Development of a nomogram prediction model for early identification of persistent coronary artery aneurysms in kawasaki disease

Due to the unsatisfactory efficiency of current predictive models of CAA lesions in KD, this study sought to establish a reliable nomogram prediction system to assess the possibilities of CAA persistence associated with KD. The study was a prospectively designed observational study, and all participants were enrolled from January 2016 to December 2020.

This prospectively designed, observational, single-center study included all patients with KD who visited West China Second University Hospital, Sichuan University from January 2016 to December 2020. Participants were followed up for at least 1 year; and during the follow-up, they underwent at least three echocardiographic assessments. This study was conducted under strict adherence to the STROBE statement. The Ethics Committee of the West China Second Hospital of Sichuan University approved this study (approval number 2014-034). Informed written consent was obtained from the parents after the nature of this study had been fully explained to them. All questionnaires were pretested and revised accordingly. Two pre-trained physicians collected data and followed up on participants. Any related questionnaires were double-checked to assure accuracy. All clinical data were confirmed by electronic medical records and the study’s follow-up database.

To recruit candidates for further analysis, we used the following inclusion criteria: (1) Confirmed diagnosis of KD by two physicians using the diagnostic criteria for complete or incomplete diagnostic standards recommended by the Japanese Circulation Society (2013) for diagnosis, treatment, and long-term management of KD; (2) identification of CAA either by transthoracic echocardiography or transcatheter angiography and identification of CAA onset within the acute or subacute phase of KD; (3) well-collected and completed programmed questionnaires, basic essential information, clinical manifestation, hematological examination results, therapeutic procedure, echocardiography results, and follow-up outcomes; (4) completed baseline hematological examinations during acute phase of KD before IVIG administration; (5) age of 2–6 years, which is the age range with the highest incidence of KD, to balance bias from high risk ages; and (6) absence of CAA lesions from previous episodes of KD, if patient had previous more than one episode of KD. The exclusion criteria included the following: (1) presence of cardiovascular malformations; (2) history of an autoimmune disease prior to KD onset; (3) treatment with anticoagulant or antiplatelet medication prior to KD onset; (4) history of cardiac surgeries; (5) suspicion of myocarditis prior to KD onset; (6) use of glucocorticoids prior to IVIG administration; (7) use of monoclonal antibodies, including tumor necrosis factor (TNF)-α and interleukin (IL)-6 antibodies; (8) diagnosis of macrophage activation syndrome or hemophagocytic lymph histiocytosis secondary to KD; (9) new onset of coronary artery lesions after persistent CAA; and (10) absence of echocardiographic records for ≥ 1 year.

All patients with KD were treated with high-dose IVIG (2 g/kg given as a single intravenous infusion) and 30–50 mg/kg/day high dose aspirin. Those with recrudescent or persistent fever ≥ 36 h after the first dose of IVIG infusion were treated with a second dose of 2 g/kg IVIG. Methylprednisolone (30 mg/kg/day for three consecutive days) followed by oral prednisone tapered over 7 days would be considered after the second IVIG administration. The first day of fever was defined as day 1 of disease onset, and IVIG resistance was defined as persistent or recurrent fever (temperature ≥ 38.0 °C orally) or presence of other clinical signs of KD for at least 36 h but not longer than 7 days after initial IVIG administration. The patients were discharged from the hospital after temperature remained normal for > 48 h and hematological examination findings returned to normal. Follow-up began from day of hospital discharge. All participants underwent echocardiographic evaluations after 2 weeks, 1 month, 2 months, 3 months, 6 months, and 12 months; findings from these evaluations were documented from the end of the subacute phase.

Two well-trained pediatric physicians performed all echocardiographic evaluations. The physicians involved in the evaluation were blinded to the clinical presentations of participants. The first echocardiographic evaluation was performed before IVIG administration, and the second evaluation was performed during the subacute phase or before hospital discharge. We defined the morphology and luminal dilation of the coronary artery, which were documented as aneurysm size, as the most severe condition between the acute and subacute phases of KD. At least two echocardiographic evaluations were performed to match the minimal requirement for first-month follow-up. Additionally, the first observation of pre-existing CAA recovery to a normal size was considered as the time for regression. If CAA persisted for 12 months, it was defined as a persistent lesion. Based on the Japanese Circulation Society criteria [24], CAA was defined as a coronary artery branch internal luminal diameter of > 3 mm in a child under the age of 5 years, an arterial diameter of > 4 mm in a child ≥ 5 years of age, or when an arterial segment was 1.5 times its adjacent segment. The CAAs were classified as small aneurysms (localized dilatation with ≤ 4 mm internal diameter), medium aneurysms (aneurysms with an internal diameter from > 4 mm to < 8 mm), and giant aneurysms (aneurysms with an internal diameter of ≥ 8 mm).

All parameters involved were used to predict the risk of CAA persistence after 1 year of follow-up. First, the basic clinical characteristics and hematological examination results of participants were recorded (Table 1). Thereafter, univariable analysis was performed between patients with CAA persistence and patients with regression. Multivariable analysis was completed using logistic regression to identify the independent factors among the significant results obtained from the univariable analysis. Nomograms were constructed based on findings from the univariable and multivariable analyses. The final prognostic nomogram model was selected using a backward step-down selection process and the Akaike information criterion. Thereafter, receiver operating characteristic curves (ROC) were used to identify the predictive value of the risk factors and nomogram scores that were found by comparison. Decision curve analysis (DCA) was performed to evaluate the efficacy of nomogram prediction. Finally, calibration plot was used to test the accuracy of Multi_nomo and Uni_nomo scores.

All data analyses were performed using SPSS 22.0 (SPSS Inc. Chicago, Illinois, United States of America). Continuous variables are presented as mean ± standard deviation, whereas categorical variables are expressed as frequencies. Differences between two groups were assessed using independent t-tests or Mann-Whitney U tests for continuous variables and chi-squared test or Fisher’s exact test for categorical variables. ROC analysis was used to determine the predictive validity of candidate risk factors for persistent CAA. Statistical significance was defined by p-values < 0.05. Nomograms and DCA were established and performed, respectively, using the rms and rmda packages of R version 3.3.2 (http://​www.​r-project.​org).