Results
Study populationIn total, 225 patients were identified as having undergone PET/CT scans as part of an FUO workup. Of them, 97 did not meet study inclusion criteria (Fig. 1). Of the 128 patients who met eligibility criteria and were included within the cohort, 80 (62.5%) were males. The mean age of the patients was 59 ± 20.3 years (range: 18-93). The mean duration of fever (prior to the performance of PET/CT) was 38 days. Twenty-nine (22.8%) of the patients were considered immunosuppressed, due to concomitant treatment with corticosteroids (19 patients) or other immunosuppressive drugs, some of whom were also receiving corticosteroids (19 patients), or by having medical conditions that cause immunosuppression. Twenty-two (17%) patients had implanted prosthetic devices, including vascular implants, pacemakers and heart valves. Table 1 presents the demographic and clinical characteristics of the study population, according to whether PET/CT provided contributory results.
Final diagnoses By the completion of the work-up, 95 patients had (74.2%) received a final diagnosis. Most (64.2% of the diagnosed patients and 47.7% of the study population) were diagnosed with infectious diseases (Table 2). These included pulmonary infections (n=12), endovascular infections (n=7), endocarditis (n=7), bacteremia (n=4), and osteomyelitis (n=2). Twenty-one (16.4%) of the patients were diagnosed with inflammatory diseases, including arthritis (n=3), adult onset Still’s disease (n=3), and vasculitis (n=3). Malignancies were diagnosed in 12 patients (9.4%), nine (75%) of which were hematologic, and included lymphoma (n=6) and leukemia (n=3). One patient (0.8%) was diagnosed with Kikuchi disease. No compatible diagnosis was made for 33 (25.8%) patients at 3 months following admission. The different diagnoses are presented in Table 2. Figures 4-7 show PET/CT scan images of four patients with FUO who were included in the study.
Contribution of PET/CT to FUO workup Overall, 86 (67.2%) of the 128 included patients had positive findings in PET/CT scans. In 61 (47.7%) patients, the results were deemed true positive for leading to the final diagnosis. The remaining 25 (19.5%) had false positive results (Fig. 2).
Forty-two (33%) of the included patients did not have pathological FDG uptake findings on PET/CT. Indeed, twenty-four (18.8% of the study population and 57.0% of patients without diagnostic findings on PET/CT) did not have a final diagnosis at the end of the workup or at 3 months follow-up and were thus considered true negative. The remaining 18 (14%) patients without a pathological FDG uptake on PET/CT were eventually diagnosed using other modalities and were therefore considered false negative (Fig. 2).
Findings yielded a sensitivity of 77.2%, specificity of 49%, positive predictive value of 70.9%, and negative predictive value of 57.1%.
PET/CT contribution according to disease category Of patients with final diagnoses of inflammatory and infectious conditions, the rates of true positive results were 60.0% and 61.7%, respectively and the rates of false negative results were 30.0% and 20.0%, respectively. Of the patients who were given a final diagnosis of malignancy, 8.3% of them had false positive results. None had false negative results (Fig. 3).
Parameters associated with true positive resultsIn a univariate analysis, weight loss, low concentrations of hemoglobin and low levels of transferrin were associated with true positive diagnoses (Table 1). In a multivariate analysis, hemoglobin was inversely associated with contributory PET/CT tests (odds ratio (OR): 0.597, 95% confidence interval (CI) 0.412-0.866, p=0.006). Weight loss, on the other hand, was positively associated with contributory PET/CT tests (OR: 3.605, 95% CI 1.123-11.576, p=0.031). The association of low-level transferrin with a diagnostic PET/CT was not significant: (OR: 0.992, 95% CI 0.981-1.003, p= 0.13) (Table 3).