Association or causation: Symptoms and rare disease

Abstract

On pp x-y, Knerr et al have expanded our knowledge of a rarely found deficiency, that of mitochondrial branchedchain amino acid transferase 2 (BCAT2). They describe five new cases to add to the one already reported. In BCAT2, there is a distinctive biochemical profile: elevated branched-chain amino acids, normal branched-chain α-keto acids, and allo-isoleucine. This is easily detected by tandem mass spectrometry. The cases described were found because of symptoms in three, family testing in one, and in a baby identified by newborn screening who was immediately placed on a low branched-chain amino acid diet. The cases with symptoms had either autism, intellectual delay, or both. One suffered from additional hereditary spastic paraparesis. That the NBS-detected baby was treated has complicated the already tenuous interpretation of the symptoms—are they causally related to the underlying biochemical defect or not? This is not a new problem in rare disorders. Soon after newborn screening was first introduced, a “new” disorder of histidinemia came to the fore. This was thought to cause intellectual and speech problems. Treatment was a low histidine diet. Many affected babies were treated. In some patients, investigations even included liver biopsy. Soon, there were suggestions that the disorder was benign, but this took years to be substantiated. In the other direction, some disorders initially thought probably benign were later found to have an adult onset phenotype. One example is 3-Methylglutaconic aciduria type I. With expanded newborn screening, the problem has intensified. One prime example of a usually symptom-free condition is 3-methylcrotonyl carboxylase deficiency (3MCCC deficiency), the most common organic aciduria found on screening but extremely rarely identified previously. However, in this case, about 5% of affected patients can have acute symptoms in infancy. In 1965, the eminent epidemiologist Bradford Hill published nine “viewpoints” to determine if and when observations of associations should infer causation. In the molecular scientific world of today, this issue is more complicated, but it is still a very useful paper to revisit. The “viewpoints” include the strength of the association, consistency, specificity, biological gradient (eg, dose-response), and plausibility. However, Hill s nine aspects of association were never intended to be viewed as rigid criteria or as a checklist for causation, and not all are entirely applicable to rare inborn errors. It is important that we do not ignore something potentially serious in what appears on first encounter as possibly benign. However, it is also equally important that, by reporting of symptoms without any caveats, authors do not initiate assumptions that are not supported by evidence. So, the correct wording of conclusions in the text and in the abstract is important—especially as some people appear only to read abstracts. Findings of, for example, apparent treatment effects must not be overstated. (Fortunately, this is not the case in this current BCAT2 paper.) Now may be a good time to start to develop some guidelines appropriate for assessing the problem of possible causation in rare inborn errors of metabolism.