Time's up: Using data-driven phenotype-severity metrics not time to map progression in the dementias

Temporal measures, such as time from diagnosis or symptom onset are often used to track disease severity in neurodegenerative diseases. Due to variations in symptom awareness, clinical presentation timing, diagnostic delays, and disease progression rates, these temporal proxies introduce substantial variance and bias, making it very difficult to map progression clearly and accurately, and to severity-match across contrastive patient groups. To address this challenge, we explored a data-driven approach to derive a transdiagnostic severity metric that is independent of time and, instead, treats temporal metrics as observed, dependent data. We analysed data from the Genetic Frontotemporal Dementia Initiative (GENFI 1 and 2). We entered neuropsychological scores for symptomatic individuals including any visits prior to conversion from at-risk to symptomatic (n = 265, 522 visits) in an unrotated principal component analysis to derive a transdiagnostic phenotype-severity model. A single component emerged (Kaiser-Meyer-Olkin = 0.92), explaining 65% of the variance, with all neuropsychological assessments loading highly. This global severity component fitted the data equally well across genetically or clinically defined groups, as well as severity levels. The severity measures validity was supported by a clear relationship with the Clinical Dementia Rating scale, and its stability was confirmed when a much broader range of neuropsychological and behavioural measures were included. Additionally, the severity score accounted for a high portion of the total variance in neuropsychological test scores, substantially more than the low proportion accounted for by standard temporal measures. To derive a time-efficient sub-battery, we demonstrated that three neuropsychological assessments (Digit Symbol, Verbal fluency (letters) and Trail Making Test- Part B were able to explain the majority of unique variance in cognitive severity. Finally, by treating time as an observed dependent variable, we showed that the baseline velocity (change in severity measure over time) varied by genetic group, with progranulin mutation carriers being the fastest. This data-driven approach provides an objective, precise measure of disease severity and progression, and it may shed new light on when clinical heterogeneity reflects distinct subtypes rather than differences in disease stage.