Abstract
The purpose of this study was to evaluate the clinical significance of 18F-FDG uptake by the spleen in patients with autoimmune disease. Methods: We retrospectively reviewed Severance Hospital’s electronic medical records of patients hospitalized for the evaluation of fever who underwent 18F-FDG PET/CT. We found 91 patients with autoimmune diseases and 101 patients with localized infection. 18F-FDG uptake was assessed by measuring SUV in the spleen and liver. The spleen-to-liver ratio of the SUVmean (SLRmean) was calculated. Clinical and laboratory parameters were collected and evaluated for association with SLRmean. In-hospital mortality was defined as all-cause mortality during hospital admission for fever. Results: SLRmean was significantly higher in autoimmune disease than in localized infectious disease (1.28 ± 0.43 vs. 0.91 ± 0.21, P < 0.001). In autoimmune disease, SLRmean was correlated with monocytes, aspartate aminotransferase, alanine aminotransferase, albumin, and ferritin. Analysis of receiver-operating-characteristic curves revealed that in comparison with laboratory parameters, SLRmean had the highest performance in differentiating autoimmune from localized infectious disease. Multivariate logistic regression analysis demonstrated that high SLRmean and low platelets were significantly associated with in-hospital mortality in febrile autoimmune disease. Conclusion: These findings suggest that spleen glucose metabolism is increased in febrile autoimmune disease. Spleen 18F-FDG uptake may provide information useful in differentiating febrile autoimmune disease from localized infectious disease and predicting clinical outcomes in febrile autoimmune disease.
The radiotracer 18F-FDG, which is used in PET/CT, is an analog of glucose; its concentration reflects regional glucose uptake in tissue (1). Increased 18F-FDG uptake indicates high glycolytic activity and is associated with various conditions that have active cellular metabolism, such as cancer and localized infection (2). Inflammation is also accompanied by high glycolytic activity, and several studies have investigated the role of 18F-FDG PET/CT in the diagnosis of autoimmune disease, such as arthritis, myositis, and vasculitis (3,4). In autoimmune disease, specific patterns of 18F-FDG uptake are detected in multiple target organs, such as joints, muscles, and arteries, which can be helpful for diagnosis. The pattern of 18F-FDG distribution in multiple sites throughout the body is consistent with the systemic nature of autoimmune diseases.
Autoimmune diseases are accompanied by activation of systemic inflammation, whereas most infection is caused by local invasion of microorganisms. The general principles of treatment can be very different depending on the underlying condition; immune suppression can be therapeutic in autoimmune diseases, though potentially deleterious in infection. Despite the distinct pathogeneses and treatments, both diseases share common clinical features, including leukocytosis and elevated acute-phase reactants, such as erythrocyte sedimentation rate and C-reactive protein (5). Fever can be the main clinical feature of both autoimmune and infectious diseases, posing diagnostic challenge for distinguishing autoimmune diseases from infection.
The spleen filters blood, monitors blood-borne antigens, and is involved in innate and adaptive immune responses. Leukocytes in the spleen include various subsets of T and B cells, dendritic cells, and macrophages. Recent findings suggest that proliferating effector T cells require high metabolic flux through growth-promoting pathways (6). Energy metabolism of stimulated lymphocytes is shifted from oxidative metabolism to glycolysis to provide additional energy for activation and proliferation (7). We hypothesized that spleen glucose metabolism is different in autoimmune and localized infectious diseases as a consequence of systemic inflammation in autoimmune disease. Hence, we evaluated the correlation between spleen 18F-FDG uptake and laboratory data, the usefulness of 18F-FDG uptake in differentiating between autoimmune and localized infectious diseases, and the role of 18F-FDG uptake in predicting patient outcome.
