Clinical Differentiation of Severe Fever with Thrombocytopenia Syndrome from Japanese Spotted Fever

Severe fever with thrombocytopenia syndrome (SFTS) and Japanese spotted fever (JSF; a spotted fever group rickettsiosis) are tick-borne zoonoses that are becoming a significant public health threat in Japan and East Asia. Strategies for treatment and infection control differ between the two; therefore, initial differential diagnosis is important. We aimed to compare the clinical characteristics of SFTS and JSF based on symptomology, physical examination, laboratory data, and radiography findings at admission. This retrospective study included patients with SFTS and JSF treated at five hospitals in Nagasaki Prefecture, western Japan, between 2013 and 2020. Data from 23 patients with SFTS and 38 patients with JSF were examined for differentiating factors and were divided by 7:3 into a training cohort and a validation cohort. Decision tree analysis revealed leukopenia (white blood cell [WBC] < 4000/μL) and altered mental status as the best differentiating factors (AUC 1.000) with 100% sensitivity and 100% specificity. Using only physical examination factors, absence of skin rash and altered mental status resulted in the best differentiating factors with AUC 0.871, 71.4% sensitivity, and 90.0% specificity. When treating patients with suspected tick-borne infection, WBC < 4000/µL, absence of skin rash, and altered mental status are very useful to differentiate SFTS from JSF.


Introduction
Severe fever with thrombocytopenia syndrome (SFTS) is a tick-borne infectious disease caused by the SFTS virus (SFTSV or Dabie bandavirus, a novel genus Banyangvirus of the family Phenuiviridae that was first reported in China in 2011 [1,2]. Since then, many cases have been reported in west Asia, China, Korea, and Japan [3] and are classified as viral hemorrhagic fever [4]. The mortality rate of SFTS ranges from 5-40% [4][5][6][7][8]. Around

Study Population and Clinical Data
We retrospectively reviewed the medical records of patients aged ≥20 years diagnosed with laboratory-confirmed SFTS and JSF between January 2013 and December 2020 at five hospitals (two tertiary medical institutions: Nagasaki University Hospital and Sasebo City General Hospital; three secondary medical institutions: Isahaya General Hospital, Sasebo Chuo Hospital, and Nagasaki Rosai Hospital) in Nagasaki Prefecture, Japan. The medical records of all enrolled patients were reviewed. Data on baseline characteristics, clinical presentations, laboratory findings, and radiological findings at admission were collected from the medical records. In-hospital complications and outcomes were assessed.

Definitions
All definitions used in the study were established before the data analyses. An episode of SFTS was defined as SFTSV, in which viral RNA was confirmed by real-time PCR analysis of a serum sample, as previously described [31]. The diagnosis of JSF was defined as either a 4-fold increase in immunoglobulin G (IgG) in the serum sample with immunofluorescence assay, PCR analysis of a blood sample, or tissue biopsy (eschar) [32]. Altered mental status was defined as Japan coma scale ≥1 points [33]. Acute kidney injury (AKI) was diagnosed according to the Kidney Disease: Improving Global Outcomes definition, as described previously [34]. Chronic kidney disease (CKD) was defined as a reduced glomerular filtration rate (<60 mL/min/1.73 m 2 ) for more than 3 months. [35] Immunosuppressants included corticosteroids and other immunosuppressive agents. Disseminated intravascular coagulation (DIC) scores were calculated using the Japanese Association for Acute Medicine (JAAM) DIC scoring system [36]. Hemophagocytic lymphohistiocytosis (HLH) was defined using HLH-2004 criteria [37].
Complications and treatments were respectively determined as those that developed or were initiated after the initial visit or the date of hospital admission.

