We report a neuroanatomical profile of individuals with PWS that presents a unique pattern of regional brain volumes in each genetic subtype of this complex syndrome. These preliminary findings are the first structural neuroimaging findings to support potentially separate mechanisms underlying the behavioral differences seen in these genetic subtypes. Compared to HWC, PWS participants as a whole had lower GMV localized to the prefrontal, orbitofrontal, and temporal cortices, as well as the caudate, hippocampus, and parahippocampus, even when controlling for absolute gray matter volume differences between these groups. We also show, for the first time in a group of PWS subjects, less WMV in the brainstem, cerebellum, and white matter surrounding the thalamus, as well as frontal white matter. In addition, consistent with previous fMRI reports, we found that individuals with the DEL subtype of PWS had lower GMV primarily in the prefrontal, medial frontal, and temporal cortices compared to HWC, while individuals in the UPD subtype had lower gray matter volumes primarily orbitofrontal and limbic gray matter compared to HWC. These findings remained significant even when controlling for age, sex, BMI, and total GM volume as covariates in the analysis.
Overall, individuals with PWS in our study had less gray matter volume in the prefrontal cortex, orbitofrontal cortex, limbic, and temporal cortices than HWC. These confirm and extend reports from a recent VBM study which reported less gray matter volume in the orbitofrontal cortex (OFC) in PWS individuals compared to healthy-weight controls 18
. The OFC is a key brain structure for incorporating reward information from food-related sensory and visceral networks 29; 30
. Studies on food-related behavior in both obese 31
and PWS 32
individuals have characterized dysfunction within the satiety system 32; 33
in the prefrontal cortex, namely the OFC. In addition, our structural findings of gray matter differences overlap with our functional neuroimaging results from a previous study in which we found that the PWS group exhibited greater activation to food pictures in the post-meal condition compared with the pre-meal condition in the orbitofrontal cortex, medial PFC, insula, hippocampus, and parahippocampal gyrus 34
. In PWS there is an onset of insatiable appetite relatively early in life, with the resultant risk of severe and life-threatening obesity 35
. Thus, less local gray and white matter volume in the prefrontal and limbic regions may underlie hyperfunction of the limbic and prefrontal cortex and may be associated with hyperphagia specific to PWS individuals; however the relationship between structural and functional differences between these PWS groups remains unclear. The PWS group had significantly less GMV in the medial prefrontal cortex (BA 6, 46), a region important in attentional control and executive function. Previous research has reported an association between specific cognitive deficits in task switching (attentional control) and PWS behaviors including repetitive questions and temper tantrums 36–38
. It may be that less gray matter volume in the medial prefrontal cortex underlies some of the cognitive deficits and corresponding behavioral deficits dependent on prefrontal networks. Overall, individuals with PWS have less temporal, frontal, and limbic gray matter volume compared to individuals with HWC that, taken in light of other functional imaging studies, may give neuroantomical basis for the behavioral deficits.
This is the first study to use VBM to study genetic subgroups of PWS individuals compared to HWC individuals. Overall, we found more extensive GMV decreases in the DEL group compared to the UPD group. These were primarily in the prefrontal cortex (BA 6, 46), the anterior cingulate cortex (BA 24) and the superior and middle temporal gyri (BA 22). Medial frontal and prefrontal regions have been associated with food processing as well as reward and affect-driven motivation 19; 34
. In a previous fMRI study of food motivation, we found that individuals in the DEL group showed a more robust, abnormal response than the UPD group in overlapping regions, namely the prefrontal cortex, anterior cingulated, and superior temporal gyri. Moreover, PET studies have documented hypermetabolism in the prefrontal, orbitofrontal, and temporal regions (primarily in the deletion subgroup)39; 40
, as well as reduced [11
C] FMZ binding to GABAA
receptors in deletion adults in the prefrontal, temporal, and cingulate cortices 41
. Although it is not possible to measure the overlap between structures in these PET studies compared to our MRI study, it must be noted that regional hypermetabolism alongside less gray matter volume is unexpected. It may be that, despite less gray matter in these frontal and temporal networks, individuals with PWS, and more often DEL individuals, may have synaptic compensatory mechanisms maintaining neuronal activity in the face of structural alterations. That said, not all studies have reported hypermetabolism in PWS; Kim et al found hypometabolism in the orbitofrontal cortex, and Lucignani et al. report a loss of GABAA
receptors in the prefrontal cortex in individuals with PWS41; 40
. Overall, reports from multiple neuroimaging modalities point towards an abnormal cortical network of food motivation circuitry in PWS, especially in the DEL subgroup.
In the present study, the UPD genetic subgroup had less GMV and WMV in the OFC, and less GMV in the head of the caudate and medial temporal cortex compared to HWC individuals. Previous fMRI data revealed that individuals with UPD had abnormal an BOLD response in the medial temporal cortex, which, taken in light of our structural finding in the UPD group, may suggesting that regions involved in memory and learning may be both structurally and functionally abnormal in this subgroup. In addition, individuals in the UPD subgroup had less GMV in the head of the caudate than HWC individuals. The caudate nucleus, part of the basal ganglia, is highly innervated by dopamine neurons, and also is involved in learning and memory, particularly feedback processing. It has been theorized that the caudate nucleus may be dysfunctional in persons with obsessive-compulsive disorder (OCD) through inability to regulate information from the OFC.
