FormalPara Key Points

The reported incidence of resistant Kawasaki disease (KD) varies widely, partly due to inconsistent definitions.

Treatment of resistant KD is important as these patients are at a greater risk of worse outcomes.

The current American Heart Association guidelines suggest that reasonable therapy for resistant KD includes a second dose of intravenous immunoglobulin, a short course of high-dose steroids, or infliximab. The level of evidence for these recommendations varies.

Other agents have been used for resistant KD, but many would require larger studies with standardised methodology to adequately assess efficacy.

1 Introduction

Kawasaki disease (KD) is an inflammatory vasculitis of medium-sized arteries. It is the leading cause of acquired heart disease in children in industrialised settings [1, 2], and may result in long-term, potentially life-threatening cardiovascular sequelae [2,3,4,5,6,7]. First described in 1967 by Tomisaku Kawasaki, a Japanese paediatrician, the aetiology of KD is unknown [8]. The consensus is that KD results from an exaggerated immune response to one or more infectious triggers in individuals with a genetic predisposition [2, 9,10,11,12,13,14,15,16,17,18,19]. The management of acute KD aims to minimise systemic and vascular inflammation to prevent cardiovascular sequelae [4, 5, 20].

Standard primary therapy consists of intravenous immunoglobulin (IVIG) and aspirin [16]. The mainstay of treatment for KD in the acute phase is high-dose IVIG, usually plus aspirin. Treatment is ideally given within the first 10 days of fever [21,22,23] and reduces the incidence of coronary artery lesions (CALs) from approximately 25% to 5% [16, 24, 25].

The use and dose of IVIG is supported by high-quality randomised controlled trial evidence [16, 24, 25], but its mechanism(s) of action are incompletely understood [24,25,26,27]. It is believed that IVIG has a generalised anti-inflammatory effect, reducing pro-inflammatory cytokines, neutralising bacterial superantigens and down-regulating endothelial activation [3, 19, 25, 27,28,29,30,31,32,33]. It is worth noting that IVIG preparations and doses vary between centres, making direct comparisons difficult [34]. Aspirin is widely used in conjunction with IVIG, but its use is controversial [35,36,37]. There is little evidence for its role in CAL prevention and lack of consensus on dosing regimens [3, 23, 35, 36, 38, 39].

Approximately 10–20% of patients do not respond to standard primary therapy. Patients with “resistant” KD are at increased risk of coronary artery damage and associated sequelae [40,41,42,43,44,45,46,47,48,49]. There is a lack of high-quality evidence on the optimal management of this group of patients. Infants, particularly those less than 3 months of age, are more likely to present with incomplete KD (i.e. not fulfilling  full diagnostic criteria) [16, 50, 51] and are also at greater risk for resistant KD [52,53,54].

Here we review the current evidence for different management options for resistant KD when primary therapy has failed. Discussion of adjunctive primary therapy, such as concurrent corticosteroids, is outside the scope of this review.

2 Defining Resistant Kawasaki Disease (KD)

Approximately 80% of KD patients defervesce following initial IVIG, together with resolution of clinical signs and symptoms [22, 25]. The remainder have “resistant” KD, although various descriptive terms are used, including refractory, recrudescent, relapsing, recurrent, IVIG-resistant or IVIG non-responsive KD, and recalcitrant systemic inflammation.

The American Heart Association (AHA) defines resistant KD as “recrudescence or persistent fever at least 36 h following completion of the first dose of IVIG” [16]. However, different definitions for resistant KD have been used, both in terms of the timing of recurrence of fever (both whether the time-period begins at the start or end of the initial IVIG and how long the period is) and fever threshold, making comparisons difficult (Table 1). These differences are likely to contribute to the wide range of reported incidence (10–38.3%) of resistant KD [24, 40,41,42, 49, 55].

Table 1 Differing definitions for “resistant KD” with accompanying resistance rates

3 Second-Line Therapies for Resistant KD

A variety of interventions have been used in the management of resistant KD, including additional IVIG, corticosteroids, biologics, other immunomodulators, and statins. The aim of treatment is to reduce systemic and local vascular inflammation, although there are no robust biomarkers of efficacy, and studies therefore rely on resolution of fever, markers of systemic inflammation, and coronary artery outcomes.

