Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease with upper and lower motor neuron involvement as its clinical hallmarks. Dating back to very early observations that a significant percentage of ALS patients develop a dementia it has been clear that extramotor involvement can occur. That such involvement is the rule rather than the exception has been underscored by findings from neuropsychological [1–4] and neuroimaging [5–9] investigations. The recent discovery that the C9ORF72 hexanucleotide repeat can cause both, frontotemporal dementia and ALS as well as overlapping syndromes, further underscores that the molecular underpinnings of ALS might damage extramotor and motor parts of the nervous system [10]. The question therefore arises, to what extent disturbances of non-motor functions are reflected in changes of cortical structure. Primary motor and extramotor involvement has been demonstrated repeatedly by voxel based morphometry (VBM) [11–13], although other studies did not report atrophy [14, 15].

VBM is a fully automated, operator-independent technique that allows the voxel-wise comparison of segmented grey and white matter images between two groups of subjects [16–19]. As such, it has the advantages of being fast and independent of any observer rating or a priori hypothesis on the location of the structural abnormality. There are several limitations of this approach, however [18, 20]. For example, VBM simultaneously reflects size, position and morphology. As nonlinear spatial normalization and spatial filtering is used to accommodate anatomical variations between brains, there is the danger of missing important differences, in particular when these are circumscribed. Moreover, VBM reveals differences in grey matter density, which may not equate in a simple way to atrophy.

In the present investigation we therefore employed an alternative to VBM, cortical thickness analysis (CTA). CTA is based on a semiautomatic/interactive segmentation of brain white and grey matter followed by a determination of the thickness of the cortical layer [21–24]. Thickness values are then entered into standard between group statistics allowing a comparison for the entire cortical surface. Neuropathological studies have described atrophy involving the precentral gyrus with loss of Betz cells and gliosis in ALS [25, 26] which lead us to suspect a decreased cortical thickness in ALS. Also, cortical thinning of the motor cortex has been described recently in ALS [27–30] and primary lateral sclerosis [31]. Verstraete et al. [28, 29] could demonstrate that cortical thinning in temporal regions was related to faster clinical progression, whereas in a recent study by Agosta et al. [30] age and cortical thinning in sensorimotor cortex was related in ALS patients but not in controls. Interestingly, no relationship was found between clinical severity and cortical thickness. Finally, in a recent study by Schuster et al. [32] clinical variables were related to cortical thinning of the primary motor cortex (PMC). Upper motor neuron (UMN) signs in the bulbar region were associated with bilateral thinning within the bulbar segment on the motor cortex, whereas UMN signs of the extremities were associated with thinning in the limb segment of the motor cortex.

In the present investigation, we asked specifically whether other areas in addition to the motor cortex are affected with regard to cortical thinning. We were further interested in whether CTA is more sensitive to cortical changes in ALS than VBM and performed CTA on a sample of ALS patients and controls that has previously been reported in a VBM study [13].

Expectedly, the current study revealed cortical thinning of the primary motor cortex bilaterally in ALS. Importantly, atrophy was also found for the postcentral gyrus bilaterally demonstrating that ALS extends beyond the motor system. These results resonate well with previous findings using voxel-based morphometry [11–13, 40–42]. The current analysis of cortical thickness also pointed to an involvement of additional extramotor areas including, for example, the cingulate cortex, as well as regions in the temporal and occipital cortex. That the CTA findings of the current study are not spurious is suggested by other recent studies [28–30] as well as by reports of functional changes implicating the same areas. For example, marked alterations of early visual event-related potential components [43] show that visual cortical areas are indeed altered in the ALS disease process. Also, impaired initiation and inhibition of movements implies the supplementary motor area and right prefrontal cortex [8, 44, 45].

These findings could guide further neuropathological investigations in ALS in order to pinpoint the underlying histological changes. A marked loss of large pyramidal cells and Betz cells in conjunction with astrocytic gliosis has been described in motor cortex in ALS [25, 46]. Other neuropathological studies have shown loss of parvalbumin-positive interneurons in ALS irrespective of the degree of loss of Betz-cells in the motor cortex [47]. Loss of GABAergic interneurons in the primary motor cortex and beyond (dorsolateral prefrontal cortex and anterior cingulate cortex) has also been documented [48]. According to the present data extramotor involvement in ALS is widespread also involving brain areas in the frontal, parietal, temporal and occipital lobe. It remains to be seen whether neuropathological changes are equivalent in all of these regions. Also, it is important to note that the patients involved in the current study were affected only mildly or moderately. Application of cortical thickness analysis in more advanced disease is therefore warranted.

A previous study using MRI-based cortical thickness analysis in ALS used a region-of-interest approach rather than a whole cortex analysis [27]. In contrast to the present results, this study found a significant thinning of the precentral but not postcentral gyrus in ALS. The thickness of the precentral gyrus was determined to be 2.15 mm in normal controls which compares to about 2.8 mm found in neuropathological studies [26]. The current data more closely resemble numbers usually found in post-mortem studies. Another recent study investigated primary lateral sclerosis and focussed on cortical thickness of the primary motor cortex in the area of the hand-knob [31]. Patients showed thinning of the motor cortex but not of the primary sensory cortex. The present investigation as well as other recent studies [28–30] significantly extends these previous findings by using a whole cortex approach supplemented by regions of interest analysis.

Interestingly, cortical thickness in the right superior frontal gyrus and in the right inferior frontal gyrus showed a negative correlation with the ALSFRS-R score. This is counter-intuitive, as lower ALSFRS-R scores indicate worse clinical status and, thus, one would expect lower scores to be associated with cortical thinning. It has to be kept in mind, however, that the ALSFRS-R score assesses global functionality of the patient with a strong focus on motor functions and that the brain areas which showed correlations in the present study were not primary motor. In fact, we did not observe any correlations of the ALSFRS-R score to cortical thickness in the precentral gyrus. A recent study investigated cortical thickness and used scores for upper motor neuron and lower motor neuron involvement [31]. In this study, a regionally specific thinning (e.g., bulbar part of the motor cortex for bulbar symptoms) was described in correlation with upper motor neuron signs.

A few brain regions showed an increase of cortical thickness in ALS, some of them even adjacent to regions showing thinning. None of the VBM studies published on ALS reported increases of grey matter density and it is unclear whether such changes were assessed. While the whole cortex statistics used in the present study were properly safeguarded against type 1 error by correction for multiple comparisons using cluster-size thresholding, it is not impossible that one or the other brain area with increased cortical thickness represents a “false positive”. However, we do not believe that this can explain the observed pattern. Interestingly, the areas with increased thickness are located mainly in temporal and medial frontal cortex (parts of the cingulate cortex). One possibility is that the observed increases are related to compensatory mechanisms. For example, we have observed an increase of cortical thickness in brain areas related to visuospatial processing and memory in subjects that did not learn to read (unpublished), which we took as indicating a greater reliance of illiterates on visuospatial and memory processes. An interplay between frontal and temporal structures is often seen in functional imaging studies of tasks involving executive and working memory functions [49–52] and changes in anatomical connectivity between frontal and temporal structures have been documented in ALS [15]. Moreover, a pattern of hypo- and hyperactivation in a motor task has been observed recently in relation to cortical atrophy in ALS [53]. If structural loss (cortical thinning) in one part of this system is related to structural gain (cortical thickening) in the connected areas remains to be tested in further studies.

The present study adds to previous investigations in showing that the cortical involvement in ALS extends considerably beyond the motor system. The task for the future is to establish a relation between this involvement and motor, cognitive, and emotional deficits in ALS. Initial steps in this direction have already been taken in a recent publication [57].