Secondary Logo

Share this article on:

Predictors of Intravenous Immunoglobulin Nonresponse and Racial Disparities in Kawasaki Disease

Clark, Daniel E., MD, MPH*; Denby, Kara J., MD*; Kaufman, Laura M., MD; Fill, Mary-Margaret A., MD; Piya, Bhinnata, MPH§; Krishnaswami, Shanthi, MBBS, MPH§; Fonnesbeck, Christopher, PhD; Halasa, Natasha, MD, MPH§

The Pediatric Infectious Disease Journal: December 2018 - Volume 37 - Issue 12 - p 1227–1234
doi: 10.1097/INF.0000000000002019
Original Studies

Background: Kawasaki disease (KD) is the most common cause of acquired heart disease in American children. Intravenous immunoglobulin (IVIG) nonresponse is a known risk factor for cardiac sequelae. Previously reported risk factors for nonresponse include age, male sex and laboratory abnormalities. We set out to identify additional risk factors for IVIG nonresponse in a racially diverse KD population.

Methods: We conducted a retrospective chart review at a referral center in the Southeastern United States of children meeting ICD-9 (International Statistical Classification of Disease and Related Health Problems) criteria for KD and being treated with IVIG.

Results: Four-hundred and fifty-nine children met inclusion criteria, 67 were excluded for subsequent rheumatologic diagnosis, unknown race, or failure to meet the American Heart Association guideline criteria. Our final cohort consisted of 392 subjects, with median age of 2.7 years, 65.1% male, 66.1% White, 24.2% Black, 4.9% Asian and 82.9% responded to a single dose of IVIG. Coronary ectasia or aneurysm developed in 27%; 7.4% developed aneurysms and 2.3% giant coronary aneurysms. Nonresponders were more likely to be Black, have higher white blood cell, erythrocyte sedimentation rate and C-reactive protein, lower hemoglobin, develop ectasia or aneurysm and require critical care and hospital readmission. Responders achieved echocardiographic normalization more often compared with nonresponders (81.3% vs. 60.9%, P = 0.002) and coronary artery pseudonormalization (87.2% vs. 69.7%, P = 0.03) at 1 year. Black nonresponders had the slowest normalization at 1 year (52.9%, P = 0.02).

Conclusions: Nonresponders have higher rates and greater severity of coronary involvement than responders. Our study uniquely demonstrates Black race as a risk factor for nonresponse and for delayed normalization of cardiac involvement at 1-year follow-up.

From the *Departments of Internal Medicine and Pediatrics, Vanderbilt University Medical Center, Nashville, Tennessee

Greater Lawrence Family Health Center, Lawrence, Massachusetts

Tennessee Department of Health

§Division of Pediatric Infectious Disease, Department of Pediatrics, Vanderbilt University Medical Center

Department of Biostatistics, Vanderbilt University, Nashville, Tennessee.

Accepted for publication February 26, 2018.

This work was supported by Investigator-initiated grant from Baxalta and the CTSA award (Clinical and Translational Science Award) UL1TR000445 from the National Center for Advancing Translational Sciences. Its contents are solely the responsibility of the authors and do not necessarily represent official views of the National Center for Advancing Translational Sciences or the National Institutes of Health.

The authors have no conflicts of interest to disclose.

Address for correspondence: Kara J. Denby, MD, Department of Internal Medicine-Pediatrics, Vanderbilt University, 1215 21st Ave South, Medical Center East, 7th Floor, North Tower, Suite II, Nashville, TN 37232. E-mail: k.denby@vanderbilt.edu.

Kawasaki disease (KD) is the most common medium-vessel vasculitis in children in the United States and the most common cause of acquired childhood heart disease in the developed world.1 , 2 In 2000, approximately 4248 hospitalizations for KD occurred in the United States with a hospitalization rate of 17.1 per 100,000 children.3 Treatment with intravenous immunoglobulin (IVIG) within 10 days of fever onset decreases coronary complications from 23% in untreated patients to 4% among treated patients.4 However, despite treatment with IVIG there is still a subpopulation that remains at risk for cardiac complications.