Comparing different studies of second-line treatment is hampered by the variation in nomenclature and definitions for resistant KD (resulting in different thresholds for treatment). In addition, different concomitant or sequential anti-inflammatory agents can confound cause and effect analyses in these studies.

3.1 Further Doses of Intravenous Immunoglobulin

Many authorities suggest a second dose of IVIG at 2 g/kg as the first choice for resistant KD, with its putative dose-response effect [16, 41, 42]. Repeat IVIG has been demonstrated as safe and effective, but has never been tested in an adequately powered randomised trial [16, 41, 42, 56]. There may be a theoretical advantage in using a different IVIG product from that used for initial therapy, as preparations from diverse donor pools may have differing antibody repertoires and/or different amounts and composition and other anti-inflammatory factors [34, 57]. The recommended dose of IVIG for treatment of resistant KD varies, with some suggesting a lower dose of 1 g/kg [41, 42], especially in settings where IVIG is expensive and may not be widely available [58, 59]. An argument against using a second dose of IVIG is that this strategy may delay the initiation of other potentially more potent and effective second-line therapies.

Comparative studies investigating the efficacy of other therapies outlined in the following sections usually compare outcomes with a second dose of IVIG.

3.2 Corticosteroids

The use of steroids as treatment for resistant KD is reasonably well-established, despite the long and chequered history of steroid therapy in KD overall [60,61,62,63,64]. Steroids are relatively inexpensive and widely available [59, 65]. The controversies regarding the role of steroids as primary adjunctive therapy will not be discussed here.

Table 2 summarises studies investigating the role of steroids in resistant KD [58, 59, 63, 66,67,68,69,70]. It is difficult to draw robust conclusions because of the small numbers in individual studies, heterogeneity in patient groups as well as different treatment preparations and treatment regimens. Overall, the evidence suggests that the use of steroids results in an improvement in inflammatory markers, rapid defervescence and possibly reduction in the incidence of CALs [69, 71,72,73,74]. The AHA recommend that a short course of high-dose steroids would be a reasonable alternative to a second dose of IVIG, or a reasonable treatment after failure to respond following two doses of IVIG [16]. The alternative recommendation by the AHA for resistant KD is that in addition to a second dose of IVIG with aspirin, a long duration of steroids could be started [16]. However, there is no clear evidence on the optimal dosing, optimal formulation, timing and duration of corticosteroids [58, 75].

Table 2 Summary of studies using steroids for resistant KD

3.3 Tumour Necrosis Factor Alpha Inhibitors

During the acute phase of KD, it is suggested that levels of pro-inflammatory cytokines, including tumour necrosis factor alpha (TNF-α), are elevated, increasing the risk of CALs and resistance to IVIG [29, 33, 61, 76,77,78,79,80,81,81]. TNF-α is an inducible pro-inflammatory cytokine produced by T cells and macrophages [22]. Excessive production of TNF-α is associated with chronic inflammation observed in many immune-modulated inflammatory disorders [82, 83]. A Cochrane review is underway reviewing the role of TNF inhibition in both acute and resistant KD [84].

TNF-α inhibitors include infliximab, etanercept and pentoxifylline. Infliximab acts specifically on TNF-α, whilst etanercept is a soluble TNF receptor with a broader action on both TNF-α and lymphotoxins [85]. Pentoxifylline, a methylxanthine derivative, blocks the production of TNF-α and leukotrienes (inflammatory mediators that are part of the arachidonic acid pathway) from monocytes and macrophages [87,88,89,89].

Infliximab is a chimeric IgG1 monoclonal antibody with potent anti-inflammatory effects arising from blockade of TNF-α activity [91,92,92]. Levels of TNF-α and other pro-inflammatory mediators decrease after infliximab in patients with resistant KD who respond, but remain elevated in those who do not [33, 93, 94]. The use of infliximab has also been studied in primary intensification regimens in acute KD [95].

Table 3 outlines studies reporting the use of infliximab in resistant KD [90, 93, 97,98,99,100,100]. The heterogeneity of the populations between Japanese and non-Japanese, as well as differing therapies received pre-infliximab makes the data difficult to compare. Overall, it appears that infliximab causes a rapid defervescence in fever, resulting in a shorter length of hospital stay, and is relatively well tolerated. In the largest randomised trial using infliximab as adjunctive primary therapy with IVIG, there was no evidence that infliximab reduced the rate of KD resistance [95]. Infliximab does not appear to prevent CAL formation nor reverse existing lesions [93, 96, 99,100,101,101]. However, relatively small numbers of patients received infliximab in these studies. Also worth noting, half of the studies outlined in Table 3 refer to the use of infliximab as third- or even fourth-line therapy, i.e. following at least a second dose of IVIG.

