Neurobiology of Aging 36 (2015) 1075e1082
Contents lists available at ScienceDirect
Neurobiology of Aging journal homepage: www.elsevier.com/locate/neuaging
Subcortical structures in amyotrophic lateral sclerosis Henk-Jan Westeneng a, Esther Verstraete a, Renée Walhout a, Ruben Schmidt a, Jeroen Hendrikse b, Jan H. Veldink a, Martijn P. van den Heuvel c,1, Leonard H. van den Berg a, *,1 a
Department of Neurology, Brain Center Rudolf Magnus, University Medical Center Utrecht, Utrecht, The Netherlands Department of Radiology, Brain Center Rudolf Magnus, University Medical Center Utrecht, Utrecht, The Netherlands c Department of Psychiatry, Brain Center Rudolf Magnus, University Medical Center Utrecht, Utrecht, The Netherlands b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 12 June 2014 Received in revised form 14 August 2014 Accepted 1 September 2014 Available online 6 September 2014
The aim of this study was to assess the involvement of deep gray matter, hippocampal subfields, and ventricular changes in patients with amyotrophic lateral sclerosis (ALS). A total of 112 ALS patients and 60 healthy subjects participated. High-resolution T1-weighted images were acquired using a 3T MRI scanner. Thirty-nine patients underwent a follow-up scan. Volumetric and shape analyses of subcortical structures were performed, measures were correlated with clinical parameters, and longitudinal changes were assessed. At baseline, reduced hippocampal volumes (left: p ¼ 0.007; right: p ¼ 0.011) and larger inferior lateral ventricles (left: p ¼ 0.013; right: p ¼ 0.041) were found in patients compared to healthy controls. Longitudinal analyses demonstrated a significant decrease in volume of the right cornu ammonis 2/3 and 4/dentate gyrus and left presubiculum (p ¼ 0.002, p ¼ 0.045, p < 0.001), and a significant increase in the ventricular volume in the lateral (left: p < 0.001; right: p < 0.001), 3rd (p < 0.001) and 4th (p ¼ 0.001) ventricles. Larger ventricles were associated with a lower ALSFRS-R score (p ¼ 0.021). In conclusion, ALS patients show signs of neurodegeneration of subcortical structures and ventricular enlargement. Subcortical involvement is progressive and correlates with clinical parameters, highlighting its role in the neurodegenerative process in ALS. Ó 2015 Elsevier Inc. All rights reserved.
Keywords: Amyotrophic lateral sclerosis Magnetic resonance imaging Longitudinal Basal ganglia Hippocampal subfields
1. Introduction Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease characterized by progressive upper and lower motor neuron degeneration (Kiernan et al., 2011). Although progressive motor neuron degeneration is the hallmark feature of ALS, widespread extramotor brain involvement can be found as well (Agosta et al., 2007; Verstraete et al., 2014). It has been shown that ALS affects a subnetwork in the brain including white matter connections with subcortical structures such as the thalamus, caudate nucleus, putamen, globus pallidus, and hippocampus (Verstraete et al., 2014). These findings may suggest involvement of these structures in the underlying neurodegenerative process in ALS. Magnetic resonance imaging (MRI) is a noninvasive, sensitive method allowing in vivo study of volume changes of structures within the brain. Most neuroimaging studies have focused on the
* Corresponding author at: Department of Neurology, G03.228, University Medical Center Utrecht, P.O. Box 85500, 3508 GA Utrecht, The Netherlands. Tel.: þ31 88 7557939; fax: þ31 30 2542100. E-mail address: [email protected] (L.H. van den Berg). 1 Authors contributed equally. 0197-4580/$ e see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.neurobiolaging.2014.09.002
cortical surface and white matter changes, aiming to capture upper motor neuron degeneration (Foerster et al., 2013; Turner et al., 2012). Postmortem studies have, however, shown prominent involvement of subcortical structures in ALS (Brettschneider et al., 2013; Geser et al., 2008; Takeda et al., 2009). In addition, studies using positron emission tomographyecomputed tomography (PET-CT), single-photon emission computed tomography (SPECT), magnetic resonance spectroscopy, and diffusion tensor imaging have also shown involvement of deep gray matter (DGM; thalamus, caudatus, putamen, pallidum, hippocampus, amygdala, accumbens), but morphometric changes have been reported sporadically (Sach et al., 2004; Turner et al., 2004; Verstraete et al., 2014). Currently, there is only 1 cross-sectional study that has reported a more detailed morphometric analysis of the DGM in ALS (Bede et al., 2013). This cross-sectional study in 39 ALS patients showed reduced volumes of the left caudate, left hippocampus, and right accumbens in ALS, underscoring the relevance of the subcortical structures in the disease process. Although, in the classical conception, the hippocampus is a structure of the cortex (3-layered cortex, not neocortex), one may also refer to it as a subcortical structure because it is located under the cortex.
