Fever of unknown origin (FUO) is a diagnostic challenge, despite recent advances in diagnostic modalities (1,2). FUO is defined as a prolonged febrile illness without a recognized cause, even after comprehensive laboratory tests and imagining workup. The current definition of FUO is as follows: i) Body temperature of ≥38.3˚C on at least two occasions; ii) duration of illness of ≥3 weeks; iii) not an immunocompromised setting; and iv) uncertain diagnosis despite thorough history taking, physical examination and laboratory assessments. These laboratory measures include the erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), hemoglobin, platelet count, leukocyte count and differentiation, electrolyte levels, creatinine level, total serum protein, protein electrophoresis, alkaline phosphatase, aspartate aminotransferase, lactate dehydrogenase (LDH), ferritin, microscopic urinalysis, blood cultures and urine cultures. The most common radiological investigation includes chest X-ray and abdominal ultrasonography. Additional tests may include the tuberculin test or an interferon-γ release assay (3,4). The cause of FUO may be multifactorial and can be subdivided into four categories: Infections, malignancies, noninfectious inflammatory processes and miscellaneous causes (5). FUO is closely related to inflammation of unknown origin (IUO), and the causes and workups are the same for both FUO and IUO (6). Early identification of the cause of FUO is important for guiding the diagnostic workup and for initiating early and appropriate treatment without notable impacts on patient care (7).

Conventional anatomic imaging modalities, such as ultrasonography, and cross-sectional imaging with CT and MRI, are used as primary diagnostic modalities to manage FUO; however, they have limited sensitivity and specificity (8). Therefore, a significant number of patients with FUO (30-50%) leave the hospital without specific diagnoses (9). Positron emission tomography (PET)/CT using ¹⁸F-fluorodeoxyglucose (¹⁸FDG) is a well-accepted clinical tool for routine use in a wide range of malignancies. ¹⁸FDG is a glucose analogue that accumulates in cells with high metabolic requirements, such as tumor and inflammatory cells. ¹⁸FDG PET/CT is currently considered a useful noninvasive imaging modality for the evaluation of patients with FUO (10-13). However, only a few studies have assessed ¹⁸FDG PET/CT in patients with FUO. Furthermore, most of these studies were performed in developed countries and the geographic area of the study strongly influenced the determination of the final diagnosis. Studies have shown that the cause of FUO in developing countries was most often infectious (14,15), whereas noninfectious inflammatory processes were the most common drivers of FUO in developed countries.

Several meta-analyses have reported the positive impact of FDG PET/CT in establishing the final diagnosis of patients with FUO/IUO. The diagnostic yield of PET CT/CT in this population is 56-60%, which is ≥30% higher than that of conventional CT (16,17). However, most studies are retrospective studies, with significant heterogeneity of baseline study characteristics, definition of FUO/IUO and imaging parameters (18,19). In certain studies, FDG PET/CT was the only imaging modality that was able to establish a final diagnosis (18,19). The vast majority of the studies investigated the diagnostic value of PET/CT, with only a few studies investigating the cost-effectiveness of FDG PET/CT in the diagnostic work-up of patients with FUO (20). Based on the previous studies, patients with FUO/IUO are a challenging, heterogeneous population with a wide variety of differential diagnoses. Furthermore, there is a lack of an established work-up strategy. According to published data, it is important to further investigate the diagnostic impact of PET/CT, the cost-effectiveness, and when and how to perform PET/CT in the setting of FUO/FUI (21).

The purpose of the present study was to evaluate the performance and benefit of ¹⁸FDG PET/CT in patients who presented with FUO at a tertiary academic general hospital in Riyadh, Saudi Arabia.

To the best of our knowledge, the present study was the first study in Saudi Arabia and the Middle East to investigate the diagnostic value of ¹⁸FDG PET/CT in identifying the underlying etiology of patients with disease with FUO. The present study investigated 105 patients with FUO and IUO, and revealed that ¹⁸FDG PET/CT helped identify the etiology of the underlying disease and establish a final diagnosis in 49% of patients. This group was classified as the true-positive group, yielding a sensitivity of 72% and a positive predictive value of 68%. PET/CT was negative in 10% of patients, in whom there was no final diagnosis and PET/CT was considered a true negative, yielding a low specificity and negative predictive value of 29 and 33%, respectively. The final diagnoses of FUO etiology in this cohort included focal or systemic infections (51%), malignancies (35%) and inflammatory noninfectious processes (14%). These results are comparable to those of published studies of FUO, in which the sensitivity varied between 26 and 75% (22-31). Determining the diagnostic value and accuracy of ¹⁸FDG PET/CT in the setting of FUO is complex and at times controversial. The sensitivity and specificity of ¹⁸FDG PET/CT are difficult to establish in FUO because of the numerous potential differential diagnoses. The most common malignancy identified in the present cohort was lymphoma and leukemia, and other few miscellaneous malignancy as outlined in Table SI. Gafter-Gvili et al (13) reported 14 cases of malignancy out of a total of 84 positive on FDG PET/CT (14%); of them, nine were non-Hodgkin's, two were Hodgkin's lymphoma, two were lung cancer and one was a sarcoma. In the present population, the prevalence of infectious disease was 51%, which is relatively higher than that in previous studies (32-34). According to several studies on FUO, the disease-associated FUO differs between developed countries and other countries. The percentages of classification of FUO were 19, 24, 12, 8 and 38% for infection, non-infection inflammatory, cancer, various etiologies and unknown, respectively in the developed countries, in contrast to 43, 20, 14, 7 and 16%, respectively, in developing countries (35). The main difference in cases of FUO between developed countries and developing countries is the high incidence of infectious disease; for instance, infectious disease was the most common cause of FUO in Iran (14) and Pakistan, in particular tuberculosis (15). The usefulness of PET/CT in the diagnostic process of FUO is also debatable. It is unknown whether only the true-positive findings can guide the final diagnosis or whether true-negative PET/CT results may rule out infection, inflammation or malignancies (36).

