Abstract
Objective: To assess the efficacy of lamotrigine, a novel antiepileptic drug that inhibits glutamate release, to retard disease progression in Huntington disease (HD).
Background: Excitatory amino acids may cause selective neuronal death in HD, and lamotrigine may inhibit glutamate release in vivo.
Methods: A double-blinded, placebo-controlled study was conducted of 64 patients with motor signs of less than 5 years’ duration who were randomly assigned to either placebo or lamotrigine and assessed at 0 (baseline), 12, 24, and 30 months. The primary response variable was total functional capacity (TFC) score. Secondary response variables included the quantified neurological examination and a set of cognitive and motor tests. Repeated fluorodeoxyglucose measurements of regional cerebral metabolism using PET also were included.
Results: Fifty-five patients (28 on lamotrigine, 27 on placebo) completed the study. Neither the primary response variable nor any of the secondary response variables differed significantly between the treatment groups. Both the lamotrigine and the placebo group deteriorated significantly on the TFC, in the lamotrigine group by 1.89 and the placebo group by 2.11 points. No effect of CAG size on the rate of deterioration could be detected.
Conclusions: There was no clear evidence that lamotrigine retarded the progression of early Huntington disease over a period of 30 months. However, more patients on lamotrigine reported symptomatic improvement (53.6 versus 14.8%; p = 0.006), and a trend toward decreased chorea was evident in the treated group (p = 0.08). The study also identified various indices of disease progression, including motor tests and PET studies, that were sensitive to deterioration over time.
Huntington disease (HD) is a relentlessly progressive, incurable autosomal dominant disease caused by CAG expansion in a novel gene.1 The protein is widely expressed in neurons2 and outside the CNS, but the mutation ultimately leads to selective neuronal loss and gliosis in restricted brain regions.3
One hypothesis, postulated more than 20 years ago, implicates slow excitotoxic cell damage in the pathophysiology of the disease.4,5 This hypothesis predicts that exposure of susceptible striatal neurons to glutamate released from corticostriatal fibers leads to neuronal demise. Excitatory amino acids, such as glutamic acid (glutamate), and their interactions with their receptors, the kainate, N-methyl-d-aspartate (NMDA), and α-amino-3-hydroxy-5-methyl-4-isoxazole propionate (AMPA) receptors, have been proposed as candidates for neuronal demise. Injections of kainic acid, a glutamate analogue, in the rat striatum destroy almost all neurons while glial cells and afferents remain intact.4,5 The NMDA receptor agonists quinolinic acid and L-homocysteic acid, as well as NMDA itself, produce striatal damage that closely mimics the structural and neurochemical changes of HD.6,7 The administration of the mitochondrial complex II inhibitors 3-nitro-propionic acid (3-NP) or malonate to rodents or primates induces a subacute degeneration of the striatum, with neuropathologic features similar to those in human HD.8,9 Neurodegeneration can be prevented by administration of lamotrigine or an NMDA receptor antagonist such as MK-801.9-11 In patients with HD, striatal NMDA, AMPA, and kainate binding sites are significantly decreased,12 whereas NMDA receptor binding was decreased in a presymptomatic person carrying the HD mutation.13 These findings support the hypothesis of excitatory amino acid-mediated, particularly NMDA receptor-mediated cell death in HD.
Lamotrigine (3,5-diamino-6-[2,3 dichlorophenyl]-1,2,4-triazine) is an approved, well tolerated novel antiepileptic drug that blocks voltage-activated sodium channels and thus inhibits glutamate release in in vitro models.14 A similar in vivo effect in humans might offer protection to the putative NMDA- and AMPA-related neurodegeneration of HD. This hypothesis is supported by animal models, whereby pretreatment of rodents with lamotrigine significantly ameliorates the neurotoxic effects of 3-NP and kainic acid.10,11 We initiated a double-blinded, placebo-controlled, randomized trial of lamotrigine to determine whether lamotrigine could retard the clinical progression of HD, as measured by the total functional capacity (TFC) scale.15 In addition, we attempted to identify genetic determinants of disease progression that could be used as covariates in the outcome analysis, as well as identify potential subgroups for future therapeutic trials. Although we failed to detect a significant benefit of the drug in retarding disease progression, our study yielded information on progression of the disease and its assessment that will be useful for subsequent studies.