1076
H.-J. Westeneng et al. / Neurobiology of Aging 36 (2015) 1075e1082
Longitudinal analyses are important for the understanding of patterns of disease progression and might result in the discovery of a more specific marker that can be used to monitor disease progression in ALS. Longitudinal neuroimaging studies may be an important tool to assess cerebral neurodegeneration in detail in a noninvasive way. For example, hippocampal atrophy has been reported in ALS but also in other neurological and psychiatric diseases (Bede et al., 2013; Thompson et al., 2004). Longitudinal analyses, in combination with a more detailed analysis of the hippocampus (hippocampal subfield segmentations), might shed more light on when the hippocampus becomes involved and which subfields in particular may be affected. We therefore studied volumes of ventricles and DGM (including hippocampal subfields) cross-sectionally and longitudinally in a large group of patients with ALS, and correlated our findings with clinical characteristics. 2. Methods 2.1. Participants All 172 participating subjects were recruited from the outpatient clinic for motor neuron diseases of the University Medical Center Utrecht in The Netherlands. Patients were classified as having definite, probable or possible ALS using the revised El Escorial criteria after excluding other conditions (Brooks et al., 2000). Subjects with a history of brain injury, epilepsy, psychiatric illness, other neurodegenerative disease (including frontotemporal lobe dementia), or structural brain disease were excluded. The median time between the first and second MRI scan was 5.5 months. Written informed consent was obtained from all participants, in accordance with the Declaration of Helsinki. 2.2. Clinical parameters Clinical characteristics, including handedness, disease duration, and survival were recorded. Functional status was evaluated using the revised ALS Functional Rating Scale (ALSFRS-R) (Cedarbaum et al., 1999). The disease progression rate was calculated using the formula (48 ALSFRS-R score)/disease duration (in months). Disease duration was evaluated from symptom onset. 2.3. Data acquisition A 3T Philips Achieve Medical Scanner was used to acquire a high-resolution T1-weighted image. Acquisition parameters were as follows: 3-dimensional fast field echo (FFE) using parallel imaging; TR/TE ¼ 10/4.6 ms, flip angel 8 , slice orientation sagittal, 0.75 0.75 0.8 mm (0.45 mm3) voxel size, field of view (FOV) ¼ 160 240 240 mm (9.216 dm3) and reconstruction matrix ¼ 200 320 320 covering the whole brain. Acquisition time was 11 minutes. The high-resolution MRI scans for scientific research were combined with a standard MRI examination, and these scans were reviewed by an experienced neuroradiologist. In the case of structural abnormalities, made scans were excluded. 2.4. Data processing Two groups of analyses were performed: cross-sectional and longitudinal. Both analyses focused on volumes of ventricles, DGM, and hippocampal subfields. For the cross-sectional analysis, subcortical volumes were automatically segmented and measured by FreeSurfer version 5.1 (Fischl et al., 2002, 2004). In addition, hippocampal subfields were automatically segmented and measured using a FreeSurfer
subroutine (Van Leemput et al., 2009). In this article, we have focused on cornu ammonis 1 (CA1), CA2/3, CA4/dentate gyrus (CA4/ DG), subiculum, and presubiculum. Fimbria and hippocampal fissure were disregarded because these are the smallest subfields and segmentation is less reliable (Van Leemput et al., 2009). To detect subtle regional volume changes of DGM, we performed a shape analysis of DGM (thalamus, putamen, caudate nucleus, nucleus accumbens, and hippocampus), previously reported to be altered in ALS (Agosta et al., 2009; Bede et al., 2013; Chang et al., 2005; Thivard et al., 2007). The shape of the above-mentioned DGM (except the hippocampus, the subfields of which were studied) was analyzed using the FSLs FIRST module version 5.0.0 (Patenaude et al., 2011). This vertex-based shape analysis was corrected for multiple testing using permutation tests with 10,000 permutations of cluster mass. The longitudinal analysis was performed using FreeSurfer, which creates an unbiased within-subject template space and image using robust inverse consistent registration (Reuter and Fischl, 2011; Reuter et al., 2010). Several processing steps, such as skull stripping, Talairach transforms, atlas registration, as well as spherical surface maps and parcellations are then initialized using common information from the within-subject template. This analysis method has been validated and shown to significantly increase reliability and statistical power (Reuter et al., 2012). 