What Caught our Eye: Kawasaki Disease...Incomplete, or...not?
Brian Feldman, MD, MSc, FRCPC
University of Toronto – The Hospital for Sick Children
I know that the rheumatology service isn’t involved in the treatment of most Kawasaki Disease patients at many centres. At SickKids, though, we are. Our team sees somewhere around 150 patients with KD a year.
I hear commonly, mostly from fellows who join us from other centres, that when it comes to KD, the faculty at SickKids tends to err on the side of over-treatment. How can one over-treat a condition for which randomized trials clearly indicate an important benefit of treatment? The newly arrived fellows imply that we over-diagnose.
Over-diagnosis and treatment with a completely benign therapy wouldn’t be a problem. However, the most commonly used therapy – intravenous immunoglobulin – is both costly and potentially harmful.
The diagnosis of Kawasaki Disease was originally made using the mucocutaneous lymph node syndrome (MCLS) criteria, of which you needed 5 of 6. As a “syndrome”, the call was easy. You either met the syndrome definition, or not.
By the early 1980s, though, it was recognized that there were children, who had developed coronary aneurysms typical for KD, who hadn’t met the MCLS criteria. They were labelled as having “atypical Kawasaki Disease.” A requirement for this diagnosis was having coronary aneurysms.
It seemed obvious, therefore, that there was some disease process that was occurring that was independent of all the phenotypic features of MCLS; there was, and is, a strong desire to figure out who is at risk for coronary lesions – whether or not they meet syndromic criteria – and treat them with proven preventative therapy. This led to the coining of the term “incomplete Kawasaki Disease,” a diagnostic label for patients with fewer than the required criteria for KD, who yet might be at risk for coronary vasculitis.
Parenthetically, I don’t like the term “incomplete.” You either have a disease, or not. For these children, the issue is (to my way of thinking) you are either at a high enough risk for coronary vasculitis that you need to be treated, or not.
The MCLS criteria, and the American Heart Association (AHA) criteria for “full” KD don’t seem to be enough to determine the risk level coronary lesions. Indeed, meta-analysis has suggested that patients having incomplete KD have a higher risk of coronary abnormalities than those with full KD. (The problem, of course, is that we have no idea about the denominator in any of the studies that made up the meta-analysis. Full KD is likely under-diagnosed in many places; there are likely millions of febrile children that could meet a diagnosis of incomplete KD that never get imaged.)
Over-treatment will likely increase with the new AHA guidelines. The 2004 AHA guidelines suggested using laboratory criteria to predict who, among the patients meeting fewer than full KD criteria, should get echocardiography. Treatment was only suggested for those with imaging evidence of coronary artery abnormalities. The new AHA guidelines have changed this recommendation; they now recommend treating all patients with incomplete clinical criteria (even just a fever longer than 7 days with inflammatory markers would do) if they have high risk, “supportive” laboratory features (3 or more of anemia, high platelet count, low albumin, high ALT, high WBC and urine leukocytosis). These are very well-meaning suggestions, but I believe they will lead to over-treatment; the evidence of just how at risk this population of patients is, is based on the wrong kinds of studies (mostly case-control studies, for which there is no proper denominator).
A real problem is that we don’t yet have very strong risk predictors for coronary lesions, at least not strong enough to make decisions about who to treat, and who not to treat. For example, the Harada score has only 33% specificity for coronary lesions in incomplete KD (in US children). In that study there were far more false positives than true positives.
So, it was with great interest that I read Xie and colleagues’ paper in Pediatric Rheumatology. In this study, from the Children’s Hospital in Hangzhou, they studied patients with “complete” KD (meeting full diagnosis, n = 410) and incomplete KD (n = 150). They were interested in predicting response to IVIG (no fever after 48 hours post-treatment) and the development of coronary artery lesions (based on lesion size, rather than z-score). As predictors they studied white blood cell count, proportion of neutrophils, hemoglobin, C-reactive protein, erythrocyte sedimentation rate, albumin, N-terminal pro-brain natriuretic peptide (NT-proBNP), IL-6 and IL-10. Over a third of the incomplete KD group developed coronary lesions; all patients were treated with IVIG and ASA.
What they found, for the incomplete KD group, was that IL-6, IL-10 and NT-proBNP were not at all predictive of the eventual development of coronary lesions. Some of the other lab tests (depending on cut-off) had either high sensitivity or specificity, but not both. For example, a CRP of >140 (mg/L) had a specificity of 90%, but a sensitivity of only 21%; an albumin of <30 (g/L) had a specificity of 94%, but a sensitivity of only 14%. Recall, all of these children with treated.
What these results mean – by my calculation – is that if you have a patient with prolonged fever and 2 of 5 criteria, in whom the background probability of developing a coronary artery abnormality after treatment is, say, 5% , and the albumin was<30 and the CRP was >140, you might predict a new risk of 26%. So, in fact, while none of the predictors was that strong, overall the results are useful.
Unfortunately, these kinds of studies don’t answer the question about treating children with features of incomplete KD – but who aren’t identified and diagnosed as such. We want to be able not to over-treat, while at the same time not missing any patients who might develop coronary artery damage (and who should, therefore, be treated). To do this, we need true population-based cohort studies, that capture febrile infants and children with all kinds of clinical and laboratory findings (i.e., the kinds of kids that present to offices and emergency departments), not treat them, and do echocardiography in follow-up on all of them.
Short of that, I hope the new AHA guidelines will work to reduce the burden of heart disease, without unduly skyrocketing costs and adverse events. But... there is at least a potential that we will do more harm than good.
BE A GUEST CONTRIBUTOR!
If you would like to contribute an article (related to pediatric rheumatology research) to the What Caught our Eye column, we would love to hear from you! Please email [email protected] with details.