Nonresponse to IVIG is a known risk factor for long-term cardiac sequelae.5 Thus, identification and treatment of IVIG nonresponders is a key aspect of managing KD. A cohort study of Japanese children with KD showed nonresponse to IVIG as the strongest independent risk factor for coronary artery aneurysms (CAAs) within 1 month after presentation with an adjusted odds ratio (OR) of 19.1 [95% confidence interval (CI): 8.73–41.98].6 One US study showed nonresponders have a higher risk of CAAs than responders (15% vs. 3%, P = 0.0008).7 Age extremes (<1-year-old, >4–5 years old),6 , 8 male gender6 , 9 , 10 and laboratory abnormalities—including higher erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), white blood cell (WBC) count, and percentage of polymononuclear cells and bands,7 , 10 , 11 lower hemoglobin,7 higher transaminases [aspartate aminotransferase (AST) and alanine aminotransferase]7 , 10 and hypoalbuminemia10 , 11—have all been identified as risk factors associated with nonresponse and increased risk for CAAs. In addition to IVIG nonresponse, delayed IVIG treatment12 and recurrent KD13 also increase the risk for CAAs. Incomplete KD (iKD) poses a significant diagnostic challenge, often leading to delay in diagnosis, and may be associated with an increased risk of cardiac complications compared with complete KD (cKD) patients.14 , 15 The American Heart Association (AHA) guidelines for the management of KD provide an algorithm to help clinicians evaluate for and diagnose suspected iKD.16 Evaluation of the 2004 recommendations showed 27% more KD patients were identified earlier and appropriately treated with IVIG, who otherwise would have been missed or had delayed treatment without the algorithm.1 , 17

Current predictive models for nonresponders are limited by lack of generalizability to different racial groups and inconsistent strength of association for laboratory predictors. This study aims to identify additional risk factors for IVIG nonresponse in a heterogeneous, Southeastern American, single-center cohort over a 14-year study period.

Back to Top | Article Outline

METHODS

Patient Population

We conducted a retrospective chart review of all hospitalizations to Monroe Carell Jr. Children’s Hospital at Vanderbilt during a 14-year study period. Patients were initially screened and identified by the following methods: all children <18 years with a KD ICD-9 code (446.1 Acute febrile mucocutaneous lymph node syndrome) or charts that mentioned KD ≥5 times in their electronic medical record using a standardized Vanderbilt research derivative. Cases were included in the analyses if they met the following inclusion criteria: (1) admitted between January 1, 2001, and December 31, 2014; (2) received at least 1 dose of IVIG; (3) discharged with a diagnosis of KD and (4) both diagnosed and treated at Monroe Carell Jr. Children’s Hospital at Vanderbilt. Cases were excluded if (1) an alternative diagnosis was revealed on follow-up through December 31, 2014 (eg, autoimmune disorders or malignancy); (2) their race was unknown and/or (3) they were classified and treated as iKD but did not meet criteria per the 2004 AHA treatment algorithm16 and did not have cardiac involvement on echocardiography (echo). This study was reviewed and approved by the Vanderbilt Institutional Review Board.

Back to Top | Article Outline

Data Collection and Storage

Demographics, clinical presentation, laboratory and echo data, treatment, readmissions and cardiology follow-up for KD were extracted from each subject’s medical record and entered using Research Electronic Data Capture.18