Evaluation of the Chest Radiograph and Computed Tomography (CT)
Two Japanese Respiratory Society (JRS) board-certified pulmonologists each with 15 years of experience reviewed all chest radiographs and CT examinations independently and arrived at a consensus. Cardiomegaly was defined as a cardiothoracic (CTR) ratio > 0.50, based on posterior-anterior (PA) chest radiographs, or a CTR ratio > 0.55 on anteriorposterior (AP) chest radiographs [31]. Abnormalities on chest CT were characterized by consolidation, ground-glass opacity (GGO), centrilobular nodules, interstitial septal thickening, and bronchial wall thickening. The presence of mediastinal lymph node enlargement (>10 mm along the short axis), pleural effusion, pericardial effusion (pericardial thickness of 4 mm or more [31]), hepatomegaly (diameter of >16.0 cm at craniocaudal line [31]), splenomegaly (width measurement of >10.5 cm [31]), and additional lung findings were also recorded.

Statistical Analysis
The results are expressed as means ± standard deviations and as percentages. Categorical variables were analyzed using Fisher's exact test, and continuous variables were analyzed using a t-test or Wilcoxon rank sum test. All tests were two-tailed, and differences were considered significant at a p-value < 0.05 or less than the value adjusted by Bonferroni correction.
All cases were divided by 7:3 into a training cohort and a validation cohort. In the training data, a decision tree was built to differentiate SFTS from JSF using variables that were significantly different on bivariate analysis and missing values less than 20%. In addition, we performed a decision tree on each continuous variable one by one to determine cut off values and applied them to categorical data.
The discriminatory power of the differentiation models was evaluated by assessing the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, positive predictive value, and negative predictive value. All analyses were performed using JMP ® Pro 16.2.0 predictive analysis software (SAS Institute Inc., Cary, NC, USA).

Ethical Considerations
The study protocol was approved by the Institutional Review Board of Nagasaki University Hospital (approval number 18121024), Isahaya General Hospital (approval number, 2020-21), Sasebo City General Hospital (approval number 2020-A024), Sasebo Chuo Hospital (approval number 2020-17), and Nagasaki Rosai Hospital (approval number 02003). The institutional review board waived the requirement for informed consent from patients included in this study because of its retrospective design. The summary and information of this study have been disclosed, following the regulations of each participating hospital.

Study Population
Data from a total of 61 hospitalized patients (23 with SFTS and 38 with JSF) at five hospitals during the study period were reviewed. All 23 cases of SFTS were confirmed by PCR testing of blood samples; 2 cases of JSF were confirmed by both the presence of increased Rickettsia japonica antibody levels and PCR test of blood and/or tissue sample,  12 cases by antibody test only, and the remaining 24 cases by PCR testing of blood and/or tissue samples. One case of JSF, diagnosed by antibodies to R. japonica and PCR of blood, was a case of internationally imported infection, with elevated antibodies to R. cornorii and R. africae other than R. japonica, and the coexistence of rickettsiosis infection was suspected.

Patients' Baseline Characteristics
The baseline patient characteristics are presented in Table 1. The elderly population was predominant in both the SFTS (mean 70.6 years) and JSF groups (mean 68.8 years). There were no significant differences between the two groups in terms of male sex or underlying diseases. Immunosuppressants were used in two cases of SFTS: corticosteroids and adalimumab. Patients who were farmers, hunters, or living or working in wooded and hilly areas comprised 50.0% of SFTS patients and 27.8% of JSF patients. Patients in both groups were diagnosed frequently in spring and summer, though incidence of SFTS was lower (8.7%) in autumn compared to that of JSF (34.2%). No patients were diagnosed during winter. Data are present as number of patients (%) or mean ± SD. Abbreviations: SFTS, severe fever with thrombocytopenia syndrome; JSF, Japanese spotted fever; ICU, intensive care unit. Significant difference level: 0.05/12 = 0.0042. Table 2 presents the clinical symptoms and physical findings of patients with SFTS or JSF at admission. Approximately 5 and 6 days in patients with SFTS and JSF, respectively, passed between onset of illness and admission. Patients with SFTS were significantly more frequently reporting altered mental status (60.9% vs. 21.1%, p = 0.003) and diarrhea (52.4% vs. 10.5%, p = 0.001) compared to patients with JSF. In contrast, skin rash (34.8% vs. 100%, p < 0.0001) and tick bites (42.9% vs. 88.9%, p = 0.001) were more frequently observed in patients with JSF. With regard to vital signs, most patients in both groups were febrile. There was no difference between the two groups in terms of respiratory rate or SpO 2 /FiO 2 ratio. The qSOFA scores were comparable between the two groups.

