Combinatorial Use of High Mobility Group Box 1 Values and Gunma Scores Improve The Accuracy of Predicting Non-Responders to Intravenous Immunoglobulin Treatment in Patients with Kawasaki Disease

  • Yuichi Nomura, Division of Pediatrics, Kagoshima University Graduate School of Medical and Dental Sciences., Japan
  • Kentaro Ueno, Division of Pediatrics, Kagoshima University Graduate School of Medical and Dental Sciences., Japan
  • Kiminori Masuda, Department of Pediatrics, Kagoshima Medical Association Hospital, Japan
  • Yasuko Morita, Department of Pediatrics, Kagoshima Medical Association Hospital, Japan
  • Daisuke Hazeki, Division of Pediatrics, Kagoshima University Graduate School of Medical and Dental Sciences., Japan
  • Taisuke Eguchi, Division of Pediatrics, Kagoshima University Graduate School of Medical and Dental Sciences., Japan
  • Yoshifumi Kawano, Division of Pediatrics, Kagoshima University Graduate School of Medical and Dental Sciences., Japan

    • Objective: Patients with Kawasaki disease (KD) who do not respond to intravenous immunoglobulin (IVIG) treatment show a high incidence of coronary artery abnormalities (CAA). How to predict non-responders and then provide a specific treatment to prevent CAA is a current topic of discussion. Kobayashi et al. have reported the usefulness of a predictive score for non-responders (Gunma score; sensitivity 86%, specificity 68%). In respect to non-responders, we have reported that an elevated value of high mobility group box 1 (HMGB1) is a potential marker. Combinational use of HMGB1 values with Gunma scores may improve the accuracy of predicting non-responders. Thus, we examined this possibility.
    Methods: Serum HMGB1 values were measured in KD patients admitted to the Kagoshima City Medical Association Hospital. Clinical data and Gunma scores were obtained from these patients.
    Results: Compared with responders (n=85), non-responders (n=16) showed significantly elevated values of WBC and CRP, along with decreased values of albumin and sodium. Gunma scores were significantly higher in non-responders than responders (5.1±2.9, 2.7±2.1, p=0.0002). Sensitivity (score>3) was 63% and specificity 68%. HMGB1 values before IVIG were also significantly higher in non-responders (14.8±11.3, 9.7±5.5, p=0.0064). Sensitivity with high HMGB1 (>15.2: mean+SD of responders) was 50% and specificity 71%. When a predicted non-responder is defined as a patient with a Gunma score of 5 or over or a high HMGB1 value, sensitivity was 81% and specificity 71%.
    Conclusion: Combinatorial use of HMGB1 values and Gunma scores improve the accuracy of predicting non-responders.