Staging Hodgkin Lymphoma Using PET - Can We Safely Exclude Bone Marrow Involvement?

Objectives: [¹⁸F]fluorodeoxyglucose (FDG)-positron emission tomography (PET) assessment for staging of Hodgkin lymphoma (HL) patients may be more useful to detect bone marrow (BM) involvement as compared to bone marrow biopsy (BMB) which only allows assessment of a single BM location. In the multicenter trials of the German Hodgkin Study Group (GHSG), BMB was performed as standard procedure in all patients and FDG-PET was acquired in a large subgroup of these patients. We compared the BMB and FDG-PET results with respect to BM involvement in advanced-stage Hodgkin lymphoma patients, to test if bone marrow involvement can be safely excluded by FDG-PET.

Methods: BMB was performed as a standard procedure in the GHSG trials HD16, HD17 and HD18. In contrast, FDG-PET was a facultative staging procedure. All available baseline FDG-PET scans were centrally reviewed and focally increased bone marrow uptake was rated as bone marrow involvement. Both the number of lesions and their localization was noted. Diffuse bone marrow uptake was recorded but not interpreted as bone marrow involvement.

Results: In 832 patients a FDG-PET scan had been performed for staging which was reviewed by the central review panel. Of 20 patients with biopsy-proven BM-involvement, 19 were detected by FDG-PET and 1 was documented to have diffuse BM uptake. FDG-PET showed bone marrow involvement in 129 patients, therefore suggesting BM involvement in 110 additional cases as compared to biopsy. A unifocal pattern was noted in 43 patients, a bifocal in 17, a trifocal in 11 and a multifocal (more than 3 lesions) in 58. Diffuse BM uptake was documented in 152 patients. As compared to biopsy results, FDG-PET showed a sensitivity of 95.0 % [95 % confidence interval (CI) 0.75, 1.00]. The negative predictive value of FDG-PET was 99.9 % (95 % CI 0.99, 1.00).

Conclusion: Baseline PET-CT imaging can be safely used to exclude BM involvement. In our cohort, FDG-PET could detect a large number of additional cases with probable BM involvement who would have otherwise been considered negative by BMB. This might indicate that conventional BMB underestimates the actual incidence of BM involvement.