Limited Prognostic Impact of Baseline Bone Marrow Infiltration Burden in Diffuse Large B-cell Lymphoma: Insights from Manual Segmentation 18F-FDG PET/CT Study

Introduction: Quantitative measures associated with 18F-FDG PET/CT scan have been shown to be prognostic for Overall Survival and Progression-Free Survival of Diffuse Large B-cell Lymphoma (DLBCL) patients, while bone marrow (BM) involvement is associated with a poorer outcome in both categories. Notably, volumetric evaluation of BM tumor burden on FDG PET/CT has rarely been investigated in the literature. Our objective was to evaluate the prognostic significance of 18F-FDG PET/CT volumetric measures of total disease (TD) and BM tumor burden, and of several clinical parameters.

Methods: We retrospectively included 184 DLBCL patients (disease stages II-IV), who underwent baseline FDG PET/CT imaging median 4 days prior to treatment (range 0-100) and were treated at the Institute of Oncology Ljubljana between January 2016 and December 2020. PET/CT scans were performed on a Siemens Biograph mCT40 PET/CT, 60 minutes post i.v. injection of 3.7 MBq/kg FDG. Manual segmentation of baseline PET/CT images was performed by an experienced nuclear medicine physician using Slicer 3D software suite. Separate segments for BM and lesion infiltrates elsewhere (XL) were delineated, employing a fixed absolute SUV threshold of 4.0 g/ml. Study end-points were prognostic assessments for complete remission (CR) after initial treatment with 6 or 8 cycles of standard chemoimmunotherapy, 3-year (OS3) and 5-year (OS5) overall survival. We included SUVmax and Metabolic Tumor Volume (MTV) of three categories: BM and XL infiltrates, and TD infiltrates corresponding to sum of BM and XL volumes. Outside of imaging domain, we considered baseline International Prognostic Index (IPI) score, stage, and MIB-1 expression.

Results: Median age was 68 years, range: 20-85, 80 were females. CR was achieved in 121/184 patients (65.7%). OS3 and OS5 rates were 74.5% and 59.4%, respectively. Using a univariate predictive model, RT was significantly associated with XL MTV (optimal cut-off <534 cm³, sensitivity 83%, specificity 48%, AUC 0.68, p=0.002), TD MTV (<228 cm³, 65%, 68%, 0.69, p<0.001), and BM SUVmax (sensitivity 84%, specificity 40%, AUC 0.62, p=0.001). Tendencies toward significance were observed for IPI (p=0.031) and stage (p=0.063), while BM MTV (p=0.18) and MIB-1 (p=0.549) did not reach statistical significance. OS3 and OS5 were significantly associated with XL MTV (optimal cut-off 198 cm³ and 198 cm³, sensitivity 62% and 62%, specificity 78% and 74%, AUC 0.71 and 0.712, p=0.001 and p=0.01, respectively), TD MTV (183 cm³ and 274 cm³, 56% and 67%, 80% and 67%, 0.69 and 0.7, p=0.002 and p=0.012, respectively), IPI (3 and 3, 70% and 62%, 68% and 81%, 0.74 and 0.77, p=0 and p=0, respectively), and stage (4 and 4, 50% and 56%, 85% and 84%, 0.65 and 0.65, p=0.003 and p=0.003, respectively). However, OS3 and OS5 were not significantly associated with BM MTV, BM SUVmax and MIB-1 (p=0.636 and p=0.943; p=0.347 and p=0.409; p=0.341 and p=0.187, respectively). There was a very strong positive correlation between TD MTV and XL MTV (Pearson's r=0.98). A moderate correlation was observed between IPI and both TD MTV and XL MTV (r=0.5 and r=0.49, respectively). Weak positive correlation was found between stage and TD MTV and XL MTV (r=0.35 and r=0.33, respectively). No correlation was seen between MIB-1 and TD MTV (r=-0.01), XL MTV (r=-0.01), stage (r=0.02), and IPI (r=-0.02).

Conclusions: We have confirmed that pretreatment TD and BM volumetric PET biomarkers effectively predict CR after the initial treatment with 6 or 8 cycles of standard chemoimmunotherapy in stage II-IV DLBCL patients. Survival expectations can be reliably anticipated based on TD volumetric PET biomarkers, IPI, and stage. Surprisingly, PET derived BM tumor burden did not emerge as a dependable biomarker for survival estimation.