Hypothalamic involvement in multiple system atrophy: A structural MRI study

Objective . To investigate hypothalamic atrophy and its clinical correlates in multiple system atrophy (MSA) in-vivo. Background . MSA is characterized by autonomic dysfunction and parkinsonian/cerebellar manifestations. The hypothalamus regulates autonomic and homeostatic functions and is also involved in memory and learning processes. Methods . 11 MSA, 18 Parkinson’s Disease (PD) and 18 Healthy Controls (HC) were included in this study. A validated and automated hypothalamic segmentation tool was applied to 3D-T1-weighted images acquired on a 3T-MRI scanner. MSA hypothalamic volumes were compared to those of PD and HC. Furthermore, the association between hypothalamic volumes and scores of autonomic, depressive, sleep and cognitive manifestations were investigated.


Introduction
Autonomic dysfunction is characteristic of multiple system atrophy (MSA), along with parkinsonism and/or cerebellar ataxia [1].In MSA, autonomic dysfunction is mainly due to the neurodegenerative process affecting the central autonomic network [2].The hypothalamus is a hub of this network and regulates many homeostatic functions, such as blood pressure, neuroendocrine regulation, and the sleep/wake cycle [3,4].More recently, a role in memory and learning has also been shown [5][6][7][8].In MSA, the hypothalamus displays characteristic glial and neuronal alpha-synuclein inclusion pathology along with neuronal loss [9][10][11].
Conversely, autonomic dysfunction in Parkinson's Disease (PD) is generally attributed to the pathological involvement of the peripheral autonomic nervous system [2].To a lesser extent, the central autonomic nuclei may also be affected by neuropathological changes in PD, and Lewy bodies are consistently found throughout the hypothalamic nuclei [12].
Although pathological involvement of the hypothalamus in MSA is recognized, in-vivo investigations are lacking.Recently, a validated and fully automated MRI-based hypothalamic segmentation tool was developed and successfully applied in healthy ageing and dementia populations [13,14].
Therefore, the study aims were to: (a) Compare hypothalamic volumes in MSA, Parkinson's disease (PD) and healthy controls (HC); (b) Investigate the associations between volumetric changes and severity in clinical ratings of depressive, autonomic, sleep, or cognitive manifestations.

Participants
Participants were recruited between June 2019 and September 2021 from the Newcastle upon Tyne NHS Clinics for Research and Service in Themed Assessment (Campus for Ageing and Vitality, Newcastle Upon Tyne, UK).Inclusion criteria were: for all participants, age between 45 and 80 years; healthy controls (HC) were included if they showed no neurological symptoms or dysfunction, or MRI structural brain abnormalities; MSA participants were included if they received a diagnosis according to Gilman et al. consensus criteria [1] less than three years earlier; PD participants were included if they fulfilled the UK Brain Bank Criteria [15].Exclusion criteria for all participants were: a diagnosis of another atypical parkinsonism, significant memory impairment (MMSE<24), meeting DSM V criteria for major neurocognitive disorder, contraindications to MRI scan, severe comorbid illness as judged by the investigator.Twelve participants with a MSA diagnosis (8 with predominant parkinsonian features and 4 with predominant cerebellar features), 19 PD participants and 18 HCs were enrolled in the study.One PD and one MSA (with predominant parkinsonian features) participants were excluded because of MRI excessive motion artifacts.All MSA participants included in the current analysis had a probable diagnosis according to Gilman's et al. criteria [1].Furthermore, at the time of writing, no changes in diagnoses were made to the participants included in this analysis.
MSA and PD participants were rated with the Montreal Cognitive Assessment (MoCA) [16], the Hospital Anxiety and Depression Scale (HADS) [17], the Unified MSA Rating Scale (UMSARS) [18] or Movement Disorders Society -Unified Parkinson's Disease Rating Scale (MDS-UPDRS) [19], and the Scales Related to Outcomes in Parkinson's disease autonomic dysfunction (SCOPA-AUT) [20], and sleep (SCOPA-SLEEP) [17].The MoCA score was corrected for the educational level, as indicated in the current version of the test, by adding one point if the participant was in education for twelve years or less.
Ethical approval was granted by the London-Surrey Research Ethics Committee Research Ethics Committee (18/LO/2123).All participants provided written informed consent according to the Declaration of Helsinki.
Total hypothalamic and subregional volumes of interest (VOIs) were adjusted for TIV (used as head size proxy) through the residuals method [22]: each region was set as the dependent variable in a regression model including all participants, and TIV was set as a covariate.The resulting model's B value was employed in the calculation of the adjusted volumes as follows: where TIV mean is the mean TIV across all participants.
Left and right subregions were summed to give the total volume for each subregion and for total hypothalamic volume.