MATERIALS AND METHODS
Patient Selection
The medical records of patients who underwent 18F-FDG PET/CT at Severance Hospital in Seoul, South Korea, from December 2005 to December 2015 were reviewed. The inclusion criteria were as follows: patients 15–75 y old, patients with documented fever (≥37.8°C) during hospital admission, and patients with a definite cause of fever. The exclusion criteria were as follows: patients with an elucidated origin of fever within 3 d of admission, patients with septic shock or sepsis, and patients with cancer. Finally, 192 patients were enrolled; 91 patients were classified as having an autoimmune disease and 101 as having localized infectious disease. Two groups of age- and sex-matched healthy controls were selected for the autoimmune and infectious disease groups. This study was approved by the Institutional Review Board of Severance Hospital (4-2016-0150), and the requirement to obtain informed consent was waived because of the retrospective nature of the study.
Clinical and Laboratory Data Collection
Clinical data collected included age, sex, and in-hospital duration. Laboratory data included white blood cell, neutrophil, lymphocyte, monocyte, and platelet counts and hemoglobin, erythrocyte sedimentation rate, C-reactive protein, blood urea nitrogen, creatinine, aspartate aminotransferase, alanine aminotransferase, total bilirubin, total protein, albumin, and ferritin levels, which were measured within 3 d of the date of the PET/CT scan. In-hospital mortality was defined as all-cause mortality during hospital admission.
18F-FDG PET/CT Image Acquisition
All 18F-FDG PET/CT scans were obtained on a dedicated PET/CT scanner (Discovery STE [GE Healthcare] or Biograph 40 TruePoint [Siemens Medical Systems]). All patients fasted for 6 h before the PET/CT scan, and a blood glucose level below 140 mg/dL was confirmed. The PET/CT scan was performed 60 min after the intravenous administration of 5.5 MBq of 18F-FDG per kilogram. The CT scan was performed at 30 mA and 130 kVp on the Discovery STE or at 36 mA and 120 kVp on the Biograph 40 TruePoint. The PET scan was performed with an acquisition time of 2.5 min per bed position in 3-dimensional mode. PET images were reconstructed using an ordered-subset expectation maximization algorithm with attenuation correction.
Assessment of SUV Acquired by 18F-FDG PET/CT Scans
Radiotracer accumulation was analyzed semiquantitatively using both SUVmean and SUVmax. Spleen 18F-FDG SUV was obtained on 3 nonadjacent slices and averaged. Bone marrow SUV was separately obtained from lumbar vertebrae 1 through 5 and averaged. In the case of bone infection directly involving the lumbar spine (n = 1) and liver abscess (n = 3), the region involved was excluded from analysis. To measure normal liver activity, 3 spheric 1-cm regions of interest were drawn on the liver: 2 on the right lobe and 1 on the left lobe. The SUVmean of the liver was defined as the mean of the 3 regions of interest. Because 2 different PET scanners were used in this study, liver was used as an internal reference organ to reduce problems related to different scanners. The spleen-to-liver ratio of the SUV (SLR) was calculated by dividing the spleen SUVmax or SUVmean by the liver SUVmean, and the bone marrow–to–liver ratio of the SUV (BLR) was calculated by dividing the bone marrow SUVmax or SUVmean by the liver SUVmean.
Statistical Analysis
Data were analyzed using SPSS, version 21 (SPSS Inc.). Continuous variables are presented as mean with SD, and categoric variables are expressed as frequencies and percentages. Continuous variables were compared using the Student t test, and categoric data were compared using the χ2 test. Correlations between laboratory variables and SLR or BLR were calculated by the Pearson correlation analysis.
The discriminative ability of SLR or BLR for autoimmune disease was analyzed using area under the receiver-operating-characteristic (ROC) curve. The area under the ROC curve was presented with 95% confidence interval (CI), and the Youden index was used to identify the maximal cutoff. Risk factors for in-hospital mortality were calculated by univariate and multivariate logistic regression analyses. For estimation of the best cutoff for SLRmean in predicting in-hospital mortality, ROC curve analysis was used. A P value of less than 0.05 was considered statistically significant.
