Progressive supranuclear palsy (PSP) is a devastating neurodegenerative disorder with poor prognosis. The development of novel therapeutics has been hindered by long delays from symptom onset to diagnosis, marked heterogeneity, and uncertainty over the optimal outcomes in early-phase clinical trials. Natural history studies can facilitate readiness for trials, with better diagnostic schema, validated biomarkers, and information on critical variability in phenotype and stage of disease.

In this thesis, I test whether clinical progression and survival differ across the phenotypic spectrum of PSP and consider new evidence for the early stages of the disease. First, using UK Biobank data, I identify subtle cognitive and motor changes in the pre-diagnostic stage. Second, I use post-diagnostic single–site data to describe characteristics and survival of different phenotypes. Third, I test the sensitivity of diverse measures to disease progression in the multi-centre observational “Progressive Supranuclear Palsy-Corticobasal Syndrome-Multiple system atrophy study” (PROSPECT-M-UK). Finally, I identify structural correlates of survival.

After an introduction to the issues and principal research questions (Chapter 1) and general methods (Chapter 2), I exploit the prospective UK Biobank cohort from 2006-2021 to demonstrate that subtle motor and cognitive markers are present at an average seven years prior to diagnosis (Chapter 3). I propose a long pre-diagnostic phase to PSP, similar to other neurodegenerative diseases, including a likely pre-symptomatic phase of several years.

In Chapter 4, I use linear mixed models of longitudinal data to estimate annualised rates of change in disease severity. Only half of patients were considered likely to have met recent major clinical trial entry criteria, and this group showed faster motor and cognitive progression than those not meeting trial criteria. Richardson’s syndrome progresses faster than other phenotypes. Subcortical phenotypes, with parkinsonism, postural instability and gait freezing features, present later but have slower motor and cognitive progression and longest survival.

In Chapter 5, I compare the progression of functional, clinical, cognitive, and magnetic resonance imaging (MRI) changes in the PROSPECT-M-UK study. I then estimate sample sizes required for clinical trials of disease modifying agents, according to phenotype and candidate outcomes. Subcortical phenotypes showed the slowest progression and longest survival. Neuroimaging would enable lower sample sizes than cognitive and functional measures, although optimal outcome measures were phenotype specific.

In Chapter 6, I identify regional atrophy patterns that predict survival. I confirmed the midbrain structural correlates of contemporary disease severity (MRI versus PSP-Rating Scale) and show that there are different anatomical correlates of the ‘temporal stage’ from onset to death in the thalamus, striatum and frontotemporal cortex. These findings suggest that an individual’s future survival time, as a function of time from onset to scanning, is related to structural change in areas that differ from the simple correlates of the PSP diagnosis.

In summary, current clinical trials are not representative of the PSP population. The variable sensitivity of outcome measures will affect the power of clinical trials, especially for phenotypes other than Richardson’s syndrome. Better recognition and customised approaches towards PSP phenotypes are required in clinical and research practice, to improve diagnosis and future treatment options.