Development and validation of an early predictive model for coronary artery lesions in incomplete Kawasaki disease

Introduction

Kawasaki disease (KD) is an acute eruptive and febrile illness marked by systemic vasculitis, predominantly affecting children under 5 years of age (1). While its incidence remains consistently low in Europe and North America, KD demonstrates a high and steadily rising prevalence among children in East Asia (2). The condition is characterized by immune-mediated inflammation of small- and medium-sized vessels, primarily involving the coronary arteries. Coronary artery lesions (CALs) represent the most severe complication and serve as a key determinant of long-term adverse outcomes in pediatric KD cases. KD has now surpassed rheumatic fever as the leading cause of acquired heart disease in children and contributes to elevated risks of ischemic heart disease in adulthood (3). The current standard treatment—a single high-dose intravenous immunoglobulin (IVIG) combined with oral aspirin—reduces CAL incidence from 25% to 3–5%, markedly suppresses systemic inflammation, and mitigates CAL progression (4,5).

At present, the diagnosis of KD is mainly guided by a series of characteristic signs and symptoms. However, some children showed only one to three main features except fever, and were classified as having incomplete Kawasaki disease (IKD). With the deepening of people’s understanding of KD, IKD has been gradually discovered. It is worth noting that IKD is not a mild KD (6). Due to its few symptoms and signs, IKD is often not recognized early and the diagnosis is delayed, and even misdiagnosed as other diseases, which leads to CAL. However, due to the incomplete clinical characteristics, lack of “gold standard” for diagnosis, and lack of specific laboratory indicators, IKD is often not recognized and treated in time in clinical practice, which leads to the occurrence of CAL. The prevalence of CAL increases in patients with IKD. Another study found that patients with IKD had a lower rate of coronary artery aneurysm resolution (7).

Clinical prediction models integrate multiple clinical features or auxiliary examinations as predictors to estimate the diagnosis or prognosis of a disease, thereby providing a rational basis for medical decision-making in its early stages (8). While predictive models for CAL in KD have become a research hotspot, studies focusing on IKD complicated by CAL remain limited. This study compared the clinical data of IKD and CKD groups, and described the clinical characteristics of IKD from the aspects of general condition, epidemiology, clinical characteristics, laboratory indicators, IVIG treatment and CAL complications. To improve the early recognition and diagnosis of IKD by clinicians, this study aimed to explore the risk factors for CAL in local IKD patients and establish a predictive model. This model provides a basis for early risk assessment and individualized treatment decision-making in children. More aggressive treatment measures can be implemented for high-risk children identified by the model. The prediction model holds important clinical significance for the diagnosis and treatment of IKD, reducing the incidence of CAL and improving the prognosis of children. We present this article in accordance with the TRIPOD reporting checklist (available at https://tp.amegroups.com/article/view/10.21037/tp-2025-531/rc).

Methods

Participants

Children with KD who were hospitalized in The Affiliated Xuzhou Children’s Hospital of Xuzhou Medical University from January 2021 to December 2023 were selected as the research objects. The epidemiological and clinical data of the children were obtained by querying the electronic medical record system. Clinical data from 191 pediatric IKD inpatients at The Affiliated Xuzhou Children’s Hospital of Xuzhou Medical University between January 2021 and December 2023 were retrospectively analyzed to develop a predictive model. An independent dataset comprising 41 IKD cases hospitalized from January to June 2024 was employed for model validation. Eligibility was determined according to the IKD diagnostic criteria outlined in the 6th edition of the KD guidelines issued by the Japanese Circulation Society (JCS) in 2020 (9). Exclusion criteria included: (I) prior administration of IVIG or hormone therapy at other hospitals; (II) hospitalization during the convalescent phase of KD for management of coronary sequelae; (III) refusal to undergo recommended auxiliary examinations; (IV) refusal to adhere to standardized KD treatment protocols; (V) presence of alternative febrile illnesses, including multisystem inflammatory syndrome in children (MIS-C), measles, scarlet fever, juvenile idiopathic arthritis, Stevens-Johnson syndrome, and infections by specific pathogens [e.g., adenovirus, enterovirus, rickettsia, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)]; and (VI) long-term use of corticosteroids or immunosuppressive agents. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of The Affiliated Xuzhou Children’s Hospital of Xuzhou Medical University (approval No. 2024-06-09-k09) and individual consent for this retrospective analysis was waived.