Back to Top | Article Outline

Definitions

KD subjects were classified as cKD or iKD according to AHA criteria.1 , 16 Responders were defined as subjects who received only 1 dose of IVIG, had resolution of symptoms with defervescence within 36 hours and no further treatment or additional admissions for their KD. Subjects not meeting these criteria are classified “non-responders” (also known as “IVIG resistant”). Ectasia was defined as a segment of the coronary artery that is 1.5 times the size of the surrounding segment with a Z score of >1.5 and <2.5.1 Aneurysm was defined as a Z score ≥2.5 or by absolute diameter per the AHA guidelines.1 , 16 Echo positivity was defined as aneurysm, ectasia or ≥3 AHA supplemental echo criteria.1 Echo abnormalities were classified into 2 groups—echos with any abnormality associated with KD and echos that strictly met AHA criteria for KD (“echo positive”). Normalization of echocardiogram was defined as resolution of any cardiac involvement related to KD. Pseudonormalization of coronary arteries was defined as an apparent normalization of coronary artery luminal dimension on echo-positive subjects. This term was used to account for normalization of luminal diameter on imaging without implying resolution of associated endothelial dysfunction as this has been shown to persist long after luminal resolution on imaging modalities, particularly in the larger aneurysms.1 , 19–21 Fever was defined as temperature of ≥100.4°F either obtained during hospitalization or reported by the patient’s family. We defined “day 1 of illness” as the first day of fever. Anemia was defined by age-adjusted standards. Subjects’ race as reported by parents or guardians was extracted from the chart and further subclassified as non-Black (White, Asian, Native American and Other) or Black. KD shock syndrome (KDSS) was defined more strictly than prior studies as subjects requiring admission to the pediatric intensive care unit for systolic hypotension.22

Back to Top | Article Outline

Statistical Analyses

We conducted descriptive analyses using SAS 9.4 (SAS Institute, Cary, NC), and the distribution of demographic, laboratory and clinical measures within the study sample were presented as proportion or median with 95% binomial or distribution-free 50th percentile CIs, respectively. Overlap of 95% CI was used to assess the proportional and median group differences at each level of selected measure. We fit a multivariable predictive model for nonresponse to IVIG treatment, using a Bayesian logistic regression approach. The model was constructed using a suite of covariates thought a priori to be predictive of nonresponse to IVIG treatment. These variables included: gender (indicator for male), race (indicator for Black), KD criteria classification (indicator for cKD), ectasia, aneurysm, hemoglobin, platelets, WBC, CRP, AST and ESR. Several predictor variables had missing values, including hemoglobin (7% missing), ESR (11%), platelets (6%), CRP (24%), WBC (6%) and AST (21%). These missing values were imputed probabilistically as the model was fit via Bayesian imputation, where we assumed the values were missing completely at random. All predictor variables were normalized by subtracting the variable mean and dividing by the standard deviation (SD) to improve model fit and interpretation. The model was fit via Markov chain Monte Carlo using the No U-Turn Sampler23 as implemented by the PyMC3 software package.24 All model coefficients were given vague normal priors with an SD of 10. The model was checked for goodness of fit using posterior predictive checks, comparing data simulated from the fitted model to the observed data values; no evidence of lack of fit was observed.

We compared the proportion of children whose echos became normal or pseudonormal within 1 year and generated Kaplan–Meier curves using subject echos over time to compare the time to both echo normalization and pseudonormalization at 2 years among both racial and responder groups. The group difference was tested by a log rank test.

Back to Top | Article Outline

RESULTS

Study Cohort

Population

A total of 1061 unique patient charts were screened for eligibility, with 459 subjects meeting initial inclusion criteria; however, 67 of 459 (14.6%) subjects were excluded from analyses because of unknown/declined race (n = 6), subsequent diagnosis of juvenile idiopathic arthritis (n = 3) or failure to meet criteria for inclusion per the AHA diagnostic algorithm (n = 58; Fig. 1).16 The final cohort for analyses consisted of 392 subjects. The median age was 2.7 years and 255 (65.1%) were male (Table 1).

TABLE 1

TABLE 1

FIGURE 1

FIGURE 1

Back to Top | Article Outline

Clinical Features and Outcomes

The most common clinical feature of KD was conjunctival injection (95.7%), followed by rash (93.9%), mucous membrane changes (91.4%), extremity changes (75.8%) and unilateral cervical lymphadenopathy ≥1.5 cm (32.3%); these findings were consistent with prior literature.1 A total of 325 of 392 (82.9%) patients were classified as responders. The vast majority of subjects (85.5%) were treated within 10 days of disease onset, 6.1% were readmitted and 2.8% presented with KDSS.