In contrast to a previous study, in which age >50 years, CRP level >30 mg/l and absence of fever were predictive of true-positive ¹⁸FDG PET/CT results (37), the present study revealed no statistically significant association between PET/CT results and age, sex, WBC, CRP, ESR, LDH or ferritin levels among different diagnostic groups. This lack of association may be due to the wide variety of disease distribution and prevalence in FUO. In a previous study of more specific inflammatory diseases, such as rheumatoid arthritis, an association existed between CRP levels and ¹⁸FDG uptake (38). When searching for clinical and laboratory tests associated with true-positive PET/CT results, elevated WBC counts and elevated serum levels of ferritin were associated with infection (P=0.018 and P=0.034, respectively) in the present study. Conversely, a previous study revealed no significant association between WBC values and ¹⁸FDG PET/CT scan results, the clinical impact of ¹⁸FDG PET/CT or the provision of additional information. Other studies showed that anemia, lymphadenopathy and male sex correlated with diagnostic ¹⁸FGD PET/CT results (13,23).

A significant number of false-positive results were reported in the present patients (23%), which reduced the positive predictive value to 68% and the overall diagnostic value in the study to 58%. This observation was most likely related to the normal physiological distribution of ¹⁸FDG (in the kidney, urinary bladder, and small and large intestines). The discrimination between pathological and normal physiological uptake in such organs is difficult, which may explain the relatively high number of false-positives in the present study. In addition, ¹⁸FDG uptake in the bone marrow and lymph nodes may be remarkable. Some investigators have suggested reclassifying ¹⁸FDG uptake in the setting of FUO as a nonspecific sign of inflammation and considering the scan in such cases a true negative. This reclassification could greatly improve the positive predictive value and accuracy (39). Several issues may cause false-negative findings as well. For instance, small structures are not readily identified with PET/CT, as these small structures (such as temporal arteries) are beyond the resolution of PET/CT. The detection limit is ~10 mm. Furthermore, certain cancer types, such as prostate carcinoma and hepatocellular carcinoma, are not very FDG-avid (40).

Multiple previous studies have reported the value of ¹⁸FDG PET/CT. However, comparing these studies is difficult, since the majority of these studies (including the present study) were retrospective with different inclusion and exclusion criteria with a possibility of selection bias. In addition, patients who undergo normal conventional imaging are more likely to have PET/CT scans than patients with positive conventional imaging results. In addition, certain studies included immunocompromised patients and comparing these patients with non-immunocompromised patients is difficult (41). Furthermore, in most studies, the definition of FUO was unspecified (42,43). One meta-analysis determined that only a true-positive scan is helpful in diagnoses (44). In the present study, 51 patients (49%) had true-positive scans. However, this diagnostic approach has been questioned in a more recent meta-analysis, which suggested that true-negative scans must also be considered helpful, since patients with true-negative scans tend to have favorable prognoses and high spontaneous remission rates (45). Large, prospective, multicenter studies are needed to better investigate the role and value of ¹⁸FDG PET/CT in patients with FUO.

The present study has certain important strengths and limitations. The study was a retrospective, small, single-tertiary care center study, and thus, the possibility of referral bias cannot be excluded. There was no structured protocol for a work-up of FUO. Both the diagnostic workup before ¹⁸FDG PET/CT and the timing of the scan itself were performed according to the preference of the referring physicians. The significant variations among patients may have affected the disease prevalence and diagnostic accuracy of the test, which is one limitation of the study; the specificity was relatively low, only 29%, most likely due to heterogeneity of the population of the present study. The gold standard approach for patients with FUO/IUO is not well defined because of different patient populations among studies and the wide variety of etiologies. Finally, another limitation is that the cost of FDG PET/CT is relatively high and there is limited availability.

In conclusion, the present findings demonstrated that ¹⁸FDG PET/CT substantially contributed to the establishment of a final diagnosis of FUO in a high percentage of patients (72%). ¹⁸FDG PET/CT can detect metabolically active foci that may indicate infection, malignancies and inflammatory noninfectious processes, all of which may be the underlying etiology of the fever. PET/CT may be helpful by eliminating additional examinations that rule out focal processes; however, identifying focal ¹⁸FDG activity may indicate false-positive findings and initiate additional examinations. Based on the present results and previous studies, it appears that ¹⁸FDG PET/CT scans have a potential role in the initial diagnostic workup for investigation of the etiology of FUO. However, larger, prospective, multicenter studies with more specific referral criteria are warranted to better investigate the role and cost-effectiveness of ¹⁸FDG PET/CT scans in the management of FUO.