Dual Time Point FDG-PET/CT Imaging Can Distinguish Between Hodgkin Lymphoma and Diffuse Large B-Cell Lymphoma

Introduction: 18F-fluorodeoxyglucose-positron emission tomography/computed tomography (PET/CT) has become a standard of care imaging modality for multiple types of lymphoma, including Hodgkin lymphoma (HL) and diffuse large B-cell lymphoma (DLBCL). However, many applications of PET/CT in lymphomas have not yet been explored. We investigated whether dual time point (DT) PET/CT would detect more sites of disease due to improved tumor to background ratio. It has also been reported that inflammatory lesions have decreased FDG accumulation over time, whereas malignant lesions accumulate FDG over time, becoming more FDG-avid. Histologically, HL lesions are predominantly composed of inflammatory cells and relatively few tumor cells. This tumor composition calls into question whether HL will behave as a malignancy or as an inflammatory lesion on DT PET/CT imaging compared to DLBCL and whether changes in FDG accumulation over time on DT PET/CT imaging will differ between HL and DLBCL.

Methods: Patients (pts) with untreated, biopsy-confirmed, nodular sclerosing HL and DLBCL were prospectively enrolled on this clinical trial (clinicaltrials.gov NCT01004718). Pts with uncontrolled diabetes, active infection, and primary mediastinal large B-cell lymphoma were excluded. Following injection of FDG, pts underwent PET/CT at 60 minutes (DT60) and 180 minutes (DT180). Any lesion with a Deauville Score of 4-5 was considered positive (i.e. FDG uptake > liver at any site) (Meignan, Leuk Lymphoma 2012). SUVmax, SUVmean, and tumor to background ratio, defined as lesion SUVmax / liver SUVmean, were calculated for each lesion. Software using an automatic adaptive thresholding method (ROVER, ABX GmbH) was also used to measure the metabolically active volume (MAV) of lesions on PET images. Mean metabolic volumetric product (MVPmean) was calculated as MVPmean = SUVmean * MAV for each lesion, and total MVPmean for a given pt was calculated as the sum of MVPmean for all lesions.

Results: 160 total lesions (78 lesions with DT60 and 82 lesions with DT180 images) were evaluated in this prospective study of 20 paired PET/CT scans in 10 pts. Five pts had HL and 5 had DLBCL. Median age was 51; 5 pts were men. Pts with DLBCL had higher SUVmax values at both 60 and 180 minutes than HL (DT60: DLBCL 22.4 v HL 10.8, p = 0.0007, t test; DT180: DLBCL 31.6 v HL 14.6, p = 0.0006, t test) as well as a higher average SUVmean than HL at both time points (DT60: DLBCL 9.2 v 4.5, p = 0.009 t test; DT180: DLBCL 12.7 v 5.8, p = 0.02 t test). We also observed that the average %change in tumor to background ratio for HL increased from DT60 to D180 by 75% whereas the %change in tumor to background ratio for DLBCL increased by 34% (p = 0.0001, t-test). DT180 detected 4 additional lesions that were not apparent on DT60; 3 lesions were in HL and 1 in DLBCL. The locations of these lesions were mesenteric, lung, external iliac, and subclavian regions. For each pt, %change in total MVPmean between DT60 and DT180 was higher in DLBCL (p=0.02, rank sum). Of interest, the %change in MAV between DT60 and DT180 decreased in HL compared to slight increase in DLBCL (HL 12% decrease v. DLBCL 2% increase in MAV). Thus, changes in MAV accounted for the difference detected in %change in total MVPmean between these histologies. There was no difference between HL and DLBCL in %change in SUVmean or %change in SUVmax.

Conclusions: We report the first prospective study of dual time point PET/CT imaging of lymphomas, as well as the first study to specifically compare dual time point imaging of DLBCL and HL. Our results indicate that lesions of DLBCL are generally more FDG-avid than HL when measured as either SUVmax or SUVmean at 60 and 180 minutes. When measured as %change in total MVPmean, DLBCL accumulates significantly more FDG than HL, consistent with the differences in histologic composition of these tumors. Computerized algorithms for generation of MAV and SUVmean add additional information not reflected in measurement of SUVmax alone that may be useful for distinguishing between tumor histologies. Given variation between histologies, optimal timing of image acquisition for detection of lesions may be improved by choosing an imaging time point that optimizes tumor to background ratio. Tumor to background ratio is improved, especially for HL, with use of imaging that occurs at 180 minutes, and we demonstrate the ability to detect additional lesions not detected at 60 minutes.