Assessment of Bone Marrow Involvement by FDG-PET in Patients with Follicular Lymphoma

Several studies evaluating the role of FDG-PET in assessing bone marrow (BM) involvement in patients with lymphoma have been published. However, they have relied on visual analysis alone which has proved inadequate. The aim of this study was to assess the value of semi-quantitative assessment in identifying BM involvement in patients with follicular lymphoma (FL).

This is a single centre retrospective analysis of 41 patients with grade 1–3a FL (female: 51%; median age: 64 years, range: 30 – 87) who underwent FDG -PET/CT and unilateral bone marrow biopsy (BMB) as part of the staging assessment. Thirty patients were newly diagnosed and 11 had their staging investigations at relapse. Amongst the latter, none of them had received any treatment in the previous 3 months. BM uptake was assessed both visually and with semi-quantitative analysis. On visual analysis, BM uptake was compared to the liver as the reference. Semi-quantitative analysis involved measuring the maximum standardized uptake values (SUVmax) at the sternum, both iliac blades and the T12 vertebra. An average of these 4 measurements was recorded as SUVav. The highest SUVmax recorded in any of the 4 bone sites, the SUVav, and the ratios SUVav/mediastinal blood pool (MBP) and SUVav/liver were compared by t-test according to whether patients had a positive or a negative BMB. Optimal SUVmax, SUVav, SUVav/liver and SUVav/MBP cutoffs were defined by receiver operating characteristic (ROC) curves.

Sixteen patients (37%) had BM involvement on the BMB but 11 of them did not have increased uptake in the BM compartment on visual analysis of the FDG-PET. In contrast, 7 patients had BM involvement (focal lesions in 6, diffuse involvement in 1) on visual analysis of the FDG-PE but 2 had a negative BMB. All semi-quantitative measurements were significantly higher in patients with a positive BMB in comparison with those with a negative BMB (table). The optimal cutoff values were ≥2. 5 for SUVmax, ≥2 for SUVav, ≥0. 75 for SUVav/liver, and ≥1 for SUVav/MBP. The sensitivity and the specificity for the prediction of BM involvement utilising SUVmax ≥2. 5 at a single site was 56% and 84%. However, the SUVav ≥2 had a better sensitivity (63%) whilst maintaining a high specificity of 96%. The sensitivity was still further improved by normalizing these values as a ratio of the uptake: SUVav/liver ≥0. 75 had a sensitivity of 81% and a specificity of 80%, whereas the sensitivity and specificity of SUVav/MBP ≥1 were 88% and 80%, respectively.

Visual assessment of FDG-PET has a high specificity but a low sensitivity to detect BM involvement in FL patients. In contrast, semi-quantitative assessments can improve the sensitivity of FDG-PET to predict BM involvement from 31% to 88%. Were these results confirmed in further prospective studies, they may help to predict BM involvement in FL patients, which may result in sparing selected patients from having a BMB.

Gribben:Celgene: Honoraria; Roche: Honoraria; Pharmacyclics: Honoraria; GSK: Honoraria; Mundipharma: Honoraria; Gilead: Honoraria.