2.5. Statistical analysis 2.5.1. Cross-sectional analyses ManneWhitney U and Fisher exact tests were used to compare demographic and clinical data. Cross-sectional volume differences were compared between ALS and healthy controls using an analysis of covariance (ANCOVA). All of these analyses were adjusted for age and gender. Because of possible non-normality, we applied permutation tests with 10,000 permutations according to participant groups. 2.5.2. Relationship between imaging and clinical characteristics The relationship between cross-sectional clinical data and volumes in ALS patients was assessed using a combination of an ANCOVA and principal component analysis (PCA). PCA is a method that is used to describe a large group of variables by a smaller group of principal components (PCs), each consisting of a number of interrelated variables. Of all subcortical structures (n ¼ 20), 3 PCs were retained (Supplementary Fig. 1). These PCs were linearly regressed using clinical data (using ANCOVA) to assess the relationship between clinical data and volumes (represented by PCs), thereby adjusting for age and gender. This method was used because it takes advantage of the underlying correlation of different subcortical structures (e.g., the ventricles are actually part of 1 system; see also Supplementary Fig. 1) and increase both power and robustness by grouping variables that share common features (in PCs), reducing the number of comparisons, and minimizing possible effects of multicollinearity. The relationship between the PCs of the subcortical structures and survival was studied using univariate and multivariate Cox proportional hazards models. Known independent predictors of survival (age at onset and bulbar onset) were included in the multivariate analysis. Gender was not statistically significantly associated with survival and was therefore excluded from the multivariate survival analysis. 2.5.3. Longitudinal analyses A linear mixed-effects model (LME) was used to assess the rate of change of subcortical volumes in ALS patients over time while accounting for random between-subject variation (Bernal-Rusiel
H.-J. Westeneng et al. / Neurobiology of Aging 36 (2015) 1075e1082 Table 1 Demographic and clinical characteristics of study subjectsa ALS patients
Age Gender (male:female) Handedness (left:right)c ALSFRS-R score Site of onset (%) Bulbar Cervical Lumbosacral Disease progression rate (points decrease/month)d El Escorial criteria (%) Definite Probable Probable laboratory supported Possible Disease duration (months) Type (sporadic:familial ALS) C9orf repeat expansion Died
Healthy controls
Baseline (n ¼ 112)
Follow-up (n ¼ 39)
Baseline (n ¼ 60)
60.4 (24e78) 86:26 18:86 41 (29e48)
59.9 (24e78)b 60.4 (29e76) 30:9 46:14 5:32 7:44 36 (25e46)
24 (21) 55 (49) 33 (30) 0.5 (02.6)e
3 (8) 24 (62) 12 (31) 0.6 (0.5 to 3)f
13 (12) 48 (43) 35 (31)
4 (10)b 19 (49)b 10 (26)b
16 (14) 14.2 (4e75) 105:7 7/112 77/112
6 (15)b 18.5 (10e82) 37:2 5/39 23/39
0/60
Key: ALSFRS-R, revised ALS Functional Rating Scale. a Values are in median (range) unless otherwise specified. b At baseline. c Missing data for ALS patients (n ¼ 8), follow-up ALS patients (n ¼ 2), healthy controls (n ¼ 9). d Points decrease are points decrease on the ALSFRS-R score. e Disease progression rate for baseline was calculated using the formula (48 ALSFRS-R score)/disease duration (in months). f Disease progression rate for follow-up was calculated using the formula (ALSFRS-R score on baselineeALSFRS-R score on follow-up)/time since baseline MRI (in months).
