Unsupervised clustering reveals longitudinal psychiatric signatures in hd

BackgroundWhile Huntington’s disease (HD) is diagnosed by motor onset, psychiatric disturbances may present years prior to formal diagnosis, bearing a significant burden on daily functioning. However, there is great inter-individual heterogeneity in psychiatric expression and evolution over time. As such, the present study strives to discern longitudinal psychiatric signatures that may inform patterns of HD progression.MethodsForty-seven HD gene-expansion carriers (23 premanifest, 24 manifest) underwent psychiatric evaluation with the short-Problem Behavior Assessment (PBA-s) for a maximum total of six longitudinal visits. Unsupervised clustering of weighted PBA-s scores was performed with the Disease Trajectories (DT) analysis software based on dynamic time warping, which allows the non-linear alignment of sequences that may vary in speed, but conceal similar temporal characteristics. Next, clusters with shared psychiatric evolution were assessed for group differences in diagnostic status and severity of psychiatric features.ResultsDT analysis identified eleven clusters, of which two psychiatric patterns (N≥5) of mixed diagnostic status were further analyzed. Clusters were defined by non-depressive and non-irritable temporal signatures, respectively. Meanwhile, both clusters increased in clinically-relevant apathy and executive dysfunction over time.ConclusionThe present study underscores the inherent heterogeneity in HD, where psychiatric signatures may predict common disease trajectories, even in prodromal stages. Through the incipient detection of shared patterns of symptom evolution, these findings model the applicability of the DT analysis software in the clinical context, lending itself to the identification of specific profiles. A stratification of the patients could allow increasing both the likelihood of successful personalized therapeutic approaches and the sensitivity in the detection of biomarkers in clinical trials.