**4. Correlation with phenotype**

#### **4.1. ALS subtypes**

The pattern of motor neuron involvement in ALS is highly heterogeneous. Depending on the relative upper and LMN involvement and depending on the neuroanatomical region within the motor system with the most extensive pathology, different subtypes of motor neuron degeneration have been defined. While in some patients UMN features (spasticity, hyperre‐ flexia) dominate the clinical picture, LMN features prevail in other patients. It remains unclear if the pure UMN disorder, PLS (primary lateral sclerosis), and the pure LMN disorder, progressive muscular atrophy (PMA), should be regarded as separate disease entities or merely as the extreme ends of the ALS spectrum [25]. Similarly, the site of onset is also highly variable, with onset in a limb ('spinal onset') being more frequent than onset in the bulbar musculature ('bulbar onset') [25]. The neuroimaging signature of these subtypes has not yet been extensively studied using PET. The challenge will be to find commonalities and differ‐ ential representations of the different endophenotypes.

#### *4.1.1. PLS and Mills' syndrome*

PLS is a variant of ALS with selective UMN signs for several years [76]. Mills' syndrome is an unusual unilateral variant of PLS, which eventually spreads to the contralateral side after a variable time period [25]. On postmortem examination, there is no difference in the essential pathological processes (e.g. TDP-43 positive intraneuronal inclusions) [77]. However, to explain this phenotypic variability, there must be a difference in the focal initiation and spreading pattern of the neurodegeneration. PET imaging has several advantages to assess this question, like in vivo usability and availability of specific tracers. Unfortunately, since no large studies investigating PET in PLS/Mills' patients have been undertaken so far, we need to rely on a handful of small case series. Four studies in PLS patients performed either FDG PET or [11C]-flumazenil PET and mainly found similar abnormalities as seen in ALS patients [44, 51, 78, 79]. However, in one study involvement of the primary motor cortex seemed to be more severe than in pure ALS [79]. On the other hand, some specific regions like the prefrontal cortex and posterior cingulate seem to be spared in PLS [44, 78]. So, based on these PET findings, neuropathology in PLS may be more restricted to the motor cortex.

Only five cases of Mills' patients with PET imaging have been reported so far. While one patient had an asymmetric involvement of the motor cortices, an almost unilateral pattern of hypo‐ metabolism (FDG PET) or hypercaptation (TSPO radioligand) has been noticed in the four remaining patients [80–82]. So, compared to PLS, neuropathology in Mills' syndrome seems to be focused even more on one unilateral motor cortex, suggesting a more restricted contig‐ uous spread of disease in this endophenotype.

#### *4.1.2. PMA*

PMA is a motor neuron disease with selective involvement of the LMN, and probably has to be seen as an unusual variant of ALS [25]. Literature regarding FDG PET in PMA patients is very limited. Only two early studies performing FDG PET in ALS patients performed a subanalysis in patients with only LMN signs. In keeping with the clinical absence of UMN signs, no to only very mild cerebral hypometabolism is present in these patients [35, 36]. Larger studies are required to establish if peri-Rolandic hypometabolism is present in a proportion of PMA patients and if this predicts progression to ALS or disease outcome.

In comparison, a few MRI-based imaging studies investigating UMN involvement in PMA remained inconclusive as well. One study found no evidence for thinning of the motor cortex on high-resolution MRI [83]. While one study found modest though clear abnormalities of the CST on DTI imaging [84], this was contradicted by another earlier study [85]. Finally, an fMRI study revealed modest prefrontal activation abnormalities in PMA patients [86].

So based on several imaging modalities, it is so far unclear whether significant measurable UMN involvement is present in all PMA patients.

#### *4.1.3. Spinal versus bulbar*

**4. Correlation with phenotype**

34 Update on Amyotrophic Lateral Sclerosis

ential representations of the different endophenotypes.

uous spread of disease in this endophenotype.