et al., 2012). Age and gender were included as covariates in this analysis. All tests were 2-tailed, and p-values
Contents lists available at ScienceDirect
Neurobiology of Aging journal homepage: www.elsevier.com/locate/neuaging
Subcortical structures in amyotrophic lateral sclerosis Henk-Jan Westeneng a, Esther Verstraete a, Renée Walhout a, Ruben Schmidt a, Jeroen Hendrikse b, Jan H. Veldink a, Martijn P. van den Heuvel c,1, Leonard H. van den Berg a, *,1 a
Department of Neurology, Brain Center Rudolf Magnus, University Medical Center Utrecht, Utrecht, The Netherlands Department of Radiology, Brain Center Rudolf Magnus, University Medical Center Utrecht, Utrecht, The Netherlands c Department of Psychiatry, Brain Center Rudolf Magnus, University Medical Center Utrecht, Utrecht, The Netherlands b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 12 June 2014 Received in revised form 14 August 2014 Accepted 1 September 2014 Available online 6 September 2014
The aim of this study was to assess the involvement of deep gray matter, hippocampal subfields, and ventricular changes in patients with amyotrophic lateral sclerosis (ALS). A total of 112 ALS patients and 60 healthy subjects participated. High-resolution T1-weighted images were acquired using a 3T MRI scanner. Thirty-nine patients underwent a follow-up scan. Volumetric and shape analyses of subcortical structures were performed, measures were correlated with clinical parameters, and longitudinal changes were assessed. At baseline, reduced hippocampal volumes (left: p ¼ 0.007; right: p ¼ 0.011) and larger inferior lateral ventricles (left: p ¼ 0.013; right: p ¼ 0.041) were found in patients compared to healthy controls. Longitudinal analyses demonstrated a significant decrease in volume of the right cornu ammonis 2/3 and 4/dentate gyrus and left presubiculum (p ¼ 0.002, p ¼ 0.045, p < 0.001), and a significant increase in the ventricular volume in the lateral (left: p < 0.001; right: p < 0.001), 3rd (p < 0.001) and 4th (p ¼ 0.001) ventricles. Larger ventricles were associated with a lower ALSFRS-R score (p ¼ 0.021). In conclusion, ALS patients show signs of neurodegeneration of subcortical structures and ventricular enlargement. Subcortical involvement is progressive and correlates with clinical parameters, highlighting its role in the neurodegenerative process in ALS. Ó 2015 Elsevier Inc. All rights reserved.
Keywords: Amyotrophic lateral sclerosis Magnetic resonance imaging Longitudinal Basal ganglia Hippocampal subfields
1. Introduction Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease characterized by progressive upper and lower motor neuron degeneration (Kiernan et al., 2011). Although progressive motor neuron degeneration is the hallmark feature of ALS, widespread extramotor brain involvement can be found as well (Agosta et al., 2007; Verstraete et al., 2014). It has been shown that ALS affects a subnetwork in the brain including white matter connections with subcortical structures such as the thalamus, caudate nucleus, putamen, globus pallidus, and hippocampus (Verstraete et al., 2014). These findings may suggest involvement of these structures in the underlying neurodegenerative process in ALS. Magnetic resonance imaging (MRI) is a noninvasive, sensitive method allowing in vivo study of volume changes of structures within the brain. Most neuroimaging studies have focused on the
* Corresponding author at: Department of Neurology, G03.228, University Medical Center Utrecht, P.O. Box 85500, 3508 GA Utrecht, The Netherlands. Tel.: þ31 88 7557939; fax: þ31 30 2542100. E-mail address: [email protected] (L.H. van den Berg). 1 Authors contributed equally. 0197-4580/$ e see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.neurobiolaging.2014.09.002
cortical surface and white matter changes, aiming to capture upper motor neuron degeneration (Foerster et al., 2013; Turner et al., 2012). Postmortem studies have, however, shown prominent involvement of subcortical structures in ALS (Brettschneider et al., 2013; Geser et al., 2008; Takeda et al., 2009). In addition, studies using positron emission tomographyecomputed tomography (PET-CT), single-photon emission computed tomography (SPECT), magnetic resonance spectroscopy, and diffusion tensor imaging have also shown involvement of deep gray matter (DGM; thalamus, caudatus, putamen, pallidum, hippocampus, amygdala, accumbens), but morphometric changes have been reported sporadically (Sach et al., 2004; Turner et al., 2004; Verstraete et al., 2014). Currently, there is only 1 cross-sectional study that has reported a more detailed morphometric analysis of the DGM in ALS (Bede et al., 2013). This cross-sectional study in 39 ALS patients showed reduced volumes of the left caudate, left hippocampus, and right accumbens in ALS, underscoring the relevance of the subcortical structures in the disease process. Although, in the classical conception, the hippocampus is a structure of the cortex (3-layered cortex, not neocortex), one may also refer to it as a subcortical structure because it is located under the cortex.
