DCM based markers of effective connectivity between pre spec

DCM-based markers of effective connectivity between pre-specified regions of the motor network also did not indicate any compensation. However, it should be noted that DCM is limited by the necessity to specify regions a priori, and untested compensatory mechanisms involving phalloidin areas not incorporated in the pre-defined list of regions of interest would remain undetected. In addition, a compensatory mechanism may become apparent only at a time window very close to the emergence of phenotypical motor abnormalities or may not be operative at all for the motor system. This is consistent with a concept of basal ganglia function that postulates a basic, fundamental role in movement sequencing and postural adjustments in anticipation of volitional movements that can neither be replaced nor compensated for once damaged beyond a certain threshold, thus resulting in increasingly degraded motor performance as the degenerative processes progress. We investigated four different (anatomical) markers of structural disease load and their relationship with brain activity and task performance. Whole brain gray matter as a percentage of total intracranial volume was the marker of structural disease load most often associated with compensation-like changes in these relationships in the rsfMRI analyses. In previous analyses involving many of the same participants, gray matter degeneration accelerates close to disease onset (Tabrizi et al., 2013). However, for both task activation and resting state analyses, significant compensatory and non-compensatory effects were observed using all four measures of structural disease load. In this study, we treated each measure of disease load as statistically independent. Given that the striatum, gray matter and white matter all degenerate, albeit at different rates, during the premanifest stage of HD, they are likely statistically dependent. For future studies, it may prove more useful to integrate all four measures within a multivariate analysis to account for the relationships between the individual measures. Our definition of compensation was in part derived from the compensation criteria detailed in Cabeza and Dennis Cabeza and Dennis (2013); in particular, that for successful compensation to be present, an increase in activation should be positively associated with an increase in task performance (positive relationship). However, it is possible that there are alternative definitions and underlying mechanisms of compensation that may also be appropriate for future investigation, which are not consistent with our operational definition. For example, compensatory processes may be driven by the downregulation of pathologically high signals or the potential disengagement of brain regions. These mechanisms would be reflected within our model as negative correlations between brain activity (within certain regions) and performance conditional on structural disease load. Furthermore, the current study highlights compensatory activity in regions such as the FFG and the hippocampus which are not routinely associated with general cognitive processing it is this potential recruitment of alternative pathways that we will look to investigate further in future studies with longitudinal data, in addition to the negative correlations and the changes in these associations over time. We recognize that our study has a number of limitations. Despite a large sample size and our a priori definition of a compensation model, we found comparatively little evidence for widespread neural compensation in our presymptomatic HD gene carriers. This may reflect a true negative finding that there is little underlying compensation in the presymptomatic phase of neurodegeneration occurring during the states we chose for fMRI measurement. The tasks we examined (working memory and motor tasks) showed no large difference in behavior between healthy and premanifest HD groups. Such findings are consistent with neural compensation, but are of course also consistent with the possibility that compensation in those particular tasks does not play a large role. Alternatively, it may be a false negative finding; that despite our large sample size, statistical power may be inadequate to detect compensation. In particular, our operational definition of compensation hypothesizes an interaction between different measures in the context of the linear statistical models employed here. Power to detect such interactions is inherently much lower than that required to detect so-called main effects of individual variables. We have argued that such interactions are a necessary, but previously overlooked aspect of the definition of disease compensation. Future work will be able to explicitly investigate whether the compensation effects identified here replicate in the same or different cohorts; and investigate whether a single weighted measure of structural disease load across the brain may be more appropriate than the four measures (necessitating correction for multiple comparisons) used here.