**2. FDG PET imaging in ALS**

#### **2.1. Early FDG PET and perfusion SPECT studies revealed widespread cortical abnormalities**

In the 1980s and 1990s of the previous century a mere handful of studies investigated a small number of ALS patients using 18F-2-fluoro-2-deoxy-D-glucose PET (FDG PET) imaging. The common denominator of their findings was a generalized cerebral hypometabolism [35–37]. Despite the low spatial resolution of the PET scanners at the time, some studies did reveal a predominance for the motor cortex [36] and even provided preliminary evidence for frontal hypometabolism [37, 38] which was even related to clinical cognitive impairment in one study. A few studies assessing cerebral perfusion using SPECT imaging confirmed a predominant involvement of the motor cortex with extensive frontal hypoperfusion, often related to cognitive impairment [39–41]. These pilot studies were instrumental in redirecting the view of ALS as a disease exclusively affecting motor neurons towards a multisystem neurodege‐ nerative disease as generally accepted nowadays.

#### **2.2. Regions involved on recent FDG PET imaging**

More recent FDG PET studies in ALS patients, using the next-generation PET scanners with higher spatial resolution, could confirm the presence of hypometabolism in the primary motor cortex, which is hence thought to be the signature of ALS on FDG PET imaging [42–44]. Besides the primary motor cortex, other peri-Rolandic regions like the premotor cortex and primary sensory cortex have been found to be affected on FDG PET [42–45]. In keeping with the clinical and pathophysiological overlap with FTD, several prefrontal (dorsolateral prefrontal cortex, orbitofrontal cortex, anterior frontal cortex) and temporal (anterior temporal lobe, fusiform gyrus) regions frequently display hypometabolism in ALS patients [42–45]. Some studies even report hypometabolism of primary and associative visual cortices in ALS patients [42, 45]. Examples of typical FDG PET patterns seen in ALS are given in **Figure 2**. Patient 1 is an example of an ALS patient with modest hypometabolism in the peri-Rolandic areas on FDG PET. In patient 2, extensive hypometabolism in the motor cortex can be noted. In about 10% of patients, extensive regions of hypometabolism are also present in the frontal and/or anterior temporal lobes. Patient 3 is an example of an ALS patients with extensive areas of hypometabolism in the frontal areas. Patient 4 has pronounced anterior temporal hypometabolism.

**Figure 2. Commonly affected regions on FDG PET in ALS patients**. Three regions frequently and typically display hypometabolism on FDG PET in ALS patients; motor cortex, prefrontal cortex and anterior temporal lobe. The three left images depict the stereotactic surface projections of brain FDG PET uptake. The three images on the right show the corresponding Z-score images (comparing patient to healthy volunteers). In patient 1 and 2, hypometabolism in the Rolandic area (including parts of the motor cortex) is noted. While this is only mild in patient 1, patient 2 has extensive hypometabolism. In patient 3, an obvious hypometabolism in the prefrontal cortex is noted. In patient 4, extensive hy‐ pometabolism of the anterior temporal lobe is noted.

Recently, several studies also reported the presence of (relative) hypermetabolism on FDG PET in certain regions. Hypermetabolism in ALS patients seems to be most obvious in the infra‐ tentorial region, like midbrain, pons and cerebellum [44–46]. This hypermetabolism is thought to be the reflection of increased astrocytosis along the course of the CST [45]. Also, hyperme‐ tabolism in mesial temporal structures, like hippocampus and amygdala, has been reported [44, 45].

