If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of General Internal Medicine, University Hospitals Leuven, Leuven, BelgiumDepartment of Microbiology, Immunology, and Transplantation, KU Leuven, Leuven, Belgium
Department of General Internal Medicine, University Hospitals Leuven, Leuven, BelgiumDepartment of Microbiology, Immunology, and Transplantation, KU Leuven, Leuven, Belgium
Department of General Internal Medicine, University Hospitals Leuven, Leuven, BelgiumDepartment of Microbiology, Immunology, and Transplantation, KU Leuven, Leuven, BelgiumEuropean Reference Network for Immunodeficiency, Autoinflammatory, Autoimmune and Pediatric Rheumatic disease (ERN-RITA), Utrecht, the Netherlands
Department of General Internal Medicine, University Hospitals Leuven, Leuven, BelgiumDepartment of Microbiology, Immunology, and Transplantation, KU Leuven, Leuven, BelgiumEuropean Reference Network for Immunodeficiency, Autoinflammatory, Autoimmune and Pediatric Rheumatic disease (ERN-RITA), Utrecht, the Netherlands
18F-FDG PET had a sensitivity of 93%, a specificity of 35%, and made a positive contribution to the diagnosis in 25%.
•
IUO had a higher likelihood of contributive 18F-FDG PET imaging compared to FUO.
•
Among those with contributive 18F-FDG PET imaging, giant cell arteritis and polymyalgia rheumatica were more frequent in the IUO group.
•
The differences in the yield of 18F-FDG PET between IUO and FUO are most likely due to a distinctive diagnostic spectrum.
Abstract
Objective
18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET) is an important imaging technique in the workup of fever of unknown origin (FUO) and inflammation of unknown origin (IUO). Studies comparing the diagnostic yield of 18F-FDG PET between both entities are lacking.
Methods
Retrospective analysis of FUO/IUO patients who underwent 18F-FDG PET between 2000 and 2019 in the University Hospitals of Leuven (Belgium). 18F-FDG PET images were assessed for accuracy and contribution towards the final diagnosis. Logistic regression was performed to evaluate the association between meeting FUO or IUO criteria and diagnostic contribution of 18F-FDG PET with and without adjustment for confounders.
Results
Out of 604 patients, 439 (73%, mean age 56 years, 43% female) underwent 18F-FDG PET imaging, including 349 (79%) classified as FUO and 90 (21%) as IUO. Noninfectious inflammatory disorders were significantly more frequent in the IUO group (37% versus 25%; P = 0.03). 18F-FDG PET imaging had a sensitivity of 93% (89–96%), a specificity of 35% (29–42%), and made a positive contribution to the final diagnosis in 25% (21–29%) of cases. IUO was significantly associated with contributive 18F-FDG PET imaging compared to FUO (aOR 2.21 [95% CI 1.31–3.72]; P = 0.003). Among those with contributive 18F-FDG PET imaging, giant cell arteritis (IUO 25% versus FUO 12%) and polymyalgia rheumatica (IUO 17% versus FUO 1%) were numerically more frequent in the IUO group.
Conclusion
The diagnostic contribution of 18F-FDG PET was higher among those with IUO, most likely due to differences in diagnostic spectrum.
1. Introduction
18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET) imaging often plays an important role in the workup of fever of unknown origin (FUO) and inflammation of unknown origin (IUO), especially in those with few or no diagnostic clues. FUO is defined as an illness duration of ≥3 weeks with fever ≥38.3 °C on ≥3 occasions and no diagnosis despite baseline investigations. IUO, which was first described in 2009, is defined similarly but instead of high-grade fevers these patients have ≥3 occasions of elevated inflammatory markers, for which there is currently no universally accepted cut-off value [
The value of 18(F)-FDG-PET/CT in identifying the cause of fever of unknown origin (FUO) and inflammation of unknown origin (IUO): data from a prospective study.
A recent meta-analysis showed that 18F-FDG PET had a sensitivity of 84 (95% CI 79–89)% and specificity 63 (95% CI 49–75)% in a case mix of FUO and IUO [
]. It is important to acknowledge that frequently reported measures of diagnostic accuracy, including sensitivity and specificity, are less appropriate in FUO and IUO as a diagnostic gold standard is lacking. Furthermore, a diagnostic test may reveal abnormalities consistent with a certain diagnosis without contributing to the final diagnosis. For example, FDG uptake at the lymph nodes, spleen and bone marrow are consistent with adult-onset Still's disease, but these abnormalities do not directly assist in establishing a diagnosis [
]. For this reason, previous studies have used different outcomes that better reflect the value of 18F-FDG PET in clinical practice among patients with FUO and IUO, including diagnostic yield (rate of true positive PET scans [
]), direct diagnostic contribution and indirect diagnostic orientation (rate of PET scans indicating a final diagnosis directly or indirectly by guiding additional tests [
Diagnostic value of 18F-FDG PET/CT vs. Chest-Abdomen-Pelvis CT scan in management of patients with fever of unknown origin, inflammation of unknown origin or episodic fever of unknown origin: a comparative multicentre prospective study.
