A letter in Nature Reviews Drug Discovery has argued that it is time to reform the taxonomy of disease. The authors are Ismail Kola, head of a pharmaceutics company, and Sir John Bell, Regius Professor at Oxford.
The taxonomy of disease (technically called nosology) is an old one. Of course the Greeks, notably Galen, had their version, and in the 17th century it was a hot topic as well. The problem seems to be that there is a conflict between the phenomenology of disease, and the etiology. That is, we classify on a mixture of what seems to us to appear as a single thing, and on the basis of what we know about causes.
In my forthcoming book with Malte Ebach on Natural Classification, we term the former classification by analogy and the latter classification by homology, using the biological terms as general tags. Classifying natural phenomena based on our dispositions and predilections ultimately tells us mostly about how we react to stimuli. Classifying on the basis of known causes allows us to properly “cut nature at its joints”. But we do not have direct access to causes before we achieve successful theory and modelling, so instead we have to deal with shared properties.
And here is the conundrum. Which properties will tell us about the world, and which about ourselves? The answer is that we do not know, to begin with. Kola and Bell use a number of examples and approaches to suggest how to reform nosology: nearly all of them involve the identification of molecular and genomic pathways in order to find therapies. This may be a useful initial step; but if we take molecular biology to be primary we may be begging some questions.
The authors mention the new version of the DSM, and it is a salutary lesson in the failure to distinguish between phenomenology and etiology. Initially, I am told by Dominic Murphy at the University of Sydney, who specialises in this topic, the DSM was largely based on (largely Freudian) phenomenology with the hope that these categories of mental disorders would be replaced with more biological causal types over time. Instead what happened was that as etiologies were discovered they got added, but that phenomenological classes were also added, mostly because there was a need for a diagnosis that could attract government support either for researchers or patients. Such “diseases” as “autism spectrum disorder” or ASD are heterogeneous grab bags of phenomenal behaviours that seem like each other. Anyone who has a child who falls under that category knows how ridiculous the ASD class is. DSM-V will collapse all variations such as Asperger’s Syndrome into that one class. [Use of the rider "syndrome" in nosology is a fine guide to it being a class based on apparent similarities and not any causal criteria.]
Nobody starts investigating a phenomenon in the absence of prior beliefs and experience, so there is nothing wrong with beginning with phenomenal and analogical classifications of this kind. The problem is that medicine seems to be content all too often to rest with them.