Role of FDG PET/CT to Detect Bone Marrow Involvement in the Initial Staging of Aggressive Non-Hodgkin Lymphoma

BACKGROUND:

According to the Lugano classification (Cheson et al., JCO 2014, 32: 3059-3067) fludeoxyglucose positron emission tomography combined with computed tomography (FDG PET/CT) is the standard for evaluation and staging of aggressive non-Hodgkin Lymphoma (NHL). It is a matter of debate whether FDG PET/CT is sufficient to determine bone marrow (BM) involvement. We performed a pooled analysis of data from the PETAL (EudraCT 2006-001641-33) and OPTIMAL>60 trials (EudraCT 2010-019587-36) as prospective, open-label, randomized, multicenter Phase-III trials to determine whether initial staging with FDG PET/CT would allow non-invasive diagnosis of BM involvement and thus could avoid established but potentially painful and unpleasant BM biopsy (BMB).

PATIENTS AND METHODS:

Patients treated within the trials were included if both FDG PET/CT and BMB were performed during initial staging. Only patients with a biopsy-proven, centrally confirmed diagnosis of diffuse large B-cell lymphoma, primary mediastinal B-cell lymphoma or follicular lymphoma grade 3b were included. FDG PET/CT images and BM findings were blinded for central review by board certified PET/CT readers. Discordant findings were documented and resolved after unblinding by interdisciplinary discussion using findings of complementary imaging and/or subsequent PET examinations. Based on literature (Berthet et al., J Nucl Med 2013), a newly defined gold standard was used to confirm BM involvement. It includes a positive BMB or a positive FDG PET confirmed by targeted biopsy, complementary CT imaging or targeted MRI, or concurrent disappearance of focal FDG-avid BM lesions with other lymphoma manifestations during immunochemotherapy.

RESULTS:

Out of 1,976 patients a total of 930 patients met the inclusion criteria. BMB confirmed BM involvement in 85 of 930 patients (9%). According to FDG PET/CT, BM was involved in 185 out of 930 patients (20%) with 36 discordances (4%) between negative initial FDG PET/CT and positive BMB ("false negatives" with respect to previous reference standard, PRS). Discordances between positive initial FDG PET/CT and negative BMB ("false positives" by PRS) occurred in 136 patients (15%). If only BMB is used as in the PRS, the negative predictive value (NPV) of FDG PET/CT was 709/745=95% (95% confidence interval [95%CI]: 93%-97%) with a sensitivity (Sens) of 58% (95%CI: 46%-68%) and a specificity (Spec) of 84% (95%CI: 81%-86%). After discussion of these cases, 185 out of 221 cases with BM involvement were identified as true positive, resulting in a Sens of 84% for FDG PET/CT (95%CI: 78%-88%). By means of FDG PET/CT 745 cases were negative of which 709 were confirmed as true negatives, resulting in an NPV of 95% (95%CI: 93%-97%). All 185 cases with positive FDG PET/CT were evaluated as true positive for BM involvement in the unblinded synopsis of all findings, resulting in a positive predictive value (PPV) of 100% (95%CI: 98%-100%). All 709 negative FDG PET/CT findings were finally confirmed as such, so Spec was 100% (95%CI: 99%-100%). Thus, the prevalence in our total cohort analyzed was 24%, since 221 out of 930 cases had confirmed BM involvement. Diagnostic performance parameters for BMB as compared to the newly defined gold standard were Sens 85/221=38% (95%CI: 32%-45%), Spec 709/709=100% (95%CI: 99%-100%), PPV 85/85=100% (95%CI: 96%-100%) and NPV 709/845=84% (95%CI: 81%-86%), respectively. Differences between the PRS in the diagnosis of BM involvement by only BMB and the newly defined gold standard result mainly from false negative BMB due to sampling errors, which are detected by FDG PET/CT.

DISCUSSION:

In the majority of patients with aggressive B-cell lymphoma, routine BMB does not give any additional relevant diagnostic or prognostic information over FDG PET/CT alone and could therefore be omitted. BMB should only be performed if results would have direct therapeutic impact e.g. in patients with limited stage disease and lack of further risk factors according to the international prognostic index (IPI).