Proton Magnetic Resonance Spectroscopy of the Motor Cortex in 70 Patients With Amyotrophic Lateral Sclerosis

Objective  To evaluate proton magnetic resonance spectroscopy for detection and monitoring of upper motoneuron degeneration in patients with amyotrophic lateral sclerosis.

Methods  Seventy patients with amyotrophic lateral sclerosis according to the El Escorial criteria were compared with 48 healthy control subjects. Single-volume proton magnetic resonance spectroscopy (echo time, 272 milliseconds; repetition time, 2000 milliseconds) was performed in both motor cortices for detection of N-acetylaspartate (NAA), phosphocreatine + creatine ([P]Cr), and choline-containing compounds (Cho) to calculate the metabolite ratios NAA/Cho, NAA/(P)Cr, and Cho/(P)Cr. In addition, absolute metabolite concentrations of NAA, (P)Cr, and Cho were obtained in 30 patients and 15 controls with the unsuppressed water signal used as an internal reference.

Results  Absolute concentrations of NAA (P<.001) and (P)Cr (P<.05) were reduced in motor cortices of patients, whereas Cho concentrations remained unchanged. The NAA/Cho and NAA/(P)Cr ratios were reduced in all El Escorial subgroups (P<.001). The Cho/(P)Cr ratio was elevated in patients with definite amyotrophic lateral sclerosis (P<.05). Metabolite ratio changes corresponded to the lateralization of clinical symptoms and were weakly correlated with disease duration and disease severity. In follow-up observations of 16 patients during a mean (±SD) of 12.1 ± 8.7 months, NAA/Cho dropped by 9.1% (P<.01), and Cho/(P)Cr increased by 7.0% (P<.01). Changes of metabolite ratios were significantly correlated with progression of disease severity.

Conclusions  Measurement of NAA concentrations and NAA/Cho ratios appear to be most suitable for detection of motor cortex degeneration by single-volume proton magnetic resonance spectroscopy. Reduced NAA/Cho ratios correspond to aspects of the clinical presentation and reflect disease progression in follow-up measurements.

AMYOTROPHIC lateral sclerosis (ALS) is a neurodegenerative disorder characterized by progressive degeneration of motoneurons in the motor cortex, brainstem, and spinal cord.¹ Whereas lower motoneuron involvement in ALS is easily detected by electromyography, electrophysiologic examination of central motor pathways and conventional magnetic resonance imaging provide only limited information with respect to upper motoneuron (UMN) degeneration or bulbar involvement.^2-5 Proton magnetic resonance spectroscopy (¹H-MRS) has been introduced as a method to detect and to quantify subtle neurodegenerative changes not apparent on conventional magnetic resonance imaging. It has been repeatedly shown that ¹H-MRS demonstrates metabolite changes in the motor cortex and brainstem of patients with ALS.^6-15 Since these changes have been attributed to neuronal loss, ¹H-MRS may be used to detect cortical and bulbar neurodegeneration.

Even more important, ¹H-MRS findings may serve as surrogate markers for ALS disease progression in therapeutic trials.¹⁶ However, most ¹H-MRS studies in ALS have been performed in a limited number of patients, and the relationship between spectroscopic findings and the heterogeneous clinical presentation of the disease is far from clear. In addition, little is known about long-term changes in ¹H-MRS findings in the course of ALS. To resolve these questions, we conducted a follow-up and extension of our previously published study¹⁵ analyzing single-volume ¹H-MRS for the detection and monitoring of neuronal degeneration in the motor cortex of patients with ALS. Seventy patients were investigated for up to 2 years and compared with 48 healthy controls. For better interpretation of ¹H-MRS findings, metabolite ratios and absolute metabolite concentrations were determined.

Seventy patients with ALS according to the El Escorial criteria¹⁷ were recruited from the Department of Neurology at the University Hospital of Bonn, Bonn, Germany (mean [±SD] age, 55.4 ± 12.7 years; 39 men and 31 women). No patient had a history of any other neurologic disease. At the time of their first ¹H-MRS examination, 15 patients were classified as having suspected ALS, 18 patients as possible ALS, 10 patients as probable ALS, and 27 patients as definite ALS. In 16 patients, follow-up investigations were performed within 4 to 24 months (3 with suspected ALS, 3 with possible ALS, 4 with probable ALS, and 6 with definite ALS). Mean duration of the observation period was 12.1 ± 8.7 months. All but 2 of these patients received riluzole treatment during the whole observation period. Forty-eight healthy volunteers (aged 51.8 ± 17.6 years; 31 men and 17 women) without history of any neurologic disease served as control subjects. All patients and control subjects gave informed consent.