*4.1.2. PMA*

*4.1.1. PLS and Mills' syndrome*

The pattern of motor neuron involvement in ALS is highly heterogeneous. Depending on the relative upper and LMN involvement and depending on the neuroanatomical region within the motor system with the most extensive pathology, different subtypes of motor neuron degeneration have been defined. While in some patients UMN features (spasticity, hyperre‐ flexia) dominate the clinical picture, LMN features prevail in other patients. It remains unclear if the pure UMN disorder, PLS (primary lateral sclerosis), and the pure LMN disorder, progressive muscular atrophy (PMA), should be regarded as separate disease entities or merely as the extreme ends of the ALS spectrum [25]. Similarly, the site of onset is also highly variable, with onset in a limb ('spinal onset') being more frequent than onset in the bulbar musculature ('bulbar onset') [25]. The neuroimaging signature of these subtypes has not yet been extensively studied using PET. The challenge will be to find commonalities and differ‐

PLS is a variant of ALS with selective UMN signs for several years [76]. Mills' syndrome is an unusual unilateral variant of PLS, which eventually spreads to the contralateral side after a variable time period [25]. On postmortem examination, there is no difference in the essential pathological processes (e.g. TDP-43 positive intraneuronal inclusions) [77]. However, to explain this phenotypic variability, there must be a difference in the focal initiation and spreading pattern of the neurodegeneration. PET imaging has several advantages to assess this question, like in vivo usability and availability of specific tracers. Unfortunately, since no large studies investigating PET in PLS/Mills' patients have been undertaken so far, we need to rely on a handful of small case series. Four studies in PLS patients performed either FDG PET or [11C]-flumazenil PET and mainly found similar abnormalities as seen in ALS patients [44, 51, 78, 79]. However, in one study involvement of the primary motor cortex seemed to be more severe than in pure ALS [79]. On the other hand, some specific regions like the prefrontal cortex and posterior cingulate seem to be spared in PLS [44, 78]. So, based on these PET

Only five cases of Mills' patients with PET imaging have been reported so far. While one patient had an asymmetric involvement of the motor cortices, an almost unilateral pattern of hypo‐ metabolism (FDG PET) or hypercaptation (TSPO radioligand) has been noticed in the four remaining patients [80–82]. So, compared to PLS, neuropathology in Mills' syndrome seems to be focused even more on one unilateral motor cortex, suggesting a more restricted contig‐

PMA is a motor neuron disease with selective involvement of the LMN, and probably has to be seen as an unusual variant of ALS [25]. Literature regarding FDG PET in PMA patients is

findings, neuropathology in PLS may be more restricted to the motor cortex.

**4.1. ALS subtypes**

In most ALS patients, the disease starts with asymmetric weakness of a limb, and hence, this classical form of ALS is called 'spinal onset ALS'. In about 20% of ALS patients, however, weakness starts in the bulbar muscles, this form is called 'bulbar onset ALS' [25]. While both endophenotypes clearly have a distinct disease initiation and disease course, they eventually converge into a common phenotype of generalized weakness. The pathological substrates underlying these initial differences are largely not understood. PET imaging has tried to elucidate some aspects of this enigma.

While one study using FDG PET suggested a differential pattern of involvement between spinal and bulbar onset ALS in frontal and parietal regions [45], this was not confirmed by others [42]. A study investigating perfusion (regional cerebral blood flow, rCBF) in ALS patients reported a significantly lower rCBF of the frontal lobe in bulbar onset patients compared to spinal onset [87]. Based on TSPO PET imaging, one study found evidence for increased neuroinflammation in the brainstem of bulbar onset ALS patients, whereas neuro‐ inflammation in the motor cortex seemed to be less pronounced [70]. So, the specific functional imaging correlate of these two endophenotypes has not been clearly established yet.