]), and diagnostic necessity (rate of PET scans which were required during the diagnostic workup to establish a diagnosis or negative PET scans for patients who had resolution of FUO/IUO) [
The value of 18(F)-FDG-PET/CT in identifying the cause of fever of unknown origin (FUO) and inflammation of unknown origin (IUO): data from a prospective study.
The value of 18(F)-FDG-PET/CT in identifying the cause of fever of unknown origin (FUO) and inflammation of unknown origin (IUO): data from a prospective study.
] reported that absence of fever was associated with a higher likelihood of a diagnostic 18F-FDG PET/CT, but the reasons for this finding were unclear. In addition, a recent meta-analysis showed that IUO was associated with a higher prevalence of noninfectious inflammatory disorders compared to FUO [
]. Differences in diagnostic outcomes could influence the value of 18F-FDG PET. In this study, we aimed to explore differences in diagnostic contribution of 18F-FDG PET between FUO and IUO patients.
2. Methods
2.1 Study setting and patient selection
Patients evaluated for prolonged fever or inflammation between January 2000 and December 2019 by the general internal medicine department of the University Hospitals Leuven (Belgium), a 1949-bed tertiary care center, were prospectively listed in a registry. We retrospectively identified adult (≥18 years) patients who met the criteria for FUO or IUO and underwent 18F-FDG PET imaging during workup in our center (when several 18F-FDG PET scans were performed, the first scan performed during a febrile or inflammatory episode was used) [
]. In accordance with the 2000 FUO definition, patients with known human immunodeficiency virus (HIV) infection, nosocomial fever, or an immunocompromised state (neutropenia with granulocyte count <0.5 × 103/mcL, hypogammaglobulinemia with IgG<7.5 g/L, or treatment with an equivalent of more than prednisolone 10 mg or other immunosuppressive agents for at least two weeks prior to the onset of FUO/IUO) were excluded [
]. We also excluded undiagnosed patients with less than 6 months of follow-up and patients with habitual hyperthermia. Several FUO studies continue to include the latter as a miscellaneous cause, but it should be considered a paraphysiological phenomenon (e.g., relating to an established set of principles governing a physiologic state) rather than a disease [
FUO was defined as an illness duration of more than 3 weeks with fever ≥38.3 °C on ≥3 occasions and no diagnosis despite a thorough center-specific baseline investigation (Supplementary Table S1). IUO was defined similarly but with CRP levels of ≥30 mg/L on ≥3 occasions and without high-grade fevers (≥38.3 °C on ≥3 occasions). The cut-off was based on previous research [
], and was set at a higher value compared to more recent studies to exclude common conditions that may be associated with chronic and minor elevations of inflammatory markers such as osteoarthritis and obesity. The symptom duration was defined as the time between the reported onset of symptoms and first presentation in our center after fulfillment of the FUO or IUO criteria. Recurrent FUO or IUO was defined according to the definition by Knockaert et al. as at least 2 episodes of fever (FUO) or inflammation (IUO) with a fever or inflammation-free interval of at least 2 weeks not related to FUO/IUO treatment [
]. Continuous fever or inflammation was defined as the absence of a recurrent pattern. The final diagnosis was assessed by two investigators (AB and SV) based on electronic health record review and included in the final analysis only if it was considered definite (confirmed by serology, histology, or culture) or probable (based on classification criteria, exclusion of other causes, response to therapy, or disease course).
2.3 18F-FDG PET acquisition
Patients were standardly fasted for at least 6 h before intravenous injection of 18F-FDG, and glycemia levels were determined in all patients (< 140 mg/dl). A whole-body 18F-FDG PET scan was performed 45–60 min after tracer administration. 18F-FDG PET scans were performed between 2000 and 2019, consecutively acquired on an Emission Computer-Assisted Tomography, high-resolution (ECAT HR +) 18F-FDG PET camera, Hirez Biograph 16 18F-FDG PET/CT or Truepoint Biograph 40 18F-FDG PET/CT (Siemens, Knoxville, TN, USA). Because of scan duration, HR+ data were not corrected for attenuation using transmission scanning. On the 18F-FDG PET/CT systems, either a low-dose non-diagnostic CT scan or a diagnostic CT scan with intravenous contrast was performed immediately before 18F-FDG PET acquisition. Non-attenuation corrected (non-AC) 18F-FDG PET images were available for all included patients and attenuation-corrected (AC) 18F-FDG PET images were only available for the patients who were scanned with a 18F-FDG PET/CT system (n = 187/439). 18F-FDG PET data were corrected for scatter and randoms. Data were reconstructed using iterative ordered subset expectation maximization reconstruction, with parameters optimized over the years (FWHM HR+ 6–7 mm, for the Truepoint 5 mm).