RESULTS
Baseline Characteristics of Patients
The mean age was 48.1 y in the autoimmune disease group and 55.0 y in the infectious disease group (P = 0.005). Sixty-four patients (70.3%) were female in the autoimmune disease group, whereas 48 patients (47.5%) were female in the infectious disease group (P = 0.001) (Table 1). During the admission period, 14 patients died: 12 in the autoimmune disease group and 2 in the infectious disease group. The autoimmune disease group had lower platelets, lymphocytes, monocytes, erythrocyte sedimentation rate, total protein, and serum albumin than the infectious disease group (Table 1). However, the autoimmune disease group had higher aspartate aminotransferase, alanine aminotransferase, and ferritin than the infectious disease group.
Baseline Characteristics and Comparison of Patients with Autoimmune and Infectious Diseases
Spectrum of Disease
Most patients were diagnosed with hemophagocytic lymphohistiocytosis, systemic lupus erythematosus, Kikuchi disease, adult-onset Still disease, vasculitis, or rheumatoid arthritis (Table 2). In patients with infectious disease, the most prevalent regions affected were the chest, followed by the abdomen, bone, biliary system, urinary tract, brain, and orofacial area (Table 2).
Clinical Spectrum of Autoimmune and Infectious Diseases
Comparison of SLR and BLR
Patients with autoimmune disease had higher SLRs than patients with localized infectious disease (SLRmax, 1.46 ± 0.49 vs. 1.09 ± 0.25, P < 0.001; SLRmean, 1.28 ± 0.43 vs. 0.91 ± 0.21, P < 0.001 [Fig. 1]). Also, patients with autoimmune disease had higher mean BLRs than patients with localized infectious disease (BLRmax, 1.36 ± 0.51 vs. 1.18 ± 0.37, P = 0.004; BLRmean, 1.19 ± 0.46 vs. 0.98 ± 0.31, P < 0.001 [Fig. 1]). Representative images (Fig. 2) demonstrate that diffuse 18F-FDG uptake in the spleen is detected with strong contrast to liver 18F-FDG uptake in autoimmune disease. Because the autoimmune and infectious disease groups had different mean ages and sex distributions, we selected 2 healthy control groups matched by age and sex to compare with the autoimmune disease group (control group A) and the infectious disease group (control group B). There was no difference in SLRs or BLRs between the two control groups, suggesting that SLRs and BLRs were not affected by age or sex in our control groups. Interestingly, SLRs and BLRs were higher in the infectious disease group than in the healthy control groups, suggesting that, even in localized infectious disease, spleen and bone marrow glucose metabolism was increased (Fig. 1).
18F-FDG uptake in patients with autoimmune disease (n = 91), patients with infectious disease (n = 101), and age- and sex-matched healthy controls (n = 50). SLRs and BLRs were compared among autoimmune disease controls (A), infectious disease controls (B), patients with infectious disease (C), and patients with autoimmune disease (D). Error bars indicate 95% CI. ns = not significant. **P < 0.01.
Representative 18F-FDG PET/CT images in age- and sex-matched healthy controls for autoimmune disease (A) and infectious disease (B) and patients with autoimmune disease (C) and infectious disease (D).
Association of Laboratory Variables with SLR and BLR
In patients with autoimmune disease, SLRmean correlated positively with aspartate aminotransferase, alanine aminotransferase, and ferritin and negatively with serum albumin and monocytes. SLRmean had the highest correlation with ferritin (r = 0.649, P < 0.001). BLRmean correlated positively with white blood cells, neutrophils, C-reactive protein, and ferritin (Table 3). However, in localized infectious diseases, SLRmean correlated positively with C-reactive protein and negatively with total protein and serum albumin. BLRmean correlated negatively with serum albumin (Table 3). We also analyzed the association of laboratory variables with SLRmax and BLRmax. Although not included in Table 3, the correlations between laboratory values and SLRmax and BLRmax were similar to the correlations between laboratory values and SLRmean and BLRmean included in the table.
Correlation Between Laboratory Variables and SLRmean and BLRmean in Autoimmune and Infectious Diseases
Differentiation Between Autoimmune and Localized Infectious Diseases
ROC analysis was performed to compare the laboratory variables and SLRs and BLRs for differentiation between autoimmune and localized infectious diseases (Table 4). SLRs had higher ROC values than BLRs, and an SLRmean cutoff of 1.06 had a sensitivity of 62.6% and specificity of 84.2%, with the highest ROC value compared with hemoglobin, platelets, lymphocytes, monocytes, erythrocyte sedimentation rate, aspartate aminotransferase, alanine aminotransferase, total protein, albumin, and ferritin (area under the ROC curve, 0.782; 95% CI, 0.717–0.839; P < 0.001).