Outcomes and Treatments
Supplementary Table S1 shows the outcomes, complications, and treatments of the patients with SFTS and JSF. The mortality rate of patients with SFTS (26.1%) was higher that of patients with JSF (0.0%) (p = 0.002). The length of hospital stay was longer (mean; 34.5 days vs. 15.9 days, p = 0.015) and the ICU admission rate was higher (mean; 30.4% vs. 7.9%, p = 0.032) in patients with SFTS than those in patients with JSF.
With regard to in-hospital complications, secondary bacterial infections occurring more frequently in the SFTS group than in the JSF group (mean: 26.1% vs. 5.3%, p = 0.044) consisted of bacteremia (n = 4) caused by Streptococcus epidermidis or methicillin-resistant S. capitis, followed by pneumonia (n = 2), and urinary tract infections (n = 1) caused by Pseudomonas aeruginosa. Secondary fungal infections, although not statistically significant, were also more frequent in patients with SFTS (17.4%). These infections include invasive aspergillosis, disseminated trichosporonosis, candidiasis, and pulmonary cryptococcosis. One case of SFTSV meningitis and another case of rickettsial meningitis were observed in the study population. A total of 11 (47.8%) and 16 (69.6%) cases of patients with SFTS developed hemophagocytic lymphohistiocytosis (HLH) or disseminated intravascular coagulation (DIC), respectively. The diagnosis of 9 of 11 cases of HLH in patients with SFTS was made by bone marrow biopsy.
All patients with SFTS were administered antibiotics against rickettsial infections at admission until the SFTSV diagnosis was confirmed. Antiviral agents such as favipiravir and ribavirin were used in five (27.3%) and two (8.7%) patients with SFTS, respectively, based on clinical trials or off-label use [26,38]. All patients with JSF received tetracyclines, including minocycline in 36 cases, doxycycline in 1 case, and a combination of minocycline and doxycycline in 1 case. A total of 12 (31.6%) of the patients with JSF were treated with a combination of quinolones including levofloxacin (n = 9) and ciprofloxacin (n = 3). Corticosteroids, G-CSF, immunoglobulin, and thrombomodulin were more frequently administered to patients with SFTS than to those with JSF, reflecting the severity of the disease. Mechanical ventilation was performed in 27.3% of patients with SFTS, which was higher compared to use in patients with JSF (5.3%).

