‘Pathology is disease’ Parkinson's mythology: The ‘brain-first-body-first’ case study

Can you think of a scientific hypothesis that has been tested and rejected in your field of medicine?

[There is no rush. Give yourself some time.]

If you couldn’t think of any, you may be a neurologist, with interest in caring for patients with Parkinson's disease (PD).

Neurologists along with psychiatrists are the two medical specialties most resistant to the Popperian model of science according to which all hypotheses are false until proven otherwise. Here is our modus operandi: come up with an idea, congratulate oneself on its brilliance, and then try to prove how closely it matches the factual reality.

No contrary evidence has ever successfully debunked a hypothesis in neurodegeneration research. Our ideas are hypotheses in name but doctrines in practice. Lewy bodies and neurites composed of abnormally aggregated alpha-synuclein are widely accepted as the cause of PD–the synucleinopathy hypothesis. Their sequential spread from one brain region to the next—the Braak hypothesis. Their replication like viral strains—the prion hypothesis.

It was only a matter of time before someone suggested that the location of the first Lewy bodies could define PD subtypes—the brain-first/body-first hypothesis.

In an opinion piece in this issue of the Journal of Parkinson's Disease, Leonidas Stefanis and Per Borghammer add a wrinkle to the brain-first/body-first hypothesis by proposing to make age a variable that can separate the centrifugal brain-first PD subtype (pathology ‘onset’ in olfactory bulb or amygdala) from the centripetal body-first subtype (pathology ‘onset’ in gut, skin, kidney, or any other non-brain tissue). ¹ After claiming that a “wealth of clinical, neuroimaging and neuropathological data support this categorization” (none is prospective and clinical correlation is meagre and inconsistent), the authors propose that body-first PD occurs at a later age than brain-first PD, and therefore “age of onset could be used as a proxy for this classification […to] aid in understanding pathogenetic mechanisms behind brain-first and body-first PD.” In other words, we could modify the brain/body dualism into ‘early-age-brain-first’ and ‘late-age-body-first’ subtypes. The logic is that if aggregated alpha-synuclein is involved in the death of neurons, the closer the pathology is found to the brain at onset, the faster the substantia nigra degenerates, and the shorter the time to disease onset.

Theory does not meet evidence. There is no face validity to the ‘early-age-brain-first’ idea. It is well established that most cases of early-onset PD are associated with slower progression, lower risk of cognitive impairment, delayed disability, and longer survival than late-onset PD. ² And contrary to a ‘late-age-body-first’ subtype, disease progression is actually faster among those with older age at onset, ³ which is problematic for a model in which pathology, the presumed villain, begins far from its ultimate victim, the brain. A villain that starts farther away should take longer to commit the crime.

If early onset implies early brain pathology (‘brain-first PD’), but early-onset PD progresses more slowly than late-onset PD, then pathology is more likely part of a homeostatic response than a disease-causing toxin. Supporting this view is the observation that Lewy pathology and neuronal count in the substantia nigra are positively correlated: the more Lewy pathology, the more neurons remain. ⁴

Leaving aside the discrepancies between the epidemiological and pathological data, there is also a practical problem with Stefanis and Borghammer's argument. Could we see a future in which clinicians adopt a particular age cutoff around which they can issue such pronouncements at the bedside as, “Mr Smith, your relatively early age at onset suggests you may have a ‘brain-first’ type of Parkinson's”? We don’t know what that would mean to a patient and how it would inform treatments.

Even if a population-based study has shown no clinical or demographic patterns distinguishing brain- versus body-first presentations, including age, ⁵ Stefanis and Borghammer operate under the logic that the clinical and biological diversity of PD can still be reduced to centripetal and centrifugal patterns of pathological alpha-synuclein distribution, even if disease duration and multiple other factors could explain the differences. ⁶

This is a fictitious world dictated by the Braak staging. ⁷ Braak et al. popularized the idea of pathology ‘propagation’ by making static data—of 41 brains from people with PD and 69 with Lewy pathology without history of PD—appear dynamic. Guided perhaps by a form of pareidolia, the Braak team envisioned pathology moving centripetally, from the lower brainstem to the cortex, falsely equating the magnitude of pathology spread with the severity of disease. In their own data, there was no relationship between Braak stage and clinical stage (Hoehn and Yahr), or even between Braak stage and age. ⁸

Neurologists fell in love with the elegance of the Braak staging and blindly accepted it as a ‘hypothesis’ without acknowledging its severe limitations.

Few ideas run deeper in the neuromythology of PD research than the Braak hypothesis—that pathology spreads in a fixed pattern and dictates disease expression. It is understandable, therefore, that Stefanis and Borghammer treat pathology as equal to disease by using these two terms interchangeably (as in, for example, “…the disease then propagates through the autonomic connectome to the lower brainstem…”). But alpha-synuclein pathology is more commonly a feature of normal aging than of disease. Already in 1988 William Gibb and Andrew Lees showed that while the prevalence of Lewy pathology increases with age, it far outpaces the incidence of PD—ten times more individuals die with Lewy pathology without ever developing PD. ⁹ Gibb and Lees’ main conclusion, that “the Lewy body is a protective mechanism to shield the neuron from toxic insults,” seems quaint nearly four decades later. In the reality we inhabit, we have made Lewy pathology not just a marker of PD, but its very maker.

Alpha-synuclein is in nearly every tissue of the human body—even in red blood cells. As part of the innate immune system, it responds to potential pathogenic exposures, whether toxic, infectious, or biological, by undergoing a reactive phase transition: from soluble, monomeric alpha-synuclein to insoluble, pathological aggregates, a process in which there is a loss of the former, the functional peptide. ¹⁰ This 140–amino acid protein has been conserved since prehistoric genomes, suggesting that alpha-synuclein aggregation, wherever it occurs, is an evolutionarily adaptive response to preserve homeostasis, not a mechanism to accelerate aging.

Pathology does not mean disease. Most individuals with pathology will never have disease, a ‘resilience’ (in the ‘proteinopathy’ jargon) that may most depend on the brain's ability to sustain high levels of the soluble monomeric proteins from which pathology forms. ¹¹ But if disease is defined by the presence of pathology alone, such a definition reinforces itself and becomes refractory to falsification: wherever there is pathological alpha-synuclein, the disease exists, or will one day emerge if people live long enough.

Alzheimer's neuroscientist Karl Herrup once playfully noted that observing Alzheimer's pathology before Alzheimer's dementia is about as meaningful as noting that most people with the disease have gray hair—an example of the logical fallacy of inferring causality from only temporal sequence.^12,13 Variable A appearing before variable B does not make A causal to B, even if someone might facetiously defend against this criticism by arguing that gray hair is a necessary but not sufficient pathogenic mechanism, as might be used in defense of the pathology-is-disease mythology.

Pathology preceding disease fulfills only one (temporality) of 9 Bradford Hill criteria required to demonstrate causality. ¹⁴ There is no dose-response effect as more pathology does not mean worse disease (strength of association and biological gradient); no replication of findings across studies (consistency); no distinctive behaviors associated with a given pattern of pathology (specificity); no well-defined, reproducible mechanism linking pathology to neurodegeneration (plausibility); no universal expression of PD when pathology is present or prevention of PD when pathology is absent (coherence); no improvement of PD with removal of pathology (experiment); and no demonstration of a causal relationship between pathology and disease in any other condition (analogy).

The more we equate pathology with pathogenesis, biology, staging, risk, and disease, the more incomprehensible and unproductive our approach to disease modification becomes, and the lower the odds that actual patients will ever benefit –at any age of onset.