Detailed volumetric analysis of the hypothalamus in behavioral variant frontotemporal dementia

Behavioral variant frontotemporal dementia (bvFTD) is a neurodegenerative disorder characterized by atrophy in the frontal and temporal lobes and progressive behavioral and cognitive impairment [1]. Although the majority of cases are sporadic, about 10–20 % are due to an autosomal dominant mutation in one of three genes: microtubule-associated protein tau (MAPT), progranulin (GRN), and chromosome 9 open reading frame 72 (C9orf72) [2]. One of the characteristic symptoms of bvFTD is the development of abnormal eating behaviors such as hyperphagia, changes in food preference, and craving for sweet foods, which are found in the majority of patients [3‐5], and have been shown to help discriminate bvFTD from Alzheimer’s disease [6, 7]. However, the neuroanatomical correlates of abnormal eating behavior in bvFTD are unclear. Previous studies have suggested the importance of an orbitofrontal–insular–striatal brain network [8, 9] but one study has also investigated the role of the hypothalamus, finding a correlation of abnormal eating behavior with lower volumes of the posterior hypothalamus [10].

The hypothalamus is the regulatory center for feeding and satiety [11]. It is composed of a number of different subnuclei and is highly interconnected with other parts of the central nervous system, particularly the brainstem, limbic system, and cortex [12, 13]. However, the hypothalamus is difficult to identify on magnetic resonance imaging and a detailed anatomical analysis of subdivisions of the hypothalamus has not yet been performed in bvFTD. In this study, we aimed to develop an optimized manual segmentation of the hypothalamus and its subunits using a novel protocol, and then use this to investigate patterns of atrophy in bvFTD and specifically whether differences could be seen in different genetic mutations.

Demographic characteristics are reported in Table 1: no significant differences were found in gender, age, or education between controls and bvFTD, but the patient group scored significantly lower on the MMSE. All patients scored abnormally on the eating disturbance subscale of the CBI-R: 12 patients scored in the mild to moderate range (1–50 %) and 6 patients scored in the severe to very severe range (51–100 %).

No significant differences were seen in demographics, disease duration, MMSE, FRS, or total CBI-R score between the MAPT and C9orf72 subgroups. Looking at the individual eating disturbance subscores of the CBI-R, there was a trend for higher scores in the MAPT group compared with C9orf72 in wanting to eat the same foods repeatedly [mean (SD) score 2.6 (1.3) versus 1.3 (1.4)], with the opposite trend (higher score in C9orf72) in decline in table manners [mean (SD) score 2.7 (1.8) versus 1.0 (1.1)].

The bvFTD group showed a 17 % lower total hypothalamic volume compared with controls (mean (SD) 783 (113) versus 944 (73) mm³, p < 0.0005, Mann–Whitney U test) with a similar reduction in both the right (17 %) and left (18 %) hypothalamus compared with controls. The MAPT mutation group showed a non-significant lower right and left hypothalamic volumes on both sides compared with C9orf72 (10–13 % difference) (Table 2).

The subsegmentation analysis revealed significant differences in the total (left and right combined) volumes of the superior regions (both anterior and tuberal) as well as the posterior region (superior tuberal 22 %, posterior 24 %, and anterior superior region 41 % smaller than controls), with a similar pattern when looking at the individual right and left volumes. No significant differences survived correction for multiple comparisons in the inferior regions (anterior and tuberal) between the bvFTD group and controls (Table 2).

Looking at the individual MAPT and C9orf72 groups, only the MAPT mutation carriers showed significant differences from controls when corrected for multiple comparisons with superior (anterior and tuberal) and posterior regions being smaller. The C9orf72 group showed a trend to smaller anterior superior (p = 0.009) and left superior tuberal regions (p = 0.047). Direct comparisons of the genetic groups revealed a trend for a smaller right superior tuberal region (16 %, p = 0.036) and posterior region (24 %, p = 0.066) in the MAPT group compared with C9orf72.

In the total bvFTD group, patients who scored in the severe to very severe range of the CBI-R eating disturbance subscale had a trend to a lower total hypothalamic volume [740 (89) mm³] than those in the mild to moderate range [805 (122) mm³]. This trend for a lower volume in those scoring in the severe to very severe range was also seen in the superior tuberal region [205 (36) versus 236 (37) mm³]. Similarly, in both genetic groups, there was a non-significantly lower total hypothalamic volume in those scoring within the severe to very severe range: in the MAPT group 721 (120) versus 773 (147) mm³, in the C9orf72 group 794 (35) versus 884 (60) mm³.

Using a novel segmentation protocol for the hypothalamus and its subunits based upon registered volumetric T1 and T2 MR images, we have shown that the hypothalamus is substantially smaller in patients with bvFTD compared with controls, particularly in the superior and posterior regions. There was also a trend for a smaller hypothalamus, particularly in the superior tuberal region, in those who had severe eating disturbance. Furthermore, there is significant atrophy in the MAPT mutation group (in superior and posterior areas), but no significant differences from controls in the C9orf72 mutation group.

