Fever pattern and C-reactive protein predict response to rescue therapy in Kawasaki disease.

Background: Evidence to guide rescue therapy in refractory Kawasaki disease (KD) is lacking. The aim of this study was to determine the most important variables in predicting non-response to rescue therapy in refractory KD.

Methods: We retrospectively analyzed 171 patients diagnosed with refractory KD resistant to initial i.v. immunoglobulin (IVIG). Participants received rescue therapy consisting of IVIG monotherapy or IVIG plus prednisolone. Characteristics and laboratory variables were compared between rescue therapy non-responders and responders. Multivariate logistic regression analysis was performed to determine the independent predictors of non-response to rescue therapy.

Results: Among the 171 participants, 54 (31.6%) were non-responders to rescue therapy. On univariate analysis, fever pattern after initial IVIG, day of illness at rescue therapy, rescue therapy regimen and six laboratory variables (pre-IVIG sodium, C-reactive protein [CRP]; post-IVIG white blood cell count, platelet count, sodium, CRP) were useful in discriminating between non-responders and responders. These nine variables were included in multivariate logistic regression analysis. Persistent fever after initial IVIG (aOR, 2.39; 95%CI: 1.07-5.37) and post-IVIG CRP (aOR, 1.09; 95%CI: 1.02-1.17, per 1 mg/dL increase) were identified as independent predictors of non-response to rescue therapy. IVIG rescue monotherapy (aOR, 3.05; 95%CI: 1.05-8.84) also predicted non-response after adjusting for fever pattern and post-IVIG CRP.

Conclusions: Persistent fever and elevated CRP after initial IVIG are predictive of non-response to rescue therapy for refractory KD. For patients at high risk of non-response, IVIG plus prednisolone, or even further intensified rescue therapy regimens may be preferable.