Statistical analysis
Demographic and clinical characteristics were summarized as medians and interquartile ranges (IQRs) for continuous variables and as percentages for categorical variables.
All residuals-adjusted hypothalamic volumes were found normally distributed through distributions and P-P plots inspection and the Kolmogorov-Smirnov test (all p s > 0.05), in all three subgroups separately.Given the relatively limited number of participants in each group, only the a-priori hypothesis of volumetric changes in MSA was tested.A general linear model (ANOVA-type) analysis followed by pre-planned contrasts (MSA vs HC and MSA vs PD) was performed for total hypothalamic volume and each hypothalamic subregion: six models were run, and each was accompanied by two two-tailed comparisons.Given the exploratory nature of the analysis and because of the overall number of comparisons (12, giving a probability of a false positive result below the threshold of 0.05), and that these were pre-planned contrasts, multiple comparisons p-value correction was not performed.
The association between hypothalamic volumes and clinical scores concerning autonomic, sleep, depression and cognitive functioning were carried out with Spearman rank tests, given the non-normal distribution of clinical variables.A Spearman partial correlation association was performed controlling for age to confirm significant correlations.To limit the number of tests in the association analysis, scores of cognition, depression and sleep disturbances were first correlated with total hypothalamic volume in the parkinsonism (PD + MSA) group.If a significant correlation arose, the association between scales scores and single hypothalamic subregions were investigated separately in MSA and PD.Since autonomic dysfunction in MSA and PD likely has a different pathophysiology, associations between hypothalamic volume and SCOPA-AUT scores were tested in the two cohorts separately.The significant p-value threshold was <0.05.Statistical analyses were performed with IBM Statistical Package for the Social Sciences (SPSS) version 27.

Results
Table 1 shows demographics and clinical characteristics in MSA, PD The posterior hypothalamus volume was significantly lower in MSA compared to controls (t = 2.105, p = 0.041) and PD (t = 2.055, p = 0.046) (Fig. 1).A trend towards a reduction was found in total hypothalamus in MSA vs HC (t = 1.676, p = 0.101).The remaining subregions did not display differences compared to HC and PD.Supplementary Table 1 shows the output values for the overall models and contrasts.
Total hypothalamus volume was not associated with age or disease duration in any subgroup.In the parkinsonism (MSA + PD) group, total hypothalamus volume showed a direct association with MoCA scores (rho = 0.425, p = 0.022).Among hypothalamic subregions, only posterior hypothalamus volume was associated with MoCA scores (rho = 0.718, p < 0.001, Supplementary Fig. 2), and this association was also present separately in the MSA (rho = 0.732, p = 0.010) and PD (rho = 0.605, p = 0.008) groups.These associations remained significant in partial correlation tests controlling for age (all p s < 0.05).
No significant associations were found between hypothalamic volumes and SCOPA-AUT, SCOPA-Sleep, or HADS.