Patients and methods.
Patients were recruited for the trial and considered eligible if they fulfilled the following criteria: a diagnosis of HD no longer than 5 years before recruitment; and diagnostic neurologic abnormalities on initial examination, including involuntary movements (chorea), gait abnormalities, and dysarthria. Although a family history of symptomatic HD initially was required for enrollment, after the discovery of the HD gene in 1993,1 all patients were tested for expansion of the CAG repeat in the HD gene, and only those with an expanded CAG repeat were included.
Patients were excluded who 1) had heart, kidney, or liver disease, or diabetes; 2) were initially unable to cope cognitively with the requirements of the trial, including giving informed consent; 3) were younger than 18 years of age; or 4) were taking antidepressants or neuroleptics. An ongoing episode of depression or psychosis precluded enrollment, but the cessation of depression made patients eligible for subsequent enrollment. A new episode of depression during the trial that necessitated drug treatment did not result in termination of participation in the trial. If presumed severe side effects would occur, the medication code would be broken and the trial participation would be terminated; no further trial follow-up of that patient would be considered.
Participants were recruited from the HD clinic at the Department of Medical Genetics, University of British Columbia (UBC; Vancouver, British Columbia), the Neurodegenerative Disorder Center at UBC, and from various Canadian neurology and genetics centers (listed in the Acknowledgment). Two patients from the United States volunteered for the study. All participants gave their informed consent. The study protocol was approved by the institutional medical ethics committee.
For this study, duration of disease was defined as the time since the appearance of motor abnormalities (either gait disorder, involuntary movements, or speech problems). Information was obtained from the patient, the family, or the referring physician.
Patients were enrolled between August 1991 and August 1994. The final assessment took place in early 1997. Study assessments were performed at 0 (baseline), 12, 24, and 30 months. Directly after the enrollment and the baseline assessment, patients were randomly assigned to either the lamotrigine or the placebo arm. All baseline and all final (30-month) assessments were performed at the Department of Medical Genetics in Vancouver. Assessments at 12 and 24 months were done by the participant’s neurologist if it was impossible for that person to come to Vancouver.
After baseline assessment, medication was started at 25 mg of active medication and incrementally increased every 2 weeks, until after 12 weeks a full dose of 400 mg/day was achieved. If unacceptable side effects occurred, the dosage was reduced and then increased in approximately a week. People who could not tolerate the full dose were allowed to continue participation on their reduced, maximum tolerable dose. Medication compliance was monitored by self-report and pill count.
Patients were randomized after baseline assessment, taking into account age and sex of affected parent. Randomization was performed by one of us (P.G.) who did not participate in patient assessment or monitoring during the course of the trial and was unaware of any individual patient’s characteristics besides age and sex.
Identical pills that contained either active drug (lamotrigine) or placebo were provided, packed, and coded by the Glaxo Wellcome Company (Missisauga, ONT), the proprietary developer and manufacturer of lamotrigine, at the time of the trial.
Primary response variable.
The TFC scale15 was chosen as the primary response variable because the scale offers a semiquantitative assessment of daily life function, ranging from 0 (worst) to 13 (best). Moreover, the TFC has been shown to be sensitive to deterioration from early to mid-stage disease.16,17 The assessment of functional capacity was performed for all examinations by the nurse coordinator (M.A.H.). Information was obtained from the patient and the caregivers (if present).
Additional clinical and neuropsychological response variables.
The quantified neurological examination (QNE) was used18 to detect abnormalities on neurologic examination; a score of 0 means no abnormalities were detected. The examination was performed by one of the participating neurologists (B.K., M.M., L.R., B.S.).
Participants underwent a cognitive assessment that included the following tests: the Mini-Mental State Examination (MMSE); the Wechsler Adult Intelligence Scale–Revised (WAIS-R); Wechsler Memory Scale (WMS) Paired Associates from form 1 and 2 (total number correct), as well as the paragraph recall and the word fluency test. Using these tests, we were able to estimate general intelligence (WAIS-R), verbal memory functions (Paired Associates), and a component of frontal lobe function by spontaneous generation of words (word fluency). Two tests for motor function were taken during the cognitive evaluation. The Kløve Grooved Pegboard test was used as a measure of manual dexterity (the number of pegs placed in 1 minute for each hand). The Finger Tapping test was used as a measure of motor speed. In addition, the patients were asked to complete the Beck Depression Inventory to assess mood, but this test was added in the course of the first year of the study after the first patients were enrolled. Therefore, Beck’s questionnaires were not completed in every assessment.
