Beyond biochemical recovery: rethinking liver dysfunction in Kawasaki disease

Liver dysfunction is among the most common extracardiac manifestations of Kawasaki disease (KD); however, its clinical importance remains poorly understood (1,2). Abnormalities in serum aminotransferases, bilirubin, γ-glutamyl transferase, and other liver-related parameters are frequently observed during the acute phase, but severe or persistent hepatic injury is uncommon in most patients (1,2). Contrarily, the main clinical concern in KD remains coronary artery involvement, as coronary abnormalities determine long-term cardiovascular risk and represent the primary target of established treatment strategies (3). This difference in prognostic importance creates fundamental uncertainty regarding hepatic involvement: should liver dysfunction be regarded as a condition requiring specific treatment or is it mainly a biochemical manifestation of systemic inflammation that improves with appropriate KD management?

In the current issue, Wang et al. (4) compared compound glycyrrhizin (CG) and magnesium isoglycyrrhizinate (MgIG) in children with KD and elevated liver enzymes using a retrospective propensity score-matched cohort. Their study is the first to directly compare these two hepatoprotective agents in this setting. The authors found that serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) markedly decreased in both groups. Although the unadjusted partial biochemical response rate was higher with CG, the treatment group did not exhibit an independent association with response following multivariable adjustment. Earlier initiation of hepatoprotective therapy was the only factor markedly associated with biochemical improvement.

The authors appropriately acknowledge that such findings are preliminary and hypothesis-generating and that biochemical improvement cannot be assumed to translate into clinical benefit (4). Although the findings are of interest, they should be interpreted with careful distinction between comparative effectiveness and therapeutic efficacy. The study compared one active hepatoprotective treatment with another (4). It did not compare CG with standard KD therapy alone, MgIG with standard KD therapy alone, or either drug with no hepatoprotective treatment. Consequently, the observed reductions in ALT and AST in both groups do not establish that either drug caused the improvement. Furthermore, it cannot determine whether hepatoprotective therapy itself provides benefit beyond the expected course of KD following treatment of the underlying inflammatory disease.

This distinction is particularly important in KD as liver enzyme abnormalities often improve following intravenous immunoglobulin (IVIG) therapy (5). In routine clinical practice, many patients with elevated ALT or AST at presentation exhibit substantial biochemical improvement without receiving a specific hepatoprotective drug. The lack of a group treated with IVIG and standard care alone makes it impossible to separate the effects of CG or MgIG from spontaneous biochemical recovery, regression toward the mean, or improvement secondary to systemic inflammatory suppression. Future investigations should first determine whether hepatoprotective therapy adds benefit to standard KD treatment before identifying which hepatoprotective agent is preferable.

The choice and definition of outcomes also need to be cautiously interpreted. Partial biochemical response was reported in 86.25% and 69.39% of the CG and MgIG groups, respectively (P=0.02), whereas complete normalization occurred in 45.00% and 30.61%, respectively (P=0.11) (4). Notably, partial response was defined as a reduction in ALT and AST to <2 × the upper limit of normal, thereby including patients who achieved complete normalization. Thus, differences in statistical significance between these nested outcomes should be interpreted in light of this overlap. More importantly, ALT and AST are useful markers of hepatocellular injury; however, they are surrogate biochemical endpoints. As with other surrogate endpoints, improvement in these laboratory parameters alone does not necessarily translate into meaningful clinical benefit (6). In KD, the most clinically important outcomes include the resolution of fever and inflammation, responsiveness to the initial IVIG therapy, avoidance of rescue treatment, prevention of coronary artery abnormalities, and long-term cardiovascular prognosis (3). It remains unclear whether accelerating aminotransferase normalization influences any of these outcomes.

This issue is not merely semantic. A statistically significant reduction in a laboratory parameter may have limited clinical relevance if the abnormality would have spontaneously resolved without intervention and if its improvement does not affect symptoms, treatment intensity, length of hospitalization, coronary artery risk, or long-term health outcomes. Conversely, even modest or transient liver test abnormalities may be clinically informative if they identify a subgroup with a more pronounced systemic inflammation, a higher likelihood of IVIG resistance, or a greater risk of coronary complications. Therefore, the clinical value of hepatic involvement may lie less in its role as a treatment target and more in its potential role as a marker of disease phenotype or severity (7,8).