Allocation criteria

The training and test sets by time period for temporal validation. According to the 2020 JCS/Japanese Society of Cardiovascular Surgery (JSCS) guidelines on the diagnosis and management of cardiovascular sequelae in KD (10), a body surface area-corrected Z-score greater than or equal to 2.5 is defined as a CAL. A total of 191 IKD cases were stratified into a CAL group (n=89) and a non-CAL group (n=102). These data were utilized as the training cohort for constructing a predictive model for CAL occurrence in IKD. Additionally, 41 IKD pediatric inpatients admitted between January and June 2024 were categorized, following identical criteria, into a CAL group (n=16) and a non-CAL group (n=25), forming the testing cohort for model validation.

Clinical data collection

All children with KD were evaluated by a senior cardiologist through medical history inquiry, clinical feature observation, physical examination, as well as ordering auxiliary tests and formulating diagnostic and therapeutic plans. The collected clinical data included the following: (I) general information: patient hospitalization ID, name, gender, age, and disease onset timing; (II) clinical presentation included fever duration; peripheral extremity alterations—such as erythema and edema during the acute stage or periungual desquamation in the convalescent phase; bilateral conjunctival injection; oropharyngeal changes including erythematous lips and strawberry tongue; rash; localized erythema and induration at the Bacillus Calmette-Guérin (BCG) inoculation site; perianal desquamation; and cervical lymphadenopathy; (III) therapeutic interventions comprised fever duration prior to IVIG administration, IVIG resistance, glucocorticoid application, and infliximab therapy; (IV) laboratory parameters, assessed using samples obtained upon admission, include: C-reactive protein (CRP), procalcitonin (PCT), erythrocyte sedimentation rate (ESR), white blood cell count (WBC), platelet (PLT) count, neutrophil percentage (Neu%), hemoglobin (Hb), aspartate aminotransferase (AST), alanine aminotransferase (ALT), albumin (ALB), total bilirubin (TB), direct bilirubin (DB), alkaline phosphatase (ALP), lactate dehydrogenase (LDH), creatine kinase (CK), interleukin 2 (IL2), interleukin 6 (IL6), interleukin 10 (IL10). (V) Coronary artery involvement was evaluated through echocardiographic findings during hospitalization.

Treatment

The initial regimen comprised IVIG at 2 g/kg administered via continuous infusion over 10–12 hours, in combination with oral aspirin at a dosage of 30–50 mg/(kg·d). Upon resolution of fever, normalization of inflammatory markers, and alleviation of clinical manifestations, aspirin was tapered to 3–5 mg/(kg·d) administered at draught.

Statistical analysis

Statistical analyses for the IKD and CKD groups were conducted using SPSS 26.0. To handle missing values, all of which had a missing rate of less than 1% for any variable, we used Multiple Imputation by Chained Equations (MICE) to avoid potential bias associated with complete-case analysis. The Kolmogorov-Smirnov test assessed the normality of measurement data. Data conforming to a normal distribution were summarized as mean ± standard deviation and compared between groups using the independent samples t-test. Non-normally distributed data were presented as median (P₂₅–P₇₅) and analyzed via the Mann-Whitney U test. Categorical variables were described using absolute counts and percentages (%), with inter-group differences evaluated through the χ² test. Statistical significance was defined by a P value <0.05. Additional analyses on the training set were performed using R version 4.0.4 (https://www.R-project.org). Least absolute shrinkage and selection operator (LASSO) regression was applied for variable selection, incorporating 10-fold cross-validation to optimize model performance. Variables identified through this method were incorporated into a predictive model, visualized via a nomogram. Model performance was assessed through the area under the curve (AUC), 95% confidence interval (CI), sensitivity, and specificity derived from receiver operating characteristic (ROC) curve analysis. Validation of the model was subsequently conducted using the testing set.