CAG size determination.
Direct assessment of the CAG repeat, excluding the adjacent CCG repeat, was performed by PCR amplification, as described previously,19 and standardized against sequenced clones of known CAG size.20
PET.
PET measurements of regional metabolic rate were performed using 18F-deoxyglucose (FDG). The protocol has been described previously.21 To summarize, all emission scans were performed on the UBC/TRIUMF PETT VI system in high-resolution mode with an in-plane (transverse) resolution of approximately 9 mm (full-width half-maximum). Before the injection of 3 to 5 mCi FDG, two transmission scans were done, separated axially by 7 mm (one-half the distance between slices). Two emission scans of 15 minutes each were performed in positions corresponding to the transmission scans, starting 40 minutes after FDG administration. Head positioning was maintained by a molded plastic face mask. During the uptake phase, the patients’ eyes were covered and the room darkened. Room noise was kept at a minimum. Timed arterialized venous blood samples were drawn sequentially, starting at the time of injection and continuing until the end of the second emission scan. The plasma aliquots were counted in a well counter to determine 18F activity. Plasma glucose was measured before FDG administration and at intervals during the procedure. The tomograph was calibrated using a uniform phantom containing a known concentration of 18F. Regional metabolic rates were calculated using Brooks’ form of Huang’s four-constant adaptation of the original Sokoloff model.22-24 The value of 0.418 was used for the lumped constant, with literature values for the gray matter rate constants.23
Selection of structures and placement of regions of interest.
Four circular regions of interest (ROIs; sized 1.2 cm2) were positioned in two adjacent planes on the right and left caudate head. Sampling was performed in the lower, rostral parts of the caudate because this subdivision shows less marked atrophy in postmortem studies than more caudal parts.3 Two ROIs were placed in one plane on the right and left thalamus; 2 ROIs in one plane on the right and left cerebellar hemispheres; and 19 ROIs were positioned in various planes on separate cortical areas; all were in accordance with predefined templates. Of these cortical ROIs, 10 were placed on frontal lobe cortical areas in the right and left hemisphere, and in the mid-sagittal plane, 7 on parietal and occipital cortical areas, and 2 on temporal cortical areas. The regional cerebral metabolic rates of glucose were averaged to obtain a mean regional value (basal ganglia, thalamic, frontal, parietal, occipital, temporal, cerebellar). The cerebellum was not equally well represented in all scans. Because the data were analyzed using a repeated-measures analysis of variance (ANOVA), no standardization was done.
Sample size considerations and statistics.
Sample size considerations were based on the design and outcome of a previous drug study in patients with HD that used baclofen and the same TFC score that we selected as the primary response variable.25 In the design of this baclofen study, a yearly deterioration in TFC score of 1.25 points was assumed,16,25 with an SD not exceeding 2.0 points at 30 months. The actual yearly deterioration amounted to 0.68 ± 0.57 points with, at 30 months, an overall change from baseline of 1.70 ± 1.43 points. In the calculations for the lamotrigine trial, we assumed a final difference between those on placebo and those on medication of 0.9 points (approximately 50% less deterioration in the treated group), and an SD of 1.5 points. At a one-tailed α level of 0.05 and a power of 70% (a β of 0.30), a total of 26 patients were needed in each group to detect statistical differences. To increase the power to 80%, 35 patients per group were required. To detect differences of up to 0.45 points (70% power), 137 participants per group would be required. However, we expected to detect biologic baseline variables that would correlate with the rate of TFC decline and, therefore, could be used as covariates in the post-trial analysis.
Baseline comparisons of demographic variables was performed by simple factorial ANOVA or chi-square. Sequential measurements on individual patients were analyzed, using repeated-measures ANOVA with polynomial contrasts because this increases statistical power. This design, however, did not allow us to perform an intention-to-treat analysis. Correlations between various measures were calculated as Pearson (parametric) or Spearman (nonparametric—for non-normally distributed data) rank correlation coefficients. For simple group comparisons, t-tests, or in the case of skewed data, nonparametric tests, were used. In all instances, α was set at 0.05 (two-tailed); no adjustment was made to provide for multiple comparisons. Results of assessments are expressed as means (95% CI), unless specified otherwise.