Back to Top | Article Outline

Complete Versus Incomplete KD

In comparing complete and incomplete groups, 313 subjects met criteria for cKD and 79 (20.2%) subjects met criteria for iKD (Table 1). Patients with iKD were statistically more likely to have lower hemoglobin, lower albumin, ectasia or aneurysm on initial echo, echo meeting AHA criteria for KD involvement, ectasia or aneurysm at any time point and giant aneurysms. They were also more likely to have KDSS and longer duration of illness before treatment with IVIG compared with subjects with cKD (Table 1). The rates of treatment with IVIG within 10 days of illness onset were not statistically different between groups.

Back to Top | Article Outline

Responders Versus Nonresponders to IVIG

Responders accounted for the majority of subjects (325, 82.9%), with 67 (17.1%) subjects classified as nonresponders. Nonresponders were more likely to be Black, have higher ESR and CRP, AHA echo positivity and ectasia or aneurysm compared with responders (Table 1). In addition, nonresponders were more likely to require intensive care and readmission (Table 1).

Back to Top | Article Outline

Multivariable Logistic Regression Model

The multivariable model (Fig. 2) was found to be predictive of IVIG nonresponse, yielding an estimated Brier score of 0.138 [95% Bayesian credible interval (BCI) = 0.125–0.134] and receiving operator characteristic area of 0.71. The performance of the model was influenced by several variables that were related to a higher probability of nonresponse. The most important factors included low hemoglobin, with a mean OR of 0.61 (95% BCI = 0.30–0.98) and a posterior probability (Pr) of 0.95 that the true OR is less than 1; black race [OR = 1.70, 95% BCI = 0.74–2.75, Pr (OR > 1) = 0.95]; high CRP [OR = 1.92, 95% BCI = 0.84–3.12, Pr (OR > 1) = 0.96]; low platelets [OR = 0.62, 95% BCI = 0.29–1.05, Pr (OR < 1) = 0.95] and high WBC [OR = 1.93, 95% BCI = 0.84–3.18, Pr (OR > 1) = 0.97]. Because all covariates were normalized, the ORs for continuous variables should be interpreted as the relative odds for an increase of 1 SD in the value of that variable; in other words, the odds of nonresponse are multiplied by the OR for each SD increase.

FIGURE 2

FIGURE 2

Back to Top | Article Outline

Black Versus Non-Black Race

Compared with non-Black, Black subjects (95, 23.2%) were more likely to have higher ESR, lower hemoglobin and lower albumin (Table 2). Black subjects were also more likely to be nonresponders. There was no significant difference in rates of initial positive echo, ectasias, or aneurysms between the groups (Table 2).

TABLE 2

TABLE 2

Back to Top | Article Outline

Echocardiographic Findings

In the overall population, ectasia or aneurysm was present on 22.7% of initial echo and 27.0% of subjects had ectasia or aneurysm present on at least 1 echo. CAAs occurred in 7.4% and giant coronary aneurysms in 2.3%. Echos were followed for a period of 2 years and compared by responder status and racial subgroups. A total of 200 children developed an abnormal echo (any cardiac involvement, including full AHA criteria) and 98% were followed for at least 2 years. Normalization of echo abnormalities occurred in 76.5% (150/196) at 1 year. Responders were more likely to achieve echo normalization by 1 year than nonresponders (81.3% vs. 60.9%, P = 0.002). Black subjects trended toward slower normalization overall compared with non-Black subjects (67.4% vs. 79.6%, P = 0.08). This was driven by the Black nonresponder subgroup, which had the slowest normalization compared with non-Black responders (52.9% vs. 83.1%, P = 0.004). Black nonresponders took a median of 13 weeks post-IVIG to normalize (95% CI: 3.42–40.00) compared with 7.7 weeks (95% CI: 5.14–22.43) in Black responders, 5.6 weeks (95% CI: 3.71–6.71) in non-Black responders and 9.4 weeks (95% CI: 3.0–52.29) in non-Black nonresponders (P = 0.01; Fig. 3A).

