Discussion
In this retrospective study, PET/CT proved helpful in diagnosing almost half of all patients who underwent scans as part of FUO work-up. The management of patients with FUO has evolved considerably since this entity was initially described. Nevertheless, many patients remain undiagnosed despite extensive testing. Review of the current literature suggests that PET/CT is a useful part of work-up though reports vary widely, citing numbers which range from 42% to 72% of cases in which they assisted in the diagnosis [5-8,12-21]. The variability of results reported can be explained, in part, by differences in the methodologies employed, definition of FUO, the varying work-up algorithms and the fluctuating place of PET/CT within the algorithms, technique and model of the PET scanners, and how different groups define whether PET/CT is helpful to establishing the diagnosis. This diversity makes it difficult to compare among the published studies.
In a prospective study by Schönau et al. [8], PET/CT scan was a mandatory part of the work-up of FUO. All patients with FUO who presented to internal medicine wards were included. PET/CT yielded true positive results in 56.7% of the patients with FUO. In contrast, we included only patients without a diagnosis after a thorough medical workup; 44.0% of them underwent CT-scans before advancing to a PET/CT scan.
Our rate of false negative results of PET/CT scans, 14.1%, is comparable to rates reported in the literature. Two thirds of these patients were diagnosed with infectious diseases and one third with inflammatory diseases (familial Mediterranean fever, adult-onset Still’s disease, polymyalgia rheumatica and temporal arteritis) while none were diagnosed with malignant diseases. Schönau et al. [8] found false negatives in only 5% of their cohort. In an early study, Gafter-Gvili et al. [7] reported a rate of false negatives of 18%; of them, 40% were eventually diagnosed with infectious diseases, and one patient with a malignant disease. This patient was subsequently diagnosed with marginal zone lymphoma based on bone marrow biopsy.
False positive results were found in one fifth of our patients. One patient was diagnosed with lymphoma, eleven were found to have infections, and 2 were diagnosed with inflammatory conditions. In the remaining 44%, no diagnosis was made at the end of the work-up or at 3 month follow-up. Non-specific findings and physiologic uptake were not considered positive results. Our false positive rates match those reported by other studies which range between 9% and 30% [3,7,22,23].
Similar to other studies, infectious diseases were the most common cause for FUO in close to half of patients, followed by inflammatory diseases which were identified in 16.4% [1,2,7,24]. PET/CT was helpful in diagnosing just over 60% of the infectious cases while it guided clinicians toward the correct diagnosis in exactly 60% of those with inflammatory diseases. Malignant causes of FUO mostly included hematologic malignancies which made up almost three quarters of these cases. For more than 90% of patients with malignant causes of FUO, the PET/CT test provided true positive results. PET/CT appears to have missed the diagnosis of cancer in only one patient, whose scan showed increase uptake in the lungs and was ultimately diagnosed with T-cell lymphoma through other diagnostic modalities. Accordingly, PET/CT demonstrated highest efficiency in diagnosing malignant causes of FUO.
In our study, weight loss and low hemoglobin levels were independently associated with PET/CT producing true positive results in yielding the cause of FUO. Likewise, Crouzet et al. [5] illustrated an association between low hemoglobin levels and true positive PET/CT scans. In addition, other studies have shown similar associations between true positives and the presence of adenopathy and elevated levels of c-reactive protein (CRP) [5,6,8]. We, however, did not find a significant association between true positive results and biomarkers of inflammation such as leukocytosis and elevated CRP. Finally, in the study published by Gafter-Gvili [7], in which PET/CT was defined as helpful if it led either to true positives or true negatives, both fever and male gender were significant predictors of PET/CT usefulness.
A key limitation of this study is its retrospective design. As a result, the diagnosis and workup of FUO were not standardized before proceeding to PET/CT. Nevertheless, our study produced reliable data on the usefulness of PET/CT in the work-up of FUO. We collected a large group of patients while excluding all those that did not meet acceptable criteria. We adhered to strict methodology in how we defined useful PET/CT scans and discounted all PET/CT findings which were not diagnostic. With our standardized protocol, we identified in which patients PET/CT may be most valuable in assisting in the diagnosis of FUO.