Discussion
In this study, we investigated in-vivo hypothalamic atrophy and its non-motor clinical correlates in MSA through structural MRI.In MSA, a significant volume reduction in posterior hypothalamus was found compared to HC and PD.Furthermore, significant associations were shown between posterior hypothalamic volume and general cognitive functioning scores in the entire parkinsonism group and in the MSA and PD groups separately.
Pathological involvement of the hypothalamus in MSA, with typical alpha-synuclein glial and neuronal inclusions and neuronal loss, was previously shown in pathological studies [9][10][11].However, in-vivo hypothalamic investigations with structural MRI were lacking.Since the hypothalamus is the central regulator of homeostatic functions, its involvement in MSA may be hypothesized even in the early stages.Our findings appear to confirm this hypothesis and show that hypothalamic grey matter loss is detectable in-vivo, even within 3 years of diagnosis of MSA as in our sample.
Associations between hypothalamic volume loss and autonomic, sleep or depressive symptoms were not found in this study.Our sample sizes may not have provided enough power to interrogate these relationships and future studies with larger samples are needed to clarify these.In contrast, an association was found between posterior hypothalamic volume and MoCA scores, a screening test for cognitive functioning.This association was present in the entire parkinsonism group and in MSA and PD subgroups separately.The posterior hypothalamus includes the mammillary bodies and lateral hypothalamus.The mammillary bodies are relay stations in the Papez circuit and are involved in memory encoding and storage.They are pathologically involved in amnestic syndromes like the Wernicke-Korsakoff syndrome [23].Recent preclinical studies have also shown an involvement of the lateral hypothalamus in memory and learning functions [5][6][7][8].Nonetheless, since cognitive functions depend on a diffuse network, we acknowledge that other factors (e.g.cholinergic, noradrenergic and/or hippocampal pathology) may contribute to cognitive dysfunction in MSA.
Several limitations should also be acknowledged.The sample sizes of the single groups were rather small, so the analysis may have missed less pronounced differences and associations due to limited statistical power.MSA participants included were either predominantly parkinsonian or cerebellar but sample size prevented further subgroup analysis.Because the analysis presented in this study was carried out when only the Gilman's diagnostic criteria for MSA were available, we are unable to comment on whether MSA participants would have fulfilled the criteria for a clinically established diagnosis according to the more recent Movement Disorders Society criteria [24], with particular reference to   the MRI marker that such level of certainty requires.However, all participants included in this analysis fulfilled the criteria of a probable diagnosis according to Gilman's criteria.Hypothalamic atrophy was only investigated in MSA due to statistical constraints related to sample sizes but hypothalamic involvement in PD has been interrogated in the past [25,26] and this topic was beyond the scope of our study.For these reasons, the ANOVA-type models were followed only by two preplanned contrasts for each variable (MSA vs PD/HC).Although this approach does not strictly require multiple comparisons correction, our results should be interpreted with caution and larger studies should aim to confirm these findings.Cognitive functioning was rated with the MoCA, which is a general cognitive screening test; no further testing of cognitive subdomains was performed.Although the MoCA score was corrected for the educational level (by adding one point to participants who were in education for twelve years or less), education years was not included in the analysis.Supine and orthostatic blood pressure and heart rate measurements were not collected systematically at the time of the MRI scan, and this prevented the association analysis with hypothalamic atrophy; thus, this should be an area of research for future studies.
In conclusion, in-vivo structural hypothalamic involvement in may be detected in MSA.The posterior hypothalamus, a structure which includes the mammillary bodies and the lateral hypothalamus, showed reduced volume compared to controls and PD.Furthermore, in MSA posterior hypothalamus volume loss was associated with worse cognitive performances.However, hypothalamic volumes were not associated with scores of autonomic, depressive or sleep symptoms.Future studies should try to elucidate the in-vivo central correlates of autonomic dysfunction in MSA by investigating the autonomic network, and the cognitive role of the hypothalamus should be further investigated.

Fig. 1 .
Fig. 1.Individual values, means and 95% confidence intervals (CI) for total hypothalamus (upper panel) and posterior hypothalamus (bottom panel) volumes in Multiple System Atrophy (MSA), Parkinson's Disease (PD) and Healthy Controls (HCs).*: significant difference between MSA vs HC/PD as found in the general linear models' contrasts.

Table 1
Demographics and clinical characteristics of Multiple System Atrophy (MSA), Parkinson's Disease (PD) and Healthy Controls (HC).