The pathophysiology of liver dysfunction in KD remains unclear (2,8). It is unlikely attributable to a single mechanism, and the available clinical findings suggest substantial heterogeneity. Possible contributors include systemic inflammatory injury, cytokine-mediated hepatocellular damage, vascular involvement, cholestatic changes, altered bile flow, gallbladder hydrops, and treatment-related hepatotoxicity (2,8). A child with predominantly hepatocellular aminotransferase elevation may have a different underlying process from one presenting with jaundice, cholestatic laboratory abnormalities, or gallbladder enlargement. This heterogeneity also complicates interpretation of treatment response. If elevated aminotransferases mainly reflect systemic inflammation, improvement would be expected following effective anti-inflammatory treatment regardless of whether a hepatoprotective drug is administered (5). If cholestatic injury predominates, ALT and AST alone may not adequately capture the relevant disease process (2). Likewise, if drug-induced liver injury contributes to hepatic involvement, modification or discontinuation of the causative medication may be more important than addition of a hepatoprotective agent (2). Therefore, a treatment effect observed in one phenotype may not apply to another. Grouping all such presentations under the broad label of “KD-associated liver dysfunction” risks treating biologically distinct conditions as if they were a single entity.

The timing of liver enzyme elevation may provide additional clues (5). Abnormal liver tests detected at initial presentation likely reflect a different clinical context from aminotransferase elevations that develop or recur during the recovery phase. In clinical practice, secondary increases in ALT or AST following defervescence should prompt consideration of aspirin-associated drug-induced liver injury as part of the differential diagnosis, particularly when the inflammatory features of KD have otherwise improved (2). As aspirin is part of the standard KD treatment in many settings, changes in aminotransferase levels during follow-up may reflect multiple processes, including recovery from inflammatory liver involvement, drug adaptation, aspirin therapy modification, ongoing KD activity, or drug-related hepatotoxicity, rather than the effects of hepatoprotective treatment alone (2). The present study investigated biochemical changes approximately 1 week after the initiation of hepatoprotective therapy but did not clearly distinguish persistent acute-phase liver abnormalities from delayed or treatment-related elevations, nor did it provide detailed information on aspirin dose, dose modification, or the temporal association between aspirin exposure and liver test abnormalities (4). Such limitations may introduce important clinical heterogeneity and complicate the interpretation of spontaneous recovery and treatment response.

The finding that earlier initiation of hepatoprotective therapy was independently associated with a greater likelihood of biochemical response is interesting, although the effect estimate was modest and its 95% confidence interval (CI) approached the null (odds ratio, 0.810; 95% CI, 0.663–0.989) (4). This association should not be interpreted as evidence of a causal benefit from early hepatoprotective treatment. Treatment timing in retrospective observational data is closely associated with many other aspects of care. Earlier initiation may reflect earlier KD recognition, hospitalization, and IVIG administration; different physician practice patterns; or less-advanced inflammatory injury. It may also reflect confounding by indication, as clinicians can delay additional therapy in patients whose liver test abnormalities seem to improve without intervention. Although multivariable adjustment was performed (4), it cannot account for all clinical factors influencing treatment decisions. Thus, the association between earlier hepatoprotective therapy and laboratory recovery may simply indicate differences in disease timing rather than an actual therapeutic window.

The propensity score-matching approach improves comparability between the CG and MgIG groups (4,9). Propensity score methods address measured differences in treatment allocation between the two active-treatment groups; however, they cannot provide evidence regarding what would have happened in the absence of either treatment (9). Moreover, a substantial proportion of patients, particularly from the CG group, were excluded during matching, with these patients demonstrating more severe biochemical abnormalities (4). The resulting matched cohort therefore represents patients within the region of overlapping propensity scores rather than the full clinical spectrum of KD-associated liver dysfunction (9). This limits the generalizability of the findings, particularly to patients with more pronounced liver enzyme elevations or more severe hepatic involvement.