2.4 18F-FDG PET assessment
18F-FDG PET images were re-evaluated by two authors (LB and AB). Next, two authors (LM and AB) determined if a 18F-FDG PET scan was true positive, true negative, false negative, or false positive. 18F-FDG PET scans showing pathological FDG uptake in at least one region were defined as 18F-FDG PET positive. Studies showing only physiological biodistribution of FDG activity were considered negative. When 18F-FDG PET imaging showed increased uptake which was confirmed to be associated with the cause of FUO, the study was defined as true positive. When 18F-FDG PET showed no abnormalities or only a nonspecific and isolated FDG uptake of the bone marrow, it was considered true negative when no localized disease process was further diagnosed, the patient did not die in the first year after the index contact, and there was no evidence of disease after a 6-month follow-up. A normally interpreted 18F-FDG PET scan was considered false negative when a localized disorder was diagnosed during follow up or the patient remained undiagnosed and died during the first year. Abnormal FDG uptake in patients without a final diagnosis or in those who only had 18F-FDG PET abnormalities not considered related to the final diagnosis were false positive.
Two authors (AB and LM) assessed if 18F-FDG PET imaging made a positive contribution towards establishing the final diagnosis by assessing for each case whether the diagnosis was directly or indirectly reached based on 18F-FDG PET imaging. For example, increased FDG uptake in a lymph node leading to a diagnostic biopsy or FDG uptake in the aorta and its branches which resulted in a diagnosis of large vessel vasculitis were considered contributive. FDG uptake in lymph nodes in a patient with adult-onset Still's disease was not considered contributive, as adult-onset Still's disease is a clinical rather than a radiographic diagnosis.
To assess inter-reader reliability of the assessment of 18F-FDG PET abnormalities, 18F-FDG PET accuracy, and 18F-FDG PET contribution, the first 50 patients were evaluated by both authors. The interrater agreement was assessed by intra-class correlation coefficients (ICC) which is used to measure the reliability of ratings in studies where there are two or more raters, with a two-way mixed effects model, where the rater effect is fixed and the subject effect is random. If the interrater agreement was sufficiently high in the samples, all other 18F-FDG PET images were randomized and scored by one of both specialists.
2.5 Statistical analysis
Categorical and continuous variables were expressed as count (percentage) and as mean ± standard deviation or median ± interquartile range (IQR) as appropriate. Binomial confidence intervals were estimated by the Clopper-Pearson method. Missing data were assumed to be missing at random and imputed with multiple imputation (20 imputations, 2.1% total missingness, imputation by classification and regression trees). Comparisons of baseline characteristics were performed by Mann Whitney U tests or Chi square tests as appropriate. The sensitivity, specificity, and contribution of 18F-FDG PET were calculated for the entire population, FUO patients, and IUO patients, with the final diagnosis at last follow up as a gold standard. Logistic regression models were fitted to analyze the association of meeting either the FUO or IUO criteria and outcomes of interest with adjustment for presence of a recurrent disease pattern, recent antibiotics or immunosuppressive treatment, presence of fever or inflammation at the time of 18F-FDG PET, and the interaction term of prior CT imaging and combined PET/CT imaging (because combined PET/CT imaging was considered particularly valuable in those without previous CT imaging). The primary outcome was the association between meeting FUO or IUO criteria and contribution of 18F-FDG PET to the final diagnosis in the entire cohort. A sensitivity analysis was performed for the primary outcome including only patients with combined PET/CT imaging. The secondary outcome included the association between contributive 18F-FDG PET imaging and the diagnostic categories. All statistical tests were performed using 2-tailed tests with significance set at the alpha <0.05 level. Statistical analysis was performed in R Studio (version 2021.09.1, The R Foundation for Statistical Computing) with inclusion of the mice, ggplot2, and psych packages. This study was approved by the Ethical Committee of the University Hospitals Leuven. Informed consent was waived because of the retrospective nature of the study and the analysis of pseudonymized clinical data.
