Kawasaki disease (KD) is a systemic vasculitis primarily affecting medium-sized arteries, especially the coronary arteries, and is a major cause of acquired heart disease in children in developed countries (1, 2). Combining high-dose intravenous immunoglobulin (IVIG) therapy with aspirin has been established as the first-line therapy for KD and can effectively reduce the occurrence rate of coronary artery lesions (CALs), from 20 to 25% to approximately 2–4% (3). Nevertheless, about 10–20% of KD patients are unresponsive to IVIG therapy, increasing their risk of CALs (4). Thus, early prediction of IVIG resistance is crucial, as these patients may benefit from prompt intensified therapy.

While the cause of KD is not yet understood, systemic inflammatory responses are pivotal in its pathogenesis and progression. Albumin, once primarily seen as an indicator of nutritional status, is now also recognized as a protein in the acute inflammatory response (5). Increasing evidence highlights the interconnection between inflammation and nutrition. Serum albumin levels were found to have a negative correlation with inflammation. Studies have indicated that hypoalbuminemia is commonly observed in patients with KD during the acute phase, primarily resulting from increased vascular permeability and serum albumin leakage, highlighting changes in albumin-derived markers (6–9). In recent years, several novel inflammatory and nutritional indices have been proposed, including the C-reactive protein-albumin-lymphocyte (CALLY) index, C-reactive protein to albumin ratio (CAR), neutrophil-to-albumin ratio (NAR), neutrophil percentage-to-albumin ratio (NPAR) and prognostic nutritional index (PNI). These markers integrate multiple clinical evaluation parameters and provide more valuable information than individual markers alone (10–12). These inflammatory factors were initially mostly used to assess the prognosis of various cancers. Such as in 2021, the CALLY index was first applied to the prognosis research of hepatocellular carcinoma, and its correlation was confirmed (13). In 2016, NPAR was first applied to assess the prognosis of rectal cancer (14). Currently, various albumin-derived nutritional inflammation indices have been linked to the risk and severity of several diseases, such as tumors, as well as neurological, gastrointestinal, and respiratory diseases (15–18). Their affordability and widespread accessibility in daily clinical practice have increased their popularity, as a higher ratio frequently suggests a poor prognosis. However, to our knowledge, there are few studies on CAR, NPAR, and PNI, and no studies on the CALLY index and NAR in KD patients. While prior studies have examined individual inflammatory markers (e.g., CRP), this is the first to evaluate composite nutritional-inflammatory indexes (CALLY, CAR, NAR, NPAR, and PNI) in predicting IVIG resistance. Therefore, our study aimed to determine if albumin-derived markers (CALLY index, CAR, NAR, NPAR, and PNI) integrating inflammation and nutrition could be the significant predictors for IVIG resistance in KD patients.

After excluding 77 cases based on the predefined exclusion criteria (Figure 1), 716 patients with KD were enrolled, consisting of 429 (59.9%) males and 287 (40.1%) females. The median age was 2.2 (1.2, 3.7) years old. Among the included individuals, 78 (10.9%) and 638 (89.1%) were diagnosed with IVIG-resistant KD and IVIG-responsive KD, respectively.

As shown in Table 1, no significant differences were observed between the IVIG-responsive group and IVIG-resistant group regarding sex, age, the occurrence of incomplete KD, and the other four typical clinical manifestations of KD, except extremity changes (p > 0.05). The IVIG-resistant group showed a significantly higher incidence of extremity changes, irritability, jaundice, tachypnea, aseptic meningitis, and KDSS than the IVIG-responsive group (all p < 0.05).

The laboratory findings are shown in Table 2. The IVIG-resistant group exhibited notably elevated CRP, AST and bile acids levels, alongside reduced lymphocytes, hemoglobin, albumin and serum sodium levels compared to the IVIG-responsive group (all p < 0.05). These findings aligned with earlier studies (21–23). Furthermore, it was noted that in the IVIG-resistant group, CAR, NAR and NPAR were significantly higher, whereas the CALLY index and PNI were significantly lower compared to IVIG-responsive group (Table 2 and Figure 2). However, the other laboratory parameters did not differ significantly between the two groups (all p > 0.05).