Comparison of ROC Curve of Spleen and Bone Marrow SUV and Laboratory Variables in Differentiating Between Autoimmune and Infectious Diseases
Risk Factors for In-Hospital Mortality in Autoimmune Disease
We next evaluated the risk factors for in-hospital mortality in patients with autoimmune disease. In univariate analysis, platelet count (odds ratio [OR], 0.993; 95% CI, 0.987–0.998; P = 0.014) and SLRmean (OR, 3.625; 95% CI, 1.013–12.967; P = 0.047) were identified as prognostic predictors. Furthermore, using ROC analysis, an SLRmean cutoff of more than 1.61 revealed the greatest risk for in-hospital mortality (OR, 5.076; 95% CI, 1.413–18.230; P = 0.012). The multivariate logistic regression model showed that platelet count (OR, 0.993; 95% CI, 0.988–0.999; P = 0.020) and an SLRmean of more than 1.61 (OR, 4.796; 95% CI, 1.228–18.723; P = 0.024) were still significantly associated with in-hospital mortality (Table 5).
Univariate and Multivariate Logistic Regression Models for Prediction of In-Hospital Mortality During Admission in Patients with Autoimmune Diseases
DISCUSSION
The spleen is the largest organ in the lymphatic system. Unlike lymph nodes, the spleen has no afferent lymph vessels but only efferent lymph vessels (8). Therefore, the spleen is specialized to monitor blood for microorganisms and blood cells, whereas lymph nodes monitor lymphatics for local inflammation. However, the diagnostic methods to evaluate spleen conditions have been limited. Although measuring spleen size is the most common method for evaluating the spleen in current clinical practice, our study is the first, to our knowledge, to show that spleen 18F-FDG uptake is elevated in autoimmune disease compared with localized infectious disease and also provides prognostication for febrile patients hospitalized with autoimmune disease.
There is a biologic rationale for using 18F-FDG uptake to evaluate spleen immunometabolism in systemic inflammation (9). In the spleen, dendritic cells and T cells interact for antigen processing and also become directly exposed to proinflammatory cytokines in the blood (10). During activation, dendritic cells and T-cell metabolism change; the aerobic glycolysis pathway is used during this high-energy-demand state (6,11,12). Recent studies demonstrated that inhibition of glycolysis or oxidative phosphorylation improves systemic inflammation in animal models of lupus as well as in human lupus (10,11,13). Monitoring spleen energy metabolism may be useful in systemic inflammatory disease for evaluating disease activity and possibly providing guidance for metabolism-targeted therapy (14).
We excluded sepsis and cancer patients because our hypothesis was to evaluate the role of spleen and bone marrow 18F-FDG uptake in febrile patients who are without a definite diagnosis. Sepsis is a diagnosis based on clinical presentation, vital signs, and identification of microorganisms and is beyond our research scope. Spleen 18F-FDG uptake in cancer is also an interesting topic, and a recent study demonstrated that high spleen 18F-FDG uptake is associated with poor prognosis in patients with cholangiocarcinoma (15). Although cancer patients can develop fever of various causes, such as chemotherapy, neutropenia, thrombosis, and infection, we did not include cancer patients because it is not common to perform PET/CT on cancer patients for fever evaluation. Because of the relatively narrow clinical focus of our study, the results from our study cannot be generalized and should be interpreted with caution when evaluating patients with fever.
Although lymphoma is known to be one of the most frequent malignant conditions associated with elevated splenic SUVs, focal mass or infiltrative patterns of 18F-FDG uptake in the spleen suggests a primary pathology (16) that is different from a diffuse uptake pattern in systemic autoimmune diseases. However, it will be difficult to differentiate diffuse infiltrative patterns of splenic lymphoma from increased spleen 18F-FDG uptake secondary to inflammation.