Table 3 Summary of the evidence for the use of infliximab for resistant KD

The AHA recommendation is that infliximab may be substituted for a second dose of IVIG, or steroids, in resistant KD [16]. A multicentre, prospective, randomised study in Shanghai determining the effect of infliximab on development of CALs in resistant KD has been completed, but the results have not yet been published [102].

There is little evidence for the use of etanercept or pentoxifylline in both acute and resistant KD aside from limited case reports [88, 89, 103, 104]. The effect of etanercept on arteritis has previously been investigated in mouse models, where it is most effective in suppressing inflammatory cytokines and reducing the extent of vasculitis [85]. Case reports of anti-TNF-α agents in resistant KD are summarised in Table 6 and the electronic supplementary material (ESM, Table 1).

3.4 Cyclosporin A

Genetic variants in the inositol-trisphosphate 3-kinase C (ITPKC) gene have been implicated in heightened susceptibility to KD and resistance to IVIG [106,107,107]. ITKPC is also involved in inflammasome activation, a component of the innate immune response that results in interleukin-1 (IL-1) production [109,110,110]. Defects in this pathway lead to increased inflammation [108].

Cyclosporin A (CsA), a calcineurin inhibitor, suppresses activity of T cells by targeting the Ca2+/NFAT signalling, thus dampening the inflammatory response [105, 111]. A trial in Japan is underway comparing the use of CsA in combination with conventional therapy in acute treatment of KD in high-risk patients [112].

In resistant KD, the limited evidence for CsA predominately comes from case series where it is used as a third-line agent [114,115,115]. Three case series are outlined in Table 4. These demonstrate that in the majority of patients, fever defervesced within 1–5 days of CsA introduction. The AHA recommend consideration of cyclosporin in patients with refractory KD where a second dose of IVIG, infliximab or a course of steroids has failed [16]. Case reports of where cyclosporin is used in resistant KD are summarised in the ESM.

Table 4 Summary of evidence of CsA for resistant KD

3.5 Methotrexate

There are a few studies which describe methotrexate use in resistant KD. In patients with rheumatoid arthritis, methotrexate administration leads to a reduction in pro-inflammatory cytokines [116, 117]. Its mechanism of action is incompletely understood, but is thought to be mediated by the intracellular accumulation of polyglutamate metabolites [118]. It is commonly used in inflammatory conditions such as rheumatoid arthritis and both small and large vessel vasculitides as a steroid-sparing immunomodulatory agent. However, the pharmacodynamics of low-dose methotrexate and its putative mechanism of action are such that it is unlikely to have significant acute anti-inflammatory activity [119]. Not surprisingly, the AHA does not make any specific recommendations relating to methotrexate for resistant KD [16]. Case series relating to the use of methotrexate in resistant KD are summarised in Table 5 [120, 121]. Further case reports using methotrexate are further described in Table 6 and the ESM.

Table 5 Summary of the evidence for the use of MTX for resistant KD
Table 6 Case reports involving less commonly used agents for resistant KD

3.6 Other Agents

There are numerous reports of the use of other agents, including other biologics, cytotoxic agents, ulinastatin, and plasma exchange, in resistant KD. These agents should be reserved for highly refractory patients who have failed other therapies. There is also increasing interest in the role of statins in the long-term management of patients with KD to minimise ongoing vascular inflammation and atherosclerosis [123,124,125,125]. Further prospective studies are needed to assess the role, efficacy, and long-term safety of these agents in the management of resistant KD.

Anakinra is a recombinant IL-1 receptor antagonist that inhibits both IL-1α and IL-1β [126, 127]. It has been used in other childhood inflammatory conditions [126, 129,130,130]. The role of IL-1 in inflammation, and more specifically, in causing coronary artery inflammation, has been demonstrated in both laboratory and animal studies [30, 131, 132]. Notably, the elevated levels of IL-1 that have been observed during the acute phase of KD may be associated with a higher risk of resistance to IVIG, as well as an increased risk of CALs [21, 28, 31, 131, 134,135,135]. There are currently studies underway investigating the role of anakinra in both acute primary treatment and resistant KD [136, 137]. The only documented use of anakinra in resistant KD is outlined in case reports (summarised in Table 6 ) [139,140,140].

