Background: FDG PET/CT and bone marrow biopsy (BMB) are considered standard procedures for the staging of patients with new, untreated FL. A key issue in FL is the early identification of patients who will fail early. We recently reported (Am J Hematology 2019) that the presence of ≥2 EN sites, spleen, bone or soft tissue involvement as detected by PET all predicted failure to achieve EFS24. In Hodgkin lymphoma and diffuse large B-cell NHL, PET has replaced the routine need for a staging BMB. However, there is no such evidence in FL. The goal of this study was to determine the value of FDG PET/CT in determining bone involvement in FL using BMB as the gold standard.
Methods: Patients were identified using the Mayo Clinic Lymphoma Database. 548 patients with newly diagnosed FL grades 1-3A between years 2003-2016, available BMB results, and PET/CT imaging at diagnosis, were included in the analysis. The presence of bone and spleen involvement on PET/CT, SUVmax and SUVmean of the axial skeleton at L3, and BMB results were recorded and compared.
Results: In all, 36% (197/548) of patients had a positive BMB, and 34% (189/548) had bone involvement detected on PET/CT. Compared to BMB, the sensitivity and specificity of PET/CT in detecting bone involvement as determined by BMB were 60% and 80%, respectively. We noted that 59 patients had focal bone involvement on PET/CT rather than a diffuse component, and found that 47% (28/59) of these patients had a negative BMB obtained in the posterior iliac crest. Excluding these patients, the sensitivity and specificity of PET/CT in detecting bone involvement were 53% and 88%, respectively (Table 1).
With respect to the spleen, 29% (157/548) of patients had evidence of splenic FL involvement on PET/CT, and of these, 69% (109/157) also had a positive BMB. The sensitivity and specificity of spleen involvement on PET/CT in predicting bone involvement on BMB were 55% and 86%, respectively (Table 2).
We recorded SUV data at L3 in the 439 patients who had either a diffuse pattern of bone involvement on PET/CT, or a PET/CT read as negative. We analyzed the positive and negative predictive values (PPV and NPV) of SUVmax and SUVmean at several cut-off points to determine whether axial bone SUV is reliable at determining patients with a positive or negative BMB. The NPV for an SUVmax of less than 2.0 was 96% (n=25 patients classified as negative). For SUVmean, the best cut-off point was at less than 1.4, where NPV was 100% (n=15 patients classified as negative). There was no logical cut-off point for a significant PPV > 95%.
Conclusion: In newly diagnosed FL, the sensitivity and specificity of bone involvement on PET/CT are insufficient for PET/CT to routinely replace BMB. However, in patients where the need for BMB at staging is being debated, certain factors on PET/CT can help facilitate this decision. The detection of focal bone lesions, especially those that may be missed on posterior iliac crest BMB, can make BMB unnecessary. If both the spleen and bone appear involved on PET/CT, this confers a relatively high chance that BMB will be positive. If SUVmax at L3 is less than 2.0, or if SUVmean is less than 1.4, the BMB will likely be negative, with a NPV > 95%. This decision-making algorithm is outlined in Figure 1 and may serve as a useful guideline for clinical trials and routine practice.
No relevant conflicts of interest to declare.
