Regional brain atrophy is an important marker of disease progression in Huntington’s disease (HD). However, multicentre studies often collect MRI data using different scanners, which can introduce variability and make it harder to distinguish genuine biological change from scanner-related effects.
This study investigated whether longitudinal ComBat, a statistical harmonisation method designed to reduce scanner-related differences, could improve the sensitivity of brain volume measurements to HD progression. Using IXITM regional segmentation algorithms and boundary shift integral, researchers analysed the caudate, putamen, whole brain and lateral ventricles in 482 healthy controls and 918 participants spanning different stages of Huntington’s disease. Data were drawn from the Track-/TrackOn-HD, Predict-HD and Image-HD studies.
While harmonisation reduced some scanner-related variability, it did not improve sensitivity to disease progression, and disease-related effect sizes were reduced in most cases after harmonisation. The findings highlight an important consideration for longitudinal neuroimaging studies: reducing technical variability may also risk reducing meaningful biological signal, underlining the importance of carefully evaluating harmonisation approaches when developing imaging biomarkers.