Decision Tree Analysis for Differentiation between JSF and SFTS
To investigate factors that could differentiate between SFTS and JSF at the initial hospital visit, all cases were divided into a 7:3 test cohort (n = 44) and a validation cohort (n = 17). A decision tree analysis was performed by applying variables with significant differences in bivariate analysis that excluded variables not available in >20% of patients for a training cohort group. In the training cohort, the final classification using decision tree analysis extracted only two variables: WBC < 4000/µL and altered mental status. A WBC < 4000/µL led to SFTS probability of 100% (16/16). In the case that the patients with WBC > 4000/µL have altered mental status, the possibility of SFTS was 14.3% ( Figure 1A). For the validation cohort, the area under the curve (AUC) by the Receiver Operating Characteristic (ROC) of the decision tree was 1.000.
Due to the small number of patients in the validation cohort, the decision tree analysis derived from all patients in the study and detected AUC, sensitivity, and specificity of each variable are shown in Supplementary Figures S1 and S2, and Supplementary Table S2.   In addition, since blood tests are not available quickly in some situations, another decision tree analysis was performed to distinguish SFST from JSF based only on physical findings with significant differences by bivariate analysis among the training cohort and is shown in Figure 2A. A decision tree analysis using only physical findings extracted two variables: skin rash and altered mental status. Absence of skin rash led to SFTS probability of 100% (12/12). When skin rash was present, altered mental status increased the SFTS probability to 50% (2/4). AUC by ROC of the decision tree was 0.871 with 71.4% sensitivity and 90.0% specificity.   Next, we evaluated each variable for which there was a significant difference in bivariate analysis and in cases with missing values less than 20% in Tables 2 and 3. In the analysis, continuous variables were changed to categorical data, with cutoff values determined for each using the decision tree analysis among the training cohort group described in the first column of Table 5. AUC, sensitivity, specificity, positive predictive value, and negative predictive values were calculated by applying these variables in the validation cohort group (Table 5). In the validation cohort, WBC count < 4000/µL (p < 0.0001, AUC 1.000), neutrophil count < 2042/µL (p < 0.0001, AUC 1.000), CRP < 4.17 mg/dL (p = 0002, AUC 0.833), and platelet count < 64 × 10 3 /µL (p = 0.003, AUC 0.818) had excellent discriminatory ability with AUC > 0.8. Table 5. Sensitivity, specificity, positive predictive value and negative predictive value for differentiating SFTS from JSF.  Abbreviations: SFTS, severe fever with thrombocytopenia syndrome; JSF, Japanese spotted fever; WBC, white blood cell; AST, aspartate aminotransferase; ALT, alanine transaminase; LDH, lactate dehydrogenase; CRP, C-reactive protein; APTT, activated partial thromboplastin; PPV, positive predictive value; NPV, negative predictive value.
Due to the small number of patients in the validation cohort, the decision tree analysis derived from all patients in the study and detected AUC, sensitivity, and specificity of each variable are shown in Supplementary Figures S1 and S2, and Supplementary Table S2.