Patients underwent physical examination at the time of each ¹H-MRS examination. The following information was obtained for further statistical analysis: (1) hemisphere dominance for motor functions as determined by handedness; (2) disease duration in months, defined as the latency between when the patient first noted weakness subsequently attributed to ALS and the ¹H-MRS examination; (3) disease severity as assessed by the score of Jablecki et al,¹⁸ which encompasses the evaluation of speech, swallowing, respiration, ambulation, muscle strength, and extremity function (a high score indicates greater impairment); (4) lateralization of clinical symptoms with respect to the severity of extremity pareses (asymmetry of clinical presentation was assumed if there was a minimum of 1 point difference in at least 1 muscle of the muscle strength subscales of the Jablecki et al score); and (5) location of disease onset differentiating bulbar- vs limb-onset ALS.

Seventy patients with ALS and 48 controls were investigated by ¹H-MRS for determination of relative metabolite concentrations of N-acetylaspartate (NAA), phosphocreatine + creatine ([P]Cr), and choline-containing compounds (Cho) by means of the NAA/Cho, NAA/(P)Cr, and Cho/(P)Cr ratios. In 30 patients and 15 controls, absolute metabolite concentrations of NAA, (P)Cr, and Cho in millimoles per liter of brain tissue were also determined with the unsuppressed water signal used as an internal reference. Of the 16 patients undergoing follow-up ¹H-MRS investigations, 8 were examined twice, 7 were examined 3 times, and 1 was examined 4 times (total of 25 follow-ups) to determine relative metabolite concentrations. The ¹H-MRS examiners (W.B. and F.T.) were not aware of clinical data.

Details of the ¹H-MRS protocol for determination of relative metabolite concentrations are given elsewhere.¹⁵ In brief, investigations were performed on 1.5-T whole-body magnetic resonance systems (Gyroscan S15/ACS II and Gyroscan ACS-NT; Philips Medical Systems, Best, the Netherlands). A spectroscopic volume of interest (40 × 30 × 25 mm) was placed anterior to the central sulcus in the motor cortex and subjacent white matter including the primary motor cortex within Brodmann areas 4 and 6 (Figure 1). The ¹H-MR spectra were acquired with a repetition time of 2000 milliseconds, echo time of 272 milliseconds, and 128 signal averages. Relative metabolite concentrations for NAA, (P)Cr, and Cho were determined by lorentzian curve fitting of the corresponding resonance in the frequency spectra. From these data, the metabolite ratios NAA/Cho, NAA/(P)Cr, and (P)Cr/Cho were determined.

Metabolite ratios were calculated in both hemispheres of all patients with ALS and 12 healthy controls. Measurements in dominant and nondominant hemispheres yielded consistent results in these 12 control subjects. We therefore assumed that ¹H-MRS findings were independent of hemisphere lateralization of motor functions, and measurements in the remaining 36 controls were thus performed only in the dominant hemisphere. Consequently, data for 140 patient and 60 control motor cortices were available.

Absolute concentrations of NAA, (P)Cr, and Cho were determined from spin-echo spectra in the volume of interest described above, with the brain water signal used as an internal reference.⁹ For this purpose, an additional acquisition of ¹H-MR spectra was obtained without water suppression, with repetition time and echo time of 3000 and 272 milliseconds, respectively, and averaged over 16 or 32 free induction decays, yielding NAA/water signal ratios. In the postprocessing of these unsuppressed spectra, polynomial baseline subtraction removed the broad lorentzian slopes of the water resonance curve underlying the much smaller metabolite peaks. To determine NAA/water concentration ratios by extrapolation to repetition time of ∞ and echo time of 0 milliseconds, T2 relaxation times and relative fractions of cerebrospinal fluid and tissue water within the volume of interest water were obtained by a biexponential fit to a series of unsuppressed spin-echo spectra with echo times of 30, 70, 136, 272, 400, 700, and 1000 milliseconds and 4 signal averages each. Errors caused by variations in the assumed T1 values of mixed gray-white matter (800 milliseconds) and cerebrospinal fluid (3000 milliseconds) were minimized by a long repetition time of 6000 milliseconds. Metabolite T1 and T2 values were not measured on an individual basis, but were taken from our comparative analysis in healthy volunteers and patients with ALS published previously.¹⁵ Metabolite concentrations were expressed as millimoles per liter of brain tissue by a mixed gray-white matter water content of 72%, corresponding to 40 mol/L.