3. Results
3.1 Patients
Out of 604 patients who fulfilled the classification criteria, 439 (73%, mean age 56 years, 43% female) underwent 18F-FDG PET imaging during the FUO/IUO workup (Table 1). Of these, 349 (79%) were classified as FUO and 90 (21%) as IUO. Patients with IUO were older (mean age 61 [±16] versus 55 [±18] years; P = 0.003) and tended to have a lower Charlson comorbidity index (2 [0–3] versus 2 [1–4]; P = 0.06) but had a similar rate of female patients and prior history of noninfectious inflammatory disorders or malignancies. Patients with IUO had a longer symptom duration (median duration 60 [29–165] days versus 42 [24–105] days; P = 0.02) and appeared to have less severe illness as illustrated by the lower number of emergency room admissions/hospital transfers (66% versus 81%; P = 0.002) and intensive care/medium care admissions (3% versus 11%; P = 0.04). There were no differences in referral pattern.
Table 1Comparison of patient, FUO/IUO, and 18F-FDG PET characteristics and diagnostic outcomes between the FUO and IUO population.
Total N = 439
FUO N = 349
IUO N = 90
P value
Patient characteristics
•
Age, mean (±SD)
56 (±17)
55 (±18)
61 (±16)
0.003
•
Female, n (%)
188 (43)
145 (42)
43 (48)
0.34
•
CCI, median (IQR)
2 (0–4)
2 (0–3)
2 (1–4)
0.06
•
History malignancy, n (%)
41 (9)
33 (9)
8 (9)
1.00
•
History NIID, n (%)
54 (12)
46 (13)
8 (9)
0.35
FUO/IUO characteristics
•
Symptom duration, median (IQR)
46 (25–120)
42 (24–105)
60 (29–165)
0.02
•
Recurrent pattern, n (%)
109 (25)
94 (27)
15 (17)
0.06
•
Temperature ≥38 °C, n (%)
385 (88)
349 (100)
36 (40)
<0.001
•
Referral, n (%)
○
External
○
Internal
250 (57) 201 (80) 49 (20)
204 (58) 165 (81) 39 (19)
46 (51) 36 (78) 10 (22)
0.26 0.84 -
•
ER or hospital transfer, n (%)
342 (78)
283 (81)
59 (66)
0.002
•
ICU or MCU, n (%)
42 (10)
39 (11)
3 (3)
0.04
•
Follow up, months, median (IQR)
19 (7–54)
16 (6–49)
22 (11–64)
0.01
18F-FDG PET characteristics
•
Prior CT imaging
317 (72)
262 (75)
55 (61)
0.01
•
Time to 18F-FDG PET, days, median (IQR)
7 (3–16)
7 (3–14)
12 (4–29)
0.02
•
18F-FDG PET combined with CT, n (%)
○
IV contrast
187 (43) 128 (68)
140 (40) 100 (71)
47 (52) 28 (60)
0.05 0.18
•
CRP at time of 18F-FDG PET(/CT), median (IQR)
73 (29–142)
77 (35–143)
66 (24–131)
0.21
•
Fever or inflammation present at the time of PET imaging, n (%)
421 (96)
338 (97)
83 (92)
0.09
•
Recent CS/IS, n (%)
45 (10)
35 (10)
10 (11)
0.89
•
Recent antibiotics, n (%)
112 (26)
102 (29)
10 (11)
<0.001
Diagnostic categories
•
Infection, n (%)
51 (12)
42 (12)
9 (10)
0.72
•
Malignancy, n (%)
66 (15)
53 (15)
13 (14)
0.99
•
NIID, n (%)
119 (27)
86 (25)
33 (37)
0.03
•
Miscellaneous, n (%)
46 (10)
36 (10)
10 (11)
0.98
•
No diagnosis, n (%)
157 (36)
132 (38)
25 (28)
0.10
Abbreviations: CCI, Charlson comorbidity index; CS, corticosteroids; CT, computed tomography; ER, emergency room; FUO, fever of unknown origin; ICU, intensive care unit; IQR, interquartile range; IS, immunosuppressive treatment; IUO, inflammation of unknown origin; IV, intravenous; MCU, medium care unit; NIID, noninfectious inflammatory disorders; 18F-FDG PET, positron emission tomography.