FIGURE 3

FIGURE 3

A total of 111 children developed echo abnormalities meeting AHA criteria (echo positive)16; 82% (91/111) of these children had pseudonormalization of their coronary arteries by 1 year. Responders were more likely to pseudonormalize than nonresponders (87.2% vs. 69.7%, P = 0.03). No racial differences were appreciated in pseudonormalization rates (Black subjects 86.7% vs. non-Black subjects 80.3%, P = 0.58), including among racial subgroups of nonresponders (non-Black nonresponders 65% vs. Black nonresponders 76.9%, P = 0.47). Similarly, there was no statistically significant racial subgroup difference in pseudonormalization rate by 2 years (P = 0.10, Fig. 3B).

Back to Top | Article Outline

DISCUSSION

In this 14-year, single-center, retrospective study of US children with KD, we uniquely identified Black race as a risk factor for nonresponse to a single dose of IVIG. We also noted elevated inflammatory markers were associated with nonresponse, consistent with prior studies.7 , 25 , 26 Interestingly, nonresponders in our cohort did not trend toward age extremes, male gender or delayed IVIG treatment—all of which have previously been shown as risk factors for cardiac involvement in KD.6–8 , 12 The frequency of nonresponse (17.1%) is within the 10%–20% rate of previously reported studies5 , 7 , 27 and is only slightly higher compared with the Pediatric Health Information System database (16.3%).28 Nonresponders in our study were more likely to develop and have more severe coronary abnormalities, with 20.9% of nonresponders having a CAA detected on echo compared with 4.6% of responders, consistent with other published studies.5 , 17 , 29 , 30 Nonresponders required more intensive care and were readmitted at higher rates. Our data support that nonresponders are at high risk for presenting with severe disease and developing CAAs. Further studies to identify these children and different targeted treatment modalities are needed.

Our finding of Black race as a risk factor for IVIG nonresponse supports the recent Pediatric Health Information System (PHIS) epidemiologic study showing that hospitals in the highest quartile of IVIG nonresponse had a statistically significant higher number of Black children compared with those at the lowest quartile of IVIG nonresponse (26.5% vs. 20.1%; P < 0.01).28 Our multivariable logistic regression model showed a posterior Pr of 0.95 that the true OR for Black race was greater than 1, indicating for the first time that Black race is a risk factor for IVIG nonresponse. Thus, this higher risk racial subgroup may warrant consideration of adjunctive therapies in the initial treatment algorithm. This is particularly concerning given African American children are disproportionately affected by KD.3 , 31 , 32

Few studies have included a significant number of Black KD patients and those that did had conflicting conclusions. A Michigan State study (n = 189, 72% Black) found that Black children were more likely to be hospitalized for KD with trends toward higher rates of coronary involvement, though statistically insignificant.31 Conversely, a study at Children’s National Medical Center (n = 302, 54% Black) found that Black race may be protective against CAAs.33 Our study found no difference in the rate of coronary artery involvement in Black children when compared with non-Black children. However, Black nonresponders had delayed normalization of KD-related echo abnormalities at 1 year (52.9%, P = 0.02) with similar rates of CAA and ectasia pseudonormalization compared with non-Black nonresponders (76.9% vs. 65.0%, P = 0.47). Thus, Black race predisposes to IVIG nonresponse with a prolonged duration of cardiac involvement compared with other races. Additional studies with racially diverse populations are necessary to confirm our results.

The explanation for the racial disparity in IVIG response is unknown, but host genetics may play an important role. It has been suggested that African-descent is associated with decreased sialylation.34 , 35 Prior immunologic work has shown nonresponders have lower levels of sialylation of the Fc (fragment crystallizable) portion of their endogenous IgG, which has been shown to be important for response to IVIG.36 Secondly, hemoglobin values were lower in Black subjects compared with non-Black subjects, which may represent an increased inflammatory state. One proposed mechanism of anemia in KD is upregulation of toll-like receptors leading to an increase in the liver-derived protein hepcidin.37 Hepcidin is speculated to be elevated in Black subjects, and levels correlate with increased levels of CRP, which trended toward significant increases among Black subjects in our cohort.38 , 39 Hepcidin-induced anemia can occur by 3 mechanisms: (1) direct inhibitory suppression of erythropoiesis; (2) intracellular iron sequestration from interaction with ferroportin and (3) decreased intestinal iron absorption in the duodenum.37 Additional studies are needed to determine if hepcidin levels could be a biomarker for nonresponse and if these levels and/or hepcidin receptors are associated with racial disparities.