#### **2.3. Detection of frontotemporal involvement on FDG PET**

so far is provided in **Figure 1**. Apart from the commonly used tracers for indirect neuronal functioning, such as glucose metabolism ([18F]-FDG) and perfusion, more specific receptor or protein deposition tracers used in other neurodegenerative diseases, like Parkinson's disease ([18F]-FP-CIT) and Alzheimer's disease ([11C]-PIB), have also been investigated in ALS patients. Recently, tracers with an affinity for specific cell types, like neurons ([11C]-flumazenil, [11C]- WAY100635), microglial cells (TSPO ligands) and astrocytes ([11C]-DED), can also highlight

**2.1. Early FDG PET and perfusion SPECT studies revealed widespread cortical**

In the 1980s and 1990s of the previous century a mere handful of studies investigated a small number of ALS patients using 18F-2-fluoro-2-deoxy-D-glucose PET (FDG PET) imaging. The common denominator of their findings was a generalized cerebral hypometabolism [35–37]. Despite the low spatial resolution of the PET scanners at the time, some studies did reveal a predominance for the motor cortex [36] and even provided preliminary evidence for frontal hypometabolism [37, 38] which was even related to clinical cognitive impairment in one study. A few studies assessing cerebral perfusion using SPECT imaging confirmed a predominant involvement of the motor cortex with extensive frontal hypoperfusion, often related to cognitive impairment [39–41]. These pilot studies were instrumental in redirecting the view of ALS as a disease exclusively affecting motor neurons towards a multisystem neurodege‐

More recent FDG PET studies in ALS patients, using the next-generation PET scanners with higher spatial resolution, could confirm the presence of hypometabolism in the primary motor cortex, which is hence thought to be the signature of ALS on FDG PET imaging [42–44]. Besides the primary motor cortex, other peri-Rolandic regions like the premotor cortex and primary sensory cortex have been found to be affected on FDG PET [42–45]. In keeping with the clinical and pathophysiological overlap with FTD, several prefrontal (dorsolateral prefrontal cortex, orbitofrontal cortex, anterior frontal cortex) and temporal (anterior temporal lobe, fusiform gyrus) regions frequently display hypometabolism in ALS patients [42–45]. Some studies even report hypometabolism of primary and associative visual cortices in ALS patients [42, 45]. Examples of typical FDG PET patterns seen in ALS are given in **Figure 2**. Patient 1 is an example of an ALS patient with modest hypometabolism in the peri-Rolandic areas on FDG PET. In patient 2, extensive hypometabolism in the motor cortex can be noted. In about 10% of patients, extensive regions of hypometabolism are also present in the frontal and/or anterior temporal lobes. Patient 3 is an example of an ALS patients with extensive areas of hypometabolism in

the frontal areas. Patient 4 has pronounced anterior temporal hypometabolism.

specific pathophysiological processes of ALS.

nerative disease as generally accepted nowadays.

**2.2. Regions involved on recent FDG PET imaging**

**2. FDG PET imaging in ALS**

28 Update on Amyotrophic Lateral Sclerosis

**abnormalities**

As explained above, a clear link between ALS and FTD has been established over the last years. This is also backed up by the early studies showing frontotemporal hypometabolism using FDG PET and rCBF measures [37, 38]. Based on the extent of frontotemporal versus motor neuron involvement, patients across the ALS-FTD spectrum can be divided into five categories [47]. Pure ALS (without any evident cognitive abnormality) and pure FTD (without any obvious motor abnormality) are located at the opposite ends of this spectrum. Patients who meet diagnostic criteria for both ALS as FTD are considered to have 'ALS-FTD'. ALS patients with mild behavioural dysfunction are classified as having ALS with behavioural impairment ('ALSbi'), whereas patients with mild executive and language dysfunction are said to have ALS with cognitive impairment ('ALSci'). Patients with a diagnosis of FTD and some motor neuron involvement are said to have 'FTD-MND'.

The signature of frontotemporal involvement on PET imaging constitutes hypometabolism mainly focused on the prefrontal cortex, often extending to the entire frontal cortex and anterior temporal cortex and even to the thalamus [48, 49]. Contrary to FTD patients, this frontotemporal hypometabolism is often more symmetric in ALS-FTD patients [49]. Impor‐ tantly, this hypometabolism is often already present in otherwise normal frontotemporal lobes on MRI, leading to the general view that hypometabolism precedes atrophy [50]. This makes FDG PET a very sensitive tool to detect frontotemporal involvement even in a very early clinical stage. Hence, frontotemporal hypometabolism on FDG PET has extensive potential to become an important diagnostic marker, and this independently from hypometabolism in the primary motor cortex.

