Integration Of CT-Based Measurements Into PET Assessment Significantly Improves The Positive Predictive Value Of Surveillance FDG PET/CT In Patients With Diffuse Large B Cell Lymphoma

Surveillance FDG PET, commonly employed in patients with diffuse large B cell lymphoma (DLBCL) who achieve remission following induction, is very sensitive but not sufficiently specific, and is associated with a high false positive (FP) rate. The current study was aimed to investigate whether the employment of specific CT morphologic measurements could reduce FP incidence and improve the positive predictive value (PPV) and clinical applicability of surveillance FDG PET/CT in detecting DLBCL relapse.

The study was approved by the Institutional Review Board (IRB) of the Rambam Health Care Campus (Haifa, Israel). The study population included patients with DLBCL who participated in the previous surveillance PET study (n=119) (Avivi I, Am J Hematol, 2013), and had at least one positive PET scan, suggestive of disease relapse (n=83). A PET scan was defined as positive in the presence of increased FDG uptake unrelated to physiological bio-distribution of the tracer or to a known benign process. CT-derived features of PET positive sites, including long and short diameters (mm), the presence of calcification and fatty hilum within lymph nodes, were assessed. Patient's outcome, focusing on relapse development, confirmed either by tissue biopsy or by consecutive imaging demonstrating progression in the presence of compatible clinical symptoms, was recorded. The performance of surveillance PET scans, with and without concurrent employment of CT-derived measurements, was compared, focusing on FP rate and PPV of these two approaches to relapse detection.

Eighty three follow-up PETs (FU-PETs) interpreted as positive for relapse, were reviewed. Seventy studies performed in 51 patients were included in final analysis, 13 studies were excluded as CT quality did not allow adequate morphologic evaluation. Twenty five of 70 FU-PET studies (36%) were true positive (TP) and 45 (64%) false positive, which is compatible with a PPV of 36%. Evaluation of CT-related parameters of PET positive sites showed that the employment of either long axis ≥15 mm or short axis ≥10 mm, significantly improved the prediction of relapse by PET (Table 1). The ratio between long and short axis measurements was not found to efficiently discriminate between FP and TP scans. Likewise, the presence of calcification or fatty hilum did not affect this discrimination. Multivariate analysis found IPI ≥2, lack of prior rituximab therapy and long or short axis measurements ≥15 and 10 mm, respectively, to be independent predictors of true positivity of FU-PET scans (odds ratio=9.84 ,8.5, and 7.67, P< 0.05, respectively).

Complementary application of long and short axis measurements ≥15 and 10 mm, respectively may allow the beneficial use of PET/CT in detection of DLBCL relapse, significantly improving the PPV of this follow-up modality.

No relevant conflicts of interest to declare.