It is well established that children with KD who fail to respond to the first dose of IVIG are more likely to develop cardiac abnormalities, which can be detrimental long term.40 , 41 Our study confirmed these findings. Both normalization of any echo abnormality and pseudonormalization of coronary abnormalities diverged by responder status by 8 weeks and persisted until 2-year follow-up, favoring responders. Future studies powered to detect racial differences in cardiac involvement are necessary to better understand how Black race affects long-term cardiac sequelae of KD.

Predicting clinical and lab features of children at high risk for IVIG nonresponse is important for understanding which children need additional monitoring, treatment and parental counseling. Thus, some international KD working groups have developed predictive models for IVIG nonresponse based on clinical, laboratory and/or imaging criteria. Unfortunately, the vast majority of these studies evaluating risk factors were done in predominately Asian populations, with limited generalizability to and/or sensitivity among US children.5 , 29 , 42 , 43 In particular, a prior study showed reduced sensitivity of the Egami score25 in Black children13 and race is not part of the predictive model derived from the UCSD (University of California - San Diego) experience.6 Multicenter studies are needed to predict and validate a score for nonresponders that is applicable to the heterogeneous US population.

Our study limitations include the retrospective design, in which clinical data were dependent on the physicians’ documentation and selection of laboratory studies was often not exhaustive. Additionally, recall bias may have affected the duration of illness, as well as the classification of children as cKD versus iKD. Sample size limited the power of our study and our ability to detect differences in long-term cardiac outcomes among racial subgroups.

Back to Top | Article Outline

CONCLUSIONS

KD is the most common cause of acquired cardiac abnormalities in American children, surpassing rheumatic fever.1 , 2 , 15 Current prediction modeling for IVIG nonresponse has limited sensitivity in the heterogeneous US population. Our study again confirms that nonresponders have higher rates and severity of coronary involvement than responders and require more intensive care and readmissions for ongoing therapy. Our study uniquely demonstrates Black race as a risk factor for IVIG nonresponse as well as for delayed resolution of any cardiac involvement at 1-year follow-up. Further studies in the United States are needed to elucidate which children are most likely to be nonresponders and to identify the optimal initial therapy for this high-risk group.