1076
H.-J. Westeneng et al. / Neurobiology of Aging 36 (2015) 1075e1082
Longitudinal analyses are important for the understanding of patterns of disease progression and might result in the discovery of a more specific marker that can be used to monitor disease progression in ALS. Longitudinal neuroimaging studies may be an important tool to assess cerebral neurodegeneration in detail in a noninvasive way. For example, hippocampal atrophy has been reported in ALS but also in other neurological and psychiatric diseases (Bede et al., 2013; Thompson et al., 2004). Longitudinal analyses, in combination with a more detailed analysis of the hippocampus (hippocampal subfield segmentations), might shed more light on when the hippocampus becomes involved and which subfields in particular may be affected. We therefore studied volumes of ventricles and DGM (including hippocampal subfields) cross-sectionally and longitudinally in a large group of patients with ALS, and correlated our findings with clinical characteristics. 2. Methods 2.1. Participants All 172 participating subjects were recruited from the outpatient clinic for motor neuron diseases of the University Medical Center Utrecht in The Netherlands. Patients were classified as having definite, probable or possible ALS using the revised El Escorial criteria after excluding other conditions (Brooks et al., 2000). Subjects with a history of brain injury, epilepsy, psychiatric illness, other neurodegenerative disease (including frontotemporal lobe dementia), or structural brain disease were excluded. The median time between the first and second MRI scan was 5.5 months. Written informed consent was obtained from all participants, in accordance with the Declaration of Helsinki. 2.2. Clinical parameters Clinical characteristics, including handedness, disease duration, and survival were recorded. Functional status was evaluated using the revised ALS Functional Rating Scale (ALSFRS-R) (Cedarbaum et al., 1999). The disease progression rate was calculated using the formula (48 ALSFRS-R score)/disease duration (in months). Disease duration was evaluated from symptom onset. 2.3. Data acquisition A 3T Philips Achieve Medical Scanner was used to acquire a high-resolution T1-weighted image. Acquisition parameters were as follows: 3-dimensional fast field echo (FFE) using parallel imaging; TR/TE ¼ 10/4.6 ms, flip angel 8 , slice orientation sagittal, 0.75 0.75 0.8 mm (0.45 mm3) voxel size, field of view (FOV) ¼ 160 240 240 mm (9.216 dm3) and reconstruction matrix ¼ 200 320 320 covering the whole brain. Acquisition time was 11 minutes. The high-resolution MRI scans for scientific research were combined with a standard MRI examination, and these scans were reviewed by an experienced neuroradiologist. In the case of structural abnormalities, made scans were excluded. 2.4. Data processing Two groups of analyses were performed: cross-sectional and longitudinal. Both analyses focused on volumes of ventricles, DGM, and hippocampal subfields. For the cross-sectional analysis, subcortical volumes were automatically segmented and measured by FreeSurfer version 5.1 (Fischl et al., 2002, 2004). In addition, hippocampal subfields were automatically segmented and measured using a FreeSurfer
subroutine (Van Leemput et al., 2009). In this article, we have focused on cornu ammonis 1 (CA1), CA2/3, CA4/dentate gyrus (CA4/ DG), subiculum, and presubiculum. Fimbria and hippocampal fissure were disregarded because these are the smallest subfields and segmentation is less reliable (Van Leemput et al., 2009). To detect subtle regional volume changes of DGM, we performed a shape analysis of DGM (thalamus, putamen, caudate nucleus, nucleus accumbens, and hippocampus), previously reported to be altered in ALS (Agosta et al., 2009; Bede et al., 2013; Chang et al., 2005; Thivard et al., 2007). The shape of the above-mentioned DGM (except the hippocampus, the subfields of which were studied) was analyzed using the FSLs FIRST module version 5.0.0 (Patenaude et al., 2011). This vertex-based shape analysis was corrected for multiple testing using permutation tests with 10,000 permutations of cluster mass. The longitudinal analysis was performed using FreeSurfer, which creates an unbiased within-subject template space and image using robust inverse consistent registration (Reuter and Fischl, 2011; Reuter et al., 2010). Several processing steps, such as skull stripping, Talairach transforms, atlas registration, as well as spherical surface maps and parcellations are then initialized using common information from the within-subject template. This analysis method has been validated and shown to significantly increase reliability and statistical power (Reuter et al., 2012). 