Results

General data analysis

A total of 814 KD children were enrolled in this study, with a male-to-female ratio of 1.65:1, comprising 191 cases (23.5%) in the IKD group and 623 cases (76.5%) in the CKD group. In the CKD group, 400 males and 223 females were included, yielding a male-to-female ratio of 1.79:1, while the IKD group consisted of 107 males and 84 females, resulting in a male-to-female ratio of 1.27:1. A statistically significant difference in gender distribution was observed between the two groups (Z=4.169, P<0.05), with the CKD group having a higher proportion of male children compared to the IKD group. The median age at disease onset for the IKD group was 1.8 years, with 55 cases (29%) occurring in children aged 0-1 years, 104 cases (54%) in those aged 1–5 years, and 32 cases (17%) in children aged >5 years. In the CKD group, the median age at disease onset was 2.0 years, including 170 cases (27%) in the 0–1 year age range, 411 cases (66%) in the 1–5 years range, and 42 cases (7%) in children >5 years. No statistically significant difference in the median age at disease onset and 0–1 year age range was found between the two groups (P>0.05). The incidence of CKD was higher in children aged 1–5 years (P<0.05), while the incidence of IKD was greater in those over 5 years of age (P<0.05) (Table 1). The overall incidence of KD reached its peak in January, July, and August, with both the IKD and CKD groups experiencing peak incidences during summer (June to August) and winter (December to February) (Figure 1).

Figure 1 The distribution of onset time between the CKD group and the IKD group. CKD, complete Kawasaki disease; IKD, incomplete Kawasaki disease.

Clinical characteristics between CKD group and IKD group

Among the 191 children with IKD, fever was present in all cases, with a median duration of 5 days at admission. The incidence of other clinical characteristics, ranked from highest to lowest, included non-purulent cervical lymphadenopathy (147 cases, 77%), erythematous, fissured lips, and strawberry tongue (100 cases, 52%), conjunctival injection (97 cases, 51%), rash (51 cases, 27%), induration and edema of the hands and feet, along with periungual desquamation (45 cases, 24%), perianal desquamation (19 cases, 10%), and erythema at the BCG inoculation site (17 cases, 9%). Among the 623 children with CKD, fever was also observed in all cases, with the median fever duration at admission being 5 days. The incidence of additional clinical characteristics, ranked from highest to lowest, was as follows: non-purulent cervical lymphadenopathy (603 cases, 97%), erythematous, fissured lips, and strawberry tongue (599 cases, 96%), conjunctival injection (595 cases, 96%), rash (459 cases, 74%), induration and edema of the hands and feet, along with periungual desquamation (429 cases, 69%), perianal desquamation (209 cases, 34%), and erythema at the BCG inoculation site (123 cases, 20%). While the median duration of fever was identical in both groups, the IKD group exhibited a significantly longer fever duration compared to the CKD group (χ²=−2.69, P<0.05). The incidence of other clinical characteristics was significantly higher in the CKD group than in the IKD group, with statistically significant differences observed (P<0.05). Beyond fever, the three most common symptoms in IKD—namely, cervical lymphadenopathy, bilateral conjunctival injection, and oropharyngeal changes (including Erythematous, lip fissures and strawberry tongue)—hold significant clinical value for diagnosis (Table 2).

Laboratory test results for CKD group and IKD group

The test results for Neu%, PCT, ALT, AST, TB, DB, ALP, IL2, IL6, and IL10 in the IKD group were significantly lower than those in the CKD group (P<0.05). Conversely, the CK index in the IKD group was higher than that in the CKD group, with a statistically significant difference (P<0.05). No significant differences were observed between the two groups for CRP, ESR, WBC, PLT, HB, ALB, and LDH levels (Table 3).

Development for CAL in IKD

A total of 191 children diagnosed with IKD were enrolled (Figure 2), comprising 89 cases (46.6%) in the CAL group and 102 (53.4%) in the non-CAL group. Compared with the non-CAL group, the CAL group exhibited a younger age distribution, prolonged febrile course, a higher proportion of male patients, and lower incidence rates of bilateral conjunctival injection and cervical lymphadenopathy. Laboratory findings revealed reduced ESR and Neu%, alongside elevated AST levels in the CAL group relative to the non-CAL group. Detailed comparisons were presented in Table 4. Clinical parameters for the subset of 41 children in the test cohort, stratified into CAL and non-CAL groups, were summarized in Table 5.

Figure 2 Flowchart of the predictive model for CAL in IKD. CAL, coronary artery lesion; IKD, incomplete Kawasaki disease; IVIG, intravenous immunoglobulin; KD, Kawasaki disease.