Back to Top | Article Outline

REFERENCES

1. McCrindle BW, Rowley AH, Newburger JW, et al; American Heart Association Rheumatic Fever, Endocarditis, and Kawasaki Disease Committee of the Council on Cardiovascular Disease in the Young; Council on Cardiovascular and Stroke Nursing; Council on Cardiovascular Surgery and Anesthesia; and Council on Epidemiology and Prevention. Diagnosis, treatment, and long-term management of Kawasaki disease: a Scientific Statement for Health Professionals from the American Heart Association. Circulation. 2017;135:e927–e999.
2. Burns JC, Glodé MP. Kawasaki syndrome. Lancet. 2004;364:533–544.
3. Holman RC, Curns AT, Ermias D, et al. Kawasaki syndrome hospitalizations in the United States, 1997 and 2000. Pediatrics. 2003;112:495–501.
4. Newburger JW, Takahashi M, Beiser AS, et al. A single intravenous infusion of gamma globulin as compared with four infusions in the treatment of acute Kawasaki syndrome. N Engl J Med. 1991;324:1633–1639.
5. Burns JC, Capparelli EV, Brown JA, et al. Intravenous gamma-globulin treatment and retreatment in Kawasaki disease. US/Canadian Kawasaki Syndrome Study Group. Pediatr Infect Dis J. 1998;17:1144–1148.
6. Kitano N, Suzuki H, Takeuchi T, et al; Wakayama Kawasaki Disease Study Group. Epidemiologic features and prognostic factors of coronary artery lesions associated with Kawasaki disease based on a 13-year cohort of consecutive cases identified by complete enumeration surveys in Wakayama, Japan. J Epidemiol. 2014;24:427–434.
7. Tremoulet AH, Best BM, Song S, et al. Resistance to intravenous immunoglobulin in children with Kawasaki disease. J Pediatr. 2008;153:117–121.
8. Nakamura Y, Fujita Y, Nagai M, et al. Cardiac sequelae of Kawasaki disease in Japan: statistical analysis. Pediatrics. 1991;88:1144–1147.
9. Loomba RS, Raskin A, Gudausky TM, et al. Role of the Egami score in predicting intravenous immunoglobulin resistance in Kawasaki disease among different ethnicities. Am J Ther. 2016;23:e1293–e1299.
10. Sleeper LA, Minich LL, McCrindle BM, et al; Pediatric Heart Network Investigators. Evaluation of Kawasaki disease risk-scoring systems for intravenous immunoglobulin resistance. J Pediatr. 2011;158:831–835.e3.
11. Ashouri N, Takahashi M, Dorey F, et al. Risk factors for nonresponse to therapy in Kawasaki disease. J Pediatr. 2008;153:365–368.
12. Muta H, Ishii M, Yashiro M, et al. Late intravenous immunoglobulin treatment in patients with Kawasaki disease. Pediatrics. 2012;129:e291–e297.
13. Nakamura Y, Oki I, Tanihara S, et al. Cardiac sequelae in recurrent cases of Kawasaki disease: a comparison between the initial episode of the disease and a recurrence in the same patients. Pediatrics. 1998;102:E66.
14. Sonobe T, Kiyosawa N, Tsuchiya K, et al. Prevalence of coronary artery abnormality in incomplete Kawasaki disease. Pediatr Int. 2007;49:421–426.
15. Witt MT, Minich LL, Bohnsack JF, et al. Kawasaki disease: more patients are being diagnosed who do not meet American Heart Association criteria. Pediatrics. 1999;104:e10.
16. Newburger JW, Takahashi M, Gerber MA, et al; Committee on Rheumatic Fever, Endocarditis, and Kawasaki Disease, Council on Cardiovascular Disease in the Young, American Heart Association. Diagnosis, treatment, and long-term management of Kawasaki disease: a statement for health professionals from the Committee on Rheumatic Fever, Endocarditis, and Kawasaki Disease, Council on Cardiovascular Disease in the Young, American Heart Association. Pediatrics. 2004;114:1708–1733.
17. Yellen ES, Gauvreau K, Takahashi M, et al. Performance of 2004 American Heart Association recommendations for treatment of Kawasaki disease. Pediatrics. 2010;125:e234–e241.
18. Harris PA, Taylor R, Thielke R, et al. Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42:377–381.
19. Denby KJ, Clark DE, Markham LW. Management of Kawasaki disease in adults. Heart. 2017;103:1760–1769.
20. Kato H, Sugimura T, Akagi T, et al. Long-term consequences of Kawasaki disease. A 10- to 21-year follow-up study of 594 patients. Circulation. 1996;94:1379–1385.
21. Dhillon R, Clarkson P, Donald AE, et al. Endothelial dysfunction late after Kawasaki disease. Circulation. 1996;94:2103–2106.