Results.
Sixty-four patients were recruited for participation, of whom 33 were enrolled in the active drug arm and 31 in the placebo arm. Fifty-five completed the trial, 28 in the lamotrigine group and 27 in the placebo group, whereas 5 patients on lamotrigine and 4 on placebo dropped out (see attrition rate, later). Figure 1 details the enrollment, assessments, and attrition for the duration of the trial. The mean dose of those who completed the trial (defined as the mean of the doses at visits at 12 and 24 months) was 334 ± 110 mg (minimum dose, 50 mg; maximum dose, 400 mg). The analysis of the effects of lamotrigine was restricted to those who completed the study. At 24 months, two patients on active medication and two on placebo failed to appear for assessment. However, in one of the two patients on placebo, a TFC evaluation was obtained by telephone close to the regular 24-month time of scheduled assessment.
Figure 1. Flow diagram of randomization and follow-up of trial participants. TFC = total functional capacity; QNE = quantified neurological examination.
Baseline characteristics.
The two treatment groups were comparable on the initial demographic variables (table 1). A total of 23 men and 32 women completed the study. The proportions of men and women in the two treatment groups did not differ significantly. Age at entry, age at onset, duration of the disease, time since diagnosis, weight at entry, CAG length in the HD gene, and the sex distribution of the affected parent were also similar.
Baseline characteristics of the participants
A strong and highly significant inverse correlation between upper CAG size and age of onset of choreic movements (r = −0.76; p < 0.0001) or age of onset of clumsiness (r = −0.76; p < 0.0001) was found. Moreover, the age of the patients at entry showed a significant inverse correlation with CAG length (r = −0.77; p < 0.0001).
The clinical assessments at entry correlated well with each other. At baseline, the TFC score correlated with QNE (r = −0.51; p < 0.0001) and with the MMSE (r = 0.46; p < 0.0001), whereas QNE and the MMSE correlated as well (r = −0.47; p < 0.0001). No significant relation was found between CAG size and any of these three baseline variables. As for other biologic factors associated with these baseline variables, age of the participant correlated significantly with QNE score at baseline (r = 0.36; p = 0.006), whereas disease duration correlated with both TFC (r = −0.33; p = 0.014) and QNE (r = 0.58; p < 0.0001), but not with MMSE (r = −0.23; p = 0.09).
Deterioration in total functional capacity score.
The means and SDs for the TFC scores are summarized in table 2, and the deterioration over time is depicted in figure 2. A significant decrease in TFC was found. However, no differential effect for lamotrigine and placebo could be detected. The within-group comparison over the total 30-month period revealed a decrease in functional capacity (F = 20.053; p < 0.001). In contrast, the between-group comparison between those taking lamotrigine and those on placebo failed to show statistical significance (F = 0.232; p = 0.53), and the interaction between time and treatment also failed to show statistical significance (F = 0.108; p = 0.96).
Clinical assessment of disease progression
Figure 2. Change over time of total functional capacity (TFC) in both groups. The 95% CIs for the mean TFC in the placebo group are indicated (gray lines). —○— = lamotrigine; – –□– – = placebo.
A tendency toward slower deterioration was shown in patients on lamotrigine compared with those on placebo, although it did not reach statistical significance. Specifically, the treated group deteriorated on average 1.89 ± 2.46 points (95% CI: 0.94 to 2.85) over the study period, at an average rate of 0.72 points per year, whereas the placebo group decreased by 2.11 ± 1.99 points (95% CI: 1.33 to 2.90) or 0.84 points per year over the same span of time. Thus, those on medication had deteriorated 0.22 points less than those on placebo, or approximately 10% less. For the whole group (n = 55), the deterioration in TFC amounted to 2.0 ± 2.2 points per 30 months (95% CI: 1.41 to 2.59), or 0.8 points per year.
Thus, although no clear benefit of the medication could be demonstrated, the TFC score appeared to be a highly sensitive instrument to detect functional deterioration in both groups. Deterioration over time, however, was markedly different for different individuals. Comparison of the TFC score at the end of the study with the score at the start of the study revealed that 13 subjects (23.6%) appeared not to deteriorate or even improved one or two points in their score. In the group on lamotrigine, four remained unchanged and four gained one point. Similarly, in the group on placebo, two remained unchanged, two gained one point, and one person gained two points.