Identification of predictors for CAL risk in IKD

A total of 27 clinical and laboratory variables were subjected to LASSO regression analysis, including demographic data (gender and age), principal clinical signs (fever duration, bilateral conjunctival injection, cervical lymphadenopathy, erythematous, fissured lips, and strawberry tongue, rash, induration and edema of the hands and feet and periungual desquamation, perianal desquamation, and erythema at the BCG inoculation site), and inflammatory and biochemical markers (CRP, ESR, WBC, Neu%, PLT, HB, PCT, ALB, ALT, AST, LDH, TB, DB, ALP, IL2, IL6, and IL10). Prior to analysis, all continuous variables were standardized (mean-centered and scaled to unit variance) to ensure an equitable penalty in the regularization process. The optimal regularization parameter (λ) was determined via 10-fold cross-validation, which minimized the binomial deviance. Applying the conservative one-standard-error rule to enhance model parsimony and stability, the final model with a λ of 0.0657 retained eight predictors with non-zero coefficients: gender, fever duration, conjunctival injection, cervical lymphadenopathy, ESR, Neu%, ALT, and AST (Figure 3).

Figure 3 Roadmap of standardized regression coefficients and cross-validation curves of variables in the LASSO regression model. (A) LASSO coefficient profile of the included predictive variables. (B) 10-fold cross-validation curve. LASSO, least absolute shrinkage and selection operator.

Logistic regression analysis of predictors for CAL in IKD

In the logistic regression analysis, eight variables identified through LASSO regression were evaluated to determine their regression coefficients and relative risks associated with CAL in IKD. As presented in Table 6, variables with P values >0.05—namely ESR, Neu%, ALT, and AST—did not demonstrate statistical significance. Conversely, fever duration, gender, conjunctival injection, and cervical lymphadenopathy yielded P values <0.05, indicating statistical significance. The results indicated that prolonged fever, male, absence of conjunctival injection, and absence of cervical lymphadenopathy independently predicted increased risk for CAL in IKD.

Construction of prediction model for CAL in IKD (nomogram)

Eight variables identified through LASSO regression were incorporated into a predictive model visualized via a nomogram (Figure 4). Each variable corresponded to a score on the upper scale (ranging from 0 to 100 points). The sum of these individual scores yielded a total score (0–240), which was then mapped onto the bottom scale (0.05–0.95) to estimate the probability of CAL occurrence in IKD cases.

Figure 4 Individual risk probability plot of CAL in IKD (nomogram). Gender (1, male; 0, female); conjunctival injection, cervical lymph node (1, occurrence; 0, non-occurrence). ALT, alanine aminotransferase; AST, aspartate aminotransferase; CAL, coronary artery lesion; ESR, erythrocyte sedimentation rate; IKD, incomplete Kawasaki disease; Neu%, neutrophil percentage.

The 8 predictors determined by LASSO regression analysis were used to establish a prediction model, and a nomogram was drawn (Figure 4). Regression coefficients and relative risk estimates were derived through logistic regression analysis. The final model was expressed by the equation:

Logit(P=CAL) = 1.12 * gender + 0.209 * fever duration − 1.156 * conjunctival injection − 1.471 * cervical lymphadenopathy − 0.015 * ESR-0.012 * Neu% + 0.006 * ALT + 0.009 * AST

This equation enabled the quantitative estimation of CAL risk among IKD patients. Positive coefficients (gender, fever duration, ALT, and AST) indicated variables associated with increased CAL probability, whereas negative coefficients (bilateral conjunctival injection, cervical lymphadenopathy, ESR, and Neu%) corresponded to a reduced likelihood.

Evaluation of nomogram predictive performance in the training set

The Nomogram developed to predict CAL risk in children with IKD demonstrated favorable predictive performance in the training set, yielding an AUC of 0.817 (95% CI: 0.757–0.878), with sensitivity and specificity values of 83.1% and 71.6%, respectively (Figure 5).

Figure 5 ROC of the nomogram in the training set. AUC, area under the curve; ROC, receiver operating characteristic.

Validation of nomogram predictive performance in the test set

Validation with the testing set confirmed the model performance, achieving an AUC of 0.847 (95% CI: 0.720–0.975), alongside sensitivity and specificity of 75.0% and 88.0%, respectively (Figure 6).

Figure 6 ROC of the nomogram in the testing set. AUC, area under the curve; ROC, receiver operating characteristic.

The calibration of the prediction model was measured using Hosmer-Lemeshow statistical test and calibration plot. The calibration curves of the training set and the validation set of the prediction model showed a good fit (Figures 7,8). Ideal stands for ideal reference line. Apparent represents the model calibration curve and Bias-corrected represents the 1,000 Bootstrap results.