2.5. Statistical analysis 2.5.1. Cross-sectional analyses ManneWhitney U and Fisher exact tests were used to compare demographic and clinical data. Cross-sectional volume differences were compared between ALS and healthy controls using an analysis of covariance (ANCOVA). All of these analyses were adjusted for age and gender. Because of possible non-normality, we applied permutation tests with 10,000 permutations according to participant groups. 2.5.2. Relationship between imaging and clinical characteristics The relationship between cross-sectional clinical data and volumes in ALS patients was assessed using a combination of an ANCOVA and principal component analysis (PCA). PCA is a method that is used to describe a large group of variables by a smaller group of principal components (PCs), each consisting of a number of interrelated variables. Of all subcortical structures (n ¼ 20), 3 PCs were retained (Supplementary Fig. 1). These PCs were linearly regressed using clinical data (using ANCOVA) to assess the relationship between clinical data and volumes (represented by PCs), thereby adjusting for age and gender. This method was used because it takes advantage of the underlying correlation of different subcortical structures (e.g., the ventricles are actually part of 1 system; see also Supplementary Fig. 1) and increase both power and robustness by grouping variables that share common features (in PCs), reducing the number of comparisons, and minimizing possible effects of multicollinearity. The relationship between the PCs of the subcortical structures and survival was studied using univariate and multivariate Cox proportional hazards models. Known independent predictors of survival (age at onset and bulbar onset) were included in the multivariate analysis. Gender was not statistically significantly associated with survival and was therefore excluded from the multivariate survival analysis. 2.5.3. Longitudinal analyses A linear mixed-effects model (LME) was used to assess the rate of change of subcortical volumes in ALS patients over time while accounting for random between-subject variation (Bernal-Rusiel
H.-J. Westeneng et al. / Neurobiology of Aging 36 (2015) 1075e1082 Table 1 Demographic and clinical characteristics of study subjectsa ALS patients
Age Gender (male:female) Handedness (left:right)c ALSFRS-R score Site of onset (%) Bulbar Cervical Lumbosacral Disease progression rate (points decrease/month)d El Escorial criteria (%) Definite Probable Probable laboratory supported Possible Disease duration (months) Type (sporadic:familial ALS) C9orf repeat expansion Died
Healthy controls
Baseline (n ¼ 112)
Follow-up (n ¼ 39)
Baseline (n ¼ 60)
60.4 (24e78) 86:26 18:86 41 (29e48)
59.9 (24e78)b 60.4 (29e76) 30:9 46:14 5:32 7:44 36 (25e46)
24 (21) 55 (49) 33 (30) 0.5 (02.6)e
3 (8) 24 (62) 12 (31) 0.6 (0.5 to 3)f
13 (12) 48 (43) 35 (31)
4 (10)b 19 (49)b 10 (26)b
16 (14) 14.2 (4e75) 105:7 7/112 77/112
6 (15)b 18.5 (10e82) 37:2 5/39 23/39
0/60
Key: ALSFRS-R, revised ALS Functional Rating Scale. a Values are in median (range) unless otherwise specified. b At baseline. c Missing data for ALS patients (n ¼ 8), follow-up ALS patients (n ¼ 2), healthy controls (n ¼ 9). d Points decrease are points decrease on the ALSFRS-R score. e Disease progression rate for baseline was calculated using the formula (48 ALSFRS-R score)/disease duration (in months). f Disease progression rate for follow-up was calculated using the formula (ALSFRS-R score on baselineeALSFRS-R score on follow-up)/time since baseline MRI (in months).
et al., 2012). Age and gender were included as covariates in this analysis. All tests were 2-tailed, and p-values