Metabolite concentration measurements were performed in both hemispheres of 30 patients with ALS and 5 healthy controls. Measurements in dominant and nondominant hemispheres yielded consistent results in these 5 controls. We therefore assumed that ¹H-MRS findings were independent of hemisphere lateralization of motor functions, and measurements in the remaining 10 controls were thus performed only in the dominant hemisphere. Consequently, data from 60 ALS and 20 control motor cortices were available.

Clinical characteristics and metabolite concentrations of patients and controls were compared by means of χ² tests or t test for unpaired samples. Clinical characteristics and metabolite ratios in different El Escorial subgroups were compared by 1-way analysis of variance (ANOVA) with subsequent post hoc Tukey tests including the covariates age, sex, and hemisphere dominance for motor functions. Sensitivity and specificity analyses were performed for a range of NAA/Cho and NAA/(P)Cr ratios and NAA concentrations to generate receiver operating characteristic curves.¹⁹ The relationship between clinical variables and metabolite ratios was tested by unpaired t test (lateralization of clinical symptoms, bulbar vs limb onset of ALS) and Pearson correlation (disease duration, severity). Changes of metabolite ratios between the first and the last ¹H-MRS investigation in patients undergoing follow-up measurements were tested for significant differences by paired t test. The relationship between metabolite ratio changes and progress of disease severity was evaluated by Pearson correlation for all follow-up investigations. P values less than .05 were considered statistically significant. Calculations were performed with the SPSS software package (SPSS Inc, Chicago, Ill). Data are presented as mean ± SD, unless otherwise indicated.

The ALS and control groups were well matched for age (P = .21) and sex (P = .65) (Table 1). Disease duration of patients with ALS at the initial ¹H-MRS investigation was 22.4 ± 20.3 months; disease severity was 10.9 ± 5.3 according to the Jablecki et al score.

When patients with ALS were analyzed according to El Escorial subgroups, there were no significant differences between patients and controls with respect to age (P = .35) and sex (P = .41) (Table 1). There were no significant differences of disease duration among different El Escorial subgroups. In contrast, disease severity was significantly different, with most severe clinical presentation in definite ALS and less severe presentation in possible ALS (ANOVA: disease duration, P = .12; disease severity, P<.001).

Group comparison of patients and control subjects showed a significant reduction in NAA concentrations (P<.001) and (P)Cr concentrations (P = .04), whereas Cho concentrations remained unchanged (P = .85) (Table 2).

When ALS subgroups according to El Escorial criteria were compared with healthy control subjects, NAA/Cho and NAA/(P)Cr ratios were significantly reduced in all El Escorial groups (ANOVA: P<.001), whereas the Cho/(P)Cr ratio was significantly elevated only in the subgroup with definite ALS (ANOVA: P = .03) (Table 3). Group differences were not influenced by the covariates age, sex, or hemisphere dominance of motor functions.

Receiver operating characteristic curves illustrated that the NAA/(P)Cho ratio and the NAA concentration had a greater accuracy in the detection of UMN abnormality than the NAA/(P)Cr ratio (Figure 2).

In patients with asymmetric clinical presentation (n = 48), NAA/Cho ratios were significantly lower in the motor cortex contralateral to the clinically predominantly affected side (P = .04). There were no such side differences for the NAA/(P)Cr ratio (P = .07) or the Cho/(P)Cr ratio (P = .29).

No significant differences were observed between patients with bulbar onset (n = 24) vs limb onset (NAA/Cho, P = .96; NAA/(P)Cr, P = .91; Cho/(P)Cr, P = .82).

There was a weak, but significant, correlation between disease duration and reduced NAA/Cho ratios (r = −0.16; P = .03; Figure 3) as well as elevated Cho/(P)Cr ratios (r = 0.18; P = .02). No significant correlation was detected between disease duration and NAA/(P)Cr ratios (P = .73).

There was a weak, but significant, correlation between disease severity and reduced NAA/Cho ratios (r = −0.24; P = .001; Figure 3) as well as elevated Cho/(P)Cr ratios (r = 0.20; P = .04-08). No significant correlation was detected between disease severity and NAA/(P)Cr ratios (P = .28).