The interrater agreement for the assessment of 18F-FDG PET abnormalities (ICC 0.845), 18F-FDG PET accuracy (ICC 0.831), and 18F-FDG PET contribution (ICC 0.834) was high. In general, 72% of patients had CT imaging of the chest or abdomen before the 18F-FDG PET scan. Prior CT imaging was even more frequent in those without combined PET/CT imaging (78%). IUO patients less frequently had prior CT imaging of the chest and/or abdominal cavity (61% versus 75%; P = 0.01) and had a longer time to 18F-FDG PET imaging (median duration 12 [4–29] versus 7 [3–14] days; P = 0.02). IUO patients tended to have more combined 18F-FDG PET/CT scans (52% versus 40%; P = 0.05). FUO patients tended to have either fever or elevated inflammatory markers more frequently at the time of the 18F-FDG PET scan (92% versus 97%; P = 0.09), but there was no difference in median CRP values. There was no difference in treatment with corticosteroids or other immunosuppressive drugs in the 2 weeks prior to 18F-FDG PET imaging (11% versus 10%; P = 0.89), but FUO patients received more antibiotics within this time frame (11% versus 29%; P<0.001).
3.2 Diagnosis
Out of 282 (64%) patients with a final diagnosis, 119 (27%) had noninfectious inflammatory disorders, 66 (15%) malignancy, 51 (12%) infection, and 46 (10%) miscellaneous conditions. Noninfectious inflammatory disorders were significantly more frequent in the IUO group (37% versus 25%; P = 0.03), and undiagnosed patients tended to be more frequent in the FUO group (28% versus 38%; P = 0.10). Infections, malignancies, and miscellaneous conditions were distributed evenly across both groups. A comprehensive overview of all final diagnoses is provided in Supplementary Table S2. The most frequently diagnosed entities included tuberculosis (n = 10), Whipple's disease (n = 6), and Q-fever (n = 5) for the infectious disorders, non-Hodgkin lymphoma (n = 29), Hodgkin lymphoma (n = 12), and myelodysplastic neoplasms (n = 6) for the malignancies, giant cell arteritis (n = 26), adult-onset Still's disease (n = 20), and ANCA associated vasculitis (n = 14) for the noninfectious inflammatory disorders, and drug fever (n = 12), venous thromboembolism (n = 6), and calcium pyrophosphate deposition (CPPD) (n = 3) for the miscellaneous disorders. No diagnosis was made in 157 (36%) of patients, of which 131 (83%) had resolution of fever and 26 (17%) persistent illness after a median follow-up of 12 (IQR 7–41) months.
3.3 Diagnostic yield 18F-FDG PET imaging
True positive 18F-FDG PET imaging was observed in 215 (49%), true negative in 74 (17%), false positive in 135 (31%), and false negative in 15 (3%) patients. Fig. 1 summarizes the sensitivity, specificity, and contribution in the entire cohort, FUO patients, and IUO patients. Overall, 18F-FDG PET imaging had a sensitivity of 93% (95% CI 89–96%), a specificity of 35% (95% CI 29–42%) and was contributive towards the final diagnosis in 25% (95% CI 21–29%) of cases. IUO patients had a numerically higher sensitivity (98% [95% CI 90–100%] versus 92% [95% CI 87–96%]) and lower specificity (24% [95% CI 12–41%] versus 38% [95% CI 31–45%]). The contribution of 18F-FDG PET to the final diagnosis was also numerically higher in the IUO group (40% [95% CI 30–51%] versus 21% [95% CI 17–26%]). Adjusted logistic regression analysis (Table 2-A) showed that patients with IUO were significantly more likely to have a contributive 18F-FDG PET scan compared to those with FUO (aOR 2.21 [95% CI 1.31–3.72]; P = 0.003). In this adjusted model, the combination of absent prior CT imaging together with combined 18F-FDG PET/CT imaging (aOR 2.13 [95% CI 1.13–3.99]; P = 0.02) was also significantly associated with contribution of 18F-FDG PET. A recurrent pattern was negatively associated with diagnostic contribution (aOR 0.50 [95% CI 0.27–0.88]; P = 0.02). A sensitivity analysis including only those with combined 18F-FDG PET/CT imaging (Table 2-B) did not show a significant difference in contribution between IUO and FUO (aOR 1.22 [95% CI 0.55–2.61]; P = 0.61).
Fig. 1Diagnostic yield of 18F-FDG PET in the entire cohort, FUO population, and IUO population. Abbreviations: FUO, fever of unknown origin; IUO, inflammation of unknown origin.
Table 2Logistic regression model for the association between meeting the FUO or IUO criteria and diagnostic contribution of 18F-FDG PET imaging with and without adjustment.
Logistic regression model adjusted for all other factors included in the crude analysis. Abbreviations: 95% CI, 95% confidence interval; CT, computed tomography; FUO, fever of unknown origin; IUO, inflammation of unknown origin; OR, odds ratio; PET,.