Figure 7 Hosmer-Lemeshow test of the nomogram in the training set.

Figure 8 Hosmer-Lemeshow test of the nomogram in the testing set.

Discussion

Long-term outcomes in KD are primarily determined by the degree of coronary artery involvement. Risk factors and predictive modeling remain central themes in KD research. A prevailing limitation of existing models lies in their restricted external applicability, often yielding suboptimal predictive accuracy when applied beyond the original study cohorts. This limitation is frequently attributed to insufficient sample sizes, substandard treatment of missing data, and inadequate evaluation of model performance (11). To address these challenges, recent investigations have adopted expanded datasets (12), integrated deep learning methodologies (13), and incorporated some emerging biomarkers (14) to enhance the generalizability and robustness of predictive frameworks.

Previous studies have identified several risk factors associated with CAL in the context of KD, including incomplete clinical presentation, prolonged fever duration, age under 6 months or over 9 years, male, hypoalbuminemia, low Hb, hyponatremia, thrombocytosis, elevated ALT and AST levels, and resistance to IVIG therapy (15-22). A Korean investigation (23) observed a temporal trend toward milder clinical features in KD and an increasing prevalence of IKD. While multiple reports have indicated a higher incidence of CAL in IKD compared to CKD (24-28), limited evidence exists regarding specific predictors and risk assessment models tailored to the CAL in IKD. In the present analysis, clinical records from 191 IKD cases were examined, and eight variables—gender, fever duration, conjunctival injection, cervical lymphadenopathy, ESR, Neu%, ALT, and AST—were identified via LASSO regression as independent predictors of CAL. A predictive model incorporating these variables demonstrated favorable performance during validation, offering a valuable tool for early risk stratification and informed clinical decision-making in IKD management.

No unified criteria have yet been established for identifying risk factors associated with CAL in IKD. This analysis identified prolonged fever duration, male sex, absence of conjunctival injection, and absence of cervical lymphadenopathy as independent predictors of CAL in IKD. Extended febrile periods have been linked to an elevated likelihood of CAL (4,22). The most recent AHA scientific statement (I) recommends prompt diagnosis of IKD to mitigate the risk of CAL progression. A meta-analysis assessing CAL risk in KD (29) reported a higher incidence of CAL in male patients presenting with fewer clinical signs. Yeo et al. (15) examined Korean infants with KD under 1 year of age and demonstrated that incomplete clinical manifestations and extended total fever duration were significantly correlated with CAL development. Nomura et al. (30), in a study involving 13,770 IKD patients, demonstrated that early IVIG administration reduced CAL incidence. Among IKD cases, those exhibiting only one or two clinical signs had a markedly higher CAL rate (6.7%) compared to those with three or four symptoms (2.6%) (P<0.0001), indicating a greater risk of CAL in presentations with minimal clinical features. CAL in KD reflects a multifaceted process involving both acute and chronic vascular inflammation (31). Incomplete clinical presentation in IKD often delays diagnosis, thereby extending febrile duration, perpetuating systemic inflammatory responses, and contributing to vascular injury.

Del Principe et al. (32) identified a potential mechanism in the pathogenesis of KD, suggesting that in genetically predisposed children, exposure to specific pathogens may trigger aberrant immune activation and subsequent inflammatory cascades. Although inflammatory biomarkers have garnered increasing clinical attention in diagnosing IKD, limited evidence exists regarding their predictive utility for early CAL identification. In the present study, LASSO regression highlighted low ESR, Neu%, and increased ALT and AST as variables associated with a heightened risk of CAL in IKD. ESR, as a highly sensitive inflammatory marker, exhibits an early and rapid rise during KD onset. According to the AHA scientific statement (4), an ESR ≥40 mm/h is recognized as a key laboratory parameter in differentiating IKD. NLR serves as an index of the dynamic interplay between systemic inflammation and immune regulation in KD. Neutrophils, predominant in the initial disease phase, contribute to endothelial injury through activation mechanisms that promote KD-associated vasculitis (33,34). Ha et al. (35) reported that IKD and CKD exhibited comparable profiles regarding complete blood count and acute-phase reactants, with both conditions marked by a sustained elevation of nonspecific inflammatory markers. Reduced ESR and Neu% in IKD may reflect delayed diagnosis due to atypical clinical presentation, potentially capturing a post-peak inflammatory phase. Some researchers (36) argue that the high CAL rate in IKD is not entirely due to treatment delay, underscoring the necessity of early detection and treatment in all suspected cases. However, their findings speculate that CKD and IKD may be distinct entities with differing pathophysiology, possibly necessitating distinct clinical pathways. The etiology of hepatic impairment in KD remains unresolved, though inflammation is considered a contributing factor. In a cohort of 37 autopsied KD cases, hepatic vasculitis was identified in 6 instances, with histopathology revealing sinus tract and portal infiltration by inflammatory cells (primarily restricted to the bile duct lumen), Kupffer cell hypertrophy, fatty degeneration, and extensive hepatic congestion (37). Elevated ALT and AST suggest intensified systemic inflammation. Prior studies have demonstrated a linear association between systemic immune inflammation index and CALs in KD patients (34). The current analysis identified increased ALT and AST levels, alongside reduced ESR and Neu%, as variables associated with a higher probability of CAL in IKD, potentially indicating that ALT and AST remain elevated during the protracted inflammatory course.