22. Kanegaye JT, Wilder MS, Molkara D, et al. Recognition of a Kawasaki disease shock syndrome. Pediatrics. 2009;123:e783–e789.
23. Hoffman MD, Gelman A. The No-U-Turn Sampler: adaptively setting path lengths in Hamiltonian Monte Carlo. J Mach Learn Res. 2014;15:1593–1623.
24. Salvatier J, Wiecki TV, Fonnesbeck C. Probabilistic programming in Python using PyMC3. PeerJ Computer Science. 2016;2:e55.
25. Egami K, Muta H, Ishii M, et al. Prediction of resistance to intravenous immunoglobulin treatment in patients with Kawasaki disease. J Pediatr. 2006;149:237–240.
26. Kobayashi T, Inoue Y, Takeuchi K, et al. Prediction of intravenous immunoglobulin unresponsiveness in patients with Kawasaki disease. Circulation. 2006;113:2606–2612.
27. Newburger JW, Sleeper LA, McCrindle BW, et al; Pediatric Heart Network Investigators. Randomized trial of pulsed corticosteroid therapy for primary treatment of Kawasaki disease. N Engl J Med. 2007;356:663–675.
28. Moffett BS, Syblik D, Denfield S, et al. Epidemiology of immunoglobulin resistant Kawasaki disease: results from a large, national database. Pediatr Cardiol. 2015;36:374–378.
29. Chen S, Dong Y, Kiuchi MG, et al. Coronary artery complication in Kawasaki disease and the importance of early intervention: a systematic review and meta-analysis. JAMA Pediatr. 2016;170:1156–1163.
30. Chen S, Dong Y, Yin Y, et al. Intravenous immunoglobulin plus corticosteroid to prevent coronary artery abnormalities in Kawasaki disease: a meta-analysis. Heart. 2013;99:76–82.
31. Abuhammour WM, Hasan RA, Eljamal A, et al. Kawasaki disease hospitalizations in a predominantly African-American population. Clin Pediatr (Phila). 2005;44:721–725.
32. Gibbons RV, Parashar UD, Holman RC, Belay ED, Maddox RA, Powell KE, et al. An evaluation of hospitalizations for Kawasaki syndrome in Georgia. Archives of Pediatric and Adolescent Medicine. 2002;156:492–496.
33. Porcalla AR, Sable CA, Patel KM, et al. The epidemiology of Kawasaki disease in an urban hospital: does African American race protect against coronary artery aneurysms? Pediatr Cardiol. 2005;26:775–781.
34. Smith KG, Clatworthy MR. FcgammaRIIB in autoimmunity and infection: evolutionary and therapeutic implications. Nat Rev Immunol. 2010;10:328–343.
35. Mahan AE, Jennewein MF, Suscovich T, et al. Antigen-specific antibody glycosylation is regulated via vaccination. PLoS Pathog. 2016;12:e1005456.
36. Ogata S, Shimizu C, Franco A, et al. Treatment response in Kawasaki disease is associated with sialylation levels of endogenous but not therapeutic intravenous immunoglobulin G. PLoS One. 2013;8:e81448.
37. Huang YH, Kuo HC. Anemia in Kawasaki disease: Hepcidin as a potential biomarker. Int J Mol Sci. 2017;18:820.
38. Karl JP, Lieberman HR, Cable SJ, et al. Randomized, double-blind, placebo-controlled trial of an iron-fortified food product in female soldiers during military training: relations between iron status, serum hepcidin, and inflammation. Am J Clin Nutr. 2010;92:93–100.
39. Smith EM, Alvarez JA, Martin GS, et al. Vitamin D deficiency is associated with anaemia among African Americans in a US cohort. Br J Nutr. 2015;113:1732–1740.
40. Hsu CH, Chen MR, Hwang FY, et al. Efficacy of plasmin-treated intravenous gamma-globulin for therapy of Kawasaki syndrome. Pediatr Infect Dis J. 1993;12:509–512.
41. Ephrem A, Misra N, Hassan G, et al. Immunomodulation of autoimmune and inflammatory diseases with intravenous immunoglobulin. Clin Exp Med. 2005;5:135–140.
42. Minich LL, Sleeper LA, Atz AM, et al; Pediatric Heart Network Investigators. Delayed diagnosis of Kawasaki disease: what are the risk factors? Pediatrics. 2007;120:e1434–e1440.
43. Uehara R, Belay ED, Maddox RA, Holman RC, Y N, Yashiro M, et al. Analysis of potential risk factors associated with nonresponse to initial intravenous immunoglobulin treatment among Kawasaki disease patients in Japan. Pediatr Infect Dis J. 2008;27:155–160.
Keywords:

Kawasaki disease; intravenous immunoglobulin; incomplete Kawasaki disease; coronary artery aneurysms

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.