(95% CI)
P
IUO (versus FUO)
2.52 (1.53–4.13)
<0.001
2.21 (1.31–3.72)
0.003
Recurrent pattern
0.48 (0.26–0.83)
0.01
0.50 (0.27–0.88)
0.02
Recent AB or CS/IS
0.94 (0.56–1.54)
0.81
1.07 (0.62–1.81)
0.81
Presence of fever/inflammation
1.16 (0.41–4.17)
0.79
1.07 (0.35–4.02)
0.91
Prior CT (+) Combined PET/CT (-)
Ref
Ref
Ref
Ref
Prior CT (+) Combined PET/CT (+)
1.14 (0.65–1.97)
0.64
1.08 (0.61–1.89)
0.80
Prior CT (-) Combined PET/CT (-)
1.76 (0.89–3.40)
0.10
1.71 (0.84–3.40)
0.13
Prior CT (-) Combined PET/CT (+)
2.34 (1.27–4.27)
0.006
2.13 (1.13–3.99)
0.02
B. Cohort with combined PET/CT imaging (n = 187)
Crude OR (95% CI)
P
Adjusted OR° (95% CI)
P
IUO (versus FUO)
1.49 (0.72–3.02)
0.27
1.22 (0.55–2.61)
0.61
Recurrent pattern
0.28 (0.10–0.66)
0.007
0.29 (0.10–0.69)
0.01
Recent AB or CS/IS
1.14 (0.54–2.33)
0.73
1.08 (0.49–2.33)
0.85
Presence of fever/inflammation
1.56 (0.22–30.85)
0.70
1.35 (0.17–27.98)
0.80
Prior CT imaging
0.49 (0.25–0.94)
0.03
0.49 (0.25–0.97)
0.04
° Logistic regression model adjusted for all other factors included in the crude analysis.Abbreviations: 95% CI, 95% confidence interval; CT, computed tomography; FUO, fever of unknown origin; IUO, inflammation of unknown origin; OR, odds ratio; PET,.
Contributive 18F-FDG PET imaging was significantly associated with a final diagnosis of malignancy (aOR 5.05 [95% CI 2.84–9.09]; P<0.001), infectious disorders (aOR 2.17 [95% CI 1.14–4.03]; P = 0.02), noninfectious inflammatory disorders (aOR 1.67 [95% CI 1.03–2.68]; P = 0.04) and tended to be associated with miscellaneous disorders (aOR 1.71 [95% CI 0.84–3.36]; P = 0.13). Among those with contributive 18F-FDG PET imaging, the leading causes included giant cell arteritis (16%), non-Hodgkin lymphoma (13%), Hodgkin lymphoma (7%), polymyalgia rheumatica (6%), and tuberculosis (6%) (Table 3). Giant cell arteritis (IUO 25% versus FUO 12%) and polymyalgia rheumatica (IUO 17% versus FUO 1%) were numerically more frequent in the IUO group, while the opposite was true for non-Hodgkin lymphoma (6% versus 16%). Out of 109 patients in whom 18F-FDG PET was contributive towards the final diagnosis, it directly indicated the diagnosis in 25 (23%) and indirectly in 84 (77%) patients. An indirect diagnosis was made by tissue biopsy (n = 66; 61%), invasive imaging with/without biopsy (n = 6; 5%), serological testing (n = 6; 5%), joint aspiration (n = 5; 5%), or genetic testing (n = 1; 1%). Among those with false positive 18F-FDG PET imaging, the most frequently observed abnormalities included localized lymph nodes (29%), pulmonary abnormalities (23%), musculoskeletal abnormalities (21%), splenic abnormalities (16%), generalized lymph nodes (14%), large bowel abnormalities (11%), and thyroid or ear-nose-throat abnormalities (8%).
Table 3Overview of the final diagnoses among those with contributive 18F-FDG PET imaging.