Several studies performing both PET imaging as neuropsychological testing in ALS patients found a correlation between extent of frontotemporal hypometabolism and neuropsycholog‐ ical performance [37, 38, 43]. A recent large study indeed confirmed that ALS patients with mild cognitive impairment ('ALSci') exhibit moderate frontotemporal hypometabolism, which is clearly more pronounced in real 'ALS-FTD' patients and less pronounced to even absent in 'pure ALS' patients [48]. So, when patients are stratified along the ALS-FTD spectrum, there seems to be a proportionate correlation between clinical cognitive involvement and fronto‐ temporal hypometabolism on PET imaging.

Frontotemporal hypometabolism on PET not only has the potential to become a diagnostic marker, it has also been shown to provide important prognostic information. In a considerably large study of ALS patients, extensive prefrontal hypometabolism was associated with significantly shorter survival [51].

#### **2.4. FDG PET as a (differential) diagnostic marker**

Extensive studies demonstrating high diagnostic accuracy are a prerequisite for any imaging biomarker to become incorporated into the diagnostic criteria of ALS. Recently, four large studies assessed the sensitivity and specificity of FDG PET in ALS patients compared to healthy controls. The first study made use of a 'region of interest method' [42]. When all brain regions were taken into account, FDG PET was able to discriminate ALS patients from controls with a sensitivity of 89% and specificity of 82.5%. When only a specific set of regions was used, an even higher sensitivity (95%) with the same specificity (82.5%) could be obtained. The second study analysed FDG PET images using a method evaluating disease-specific spatial patterns on a voxel-based manner, trying to disclose the networks with the highest diagnostic value [52]. When all relevant networks were taken into account an astonishing accuracy of 99.0% was achieved. Remarkably, when only the most discriminating network (i.e. bilateral cerebel‐ lum and midbrain) was used, accuracy was still 96.3%. The third study reported an accuracy of 89.7% using a VOI-based discriminant analysis, further increasing to 95% if a discriminant analysis based on a support vector machine (SVM) approach was used [44]. In the fourth study, a prospective validation of the diagnostic algorithm based on SVM was carried out in 105 novel cases and a sensitivity and specificity of 100% was obtained [51]. All these studies suggest that FDG PET is a very sensitive marker of ALS pathology that can be used in a clinical setting.

There is, however, one major limitation of these studies. While these studies report a high sensitivity using healthy controls, it is clear specificity may be lower since ALS patients need to be discriminated from patients with ALS mimicking diseases. Unfortunately, no large studies using FDG PET in such disease mimics exist. A study performing FDG PET in ten patients with Kennedy's disease even surprisingly revealed the presence of frontal hypome‐ tabolism [53]. This means that FDG PET may be insufficient to discriminate ALS from certain ALS mimicking diseases.

#### **2.5. FDG PET as a prognostic marker**

The signature of frontotemporal involvement on PET imaging constitutes hypometabolism mainly focused on the prefrontal cortex, often extending to the entire frontal cortex and anterior temporal cortex and even to the thalamus [48, 49]. Contrary to FTD patients, this frontotemporal hypometabolism is often more symmetric in ALS-FTD patients [49]. Impor‐ tantly, this hypometabolism is often already present in otherwise normal frontotemporal lobes on MRI, leading to the general view that hypometabolism precedes atrophy [50]. This makes FDG PET a very sensitive tool to detect frontotemporal involvement even in a very early clinical stage. Hence, frontotemporal hypometabolism on FDG PET has extensive potential to become an important diagnostic marker, and this independently from hypometabolism in the primary

