Background: Studies on the accuracy of F18-FDG PET/CT scans (PET) in osseous infections have provided mixed results, and few studies have commented on clinical utility. Our goals were to evaluate the sensitivity and specificity of PET in patients with osteomyelitis (OM) and clinical utility of PET in the diagnosis and management of OM.
Methods: We reviewed all patients with PET ordered for evaluation of OM over a 2 yr period in a tertiary care hospital; patients with simultaneous malignancy were excluded. Imaging was read by a single radiologist. An infectious disease specialist blinded to the PET result assessed the clinical course of each patient; after this review, the same specialist re-evaluated each case in light of PET results. The likelihood of an infectious diagnosis pre and post scan was divided into negative (0), possible (1), probable (2), and definite (3) based on presentation, laboratory values, microbiologic assays. Diagnostic test calculations were determined in standard manner. PET was considered clinically useful if it altered the likelihood of diagnosis or changed management.
Results: Out of 67 cases reviewed, 57 patients met criteria for further analysis. One case with an inconclusive reading was excluded from sensitivity and specificity analysis. In one case, the likelihood remained unchanged post scan but had clinical utility as the purpose was to assess disease extent for surgery. PET was pos for 31 and neg for 25 patients; the final diagnosis of osteomyelitis (likelihood category 2 or 3) was established as pos for 29 and neg for 27 patients, leading to a sensitivity 90%, specificity 85%, positive predictive value (PPV) 87%, negative predictive value (NPV) 92%. PET results affected clinical management in 38 (67%) cases and had no impact in 17 cases (33%) (Fig 1 and 2).
Conclusion: Results of PET sensitivity and specificity were similar to other studies; PET did not change suspicion of the final diagnosis or alter management in 33% of cases, the majority falling in extremes of diagnostic likelihood. When used in this format, the PPV and NPV are not optimal. Our analysis of clinical utility suggests that if OM is already strongly suspected or ruled not clinically likely, imaging with PET scan is not warranted, as the outcome is frequently no change in management.
L. Tamara, None
D. Musher, None