Associations Between Brainstem Volume and Alzheimer’s Disease Pathology in Middle-Aged Individuals of the Framingham Heart Study

INTRODUCTION

The pathophysiological process of Alzheimer’s disease (AD) starts two to three decades prior to the clinical symptoms. Several autopsy studies suggested that neuromodulatory nuclei in the brainstem may be among the first regions in the brain to accumulate tau pathology, starting as early as young adulthood [1–3]. Accrual of tau pathology has been identified in midbrain nuclei, including the substantia nigra and raphe nuclei, and in the pontine nuclei, including the locus coeruleus and pedunculopontine nuclei. Some of these changes may occur early in life; for example, the locus coeruleus has been reported to harbor hyperphosphorylated tau aggregates in approximately 50% of 30-year-old individuals [4, 5]. Similarly, raphe nucleus neurons accumulate tau deposition in precortical stages [6]. Previous studies reported lower grey matter volume in the rostral midbrain and pons in AD patients relative to cognitively normal individuals [7]. Cognitively healthy older individuals who later progressed to AD dementia also exhibited lower grey matter volume in the midbrain and in a specific pontine cluster colocalized to the locus coeruleus as compared to those who did not progress [8, 9]. These findings suggest that brainstem volumetric changes may be occurring early, starting in clinically normal individuals, and reflect underlying progression of AD pathology. As tau pathology is one of the major contributors to neurodegeneration, we sought to examine the association between brainstem volume and neocortical amyloid-β (Aβ) or tau pathology in the third generation (Gen 3) cohort of the Framingham Heart Study (FHS), comprising middle-aged clinically normal individuals.

METHODS

RESULTS

A description of the demographics of our sample is provided in Table 1. Lower midbrain volume was significantly associated with greater amygdala and fusiform tau (non-PVC), but these associations were at-trend or nonsignificant for the PVC tau data. We observed no other associations between any of the brainstem volumes and the regional tau measurements (Table 2). But lower volume of the medulla, pons, and midbrain were each associated with greater neocortical amyloid burden (Table 2 and Fig. 1). In the non-PVC data, these relationships were also observed for the medulla and pons. These results did not change when adjusting the models for global cognition (PC1, Supplementary Table 1) or APOE E4 status (Supplementary Table 2), or when removing potential influential PiB-values > 2SD from the mean (Supplementary Table 3). In separate models, striatal volume or fourth ventricular volume were not associated with any of the PET variables, thus supporting the specificity of these findings. Hippocampal volume was not significantly associated with amyloid burden or tau pathology. No other associations between hippocampal volume and tau pathology were detected. We observed no interaction between neocortical amyloid burden and any of the brainstem volumes on cortical tau burden measures (Table 2).

OPEN IN VIEWER

Fig. 1

Brainstem volumes are associated with neocortical PiB Unadjusted associations between the brainstem volume (as a % of total cranial volume) and neocortical PiB deposition (log-transformed, here referred to as amyloid FLR). FLR, fronto-lateral temporal–retrosplenial aggregate (neocortical region); PiB, Pittsburgh Compound B.

DISCUSSION

Several brainstem nuclei exhibit AD-related neurodegenerative processes early in adulthood [1], hence we examined associations between subregional brainstem volumes and the initial stages of AD pathology in a large cohort of middle-aged cognitively normal individuals. We found that lower brainstem volumes, in particular the medulla and the pons, were associated with greater Aβ deposition. Contrary to our expectations, based on the Braak staging [4] and previous MRI-work [8], we did not observe robust relationships between any of the brainstem volumes and tau deposition. Effect sizes for the amygdala and fusiform gyrus—regions with strong connections with the brainstem—are in a similar range to those of neocortical PiB, providing some initial support for the hypothesis that neurodegenerative diseases including AD start in the brainstem. Interestingly, hippocampal volume was not associated with Aβ or tau deposition in our sample. There is a plethora of studies reporting correlations between hippocampal volume and amyloid or tau pathology, or disease progression, in older or impaired individuals [20–24]. However, in younger or midlife individuals, previous studies reported no relationship between APOE genotyping and hippocampal volume [25–27], indicating that loss of grey matter tissue is a downstream effect of accumulating pathology [28]. In fact, recent work by Bussy et al. (2019) indicated that amyloid may already be accumulating at younger ages, prior to any measurable volumetric changes [25]. We observed that lower brainstem volumes were associated with initial amyloid accumulation in the neocortex, the earliest sign of preclinical AD. Based on these observations, we speculate that our associations between brainstem volumes and amyloid pathology reflect brain changes that occur prior to cortical neurodegeneration and thus may be valuable in the detection of preclinical AD. Other physiological markers of brainstem function, such as changes in heart rate variability, have also been associated with a greater risk of developing dementia. However, autopsy studies have identified a long-time lag between tau deposition in the locus coeruleus and neuronal loss [29], indicating that the observed volumetric changes in the locus coeruleus may reflect loss of projection fiber density or other elements besides locus coeruleus neuronal loss. Similarly, while tau accumulation in the raphe nucleus occurs prior to cortical tau deposition, neuronal loss in the raphe nucleus was typically detected only in later Braak stages [6, 30]. Thus, the changes in brainstem volume may reflect processes other than tau deposition in brainstem nuclei. Previous volumetric MRI studies have already reported a relationship between lower brainstem volume and worse cognition in prodromal AD, but not in normal older individuals [7, 9] and no prior study has related brainstem volumes to brain amyloid and tau burden. It is important to note that these findings were specific to the brainstem, and not observed for any of the control regions, including the hippocampus. We did not yet probe the potential subtle functional implications of these AD-related changes in brainstem volume, but plan to do so when our sample size is large enough to capture cognitive variability in this cognitively healthy middle-aged population.

Unfortunately, detecting tau aggregation in vivo in these tiny nuclei of the brainstem is not yet feasible, due to the off-target binding of current tau radiotracers to neuromelanin [31, 32], which is abundantly present in the substantia nigra and locus coeruleus. Novel emerging imaging methods may enable examination of the integrity of the locus coeruleus, substantia nigra and potentially also the other neuromodulatory subcortical systems, as well as their molecular alterations during aging or disease contributing to volumetric changes [33–35]. To better understand the temporal dynamics between volumetric changes and deposition of cortical beta-amyloid and tau, future studies should explore these associations through longitudinal studies that include repeated MRI and PET imaging.