Rituximab is a chimeric monoclonal antibody specifically directed against CD20 surface antigen present on B cells [141, 142]. It is increasingly used in other systemic vasculitides in children and is well-tolerated [82, 143, 144]. During the acute phase of KD, B cells are activated, leading to a cytotoxic antibody response, which affects endothelial cells [82]. The use of rituximab in resistant KD has been described in one case report (Table 6 ) [145].

Ulinastatin is a urinary trypsin inhibitor, which by inhibiting neutrophil elastase may have a dose-dependent effect on endothelial cell injury [15, 147,148,148]. Its efficacy is questionable, but it is used widely in Japan in both primary and adjunctive management of KD [15, 148]. Its use has been described in two case reports in combination with other agents in resistant KD (Table 6) [149, 150].

The use of plasma exchange in severe resistant KD has been described in case series, with a focus on presumed effects on reducing plasma cytokine levels and reducing CALs [152,153,154,155,156,156]. All documented case series use plasma exchange as third- or fourth-line therapy for resistant KD [90, 97, 100].

In addition to their lipid-lowering effects, the anti-inflammatory and anti-oxidant properties of statins have long been recognised [122, 125]. In KD, the proposed utility of statins is in reducing endothelial dysfunction and minimising atherosclerosis; however, there are currently insufficient data to inform timing, duration of therapy, and potential long-term side effects [123, 124, 157, 158]. There are currently studies underway investigating the safety and efficacy of atorvastatin in children with KD with persistent CALs [159, 160].

4 Limitations

There are important limitations to the data on the management of resistant KD, which make comparison between treatment options difficult. Importantly, the definition of “resistant KD” differs markedly between studies, with differences in the total number of doses or total cumulative doses of IVIG, the number of hours between the end of the initial IVIG and fever recrudescence, the anatomical site of temperature measurement, the threshold used to define fever, and whether additional features are part of the definition of resistance. Differing thresholds for treatment are a major potential source of bias.

Almost all studies are retrospective, with inherent limitations. These include issues regarding the certainty and timing of the initial KD diagnosis (particularly important if there is poor response to initial therapy). This may be compounded by variation in dosing and administration of initial treatment regimens, and differences in concomitant anti-inflammatory and/or other agents during primary therapy.

CALs and aneurysms are the most important clinical outcome. Importantly, it should be noted that some studies use the Japanese Ministry of Health (JMoH) criteria (based on internal diameter of the coronary arteries), whereas others use the AHA definition (based on coronary artery z scores). This can lead to issues of under-estimation and under-diagnosis of coronary lesions, as coronary artery z scores are considered more sensitive [161, 162]. Of note, some studies excluded patients with coronary artery abnormalities (prior to administration of their second- or third-line therapy) in their analysis. In addition, interpretation of echocardiograms may not be blinded, and these are potentially subject to bias.

Finally, the natural history of KD is that fever usually resolves after 2–3 weeks without specific treatment [16]. Therefore, defervescence after this time period may be falsely attributed to an intervention, when it may reflect the spontaneous resolution of the acute phase of KD.

5 Conclusions

Despite over 5 decades of research, the aetiological trigger and pathogenesis of KD remain poorly understood. It is clear that abnormal inflammation during the acute phase of KD increases the risk of CALs and cardiovascular sequelae. Management of resistant KD has thus focussed on downregulation of the host inflammatory response, aiming to reduce the risk of formation of new CAL and to halt progression of existing lesions.

The current evidence to guide treatment of resistant KD is of moderate to low quality, which is unfortunate given the potential severity of adverse outcomes of both KD itself and the treatments, and the cost of some proposed agents. As suggested by the AHA, [16], it seems reasonable to consider additional IVIG and/or steroids if the initial IVIG does not result in sustained defervescence of fever.

The relative infrequency of resistant KD means a prospective trial is unlikely to be sufficiently powered to provide an evidence base for appropriate treatment. The published data differ substantially between patient groups and treatment approaches, making comparison problematic and robust conclusions difficult. Expert consensus (such as the AHA guidelines) [16] together with accumulating published clinical experience will continue to guide management of resistant KD.