Several studies performing both PET imaging as neuropsychological testing in ALS patients found a correlation between extent of frontotemporal hypometabolism and neuropsycholog‐ ical performance [37, 38, 43]. A recent large study indeed confirmed that ALS patients with mild cognitive impairment ('ALSci') exhibit moderate frontotemporal hypometabolism, which is clearly more pronounced in real 'ALS-FTD' patients and less pronounced to even absent in 'pure ALS' patients [48]. So, when patients are stratified along the ALS-FTD spectrum, there seems to be a proportionate correlation between clinical cognitive involvement and fronto‐

Frontotemporal hypometabolism on PET not only has the potential to become a diagnostic marker, it has also been shown to provide important prognostic information. In a considerably large study of ALS patients, extensive prefrontal hypometabolism was associated with

Extensive studies demonstrating high diagnostic accuracy are a prerequisite for any imaging biomarker to become incorporated into the diagnostic criteria of ALS. Recently, four large studies assessed the sensitivity and specificity of FDG PET in ALS patients compared to healthy controls. The first study made use of a 'region of interest method' [42]. When all brain regions were taken into account, FDG PET was able to discriminate ALS patients from controls with a sensitivity of 89% and specificity of 82.5%. When only a specific set of regions was used, an even higher sensitivity (95%) with the same specificity (82.5%) could be obtained. The second study analysed FDG PET images using a method evaluating disease-specific spatial patterns on a voxel-based manner, trying to disclose the networks with the highest diagnostic value [52]. When all relevant networks were taken into account an astonishing accuracy of 99.0% was achieved. Remarkably, when only the most discriminating network (i.e. bilateral cerebel‐ lum and midbrain) was used, accuracy was still 96.3%. The third study reported an accuracy of 89.7% using a VOI-based discriminant analysis, further increasing to 95% if a discriminant analysis based on a support vector machine (SVM) approach was used [44]. In the fourth study, a prospective validation of the diagnostic algorithm based on SVM was carried out in 105 novel cases and a sensitivity and specificity of 100% was obtained [51]. All these studies suggest that FDG PET is a very sensitive marker of ALS pathology that can be used in a clinical setting.

motor cortex.

30 Update on Amyotrophic Lateral Sclerosis

temporal hypometabolism on PET imaging.

**2.4. FDG PET as a (differential) diagnostic marker**

significantly shorter survival [51].

Besides its capacities as a diagnostic marker, FDG PET is increasingly proposed as a potential marker for prognosis. So far, only two studies assessed the value of FDG PET in predicting the prognosis of ALS [44]. Assessing 70 ALS patients, extensive prefrontal hypometabolism was associated with a significantly shorter survival. This is in line with previous studies that reported concomitant clinical FTD or cognitive/behavioural impairment with a worse prog‐ nosis [4]. More recently, a larger prospectively collected cohort of 175 (including the initial 70 patients) was studied [51]. It was confirmed with longer follow-up that extensive hypome‐ tabolism in the prefrontal or anterior temporal lobe was associated with a shorter survival. In a Cox regression, taking into account other prognostic factors, such as age of onset, site of onset, diagnostic delay, FVC and slope in the ALS FRS-R, extensive frontotemporal hypometabolism was significantly correlated with shortened survival (**Figure 3**).

**Figure 3. Relation of ALS survival with degree of frontotemporal hypometabolism**. Kaplan-Meier survival plot of all ALS patients (ALS group 1 + ALS group 2) with (red line, n = 34) and without (blue line, n = 141) extensive hypometab‐ olism in the frontal and/or temporal cortex (n = 175, p < 0.001) (This research was originally published in *JNM*. Van Weehaege et al. prospective validation of 18F-FDG brain PET discriminant analysis methods in the diagnosis of amyo‐ trophic lateral sclerosis. *J Nucl Med*. 2016;vol:pp-pp. © by the Society of Nuclear Medicine and Molecular Imaging, Inc.) [51].