Disorder
Total n = 109
FUO n = 73
IUO n = 36
Infection
20 (18)
14 (19)
6 (17)
Aortitis (S. anginosus)
1 (1)
1 (1)
0 (0)
Bacterial translocation subobstruction
1 (1)
1 (1)
0 (0)
Bartonella
1 (1)
1 (1)
0 (0)
Bronchiectasis
1 (1)
1 (1)
0 (0)
Leishmaniasis
1 (1)
1 (1)
0 (0)
Malacoplakia
1 (1)
0 (0)
1 (3)
Osteomyelitis
1 (1)
0 (0)
1 (3)
(Prosthetic) joint infection (S. anginosus – MSSA)
2 (2)
1 (1)
1 (3)
Septic thrombophlebitis
1 (1)
1 (1)
0 (0)
Spondylodiscitis
1 (1)
0 (0)
1 (3)
Tuberculosis
6 (6)
4 (5)
2 (6)
Whipple's disease
3 (3)
3 (4)
0 (0)
Noninfectious inflammatory disorders
40 (37)
23 (32)
17 (47)
ANCA vasculitis
2 (2)
1 (1)
1 (3)
Dermatomyositis
1 (1)
1 (1)
0 (0)
Familial Mediterranean fever
1 (1)
1 (1)
0 (0)
Giant cell arteritis
18 (16)
9 (12)
9 (25)
Inflammatory bowel disease
1 (1)
1 (1)
0 (0)
Kikuchi disease
4 (4)
4 (5)
0 (0)
Polyarteritis nodosa
1 (1)
1 (1)
0 (0)
Polymyalgia rheumatica
7 (6)
1 (1)
6 (17)
Reactive arthritis
1 (1)
1 (1)
0 (0)
Sarcoidosis
3 (3)
3 (4)
0 (0)
Takayasu arteritis
1 (1)
0 (0)
1 (3)
Malignancy
34 (31)
26 (36)
8 (22)
Colorectal carcinoma
1 (1)
1 (1)
0 (0)
Cancer of unknown primary
1 (1)
1 (1)
0 (0)
Large cell carcinoma
1 (1)
1 (1)
0 (0)
Hodgkin lymphoma
8 (7)
5 (7)
3 (8)
Lung carcinoma
2 (2)
1 (1)
1 (3)
Lymphoplasmocytic lymphoma
1 (1)
0 (0)
1 (3)
Mast cell leukemia
1 (1)
1 (1)
0 (0)
Mesothelioma
1 (1)
1 (1)
0 (0)
Non-Hodgkin lymphoma
14 (13)
12 (16)
2 (6)
Prostate carcinoma
3 (3)
2 (3)
1 (3)
Renal cell carcinoma
1 (1)
1 (1)
0 (0)
Miscellaneous
15 (14)
10 (14)
5 (14)
Addison's disease
1 (1)
1 (1)
0 (0)
Cryptogenic organizing pneumonia
2 (2)
2 (3)
0 (0)
Calcium pyrophosphate deposition disease
3 (3)
1 (1)
2 (6)
Erdheim-Chester disease
2 (2)
1 (1)
1 (3)
Ganglioneuroma
1 (1)
1 (1)
0 (0)
Multicentric Castleman's disease
1 (1)
1 (1)
0 (0)
Pulmonary nodular lymphoid hyperplasia
1 (1)
1 (1)
0 (0)
Primary/secondary sclerosing cholangitis
1 (1)
0 (0)
1 (3)
Thyroiditis
3 (3)
2 (3)
1 (3)
Abbreviations: MSSA, methicillin-susceptible Staphylococcus aureus; S. anginosus, Streptococcus anginosus.
In this large cohort of FUO/IUO patients with 18F-FDG PET imaging, there were 2 key findings. First, patients with IUO had significantly higher odds of a contributive 18F-FDG PET scan compared to those with FUO. Second, IUO was caused significantly more often by noninfectious inflammatory disorders, and giant cell arteritis and polymyalgia rheumatica were more frequent among IUO patients with contributive 18F-FDG PET imaging.
Our study confirms the substantial diagnostic value of 18F-FDG PET in a large cohort of FUO and IUO patients. When compared to the meta-analysis by Kan et al., we had a somewhat higher sensitivity (93% versus 84%) but a lower specificity (35% versus 63%). Our findings are consistent with the 2017 prospective study by Schonau et al., which also reported a high number of false positive results and a lower specificity of 21.7% among 240 FUO and IUO patients [
The value of 18(F)-FDG-PET/CT in identifying the cause of fever of unknown origin (FUO) and inflammation of unknown origin (IUO): data from a prospective study.
]. This may be due to a strict definition of false positive 18F-FDG PET imaging, which also illustrates the limitations of these common measures of diagnostic accuracy when a gold standard is lacking. The authors also found that several clinical features were associated with a higher diagnostic yield, amongst which the absence of fever [
The value of 18(F)-FDG-PET/CT in identifying the cause of fever of unknown origin (FUO) and inflammation of unknown origin (IUO): data from a prospective study.
]. While this may seem surprising at first, we assumed this could be related to distinctions between FUO and IUO patients. In this study, we indeed found that IUO was associated with a significantly higher likelihood of contributive 18F-FDG PET imaging (aOR 2.21 [95% CI 1.31–3.72]). The sensitivity analysis which included only patients with combined PET/CT imaging did not show a significant difference between the IUO group and the FUO group (aOR 1.22 [95% CI 0.55–2.61]) but this analysis was underpowered. We cannot exclude that these findings may differ in a FUO and IUO population being investigated only with combined 18F-FDG PET/CT, and additional research should assess if combined 18F-FDG PET/CT imaging, which currently is the standard, also has a higher diagnostic contribution among those with IUO.
