Apathy, depression, and motor symptoms have distinct and separable resting activity patterns in idiopathic Parkinson disease.

Apathy and depression are heterogeneous syndromes with symptoms that overlap clinically. This clinical overlap leads to problems with classification and diagnosis in clinical populations. No functional imaging study has attempted to separate brain regions altered in apathy from those altered in depression in a clinical population. Parkinson disease (PD) is a disorder in which apathy and depression co-exist in a single population. We evaluate the relationship between apathy, depression, and motor severity of disease in PD, focusing on the relationship between these factors and the amplitude of the low frequency fluctuation (ALFF) in the resting state. We first evaluated if the resting ALFF signal is a reliable measure for our clinical question. For this, we develop and introduce a cross validation approach we term the "Regional Mapping of Reliable Differences" (RMRD) method to evaluate reliability of regions of interest deemed "significant" by standard voxel-wise techniques. Using this approach, we show that the apathy score in this sample is best predicted by ALFF signal in the left supplementary motor cortex, the right orbitofrontal cortex, and the right middle frontal cortex, whereas depression score is best predicted by ALFF signal in the right subgenual cingulate. Disease severity was best predicted by ALFF signal in the right putamen. A number of additional regions are also statistically (but not reliably) correlated with our neuropsychological measures and disease severity. Our results support the use of resting fMRI as a means to evaluate neuropsychiatric states and motor disease progression in Parkinson disease, and the clinical and epidemiologic observation that apathy and depression are distinct pathological entities. Our finding that "significance" and "reliability" are dissociated properties of regions of interest identified as significant using standard voxel-wise techniques suggests that including reliability analyses may add useful scientific information in neurobehavioral research.