Discussion
SFTS and JSF are tick-borne diseases that are geographically, seasonally, and clinically similar [39]. Definitive diagnosis of these diseases is time-consuming, and only a few university hospitals or specialized facilities in Japan are able to perform SFTSV PCR. Differentiating between these two infections is important because their treatment is different, and more importantly, SFTSV is capable of nosocomial transmission [27,28].
The present study revealed that SFTS and JSF did not differ in patients' underlying conditions; however, many symptoms, such as altered mental status and diarrhea were more frequently found in patients with SFTS. These symptoms may be associated with viremia in SFTS [30,31]. Skin rash and tick bite were frequently found in patients with JFS. Comparison of chest CT findings between SFTS and JSF revealed that bronchogenic infiltration, such as centrilobular nodules and bronchial wall thickening, was likely to occur more frequently in patients with SFTS, which may be due to secondary bacterial or fungal pulmonary infections [31,40]. Notably, abnormal findings on chest CT were found in around 80% of the SFTS group; ground glass opacity, and interstitial septal thickening was detected in more than 50% of patients. Hospitalized patients with SFTS or JSF frequently develop pulmonary edema that may be caused by systemic inflammation [31,[40][41][42]. Our study is the first to compare chest radiologic findings between SFTS and JSF; however, the results indicated that radiographical differentiation between these two infectious diseases was difficult.
Many laboratory data values at admission were different between patients with SFTS and those with JSF. Among them, leukopenia of WBC < 4000/µL, as supported by a previous report [8], was the most remarkable variable that was able to distinguish SFTS from JSF with sensitivity of 100%, specificity of 100%, and AUC 1.000. Peripheral blood count is a basic examination that can be performed in most facilities, and results can be obtained within one hour. Altered mental status was detected in more than 60% of SFTS patients in our study, a slight increase from the previously reported 51% [8]. The possibility of SFTS cannot be ruled out if altered mental status is present even when WBC is above 4000 /µL, based on our cohort. If a patient with suspected tick-borne disease has an elevated WBC (≥4000 /µL) but an altered mental status, it is recommended to test for SFTS at the same time as treating rickettsial disease.
We also attempted to differentiate SFTS from JSF based on physical examination alone, assuming a situation where blood tests are not available immediately. Our results showed that the presence or absence of a skin rash was the most important physical finding in differentiating between the two, and the absence of a skin rash was most likely in SFTS patients (100%). Even in the presence of a rash, altered mental status was associated with a high probability of SFTS (50%).
To our best knowledge, only one previous report discussed the differentiation between JSF and SFTS. Kawaguchi et al. [8] reported that CRP level ≤ 1.0 mg/dL was the most useful variable for differentiating SFTS from JSF. The decision tree in our study did not keep CRP in the final node; however, CRP < 4.17 mg/dL itself had sensitivity of 83.3%, specificity of 90.9%, and AUC 0.833 in our study. In Supplementary Table S1, we also performed a decision tree for each variable among all cases in our study, and the cutoff value of CRP 1.78 mg/dL (sensitivity of 87.0%, specificity of 100.0%, and AUC 0.935) for differentiation was similar to that previously reported [8]. Although the incidence of scrub typhus in our study period was low and comparisons could not be made, it is reported that scrub typhus and JSF are both rickettsial diseases and result in similar symptoms and blood test data [43,44]. Table 6 lists previous reports which differentiated between SFTS and other tick-borne infectious diseases such as JSF and scrub typhus. Some previous studies also supported our findings that leukopenia (WBC < 4000 /µL) and altered mental status are significant differentials for SFTS.
Our study had several limitations. First, we enrolled hospitalized patients from secondary and tertiary medical institutions. Among 40 cases of SFTS and 109 cases of JSF diagnosed in our area according to the Infectious Diseases Weekly Report of Nagasaki Prefecture between 2013 and 2020 [45,46], our study included 23/40 (57.5%) of patients with SFTS and 38/109 (34.8%) of patients with JSF. Selection bias, especially for patients with JSF, may have occurred because hospitalized severe cases were mainly included in the study. In fact, our data show a higher percentage with 45.2% of SFTS patients in tertiary medical institutions compared to 21.1% in secondary medical institutions. Second, the data were obtained retrospectively; therefore, missing data values regarding such characteristics as PCT, ferritin, sIL2-R, IgG, and fibrinogen were insufficient for applying to regression analysis and could not be used to determine the significance of these factors in differentiating diseases. In addition, blood tests at the patient's first visit had variations from the date of infection. This blood time-point of the infection may affect the readout values. Third, scrub typhus, another tick-borne disease, is also prevalent in western Japan, including Nagasaki prefecture, though comparisons could not be made in this study based on the low number of scrub typhus during the study period. In clinical practice, it is necessary to distinguish SFTS from other tick-borne infectious diseases prevalent in the area. A prospective study including mild cases with secured data collection from patients is crucial to substantiate our results and construct a scoring system in real-world practice. Abbreviations: SFTS, severe fever with thrombocytopenia syndrome; JSF, Japanese spotted fever; AUC; area under the curve, CI; confidence interval; WBC, white blood cell; APTT, activated partial thromboplastin; CRP, C-reactive protein; CK, creatine kinase.

Conclusions
In conclusion, when clinicians differentiate between SFTS and JSF among patients with suspected tick-borne infection, a single item, WBC < 4000/µL, strongly differentiates Viruses 2022, 14, 1807 12 of 14 SFTS from JSF. Additionally, the absence of a skin rash and altered mental status may be helpful for differential diagnosis.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/v14081807/s1. Supplemental Tables S1 and S2 and Supplemental Figures S1 and S2.  Institutional Review Board Statement: This multicenter study was conducted in compliance with the principles of the Declaration of Helsinki and was approved by the ethics committees of the participating institutions (approval number:18121024; Nagasaki University Hospital, approval number:2020-21; Sasebo City General Hospital, approval number:2020-A024; Sasebo Chuo Hospital, approval number:2020-17; Nagasaki Rosai Hospital, approval number:02003).

Informed Consent Statement:
The requirement for patient consent was waived owing to the retrospective nature of the study, which ensured anonymity.
Data Availability Statement: Data are contained within the article.