This study expands upon prior methods to evaluate the extent of neuropathologic involvement in HD brains,3,14
by clustering the evaluations of multiple regions to create scores representing the combined involvement of these ratings. A total of 523 HD brains without other concomitant neuropathologic diagnoses were evaluated by a single neuropathologist for the severity of involvement in 41 informative brain regions. Cluster analysis identified 2 main groupings: 1) a striatal cluster, representing 28 rated regions, and 2) a cortical cluster, representing 13 rated regions. Notably, the clusters show different interrelationships between HD CAG repeat size, age at onset, age at death, and the duration of the disease from onset to death, suggesting that the relative contributions of factors involved in the neurodegenerative processes may be different in these 2 clusters of brain regions. While the size of the expanded repeat is the primary driver in the striatum, there is evidence for more complex influences in the cortex.
presents a scatterplot of the 523 cases for these 2 clusters, which illustrates that cases with similar striatal involvement may have very different levels of cortical involvement, and vice versa. Similarly, depicts 2 HD brains with similar Vonsattel grade 4 striatal involvement but dramatically different cortical involvement. The 2 clusters, which explain 57.2% of the variation in the neuropathologic ratings, represent continuous quantitative measures allowing for the assessment of the relationship of multiple factors with the neuropathologic involvement in HD.
The scatterplot for the striatal and cortical cluster scores with the trend line superimposed shows that these 2 scores are correlated
These 2 Huntington disease brains have similar low grade 4 striatal involvement but present dramatically different cortical involvement
While the striatal cluster shows a strong bivariate correlation with the size of the expanded HD CAG repeat, the age at onset, and the age at death, multivariate analysis reveals that these correlations are completely dependent upon the effect of the expanded HD CAG repeat. Neither onset age nor age at death is predictive of the striatal cluster when the repeat size is in the model. Thus, for a given repeat size and duration, younger onset age individuals do not have greater striatal involvement than do those with older onset. This observation is consistent with the length of the HD CAG repeat expansion being the primary determinant of both striatal involvement and onset age. The corollary of this observation is that onset age does not influence the extent of striatal involvement independent of the relationship to repeat size. While there is substantial evidence for genetic modifiers of age at onset which are independent of the CAG repeat size,2,19
their relationship to the extent of striatal involvement will be an important area of investigation.
In bivariate analyses, the cortical cluster shows neither a correlation with the size of the expanded HD CAG repeat nor correlation to the age at onset. There is, however, a significant (p = 0.049) but modest correlation with older age at death. Notably, multivariate analysis reveals that the cortical cluster is related to the repeat size after adjustment for the disease duration, suggesting that cortical involvement is a part of the disease process that is neither determining nor determined by age at motor onset. The relationship of the cortical cluster to HD repeat size is strengthened when age at onset is also in the model, suggesting a more complex relationship among these variables in the context of cortical involvement that we discuss below.
Both the striatal cluster and the cortical cluster are associated with the HD CAG repeat size and duration, with more severe involvement associated with larger repeats and longer disease duration. Much more of the variation in the striatal cluster is explained by the HD CAG repeat, duration, and onset age (R2
= 0.313) than is explained for the cortical cluster (R2
= 0.138), suggesting that the contribution of the HD repeat size to cortical involvement may be less than for striatal involvement. The striatal cluster is strongly correlated with the Vonsattel grade,3
which is heavily weighted to the involvement in the caudate nucleus and the putamen. Conversely, the cortical cluster is more strongly correlated with the brain weight. While this cluster correlation to brain weight is not surprising given that the cortex makes up a large component of the brain, it further emphasizes the difference between these 2 cluster scores, and suggests that brain weight may be a surrogate for the cortical cluster when this type of extensive neuropathologic evaluation is not available.
While the striatal cluster is not associated with either onset age or death age after adjusting for the HD repeat size and duration, the cortical cluster is associated with either an older age at onset or age at death. Although one might expect more severe changes with younger onset, the relationship of age to cortical pathology might be due to superimposed aging effects. A number of brain imaging studies have shown that the regional atrophy in HD is correlated with clinical features, including cognitive performance and disease progression.4–6
Similarly, a significant association has been reported between motor dysfunction and postmortem cell loss in the primary motor cortex and an association of mood with cell loss in the anterior cingulate cortex.20
The relationship of regional atrophy to clinical expression emphasizes the importance for identifying genetic and nongenetic factors that influence the extent of neuropathologic involvement. These neuropathologic rating cluster scores offer an opportunity to evaluate the relationship of the extent of involvement to a variety of unbiased genome-wide measures, microarray or RNA sequencing, and chromatin immunoprecipitation sequencing for epigenetic effects which have potential to implicate specific pathways in the pathogenesis of HD.