Son et al. (38) developed a CAL risk prediction model for KD in North America by integrating demographic, laboratory, and echocardiographic parameters, achieving significant advancement in risk identification and demonstrating applicability to external populations (39,40). Drawing on this approach, previously recognized risk indicators can be translated into a clinically applicable predictive tool. The eight CAL-associated variables in IKD identified via LASSO regression are readily obtainable during routine clinical evaluation. The resulting model exhibited favorable discriminatory capacity in both the training and test datasets, with sensitivities of 83.1% and 75.0%, and specificities of 71.6% and 88.0%, respectively. Future external validations in independent medical institutions or across diverse geographic cohorts are warranted to assess the model’s generalizability.

However, the study’s design is limited by its single-center setting and relatively homogeneous sample population. Future research involving multi-center collaborations is recommended to broaden the sample base, enhance the generalizability of findings, and optimize the predictive model. The validation cohort included only 41 patients, and this modest sample size may limit the reliability of the validation findings. Second, some data were missing during the study. Although statistical methods such as stepwise regression analysis were used to handle it, it was still difficult to completely avoid the bias caused by this. The predictive value of this model for IKD combined with CAL still needs to be further verified by larger sample size, multi-center and prospective studies. At the same time, the number of indicators included in this study was limited, and more relevant indicators and risk factors should be included in the future to make the study more convincing. In this study, a retrospective study was used to establish the model, and some relevant factors could not be explored and found in depth. Limitations related to echocardiographic assessment should be considered. The non-standardized timing of assessments may have impacted the accurate tracking of CAL evolution, and additionally, the model might lack the specificity to differentiate transient dilation from persistent CAL. The lack of quantitative evaluation for inter-observer variability also poses a potential bias in classifying borderline cases, all of which could influence the model’s performance.

Conclusions

In this study, LASSO regression was employed to identify variables associated with CAL in IKD and to construct a predictive model for early identification of high-risk pediatric cases. The model enables stratification of patients to inform individualized therapeutic strategies. In children identified as high-risk, early initiation of IVIG, glucocorticoids, or immunosuppressants may contribute to improved clinical outcomes.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-531/rc

Data Sharing Statement: Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-531/dss