This study investigated the potential role of albumin-based biomarkers in contributing to IVIG responsiveness in KD. The potential predictive values of albumin-based markers including the CALLY index, CAR, NAR, NPAR and PNI, which were shown to be different between the two groups, were assessed in the models. To avoid collinearity, albumin-based markers were incorporated and assessed in the separate statistical models (Supplementary Tables 1–5), Among all the models, variables of Days of IVIG at initiation, Tachypnea, Hemoglobin, and Serum sodium were statistically significant in the binary logistic regression analysis. Therefore, we included the four variables to construct the Base Model, while Models 1–5 were, respectively, formed by incorporating CALLY index, CAR, NAR, NPAR and PNI into the Base Model to compare whether the predictive value of the model increased after the inclusion of these novel nutritional inflammation indexes, respectively. The CALLY index (OR: 0.072, 95% CI: 0.009–0.577, p = 0.013), CAR (OR: 1.207, 95% CI: 1.070–1.361, p = 0.002), NAR (OR: 6.073, 95% CI: 1.610–22.916, p = 0.008), NPAR (OR: 2.535, 95% CI: 1.648–3.898, p < 0.001), and PNI (OR: 0.953, 95% CI: 0.927–0.979, p = 0.001) were independent predictors for IVIG-resistance (Table 3, the brief table and Supplementary Table 6, the complete table).

Furthermore, the predictive power of those above albumin-derived markers for IVIG-resistance were analyzed (Tables 4, 5; Figure 3; Supplementary Tables 7–11). ROC curve analyses indicated that the CALLY index, CAR, NAR, NPAR, and PNI are predictive of IVIG-resistance, with AUC values of 0.725 (cut-off value: 0.105, sensitivity: 66.4%, specificity: 70.5%), 0.700 (cut-off value: 3.245, sensitivity: 62.8%, specificity: 69.6%), 0.580 (cut-off value: 0.485, sensitivity: 28.2%, specificity: 89.8%), 0.717 (cut-off value: 2.335, sensitivity: 47.4%, specificity: 88.7%), and 0.712 (cut-off value: 47.255, sensitivity: 42.3%, specificity: 22.7%), respectively. Meanwhile, the AUC values corresponding to the multiple comparison analysis of the predicted probabilities of each model was presented (Table 6 and Figure 4). As shown in Table 6, the AUC values of Models 1–5 have increased compared to those of the Base Model. However, the DeLong test results indicate that there was no statistically significant difference in AUC between the Base Model and Models 1–5 (Table 7).

Research on IVIG-resistant KD has long been a significant challenge for pediatricians. Over the past decades, efforts have been made to identify more effective methods for predicting IVIG resistance. Initially, single factors such as CRP and albumin were used; later, scoring systems incorporating multiple variables, such as the Kobayashi score, were developed. While single biomarkers lack comprehensiveness in capturing the complex inflammatory response during the acute phase of KD, multi-factor scoring systems can be somewhat cumbersome in clinical settings. Therefore, this study aims to explore inflammatory indicators that are both comprehensive and clinically practical for predicting IVIG-resistant KD. This was a relatively large-scale study and, to our knowledge, the first to explore albumin-related biomarkers, including the CALLY index, CAR, NAR, NPAR and PNI, in IVIG-resistant KD. In our study we compared albumin-derived nutritional inflammation indices between two groups and found that CAR, NAR and NPAR were positively correlated with IVIG resistance, while the CALLY index and PNI showed negative correlations. In conclusion, our findings suggest that the CALLY index, CAR, NAR, NPAR and PNI are associated with IVIG resistance risk.

Serum albumin produced by the liver is a negative acute-phase reactant (AFR) that decreases during immune activation, whereas CRP produced by the liver is a positive AFR that increases in inflammation or infection. Neutrophils indicate active inflammation, whereas lymphocytes serve as markers for immune regulation. During an inflammatory response, delayed neutrophil apoptosis and stem cell stimulation by growth factors result in neutrophilia and redistribution within the lymphatic system, while increased lymphocyte apoptosis causes lymphocytopenia (24). Consistent with these findings, our study demonstrated that the IVIG-resistant group exhibited elevated CRP levels and reduced lymphocyte and albumin levels compared to the IVIG-responsive group. The absence of consensus on a singular risk factor and the susceptibility of individual inflammatory parameters to external influences have led to a research focus on combined risk factor indices. These indices are considered potentially more reliable for predicting IVIG resistance in KD patients than individual parameters.

