Background We provide a summary of study-wide variables which were predictive of the time to HD diagnosis (DCL=4).

Aims Two aspects of each predictor were considered: (1) the ability of the variable at baseline (only) to predict diagnosis, and (2) the ability of the change in the variable to predict diagnosis.

Methods All analyses adjusted for age at baseline, gender, years of education, motor examiner experience and CAP. There were 44 baseline and 36 longitudinal variables from five sections (cognitive, functional, imaging, motor, and psychiatric) which were examined in the analysis. Most variables were significant as baseline predictors, whereas variables from motor, cognitive and functional sections were significant as longitudinal predictors. These include: motor score, TFC, FAS, symbol digit, stroop word, stroop interference, stroop colour, verbal fluency, trails B time, dominant index finger speeded tapping, alternating thumbs paced tapping, 2-back discriminability, chooser two-choice movement time, dominant index finger paced tapping, and negative emotion recognition. We also present an evaluation of the prognostic potential of the baseline variables with a type of longitudinal receiver operating characteristic (ROC) curve analysis.

Results Results of this analysis showed that variables from the imaging and cognitive sections had relatively high prognostic potential. These included: globus/ICV ratio, alternating thumbs paced tapping, striatum/ICV ratio, putamen/ICV ratio, caudate/ICV ratio, dominant index finger paced tapping, and symbol digit.

Conclusions Findings are discussed in terms of clinical trial design and directions for future research study in observational studies such as ENROLL.