Brain Stem Glucose Hypermetabolism in Amyotrophic Lateral Sclerosis/Frontotemporal Dementia and Shortened Survival: An ¹⁸F-FDG PET/MRI Study

DISCUSSION

This study, to our current knowledge, was the first to exploit integrated ¹⁸F-FDG PET/MRI to study the metabolic patterns of patients with ALS and FTD. In our population, MRI could be useful when differentiating ALS subjects from FTD-MND because corticospinal tract hyperintensities seem to be more frequent in the former group (40% vs. 8%). Furthermore, we have noticed the motor band sign (susceptibility-weighted sequence) in the majority (64%) of ALS subjects. This qualitative finding could complement, at a single-subject level, the diagnostic accuracy of PET regional glucose hyper- and hypometabolism.

We detected midbrain/pons and medulla oblongata increased uptake in patients as compared with controls, confirming previous PET/CT findings (15,16). Furthermore, we determined how medulla oblongata metabolism relates to patient survival, suggesting a possible prognostic value for PET/MRI.

The results of our study indicated that the uptake of all 3 brain regions investigated (midbrain/pons, medulla oblongata, and cervical spinal cord) correlates significantly, as shown by the paired correlation (Fig. 2A). An explanation for the correlation of metabolic patterns in the 3 regions is the involvement of the corticospinal and corticobulbar tracts containing the projections of upper motor neurons from the motor cortex. A bimodal distribution could be seen in the density plot distribution of controls as compared with the ALS/FTD continuum in Figure 2B. The second lower peak of the bimodal distribution was, however, due to a single, slightly outlier, control subject.

Theoretically, the expected effect of a neurodegenerative disease is a reduction in tissue metabolic rate caused by neuronal loss; indeed, frontal hypometabolism is one of the cortical signatures of the diseases along the ALS/FTD spectrum (12–14). Nevertheless, strong evidence has been brought in favor of neuroinflammation as one of the key steps in the pathogenic cascade that leads to ALS/FTD. Initially described in pathology studies, glial infiltration has later been confirmed in vivo both in models and in humans (4,5).

A possible explanation, which links neuroinflammation with neurodegeneration, involves these 2 types of glial cells, namely activated microglia and astrocytes. Activated microglia, which have been found in the motor cortex, pons, and thalamus of patients with ALS, contribute to the pathogenesis propagating and sustaining the tissue damage through the release of free radicals and other neurotoxic substances such as glutamate. Moreover, a shift from a neuroprotective to a disrupting phenotype is apparent starting in the early stages of the disease (2). Astrocytes physiologically account for a consistent share of central nervous system glucose consumption. Thereafter, they start glycolysis to provide lactate to neurons; in hindsight, astrocytes play a key role in coupling glucose metabolism with synaptic activity. In the case of neurodegeneration, glutamate excess leads to an increased glucose accumulation in astrocytes, even though neurons degenerate (26). Moreover, during the course of the disease, astrocytes replace dead neurons and axons following the corticospinal and corticobulbar tracts, further increasing the metabolic uptake of the degenerating brain region. Neuroimaging evidence for the involvement of these glial cells in the pathogenesis of ALS/FTD has been brought by studies that exploited neuroinflammation-specific ligands such as ¹¹C-l-deprenyl-D₂ (4), mapping of astrocyte activation, the activated microglia marker ¹¹C(R)-PK11195 (5,6), and translocator protein ligand ¹⁸F-DPA-714 (27), showing increased tracer uptake along the corticospinal tract.

From a metabolic perspective, these phenomena result in an increased relative ¹⁸F-FDG uptake in regions of neuronal degeneration in affected individuals compared with controls. However, when evaluating the relative weight of the 2 cell types, one must consider that the microglia seem to relate more to hypometabolic areas, as shown in frontotemporal areas of FTD patients (5,6), whereas astrocytosis drives glucose consumption and is, therefore, more likely associated with the hypermetabolic burden (26), in line with previous findings (4). We acknowledge that this evidence cannot prove the exclusive role of 1 of these 2 glial cells—a role that is based on a continuous interplay involving other cell types also—and that longitudinal studies will be needed to better characterize the interplay between neurodegeneration and hypermetabolism and differences in their behavior between upper and lower motor neurons.

An alternative reason for the finding of glucose hypermetabolism in ALS/FTD patients can be a methodologic issue involving the normalization process of ¹⁸F-FDG data. We normalized the uptake values of the regions of interest for the values of the occipital white matter to exclude interindividual differences. Normalizing for other regions would require an a priori assumption, namely that none of the background VOIs will be affected by hypo- or hypermetabolism. In the present study, occipital white matter was chosen as the reference because it seems spared from the ALS/FTD neurodegenerative process (14,24), and none of the PET/CT studies on subjects affected with the disease showed significant metabolic variations in that region.

Another issue we faced was the difference in age distribution between the 2 groups. To ignore this effect on the uptake values, we adjusted for age through a linear regression model (25) and we then used the residual values for the group analysis (Fig. 3). However, we usually expect a higher metabolism in younger patients with respect to controls. On the contrary, in this setting, although the controls were younger than the patients, they had a lower metabolism; this fact reassures us about the choice of groups selected and about the obtained results.

Although considering a small cohort, the subgroup analysis included patients with differences in ALS onset, namely spinal (n = 11), bulbar (n = 5), and behavioral/cognitive (n = 12). Although the groups with spinal or behavioral onset showed higher ¹⁸F-FDG uptake than did controls, bulbar onset did not prove significant because of the small number of individuals in that group. Interestingly, the density plot of normalized ¹⁸F-FDG uptake in FTD-MND was similar to that of patients with ALS (Figs. 2 and 4). Therefore, we can state that hypermetabolism in the corticospinal tract could be a common feature for all the ALS subgroups, thereby suggesting, once again, common metabolic and pathophysiologic patterns within the disease spectrum. Prospective studies involving a larger number of individuals with different sites of onset, including FTD patients without clinical MND, as well as comparisons between cortical and subcortical regions with lower motor neurons (to elucidate the possible metabolic relations between upper and lower motor neurons), are expected to strengthen the current results.

Bulbar-onset ALS is a prognostic factor associated with short survival (28). Here, we instead showed that the metabolic uptake of the medulla oblongata was significantly associated with short survival (Fig. 5). This finding supports the hypothesis that neuroinflammation of midbrain structures is one of the main mechanisms involved in ALS progression, highlighting a potential role for neurometabolic studies as prognostic and outcome predictors. Moreover, a previous study (17) reported a significant relationship between the whole-spinal-cord–normalized uptake value and patient survival; thus, it will be necessary to compare the metabolic patterns of these 2 regions to draw consistent evidence.