Bone marrow and spleen play different roles in systemic inflammation. Bone marrow is the site of hematopoietic progenitor cell production (17), whereas spleen is the site of interaction of activated immune cells (8,18). Although both SLRs and BLRs were higher in autoimmune disease than in localized infectious disease, SLRs were better at differentiating autoimmune from localized infectious diseases than were BLRs. This finding may be due to nonspecific progenitor cell activation in bone marrow in both autoimmune and localized infectious diseases, whereas immune cell activation in the spleen is more specific in febrile autoimmune diseases (17). In fact, an SLRmean cutoff of 1.06 showed the highest area under the ROC among all imaging and laboratory variables.
Comparison of laboratory variables with SLRs and BLRs also showed distinct patterns. In autoimmune disease, SLRmean was most strongly associated with ferritin level, whereas BLRmean was most strongly associated with white blood cell and neutrophil counts. Ferritin is an acute-phase reactant that is elevated in several autoimmune diseases such as systemic lupus erythematosus, rheumatoid arthritis, and adult-onset Still disease (19). The association between serum ferritin level and spleen glucose metabolism is an interesting novel finding, and further studies are necessary to understand their relationship to systemic inflammation.
Identification of prognostic factors is important for the proper immunosuppressive treatment of autoimmune diseases. In this study, SLRmean with a cutoff greater than 1.61 was associated with in-hospital mortality during the admission period. Increased spleen 18F-FDG uptake may represent an early sign of severe systemic inflammatory response that is not detectable by conventional laboratory parameters. Therefore, closer monitoring may be needed in autoimmune disease patients with high SLRs. Although our findings cannot be simply generalized due to heterogeneity of disease, they suggest the importance of quantitative evaluation of SLRs in clinical practice.
The strength of our study is the large number of patients who were clearly diagnosed as having either autoimmunity or infection as the cause of fever. Although previous studies have evaluated the role of PET/CT in diagnosing autoimmune and infectious diseases (3,4,20,21), they were limited by relatively small sample sizes and focused mainly on detection of localized infection or inflammation. However, previous studies demonstrated that specific patterns of 18F-FDG uptake in inflamed tissue can provide important information. In rheumatoid arthritis, 18F-FDG uptake in joints is associated with disease activity (22). In large-vessel vasculitis, the patterns of 18F-FDG uptake in blood vessels are useful for diagnosis and predict poor outcome (23). Also, hidden localized infection can easily be identified by PET/CT scanning, especially in chronic infections of bone or adjacent structures (4). Therefore, PET/CT can provide multiple useful findings in febrile patients in addition to SLR or BLR.
Limitations to our study include those that are inherent in all single-institution retrospective observational studies, including bias in patient selection and analysis. Also, only patients who had autoimmune or localized infectious diseases as the cause of fever were included. Other causes of fever, such as cancer or systemic infectious diseases, may have different spleen 18F-FDG uptake values and patterns from our study. Further studies will be required to investigate this possibility.
CONCLUSION
Spleen 18F-FDG uptake is increased in febrile autoimmune disease and is associated with an increased risk of all-cause in-hospital mortality. Evaluation of spleen glucose metabolism might be useful for differentiating systemic inflammatory diseases from localized infectious diseases when other causes of fever are excluded. Further studies to evaluate the molecular mechanism for glucose energy metabolism during systemic inflammation are necessary.
DISCLOSURE
Dr. Song’s work was supported by Basic Science Research Program (2015R1C1A1A01053140) through the National Research Foundation of Korea, funded by the Ministry of Education, Science, and Technology. Dr. Yun’s work was supported by a National Research Foundation of Korea grant funded by the Korean government (MSIP) (NRF-2011-0030086). No other potential conflict of interest relevant to this article was reported.
Acknowledgments
We thank Michael Garcia for critical reading of the manuscript.
Footnotes
Published online Sep. 29, 2016.
- © 2017 by the Society of Nuclear Medicine and Molecular Imaging.
REFERENCES
- Received for publication July 6, 2016.
- Accepted for publication September 8, 2016.
Jump to section
Related Articles
Cited By...
- No citing articles found.