In patients with ALS undergoing follow-up ¹H-MRS measurements (n = 16), NAA/Cho ratio decreased by 9% (P = .003) and NAA/(P)Cr ratio by 2% (P = .49), whereas Cho/(P)Cr ratio increased by 7% (P = .006) between the first and the most recent measurement (mean observation period, 12.1 ± 8.7 months)(Table 4). In individual patients, different patterns of NAA/Cho ratio changes emerged with respect to onset and slope of NAA/Cho deterioration (Figure 4). In late disease stages, no further decrease in NAA/Cho ratios was observed at values of about 1.7. Of 3 patients without clinical signs of UMN involvement at first ¹H-MRS measurement, 2 exhibited a marked decrease of NAA/Cho and presented clinical signs of UMN dysfunction at follow-up investigations, whereas 1 had stable ¹H-MRS measurements and no clinical UMN involvement during a period of 6 months.

There was a significant correlation between the decline of the NAA/Cho ratio and the progress of disease severity as indicated by an increase in the Jablecki et al score (r = 0.55; P<.001; Figure 5). Moreover, there was a significant correlation between the progress of disease severity and an increase of the Cho/(P)Cr ratio (r = 0.52; P<.001), whereas the NAA/(P)Cr change over time did not correlate with disease progress (P = .07).

This study is an expansion of our previous work¹⁵ on motor cortex ¹H-MRS in ALS. We confirmed our previous results of reduced NAA/Cho and NAA/(P)Cr ratios. In addition, we observed reduced absolute NAA and (P)Cr concentrations by quantitative measurements and found that NAA/(P)Cho ratios as well as NAA concentrations have a greater accuracy in the detection of UMN abnormality than NAA/(P)Cr ratios.

Most ¹H-MRS studies in ALS have demonstrated that either NAA concentrations^6,9 or NAA/(P)Cr,^7,11,12 NAA/Cho,^10,11 and NAA/([P]Cr + Cho)¹⁴ ratios are reduced in the motor cortex of patients with ALS. Since NAA is present only in neurons,²⁰ it has been concluded that these metabolite changes indicate loss or dysfunction of UMN. The present observation that metabolite changes are more severe on the hemisphere contralateral to the clinically predominantly affected extremities further supports that ¹H-MRS truly reflects neurodegeneration in ALS. There is a substantial debate on the interpretation of data obtained by different ¹H-MRS protocols. By theory, absolute quantification of NAA appears to be more attractive than relative quantification of metabolite ratios, since data are robust against changes of metabolites serving as denominators. However, because the clinical application of quantitative ¹H-MRS is time consuming and still suffers from technical limitations, measurements of relative metabolite concentrations are still in common use. In this context, the NAA/(P)Cr ratio was considered the most valuable surrogate marker, since the (P)Cr peak was thought to remain unaffected by the neurodegenerative disease process.^7,12 However, we and others observed a stronger decrease of the NAA/Cho than the NAA/(P)Cr ratio, suggesting that either Cho levels are increased or (P)Cr levels are reduced in the ALS motor cortex.^11,15 In the present study, we used ¹H-MRS with absolute quantification to interpret the changes in metabolite ratios. We found that (P)Cr concentrations were indeed reduced, whereas Cho concentrations were unchanged in the ALS motor cortex. We conclude that NAA/Cho appears to be more valuable than NAA/(P)Cr to characterize motor cortex degeneration in ALS, since the latter metabolite ratio is confounded by a parallel decrease of (P)Cr levels, leading to a decrease in sensitivity and specificity for identification of UMN degeneration.

The mechanisms leading to a decrease in (P)Cr concentrations in cortical areas of patients with ALS are unclear. Possibly as a result of the high variability of quantitative (P)Cr measurements, reduction of (P)Cr concentrations has not been observed in 2 smaller spectroscopic investigations of the ALS motor cortex.^6,9 Alternatively, reduction of (P)Cr might be due to the larger spectroscopic volume of interest used in the present study, including subcortical apart from cortical regions. However, a parallel decrease of NAA and (P)Cr concentrations without changes of Cho concentrations has also been observed in the cortex of patients with Alzheimer disease, suggesting that cortical (P)Cr reduction might indeed be found in various neurodegenerative disorders.²¹ The (P)Cr peak detected by ¹H-MRS consists of signals derived from creatine and phosphocreatine located in astrocytes, oligodendrocytes, and neurons.²² Therefore, changes in (P)Cr concentrations detected by ¹H-MRS are not specific for neuronal damage and do not show whether the equilibrium between creatine and phosphocreatine is altered or whether the concentrations of both compounds are decreased to the same degree. Therefore, decreased (P)Cr concentrations might be nonspecifically linked to neuronal death. Alternatively, it might be speculated that decreased (P)Cr indicates a disturbed cellular energy buffering and transport system in degenerating UMNs. This would be consistent with actual hypotheses that adenosine triphosphate depletion and oxidative damage might play a role in the pathogenesis of ALS.²³ In particular, a recent experimental study reported that orally administered creatine extended survival of transgenic ALS mice and had a neuroprotective effect because of a reduction of oxidative damage in spinal and cortical motoneurons.²⁴