While only 42% of patients had combined PET/CT imaging, over 80% of those without a combined PET/CT had prior CT imaging. This was related to the 20-year period in which the 18F-FDG PET scans were performed, and which was associated with relevant technical evolutions in terms of device quality and the implementation of combined 18F-FDG PET/CT with or with intravenous contrast. It is important to mention that IUO patients had significantly less prior CT imaging but also more combined 18F-FDG PET/CT imaging. The combination of no prior CT imaging together with combined 18F-FDG PET/CT imaging was also significantly associated with a higher contribution (aOR 2.13 [95% CI 1.13–3.99]), which is not surprising considering CT also has a substantial diagnostic yield in FUO [
]. However, the association between IUO and the diagnostic contribution of 18F-FDG PET remained largely unchanged after adjustment for the interaction of prior CT imaging and combined PET/CT imaging (Table 2), which suggests this difference is not related to CT imaging. In addition, a recurrent pattern was associated with a lower diagnostic contribution (aOR 0.29 [95% CI 0.10–0.69]), which may be explained by differences in diagnostic outcomes [
], but also by the fact that it may be more difficult to perform imaging during disease activity.
While it is interesting that 18F-FDG PET appears to be more frequently contributing in IUO, it is mainly relevant because it appears to indicate distinctions in the underlying disease spectrum. We found that noninfectious inflammatory disorders were significantly more often observed in IUO patients (37% versus 25%), which contrasts with a small monocentric study of 114 patients by Vanderschueren et al. which reported no differences in diagnostic categories between FUO and IUO patients [
]. In our study, comparison of specific disorders in both groups among those with contributive 18F-FDG PET imaging indicated that IUO was caused numerically more often by giant cell arteritis and polymyalgia rheumatica than FUO, but significance testing was not performed due to the small number of patients affected by specific disorders. 18F-FDG PET imaging is indeed an excellent tool to identify large vessel vasculitis and polymyalgia rheumatica, in particular among those with nonspecific clinical features [
]. These findings are also in line with the results reported by Schönau et al., as in their study IUO was also caused more often by large vessel vasculitis (21% versus 7%) and polymyalgia rheumatica (18% versus 6%) compared to FUO [
The value of 18(F)-FDG-PET/CT in identifying the cause of fever of unknown origin (FUO) and inflammation of unknown origin (IUO): data from a prospective study.
]. These findings suggest that IUO and FUO may differ significantly in their diagnostic spectrum, which may be confirmed by future research.
Strengths of our study include the very large cohort of FUO/IUO patients with 18F-FDG PET imaging, the required 6 months follow-up in those without a diagnosis, and the reassessment and classification of 18F-FDG PET scans, which were each performed by two independent authors. Our study also has several limitations. First, the retrospective design may be associated with information and misclassification bias. Second, the case-mix of FUO/IUO varies according to geographical area, which limits the generalizability of this single-center study to other populations and non-Western countries. In addition, it is important to note that there is no universally accepted cut-off value for the definition of IUO. As a result, the findings of our study are specific to the cut-off value of 30 mg/L, and possibly associated with a decrease in sensitivity and increase in specificity of 18F-FDG PET. Future research may be required to confirm these findings when lower cut-off values are applied. Finally, our study included 18F-FDG PET imaging from a 20-year period, which was associated with relevant technical evolutions.
In conclusion, 18F-FDG PET has an excellent sensitivity for both FUO and IUO, but the diagnostic performance is somewhat compromised by a high number of false positive findings and a lower specificity. Regardless, 18F-FDG PET made a diagnostic contribution to the final diagnosis in 25% of patients. 18F-FDG PET was more frequently contributive to the final diagnosis among those with IUO, who appear to have an even higher yield of this imaging technique, most likely related to a distinctive spectrum of underlying disorders.
Conflict of interest disclosure statement
The authors have no conflicts of interest to declare.
The value of 18(F)-FDG-PET/CT in identifying the cause of fever of unknown origin (FUO) and inflammation of unknown origin (IUO): data from a prospective study.
Diagnostic value of 18F-FDG PET/CT vs. Chest-Abdomen-Pelvis CT scan in management of patients with fever of unknown origin, inflammation of unknown origin or episodic fever of unknown origin: a comparative multicentre prospective study.
00031-6&id=gr1.jpg){kind=link}