Peer Review File: Available at https://tp.amegroups.com/article/view/10.21037/tp-2025-531/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tp.amegroups.com/article/view/10.21037/tp-2025-531/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was approved by the Ethics Committee of The Affiliated Xuzhou Children’s Hospital of Xuzhou Medical University (approval No. 2024-06-09-k09) and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Jone PN, Tremoulet A, Choueiter N, et al. Update on Diagnosis and Management of Kawasaki Disease: A Scientific Statement From the American Heart Association. Circulation 2024;150:e481-500. [Crossref] [PubMed]
2. Lin MT, Wu MH. The global epidemiology of Kawasaki disease: Review and future perspectives. Glob Cardiol Sci Pract 2017;2017:e201720. [Crossref] [PubMed]
3. Ae R, Makino N, Kosami K, et al. Epidemiology, Treatments, and Cardiac Complications in Patients with Kawasaki Disease: The Nationwide Survey in Japan, 2017-2018. J Pediatr 2020;225:23-29.e2. [Crossref] [PubMed]
4. McCrindle BW, Rowley AH, Newburger JW, et al. Diagnosis, Treatment, and Long-Term Management of Kawasaki Disease: A Scientific Statement for Health Professionals From the American Heart Association. Circulation 2017;135:e927-99. [Crossref] [PubMed]
5. Mori M, Miyamae T, Imagawa T, et al. Meta-analysis of the results of intravenous gamma globulin treatment of coronary artery lesions in Kawasaki disease. Mod Rheumatol 2004;14:361-6. [Crossref] [PubMed]
6. Manlhiot C, Christie E, McCrindle BW, et al. Complete and incomplete Kawasaki disease: two sides of the same coin. Eur J Pediatr 2012;171:657-62. [Crossref] [PubMed]
7. Sonobe T, Kiyosawa N, Tsuchiya K, et al. Prevalence of coronary artery abnormality in incomplete Kawasaki disease. Pediatr Int 2007;49:421-6. [Crossref] [PubMed]
8. Wolff RF, Moons KGM, Riley RD, et al. PROBAST: A Tool to Assess the Risk of Bias and Applicability of Prediction Model Studies. Ann Intern Med 2019;170:51-8. [Crossref] [PubMed]
9. Kobayashi T, Ayusawa M, Suzuki H, et al. Revision of diagnostic guidelines for Kawasaki disease (6th revised edition). Pediatr Int 2020;62:1135-8.
10. Fukazawa R, Kobayashi J, Ayusawa M, et al. JCS/JSCS 2020 Guideline on Diagnosis and Management of Cardiovascular Sequelae in Kawasaki Disease. Circ J 2020;84:1348-407. [Crossref] [PubMed]
11. Huang H, Dong J, Wang S, et al. Prediction Model Risk-of-Bias Assessment Tool for coronary artery lesions in Kawasaki disease. Front Cardiovasc Med 2022;9:1014067. [Crossref] [PubMed]
12. Li C, Zhang H, Yin W, et al. Development and validation of a nomogram-based prognostic model to predict coronary artery lesions in Kawasaki disease from 6847 children in China. Comput Methods Programs Biomed 2025;260:108588. [Crossref] [PubMed]
13. Yang L, Shen X, Liu Y, et al. Development and Validation of KCPREDICT: A Deep Learning Model for Early Detection of Coronary Artery Lesions in Kawasaki Disease Patients. Pediatr Cardiol 2025; Epub ahead of print. [Crossref]
14. Wu Y, Wang S, Zhou Y, et al. Clinical indicators combined with S100A12/TLR2 signaling molecules to establish a new scoring model for coronary artery lesions in Kawasaki disease. PLoS One 2023;18:e0292653. [Crossref] [PubMed]
15. Yeo Y, Kim T, Ha K, et al. Incomplete Kawasaki disease in patients younger than 1 year of age: a possible inherent risk factor. Eur J Pediatr 2009;168:157-62. [Crossref] [PubMed]
16. Salgado AP, Ashouri N, Berry EK, et al. High Risk of Coronary Artery Aneurysms in Infants Younger than 6 Months of Age with Kawasaki Disease. J Pediatr 2017;185:112-116.e1. [Crossref] [PubMed]
17. Sabharwal T, Manlhiot C, Benseler SM, et al. Comparison of factors associated with coronary artery dilation only versus coronary artery aneurysms in patients with Kawasaki disease. Am J Cardiol 2009;104:1743-7. [Crossref] [PubMed]
18. Nakamura Y, Yashiro M, Uehara R, et al. Use of laboratory data to identify risk factors of giant coronary aneurysms due to Kawasaki disease. Pediatr Int 2004;46:33-8. [Crossref] [PubMed]
19. Manlhiot C, Yeung RS, Clarizia NA, et al. Kawasaki disease at the extremes of the age spectrum. Pediatrics 2009;124:e410-5. [Crossref] [PubMed]
