In total, 260 patients with infantile KD were recruited in our study, with a mean age of 24 month. Sixty-six (25.4%), thirty-nine (15%) and fifty-six patients (21.5%) of the participants had coronary dilation, coronary aneurysm, and intimal thickening of coronary arteries, respectively. Single-variable analysis found age, gender, duration of fever, time of initial use of IVIG, ESR, WBC, time of platelet increase, maximum value of platelet, albumin and IgG level were associated with CAL. In multi-variable logistic analysis, younger in age, male, lower serum albumin level, and later initial use of IVIG were found to be potential risk factors of CAL in the studied population. In the sensitivity analysis that added IVIG resistance as a covariance in the models, the risk factors for coronary dilation and intimal thickening remained the same.
There are few studies that explored the association between biomarkers and CAL in KD patients. A number of them found similar results to our study, suggested that lower albumin was a risk factor of CAL in KD patients [20, 21]. Meanwhile, other risk factors have also been revealed in the current literature, such as hemoglobin, fever duration, C-reactive protein, and so on. For instance, H.J. Kim et al reported that compared to KD patients without CAL, those with CAL had significantly higher serum interleukin-6 levels (228.26 ng/mL vs. 39.18 ng/mL), and lower body mass index (15.09 vs. 16.60) [22]. Li et al. revealed that red blood cell distribution width could be used as an independent predictor of CAL in KD patients [23]. Shao et al. found Apo-A level, as a predictor of CAL in KD patients, was significantly negatively linked to CAL [24]. In our study, factors such as duration of fever and platelet level were not found to be significantly associated with CAL. The variance in ethnicity, geographic location and seasonality could partially explain the divergences in our findings. In addition, a large body of evidence involved research on potential predictive biomarkers of CAL in KD patients, which were novelty but could not be easily measured and monitored in the general medical practice. The biomarkers included adropin [20], plasminogen activator inhibitor-1 [25], tenascin-C [26], n-terminal pro-brain natriuretic peptide [27–33], circulating endothelial glycocalyx components [26], lymphocyte hydrogen sulfide production [28], to mention a few. Different from these studies, we included results from clinical blood examinations as potential risk factors, which could be easier and faster to be assessed in the clinical practice. Furthermore, extracoronary echocardiographic findings and Z-values were also found to have predictive value for CAL [28, 33, 34]. Compared to those studies, investigation on blood-based biomarkers avoided the influence of variance in equipment and technicians.
In our study, lower albumin level was found to be a risk factor of CAL. Serum albumin is commonly used as a parameter to negatively indicate the activity of inflammatory as an acute phase reactant [35]. Hypoalbuminemia in KD patients was reported to be caused by vascular leakage, which was due to the elevation of vascular endothelial growth factor. Studies have found that vascular endothelial growth factor and its receptors (Fms-related tyrosine kinase-1 and kinase insert domain receptor) were involved in the development and progression of CAL in KD patients [36–38].
Although IVIG combined with aspirin were commonly adopted as the mainstay of initial treatment which could effectively decrease the development of CAL in KD patients, the high incidence of IVIG-resistant were reported in recent years [39–43]. Approximately 10–20% of pediatric patients with KD were reported to have persistent or recrudescent fever after the initial IVIG therapy [44, 45]. Patients that were not responsive to IVIG therapy were considered as at higher risk of incidence for CAL [40, 43]. Several studies suggested potential underlying mechanism that could partially explain the phenomenon. For example, through genome-wide association studies (GWAS), a number of genetic markers associated with both IVIG resistance and CAL formation were found in KD patients, including inositol 1,4,5-trisphosphate 3-kinase C (ITPKC), caspase-3 (CASP3), FCGR2A, CD40, and interleukin1 beta (IL-1B) [46, 47]. Wang et al. found that the change of serum levels of IL-10 varied between patients with and without IVIG resistance after the usage of the medication, and the genetic polymorphisms of IL-10 were corelated to CAL [48]. However, in our study, IVIG resistance was not found to be associated with CAL. The potential reasons could be as followed. The IVIG resistance may suggest the severity of ongoing inflammation, nevertheless development of CAL might also be affected by other factors such as hemodynamics [49], of which the data were not collected in our study. Moreover, in the sensitivity analysis that involved IVIG resistance, even though the responsiveness of IVIG was not found to be associated with CAL directly, some of the other risk factors such as WBC and time of initial use of IVIG became nonsignificant after the addition of IVIG resistance. The mechanisms that lead to the variation should be investigated in future research.
It is crucial to identify the patients with elevated risk of CAL at early stage of the disease, through surveillance on specific risk factors in KD patients. Our study suggested several risk factors including age, gender and albumin, with might be useful in clinical practice because these factors are easy to be measured and are possible to be tested frequently. It could help the pediatrician identify patients that might at high risk of CAL at early stage of the disease, for example upon admission to the hospital. Meanwhile, our findings could offer the pediatrician guidance on monitoring and providing targeted preventive and therapeutic measures. It could be inferred that patients with KD could be benefited from the timely medical treatment.
Strength and limitations
Our present study provided evidence on the associations between clinical biomarkers and the incidence of several CALs. There are a plenty of strengths that made our study valuable and practical. First, although the study was retrospective, we used data from medical records, which were obtained during the hospitalization. The accuracy and reliability of the data largely reduced the potential recall bias. Second, by including a group of clinical biomarkers as potential risk factors, combined with a comprehensive involvement of three major outcomes, our study provided an overall insight of the topic. Third, KD has been more prevalent in Asia area. Our study was carried out in Asian population (Chinese), which made our result proper to be generalized in the high-risk area of Asia. Forth, we investigated clinical characteristics and biomarkers that are easy to be examined, which made our conclusion more practical to be verified in future clinical research, and hence, to be implemented in clinics. However, there are several limitations of our study. First, the duration of follow-up was short. KD could have a life-long influence on the patients’ health conditions. As reported by a Canadian study, 5% miocardial infarction patients less than 40 years old who underwent coronary angiography had lesions that were linked to sequelae of KD [50]. Therefore, it is necessary to follow up the participants as long as possible, to investigate the longitudinal risk factors for CAL. Second, the sample size of our study was relevant small. Due to the low incidence of KD in Guangdong province (approximately 5 per 100000) [7], it might be greatly time-consuming to recruit a large sample, even in multi-center research. Because of the insufficient sample, we cannot compare our findings between IVIG responsive and nonresponsive patients. The small sample size weakened the robustness of our conclusion. Third, information on some of the potential confounders that could have influence on the result were not collected in our study. The factors included family history, diet, sleep and other life-style factors, genetic data, etc. These data could not be implemented due to the retrospective feature of the study. Forth, we did not include patients in the recurrent episode. However, Nakamura et al. reported a higher risk of developing coronary artery sequelae with the recurrent episode [12]. This characteristic of our sample limited the generalization of our conclusion.