Safety also warrants cautious interpretation. All five adverse events reported in the study were episodes of hypokalemia, occurring in the CG and MgIG groups (4). Hypokalemia is a well-recognized adverse effect of glycyrrhizin-containing preparations and may result from mineralocorticoid receptor activation following inhibition of renal 11β-hydroxysteroid dehydrogenase type 2 by glycyrrhizin metabolites (10). However, the present study did not report the definition, severity, or timing of hypokalemia, nor its temporal association with IVIG administration and hepatoprotective therapy (4). This distinction is important as hypokalemia may also occur shortly after IVIG treatment in children with KD (11). However, direct comparison of its frequency is difficult, as the present study did not specify the threshold used to define hypokalemia. Therefore, the five events cannot be confidently attributed to CG or MgIG or regarded as evidence of pseudoaldosteronism. Nevertheless, the fact that all reported adverse events were hypokalemia requires attention, particularly because the study involved young children and employed broad weight-tiered dosing rather than precise weight-based dosing. Owing to the relatively short treatment duration (median, 7 days; interquartile range, 5–8 days), the absence of serious adverse events is reassuring with respect to short-term tolerability; however, the limited information on hypokalemia restricts the interpretation of the safety findings. More detailed monitoring of serum potassium and blood pressure during treatment, alongside documentation of the definition, timing, severity, and clinical course of hypokalemia, would be warranted to more adequately characterize the short-term safety profile.

An equally important direction is to determine whether liver dysfunction identifies clinically meaningful KD subphenotypes. Recent data-driven cluster analysis has revealed a hepatobiliary subtype characterized by elevated liver enzymes, the highest rate of IVIG resistance, and a relatively low incidence of coronary artery aneurysms (12). These findings suggest that hepatic involvement may reflect a distinct disease phenotype rather than simply an extracardiac manifestation of KD. If specific biochemical profiles are consistently associated with IVIG resistance or coronary artery abnormalities, liver function tests may aid risk stratification even if they do not represent direct therapeutic targets (8,12). Such a perspective shifts the research question from “How should liver enzymes be normalized?” to “What does hepatic involvement tell us about the biology, phenotype, and prognosis of KD?”

A stepwise approach may help elucidate the role of hepatoprotective therapy in KD. Initial observational studies should better characterize the temporal patterns and phenotypes of liver dysfunction (8,12). Such phenotypes could then be examined in relation to inflammation, IVIG response, coronary artery outcomes, and medication exposure (7,8,12). If a clinically relevant subgroup with persistent or symptomatic hepatic involvement is identified, prospective comparisons between hepatoprotective therapy and standard KD treatment alone would be reasonable. Only after an incremental benefit has been established should comparative trials between different hepatoprotective agents be prioritized.

The study by Wang et al. is valuable as it systematically investigates a common but insufficiently understood manifestation of KD (4). Its principal contribution, however, may lie not in demonstrating the superiority of one hepatoprotective agent over another but in providing an important foundation for future prospective studies while underscoring the current limitations of the available evidence. CG and MgIG may have similar short-term biochemical outcomes (4); however, whether either agent improves the natural course of liver test abnormalities, modifies KD disease activity, or benefits outcomes that matter to children remains unclear (6).

The more fundamental question, therefore, is not which hepatoprotective drug should be selected, but rather what clinicians are attempting to treat. If aminotransferase elevation is mainly a transient marker of systemic inflammation, normalization would be expected with the resolution of KD and may not warrant a specific intervention (5). If hepatic dysfunction identifies a high-risk inflammatory phenotype, its main value may be prognostic (7,8,12). If a subset of patients develops clinically significant cholestatic, gallbladder, or drug-induced disease, management should be tailored to the underlying mechanism rather than uniformly directed toward hepatoprotection (2). Until these possibilities are clearly distinguished and a therapeutic benefit beyond standard KD care is established, biochemical recovery should be primarily interpreted as a marker of disease resolution rather than a therapeutic endpoint in its own right.

Supplementary

The article’s supplementary files as

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.