#### **4.2. Severity of disease**

An independent objective marker for severity and spreading pattern of disease pathology is highly needed. None of the imaging biomarkers so far has been able to reflect local disease severity and disease spreading in a longitudinal fashion. A limited amount of studies inves‐ tigated the link between PET imaging and clinical scores. First, although several FDG PET studies in ALS patients found no correlation with parameters of disease severity in general, one study did reveal a correlation of prefrontal hypometabolism with a reduced ALS-FRS R (ALS functional rating scale revised version) [44]. Second, PET imaging of microglia, via TSPO radioligands, was not correlated with ALS-FRS R in two studies [68, 69]. However, one of these studies found a relation with clinical burden of UMN signs [68]. Third, abnormalities on [11C] flumazenil PET in one study was not correlated with ALS-FRS, whereas it was associated with the UMN score [74]. So in general, findings with the currently available PET targets show very little, if any, clear correlation with clinical severity of disease. Longitudinal studies are required to find out if PET imaging is of value in tracking disease progression.

## **5. Correlation with genotype**

While most cases of ALS are sporadic, about 10% are caused by a variety of genetic deficits, with alterations in the *C9orf72* and *SOD1* gene being the most frequent and most studied [88]. A few PET studies have been performed in genetic subtypes of ALS.

#### **5.1. C9orf72**

Two recent studies investigated 26 patients in total with *C9orf72*-related ALS using FDG PET and found in general more severe hypometabolism [44, 46]. Remarkably, both studies also reported hypometabolism in the thalamus and parts of the limbic system to be present almost uniquely in *C9orf72* patients. An example of FDG PET findings in a *C9orf72* ALS patient is

**Figure 4. PET abnormalities in a C9orf72 ALS patient**. Example of Z-score images of FDG PET imaging of a *C9orf72* patient with ALS revealing a remarkable hypometabolism of the thalamus, as noted on axial (A and B) and sagittal (C and D) sections.

provided in **Figure 4**. This involvement of thalamus, extrapyramidal system and limbic system is probably the neuroanatomical correlate of the increased incidence of phenocopies of several other neurodegenerative diseases in *C9orf72* mutation carriers. Among others, phenocopies of Parkinson's disease, Huntington's disease, corticobasal degeneration, Alzheimer's disease and several neuropsychiatric diseases (psychosis, schizophrenia) have been reported in *C9orf72* mutation carriers [57].

#### **5.2. SOD1 (D90A)**

the UMN score [74]. So in general, findings with the currently available PET targets show very little, if any, clear correlation with clinical severity of disease. Longitudinal studies are required

While most cases of ALS are sporadic, about 10% are caused by a variety of genetic deficits, with alterations in the *C9orf72* and *SOD1* gene being the most frequent and most studied [88].

Two recent studies investigated 26 patients in total with *C9orf72*-related ALS using FDG PET and found in general more severe hypometabolism [44, 46]. Remarkably, both studies also reported hypometabolism in the thalamus and parts of the limbic system to be present almost uniquely in *C9orf72* patients. An example of FDG PET findings in a *C9orf72* ALS patient is

**Figure 4. PET abnormalities in a C9orf72 ALS patient**. Example of Z-score images of FDG PET imaging of a *C9orf72* patient with ALS revealing a remarkable hypometabolism of the thalamus, as noted on axial (A and B) and sagittal (C

to find out if PET imaging is of value in tracking disease progression.

A few PET studies have been performed in genetic subtypes of ALS.

**5. Correlation with genotype**

36 Update on Amyotrophic Lateral Sclerosis

**5.1. C9orf72**

and D) sections.

Worldwide, the D90A is the most common *SOD1* mutation, although it is not the most studied one [89]. This mutation can be inherited in an autosomal recessive or dominant way, which is quite unusual in ALS. Two studies performing [11C]-Flumazenil-PET in a total of 21 D90A patients revealed that these patients showed a specific pattern of reduced tracer binding confined to the left frontotemporal junction and anterior cingulate gyrus, without involvement of the motor and premotor cortices [73, 74]. Larger studies on other *SOD1* mutations or other genetic subtypes of ALS are largely lacking.
