Hedgehogs Wanted

Philosopher Isaiah Berlin argued that great thinkers come in two varieties: foxes and hedgehogs. Foxes dabble, often brilliantly, in many things; hedgehogs discover and develop one big defining idea. Scientific foxes are “splitters,” focusing on nature’s differences rather than its commonalities. Hedgehogs are “lumpers” who see the big picture. Physics is famous for scientific hedgehogs like Maxwell (who in 1864 unified the previously distinct conceptual domains of electricity and magnetism in his Electromagnetic Theory) and Einstein (who unified space, time and gravity in his over-arching Theory of General Relativity).

Based on what I observed at the recent Third World Parkinson Congress in Montreal, current Parkinson’s research is dominated by splitters. At dozens of presentations, I watched some brilliant scientific foxes argue that everyone’s PD is different; that each patient’s varied package of symptoms (from traditional complaints like tremor and rigidity to “non-motor” features like sleep disorders, depression, anxiety, hallucinations, and dementia) results from a mind bogglingly complex interaction between genes, environment, lifestyle, age, and luck. As University of Vancouver geneticist Matthew Farrer put it, “there are many paths to Parkinson’s.”

Does this heterogeneous image of PD simply reflect scientific reality, or does it suggest that the PD research community lacks hedgehogs of the caliber of Einstein and Maxwell; big thinkers who can develop grand, unifying theories? As C. Warren Olanow’s beautifully crafted keynote lecture telling the history of PD made clear, it wasn’t always this way. The first part of his narrative (from 1817 to roughly 1970) was a gripping tale of heroic clinicians and scientists building a unified and coherent picture of a disease. These scientific hedgehogs included James Parkinson (the author of the Essay on the Shaking Palsy), Jean-Martin Charcot (the brilliant French clinician who nailed the clinical signs and symptoms of what appeared to be a movement disorder), Konstantin Tretiakoff (the Russian student who demonstrated that the disease was caused by damage to the substantia nigra region of the brain), Frederic Lewy (after whom “Lewy bodies” – the pathological hallmarks of PD – are named), Arvid Carlsson (who suggested that the disease resulted from a shortage of dopamine), and Birkmayer, Hornykiewicz and Cotzias (who showed that levodopa could reverse the symptoms in humans).

But once Olanow’s presentation reached 1970 and beyond, his story became more and more fragmented – with advances in epidemiology, clinical neurology, physiology, neurosurgery, genetics (including genome wide association studies), and neuropathology going on simultaneously. The aggregate picture that emerges from this research is sobering. Researchers no longer view PD as a singular dopamine-centered movement disorder, but something vastly more complicated: A malady with a highly complex etiology, multiple potential disease pathways (involving both dopaminergic and non-dopaminergic circuits), diverse clinical symptoms – from motor (e.g. tremor, rigidity and slowness) to “non-motor” (disabling fatigue, pain, constipation, sudden drops in blood pressure, sleep disorders, depression, anxiety, hallucinations, and dementia – and mixed pathology (involving both alpha-synuclein and tau proteins).

In a sense, in this paradigm every parkie has his/her own version of the disease. This picture makes perfect sense to epidemiologists and geneticists. For example, Parkinson’s Institute’s Caroline Tanner says, “the combination of genetics and environment is central to understanding PD etiology…Take head trauma as a risk for PD… By itself, concussions might be a small risk factor for PD, but when combined with a genetic predisposition – a variant in the alpha-synuclein gene that has a tiny effect in people without head injury – the risk increases 11-fold.”

But what does it mean for therapies? For one thing, it makes clinical trials much more complicated. As Haydeh Payami, Professor of Molecular Genetics at the New York State Department of Health Wadsworth Center put it, “for any given drug there are some responders and some people who do not respond, and the non-responders dilute the effect of a drug.”

But what if it were possible to preselect the responders for the trial? Then it might be possible to target drugs at clinical subtypes. Take the protective effect of smoking and coffee drinking found in epidemiological studies. Says Payami, “each one will lower the general risk of Parkinson's disease by between 20 and 30%.” But, she argued, genes based on the actions of nicotine and caffeine have potentially a much stronger protective effect, up to 80%. So, if we “personalized” treatments, taking account of people’s genetics, lifestyle and age, we might, she said, have much more success in clinical trials. True, but given the pharmaceutical companies current lack of enthusiasm for CNS drug development, is it realistic to expect them to develop a multitude of medicines, each targeted at small cohorts of patients sharing a clinical subtype?

As neuroscientist Patrik Brundin pointed out to me, good science depends on finding important distinctions and ignoring irrelevant ones. If researchers back in the 1950s had treated everyone complaining of a sore throat with antibiotics, that trial would have failed (just as neurotrophic factors failed for PD). The question is why? If researchers had (reasonably) tried to split patients into different phenotypes – whether this one had a runny nose, that one a cough, bloodshot eyes etc – that would have been a waste of time, because the relevant classificatory distinction lies elsewhere; namely that around 90% of sore throats are viral and only 10% caused by bacteria.

I detected glimmers that some researchers were pushing back against excessive splitting. Hammersmith Hospital’s David Brooks said the idea of “developing personalized treatments for people is something of a nightmare…” You can if you wish to split PD into “PD sleep, PD constipation, PD depression, PD whatever, but unless this ties in with useful (…) therapeutic guidance… you’re simply classifying disease for the sake of classifying disease.”

Brooks makes a good point. I would go further and make the following suggestion: Because Parkinson’s is much more complicated and multifaceted that was once thought, it can be argued that the field badly needs hedgehogs with big unifying ideas; ideas that identify the disease’s Achilles heel, wherever that may be. Hedgehogs were indeed hard to find at the WPC – and I don’t count Andreas Kottman’s talk about a brain protein, which has the cute name “Sonic hedgehog.” But it would be wrong to say there are no big thinkers in PD research. One hedgehog, whose name was frequently mentioned, is German neuroanatomist Heiko Braak. His staging theory, in which alpha-synuclein pathology spreads from one brain region to another, causing sequential mayhem in its wake, is a powerful organizing idea that appears to explain much of PD’s clinical complexity. At first, he met with skepticism, but today his paradigm has won many converts. And the WPC hosted some other big thinkers, whose frustration with the many failed clinical trials had inspired them to think outside the box. UK neuroscientist Roger Barker is challenging the hegemony of randomized double blind controlled trials – as is Dutch neuroscientist Bastiaan Bloem. And the UK charity Cure Parkinson’s Trust (together with neuroscientist Patrik Brundin) are seeking to revive the CNS drug pipeline by employing the provocative strategy of repurposing existing FDA-approved drugs for PD (such as the diabetes drug Exenatide).

Despite its many scientific advances, PD research may not be quite ready for its Albert Einstein. As philosopher of science Thomas Kuhn famously argued, “paradigm shifts” (like Einstein’s) only happen when a field of science is ripe for change. If we still have a lot of fundamental truths to learn about the brain, then the foxes must continue for a while. But if PD research is indeed primed for a scientific revolution – and I choose for reasons both objective and personal to believe it is – then what’s needed most now are lumpers like Braak, scientists looking to conceptually unify nature rather than splitters that keep parsing it into smaller and smaller slices.