Individuals with PWS appear to have a compulsive disorder with behaviors seen in other conditions such as autism, specifically repetitive and ritualistic behavior 42
. Moreover, a large structural imaging study in typically developing school-aged children found that there was a significant relationship between general cognitive ability (IQ) and the volume of the cerebellum and caudate 43
. Thus lower gray matter in the basal ganglia specifically in UPD individuals may explain both their obsessive-compulsive symptoms as well as play a role in the decreased cognitive abilities in this group of young and old individuals.
As a whole, we also found that PWS subjects had significantly less regional WMV in the ventral midbrain, brainstem, cerebellum, medial temporal cortex and surrounding the thalamus, along with regionally less white matter in the inferior frontal cortex. Genetic subgroups were more similar in WMV differences than they were GMV differences () compared to HWC, with the only differences between subgroups being less orbitofrontal white matter volume in the UPD group, and less parietal cortex white matter in the DEL subgroup. Gross white matter abnormalities in PWS have been reported in post-mortem studies in the brainstem 13
and cerebellum 15
. Moreover, a recent study used diffusion tensor imaging (DTI), a type of imaging that characterizes the quality of white matter tracts, in 8 PWS individuals and identified regional abnormalities in white matter development in PWS, specifically in frontal and thalamic white matter, along with the posterior limb of the internal capsule and the splenium of the corpus callosum 44
. Similar to this DTI study, fronto-thalamic regional white matter volumes were decreased in both DEL and UPD subgroups in our analysis. Fronto-thalamic white matter tracts have close connection to the limbic system through the prefrontal and cingulate cortices, and abnormalities in these regions have been linked to psychiatric dysfunction; thus, abnormalities in these regions might contribute to clinically observed psychiatric manifestations in PWS 45
. Moreover, our PWS individuals had decreased WMV in the central and posterior white matter tracts coming up through the midbrain, brainstem, and cerebellum, which may underlie some of the motor dysfunction in PWS, as similar white matter dysfunctions in these regions have been shown to play a role in motor neuron dysfunction in diseases like amyotropic lateral schlerosis 46
. Future DTI studies using larger groups of subjects might clarify tract-specific changes among in PWS subgroups.
Neuroanatomical differences between genetic subtypes of PWS may be due to several genetic mechanisms. Individuals with the DEL subtype may have a more severe phenotype due to haploinsufficiency of genes in the PWS region, resulting in a reduced expression of non-imprinted genes in these individuals. As a contrast, individuals with the UPD subtype may have a less severe neuroanatomical phenotype because they inherit two copies of maternally expressed genes, resulting in a gene dosage effect with elevated expression of these genes. Moreover, imprinted genes distal to 15q11-q13 are possibly impacted in individuals with UPD but not in individuals with a deletion. Thus, these mechanisms may lead to variability in both the patterns of neural activation we have seen in these subgroups, as well as differences in the underlying smaller regional GMV between UPD and DEL individuals.
Individuals with PWS have smaller congenital brain volume compared to normal volunteers, possibly owing to lack of paternally-expressed genes on chromosome 15q11-q13. To control for brain size differences, we included total GMV and WMV volume as a covariate where appropriate, however this may have underestimated the difference between the three groups derived from this disease. There were also differences in the numbers of males and females between the groups, and considering that sex influences gray matter volume, we can not rule out the contributing factor of sex to lower gray matter volumes in the DEL group. Despite this, our analyses account statistically for overall size differences; thus, we can more clearly interpret results, such that regions where DEL had significantly lower regional GMV compared to UPD and HWC were not just due to proportionally smaller brains. We also controlled for gender in all of our imaging analyses. Moreover, our use of a study-specific template further compensates for the gender differences among the sample, as all images are normalized to a mixed-gender template. Because our data was collected as part of several separate studies, this study lacked more detailed neurocognitive tests for 10 of our PWS subjects. Thus, we did not attempt to correlate various types of tests with our morphological data, nor could we control for IQ in our imaging analysis of HWC. For instance, individuals with the DEL subtype had a lower mean IQ than those with the UPD subtype, and had in general had less local gray matter volume compared to HWC than the UPD subtype, with a trend for lower gray matter volume in the cerebellum, which has been noted to be related to IQ in school-aged children 43
. In addition, in our secondary analysis of UPD vs. DEL subgroups we were able to control for IQ and found that the DEL subgroup had frontal and temporal gray matter volume differences similar to the original analyses; however, this did not reach a corrected threshold. Therefore, there may be differences in cognitive functioning that may further explain the differences between genetic subtypes that this study could not clarify. However, fMRI data on IQ-matched individuals (PWS and HWC) does suggest that there are neural mechanisms of hyperphagia in PWS that are not accounted for by cognitive differences between these populations 47