20. Kim T, Choi W, Woo CW, et al. Predictive risk factors for coronary artery abnormalities in Kawasaki disease. Eur J Pediatr 2007;166:421-5. [Crossref] [PubMed]
21. Xia Y, Qiu H, Wen Z, et al. Albumin level and progression of coronary artery lesions in Kawasaki disease: A retrospective cohort study. Front Pediatr 2022;10:947059. [Crossref] [PubMed]
22. Jiang S, Li M, Xu K, et al. Predictive factors of medium-giant coronary artery aneurysms in Kawasaki disease. Pediatr Res 2024;95:267-74. [Crossref] [PubMed]
23. Kil HR, Yu JW, Lee SC, et al. Changes in clinical and laboratory features of Kawasaki disease noted over time in Daejeon, Korea. Pediatr Rheumatol Online J 2017;15:60. [Crossref] [PubMed]
24. Morishita KA, Goldman RD. Current Approach to the Evaluation and Management of Incomplete Kawasaki Disease in the Emergency Department. Pediatr Emerg Care 2020;36:537-41. [Crossref] [PubMed]
25. Jiao FY, Mu ZL. Diagnosis and treatment of incomplete Kawasaki disease in children. Zhongguo Dang Dai Er Ke Za Zhi 2023;25:238-43. [Crossref] [PubMed]
26. An HS, Kim GB, Song MK, et al. The occurrence of coronary artery lesions in Kawasaki disease based on C-reactive protein levels: a retrospective cohort study. Pediatr Rheumatol Online J 2021;19:78. [Crossref] [PubMed]
27. Bressieux-Degueldre S, Gradoux E, Di Bernardo S, et al. Complete and incomplete Kawasaki disease: Clinical differences and coronary artery outcome from a national prospective surveillance study in Switzerland. Front Pediatr 2023;11:1137841. [Crossref] [PubMed]
28. Cai WJ, Ding SG. Retrospective analysis of clinical characteristics and related influencing factors of Kawasaki disease. Medicine (Baltimore) 2022;101:e32430. [Crossref] [PubMed]
29. Mehta R, Balaraman AK, Shabil M, et al. Commentary on “Biomarkers and Related Factors for the Diagnosis, Risk of Coronary Artery Lesions, and Resistance to Intravenous Immunoglobulin in Kawasaki Disease: An Umbrella Review of Meta-Analyses”. Pediatr Cardiol 2025;46:242-3. [Crossref] [PubMed]
30. Nomura Y, Yashiro M, Masuda K, et al. Treatment change and coronary artery abnormality in incomplete Kawasaki disease. Pediatr Int 2020;62:779-84. [Crossref] [PubMed]
31. Orenstein JM, Shulman ST, Fox LM, et al. Three linked vasculopathic processes characterize Kawasaki disease: a light and transmission electron microscopic study. PLoS One 2012;7:e38998. [Crossref] [PubMed]
32. Del Principe D, Pietraforte D, Gambardella L, et al. Pathogenetic determinants in Kawasaki disease: the haematological point of view. J Cell Mol Med 2017;21:632-9. [Crossref] [PubMed]
33. Chang LS, Lin YJ, Yan JH, et al. Neutrophil-to-lymphocyte ratio and scoring system for predicting coronary artery lesions of Kawasaki disease. BMC Pediatr 2020;20:398. [Crossref] [PubMed]
34. Huang T, Peng Q, Zhang Y, et al. The Systemic Immune-Inflammation Index (SII) and coronary artery lesions in Kawasaki disease. Clin Exp Med 2024;24:4. [Crossref] [PubMed]
35. Ha KS, Jang GY, Lee J, et al. Laboratory Markers in Incomplete Kawasaki Disease according to Coronary Artery Outcome. Korean Circ J 2018;48:287-95. [Crossref] [PubMed]
36. Mossberg M, Mohammad AJ, Kahn F, et al. High risk of coronary artery aneurysm in Kawasaki disease. Rheumatology (Oxford) 2021;60:1910-4. [Crossref] [PubMed]
37. Amano S, Hazama F, Hamashima Y. Pathology of Kawasaki disease: II. Distribution and incidence of the vascular lesions. Jpn Circ J 1979;43:741-8. [Crossref]
38. Son MBF, Gauvreau K, Tremoulet AH, et al. Risk Model Development and Validation for Prediction of Coronary Artery Aneurysms in Kawasaki Disease in a North American Population. J Am Heart Assoc 2019;8:e011319. [Crossref] [PubMed]
39. La Vecchia A, Stracquadaino R, Mauri L, et al. Risk factors and scores for prediction of coronary artery aneurysms in Kawasaki disease: a European monocentric study. BMC Pediatr 2024;24:139. [Crossref] [PubMed]
40. Miyata K, Miura M, Kaneko T, et al. Evaluation of a Kawasaki Disease Risk Model for Predicting Coronary Artery Aneurysms in a Japanese Population: An Analysis of Post RAISE. J Pediatr 2021;237:96-101.e3. [Crossref] [PubMed]