Albumin-derived markers (CALLY index, CAR, NAR, NPAR, and PNI), calculated based on serum albumin, CRP, lymphocyte count, and neutrophils, more comprehensively reflect the nutritional and inflammation status. These markers are crucial for evaluating the severity and predicting outcomes of inflammatory conditions (25–27). The CALLY index, introduced by Liu et al. (28), combines albumin, lymphocyte count and CRP, offering insights into nutritional status, inflammation, and immune function (14). NAR serves as a crucial index that integrates the advantages of neutrophils and albumin to provide a comprehensive assessment of systemic immunity and nutritional status. This study represents the inaugural examination of the CALLY index and NAR within the framework of KD. In our study, we found that the results of the ROC analysis demonstrated that the CALLY index had the highest AUC among the parameters evaluated, indicating that it exhibited superior diagnostic performance. However, NAR demonstrated the least effective performance in predicting IVIG resistance in KD, as indicated by its lowest AUC among the evaluated parameters. It was well recognized that during the acute phase of KD serum albumin levels and lymphocyte counts were typically reduced, whereas CRP levels, neutrophil counts, and the neutrophil percentage were elevated. Among these albumin-derived markers, the three components of the CALLY Indexes show a consistent direction of change, while the other four albumin-derived markers show a consistent direction of change for two of their components. This might be the reason why the CALLY Indexes have the highest predictive value. In the univariate analysis of the components of the five albumin-derived markers, albumin levels, lymphocyte count, CRP levels, and neutrophil percentage demonstrated statistically significant differences between the two patient groups, whereas neutrophil count did not reach statistical significance. This might partially explain the relatively lower predictive value of NAR.

To date, limited research has explored the association between CAR, NPAR, PNI, and KD. A meta-analysis showed that lower PNI or high CAR was associated with the increased risk to develop IVIG resistance (28). NPAR was identified an independent biomarker for IVIG resistance (10). Our findings confirm that CAR, NPAR, and PNI are independent risk factors for IVIG resistance, aligning with previous research. Unlike previous studies, which often analyzed these markers individually, our study comprehensively examined the association of these albumin-derived markers with IVIG-resistant KD. As of the time of writing this manuscript, our study was the first to report the association between CALLY indexes and IVIG-resistant KD. Additionally, only one previous study had investigated the predictive value of NPAR for IVIG-resistant KD, and it was based on a sample size of 438 cases, which is considerably smaller than that of our study (10). This study indicates that PNI, NPAR, and CAR are potential predictive factors for IVIG resistance in KD, a finding that aligns with the results of previous research. In this study, we also calculated the Kobayashi score for each patient. However, its sensitivity for identifying IVIG resistance in Kawasaki disease was only 21.8% (17/78), which was lower than that of other albumin-derived markers evaluated in this study. Therefore, in this study, compared with the Kobayashi score, the albumin-derived markers demonstrate greater potential for clinical application.

This study has certain limitations. First, owing to its retrospective, single-center design and the insufficient sample size of patients, and 35 patients were excluded from the study due to missing data, all of which inevitably led to a certain degree of selection bias in our research process. The potential for bias cannot be completely ruled out, which may reduce the statistical power of our findings. Second, the homogeneity of the study population in terms of racial background may limit the generalizability of our findings. For instance, the Kobayashi score, which has achieved relatively good predictive effects in previous reports, may not perform well in our study due to differences in ethnicity. Therefore, prospective multicenter studies with larger and more diverse racial samples are necessary to confirm these results. The association of hypoalbuminemia (PNI, CAR) with IVIG resistance aligns with KD’s endothelial dysfunction paradigm. For future work, we need to proactively collect data on KD patients in our hospital or from other centers for external validation, so that our research can be better applied in clinical practice.