Investigators establish 2 predictive models, which had excellent performance, and included 90 children.
Age, intravenous immunoglobulin (IVIG) resistance at the first episode, and some laboratory variables may be risk factors for IVIG resistance in children with recurrent Kawasaki disease (KD), according to the results of a study published in the Journal of Inflammation Research.
Additionally, investigators established 2 predictive models for IVIG resistance in recurrent KD, which had excellent performance.
Investigators included a total of 90 children with recurrent KD. Of the 90 children, 16 were IVIG-resistant.
They found that the individuals with IVIG resistance had higher age and IVIG resistance likely at the first episode. Those individuals also had increased C-reactive protein (CRP) and direct bilirubin levels, international normalized ratio, neutrophil count and percentage, and prothrombin time.
The individuals also had decreased lymphocyte count and percentage and serum sodium levels.
Investigators found that 5 variables, age, CRP levels, IVIG resistance at first episode, lymphocytes count, and serum sodium levels, were selected by standard Lasso (ILasso) and 4 variables, including age, CRP levels, IVIG resistance, and neutrophil percentage were selected by group Lasso (gLasso).
In the gLasso analysis, age at the first episode were associated with the outcome in a non-linear way. However, there was a linear relationship observed for CRP.
Investigators found that the risk of IVIG resistance at recurrence increased quickly with age at the first episode before about 20 months. It remained similar until 60 months and then increased again.
Receiver operating characteristics (ROC) curves suggested that both models had similar excellent discrimination in both the training cohort, as well as the validation cohort.
Investigators performed variable selection and coefficient estimations using gLasso and ILasso.
The continuous variables had linear relationships to outcomes in the ILasso analysis, while they had potential non-linear relationships with the outcome in the gLasso analysis, investigators said.
Additionally, they had non-linear relationships in the expanded to restricted cubic splines in gLasso.
Additionally, the Hosmer-Lemeshow tests suggested the 2 models demonstrated a good calibration. Also, the 2 nomograms were constructed to facilitate the potential application of the 2 models.
External validation should be performed before clinical use, investigators said.
The investigators gathered clinical and laboratory data at recurrence and compared data between individuals with and without IVIG resistance. The individuals were randomly divided into training and validation cohorts for model and development and model validation.
Additionally, model performance was evaluated by calibration and ROC curves and Hosmer-Lemeshow tests.
KD is rare, so the sample size was smaller. Children with recurrent KD were included during a long observation period of 12 years in a pediatric hospital in China.
This study was potentially the first that devolved and validated predictive models for IVIG resistance in children with KD.
However, there were a few limitations to the study, investigators said.
The biggest was the size of the sample. Additionally, the study was retrospective and performed in a single center with only internal validation.
Reference
Chen X, Gao L, Zhen Z, Wang Y, et al. Risk factors and predictive models for intravenous immunoglobulin resistance in children with recurrent Kawasaki disease. J Inflamm Res. 2022;15:2877-2889. doi:10.2147/JIR.S360802