the right lobe using a 50-cc sphere. MTV and TLG were measured by applying a threshold of 1.5×SUVmean of the liver by two readers (nuclear medicine physician and technologist) using Affinity (Hermes Medical Solutions) workstation. Cases with residual lesion uptake below the pre-defined threshold were excluded from ROC analysis. Correlation between readers was assessed with Pearson’s correlation and Bland-Altman plots. The diagnostic value of SUVmax, MTV (cc) and TLG (SUV·cc) was assessed with ROC curve analysis. Cases with residual lesion uptake below the pre-defined threshold (1.5×liver SUVmean) were excluded from ROC analysis.

Results: Of the 66 patients with post therapy DS 4 meeting the criteria, 19 were malignant (29%) and 40 (61%) had baseline scans. Median duration between the end of therapy and follow-up scan was 39 days (range: 9-99). The segmentation of tumors was straightforward with functions of constraint and eraser to exclude physiological uptake. Inter-operator Pearson’s correlation was 1.000 for both MTV and TLG and Bland-Altman plots for both MTV and TLG showed no bias between readers. 

For baseline scans, SUVmax, MTV and TLG were 22.0±6.4, 1784±4043, and 8151±8817 in the benign (n=30) versus 20.4±11.5 (p=0.65), 404±321 (p=0.3), and 2472±2101 (p=0.057) in the malignant group (n=10). 21 patients (38% malignant) with residual lesions uptake below the pre-defined threshold on post therapy scan were excluded from analysis. For post therapy scans (n=45), SUVmax, MTV and TLG were 5.1±1.3, 3.5±5.3, and 13.4±20.3 in the benign (n=34) versus 4.8±1.3 (p=0.48), 1.5±2.4 (p=0.25), and 6.3±10.4 (p=0.29) in the malignant group (n=11).

ROC analysis on SUVmax, MTV, and TLG for both baseline and post treatment scans were only statistically meaningful for baseline TLG, post treatment MTV and post treatment TLG, with corresponding AUC of 0.68, 0.71 and 0.71. On post treatment scans, the sensitivity was 73% and specificity 82% with TLG < 1.9, and sensitivity 73% and specificity 77% with MTV < 0.7 for residual lymphoma.

Conclusions: It is clinically feasibility to reproducibly segment tumors on FDG PET in the DS 4 patient group with excellent inter-reader concordant. Unexpectedly, lower MTV and TLG values correspond to residual lymphoma disease post-therapy. This result could possibly be due to post-therapy inflammatory changes if images performed too soon following the therapy. This small sample study warrants further studies with larger patient population, especially in the partial metabolic response group.