Our findings are different from the only previous study of hypothalamic volume in FTD [10], which found significant atrophy in the posterior hypothalamus. Our study differed from this in a number of aspects, both technical and clinical. Technically, we used different criteria for the segmentation of the hypothalamus, in particular, paying attention to the exclusion of the fornix, and inclusion of the supraoptic, suprachiasmatic, retrochiasmatic, and arcuate (or infundibular) nuclei. Furthermore, we were able to delineate five specific subunits of the hypothalamus, whereas Piguet et al. [10] used an arbitrary definition of anterior and posterior regions of the hypothalamus, splitting it through the middle coronal plane of their segmentation. Their ‘posterior’ hypothalamus may therefore include part of the tuberal region as we have defined it in this study. Clinically, it is unclear whether the cohorts overlap as their 18 patients are not genetically defined unlike the group studied here which contains a significant number of genetic bvFTD cases. The Piguet cohort has an earlier mean disease duration (3.3 years versus 9.1 years here), although disease severity is similar (MMSE 23.9 versus 25.0 here; CBI-R eating disturbance score 6.9 versus 7.7 here). In a separate pathological analysis, they investigated six tau-positive and six TDP-43 positive FTD cases, also finding problems more posteriorly, attributing this to atrophy in the TDP-43 group. However, their tau group contained only cases with a specific type of tau pathology, Pick’s disease (a 3-repeat tauopathy), and their TDP-43 group similarly contained only cases with one subtype (TDP-43 type B). Recent studies have made it clear that there are large differences both clinically and pathologically between the different tau and TDP-43 pathological subtypes (which number at least four in each group), and so one cannot extrapolate to a significant difference between ‘all TDP-43’ and ‘all tau’ pathology by investigating only one subtype. Our study suggests that a particular tau group (MAPT mutations) appear to have significant hypothalamic volume loss compared with a TDP-43 group (C9orf72 expansions), the opposite finding to Piguet et al. [10].

We found that patients with more severe eating disturbance had a trend to lower hypothalamic volumes, particularly within the superior tuberal region. However, the differences did not reach statistical significance and so caution should be attributed to these findings until further studies are performed. The CBI-R provides only limited information on the many different types of abnormal eating behaviors, as only four questions are asked. Eating behaviors in bvFTD are complex and varied, and include different aspects, such as carbohydrate craving, overeating, obsessions for specific foods and oral exploration of inedible objects, which may not always coexist in an individual patient [7, 27, 28]. There was a trend for a difference in the type of eating behaviors exhibited between the two genetic groups with the MAPT group scoring higher on wanting to eat the same foods repeatedly and the C9orf72 group showing a greater decline in table manners. Further studies are required using more detailed feeding questionnaires to explore these issues further [7].

This study found significant atrophy in the superior and posterior subunits of the hypothalamus which contain the paraventricular nucleus (anterior superior region), dorsomedial nucleus (superior tuberal region), and lateral hypothalamic areas (superior tuberal and posterior regions). These subnuclei are all involved in important aspects of appetite regulation and contain neuropeptide-expressing neurons and neuropeptide receptors [20, 29‐32]. Interestingly, the inferior tuberal area was not atrophic, an area which contains the arcuate (or infundibular) and ventromedial nuclei. The arcuate nucleus is the primary target of metabolic and hormonal signals from the periphery, with important connections to other nuclei in the hypothalamus particularly the paraventricular nucleus [30]. This suggests that appetite abnormalities in bvFTD could be due to changes in neuropeptides (or neuropeptide receptors) within the superior and posterior areas of the hypothalamus and/or from disruption of pathways from the arcuate nucleus to other areas of the hypothalamus. As there is differential neuropeptide expression within these nuclei, different neuropeptide levels should be impaired compared with others in bvFTD: such a hypothesis has yet to be explored.

There was a trend for greater posterior involvement in the MAPT group compared with the C9orf72 group, with both left and right posterior regions 24 % smaller in the MAPT group. As well as the posterior part of the lateral hypothalamus, the posterior region contains the mammillary bodies, which are connected to the amygdala and hippocampus, areas known to be major areas of atrophy in patients with MAPT mutations [33]. Given this finding, future studies should further investigate the relationship of memory impairment in these patients to hypothalamic atrophy.

In summary, bvFTD patients had smaller hypothalamic volumes compared with controls, with atrophy localized to subnuclei regulating food intake, reward and perception of satiety, and related to the severity of the eating disturbance. Moreover, different genetic mutations seem to have a differential impact on the hypothalamus, although further studies in larger genetic and pathological series are required to confirm this. The structural and functional connections of the hypothalamus should be further explored in bvFTD, particularly how they relate to the orbitofrontal-insular-striatal reward network previously identified. Lastly, the results of this study suggest testable hypotheses of the role of different neuropeptides in impaired appetite regulation in bvFTD.