To evaluate motor cortex ¹H-MRS for diagnostic purposes, we compared measurements in different ALS subgroups according to the El Escorial criteria. We found the most prominent reduction of NAA/Cho in patients with definite ALS, indicating that more widespread clinical involvement is associated with pronounced motor cortex degeneration. However, there was also a substantial decrease in NAA/Cho ratios in patients without an established diagnosis of probable or definite ALS. Although the diagnostic value of ¹H-MRS is limited by the overlap of metabolite ratios in the motor cortex of healthy control subjects and patients with ALS, these findings indicate that the method may increase diagnostic certainty in some patients with missing clinical UMN signs.^7,12 Because ¹H-MRS is a well-tolerated procedure that can be easily performed along with conventional magnetic resonance imaging, it might add to the spectrum of procedures used in the diagnosis of motoneuron disorders.

Unlike a previous study,²⁵ this study did not find an association of altered NAA ratios with bulbar vs limb onset of the disease. Therefore, our data do not point to a more severe UMN involvement or different pathologic process in the bulbar group. Because of considerable variation of the metabolite ratios in early disease stages, we found only weak correlations of ¹H-MRS findings with duration and severity of ALS symptoms. These findings reflect a marked interindividual variability of UMN involvement in the course of ALS, as known from pathological examinations.²⁶ Moreover, they suggest that the extent of motor cortex degeneration is not of critical importance for the overall clinical impairment in patients with ALS. Motor cortex ¹H-MRS findings are correlated with limb function, as previously shown by other investigators.^14,25

Apart from diagnostic issues, ¹H-MRS might be used for monitoring of UMN degeneration in patients with ALS.¹⁶ Herein we report ¹H-MRS findings in 16 patients with ALS followed up longitudinally for an average period of 1 year. We found a significant decline in NAA/Cho ratios along with a significant increase in Cho/(P)Cr ratios. By contrast, decrease in NAA/(P)Cr ratios failed to reach significance, indicating that NAA/Cho ratios might be superior not only for diagnostic purposes but for disease monitoring as well. Interestingly, the average 9.1% decline in NAA/Cho observed in this study resembles the annual 9.3% decrease in NAA concentrations found in cortical areas of patients with Alzheimer disease.²¹ However, as we did not perform serial ¹H-MRS measurements within predetermined observation intervals, our data on natural course of NAA deterioration in the ALS motor cortex clearly need extension and reproduction. Nevertheless, they might help to determine what observation periods are necessary to detect neuroprotective effects on ¹H-MRS findings in therapeutic trials.

As observed in our previous study, there was considerable interindividual variability for onset and slope of NAA/Cho deterioration, ranging from patients with marked early NAA/Cho reduction to patients without progressive changes of the NAA/Cho ratio up to 2 years after disease onset.¹⁵ Investigations by ¹H-MRS in patients with advanced cases of definite ALS and severe but stable disease demonstrated a deteriorated plateau of NAA/Cho ratios reaching values of about 1.7. In addition, we found a moderate correlation between decline of metabolite alterations with the clinical progression of the disease. These findings suggest that changes in NAA/Cho ratios might be taken as surrogate markers to estimate the activity of the neurodegenerative process in the motor cortex of individual patients with ALS. Therefore, ¹H-MRS appears to be a promising tool for monitoring the active phase of the ALS disease process.

Accepted for publication September 29, 2000.

This study was supported by a research grant from BONFOR, Bonn (Dr Pohl) and by research grant TR428/1-1 from the Deutsche Forschungsgemeinschaft, Bonn (Drs Träber and Schild).

Drs Pohl and Block contributed equally to this work.

We thank all of the patients, their relatives, and the volunteers for their cooperation in this study.

Corresponding author and reprints: Christoph Pohl, MD, Department of Neurology, University of Bonn, Sigmund-Freud-Str 25, D-53105 Bonn, Germany (e-mail: c.pohl@uni-bonn.de).