Baseline variables associated with rate of total functional capacity deterioration.
We were unable to identify baseline factors that clearly influenced the magnitude of functional decline. None of the following variables correlated with the deterioration in TFC score: age of the participant, age at onset of disease, duration of disease, or TFC score at baseline (results not shown). Weak correlations were found for QNE score at baseline (r = −0.25; p = 0.07) and MMSE at baseline (r = 0.29; p = 0.03).
Recalculating the repeated-measures ANOVA with baseline QNE score or MMSE as a covariate did not yield a significant effect of treatment, whereas the time effect remained strong.
The magnitude of deterioration in TFC score (i.e., the difference between 0 and 30 months) did not correlate with the length of the CAG trinucleotide repeat (r = −0.09; p = 0.52). In multiple linear regression, the addition of variables such as age, age at onset of chorea, and duration of disease did not yield a significant contribution of upper CAG length to the magnitude of TFC deterioration.
We examined whether, based on CAG size, a subgroup of favorable responders could be identified. Those treated with lamotrigine and those on placebo did not differ in CAG length (table 1). The median CAG size for the total group was 44 (range, 41 to 57). Subdividing the patients into groups with CAG length of ≥45 (n = 27) and CAG length of ≤44 (n = 28) revealed no difference between those on active medication and those on placebo. The TFC score at baseline, at 12 months, at 24 months, and at 30 months showed slightly less progression in the treated versus untreated group (F = 2.99; p = 0.09; figure 2). However, deteriorations in TFC score over that period were similar in the two groups (F = 0.12; p = 0.73; figure 2). FIGURE
Figure 3. Relationships between CAG size and age at onset of chorea (A), and CAG size and change in total functional capacity (TFC) over 30 months (B). The former correlation is significant (r = −0.76; p < 0.0001), the latter obviously is not (r = −0.09; p = 0.52). Linear regression lines are depicted. ○ = placebo; ▪ = lamotrigine.
Effects on quantified neurologic examination.
The quantified measurement of abnormal neurologic signs showed a significant deterioration over time (F = 80.969; p < 0.001), but no differences between the lamotrigine and the placebo group were detected (F = 0.835; p = 0.36; table 2). Selected items from the QNE were grouped into three separate components, the eye movement score, the motor score, and the chorea score. These subscales represent three independent factors within the QNE.18
A tendency toward a significant drug effect was noted in relation to the chorea score. In the treated group, although they displayed more choreic movements at baseline than the placebo group, the chorea score did not increase as rapidly as in the placebo group. Although the treated group deteriorated from a median of 7.5 points to a median of 10 points over the 30-month period, the placebo group deteriorated from a median of 5 to a median of 9 points (F = 3.17; p = 0.08; table 2). This finding may suggest an antichoreic effect of lamotrigine.
The other subscales did not reveal any benefit of lamotrigine for the treated group (table 2). CAG size did not correlate with the magnitude of the deterioration in QNE score over the 30-month observation period (r = 0.20; p = 0.135). Deterioration of the TFC score did correlate with deterioration in QNE score (r = −0.32; p = 0.02) and with deterioration in MMSE (r = 0.26; p = 0.05).
Symptomatic assessment.
All patients were also asked about their perception of the effects of the medication on their symptoms. Four of 27 (14.8%) patients on placebo reported symptomatic improvement on placebo. By contrast, 15 of 28 (53.6%) patients on lamotrigine reported feeling better (p = 0.006). Reports of improvement were given by patients on at least two different occasions. More precise descriptions of mode of improvement included more energy, less clumsiness, and improvement in speech, coordination (specifically balance) and mood. For those on lamotrigine, eight reported symptomatic relief at 1 month that was sustained throughout the study. Six patients reported improvement at 3 months and throughout the study, and one reported improvement at 6 months that was sustained only until 18 months into the study. By contrast, improvements in the placebo group (four people) started at different times (one at 1 month, two at 3 months, and one at 12 months) and were not sustained throughout the duration of the study.
Effects on neuropsychological tests.
In the cognitive and motor assessment, the motor tests were highly sensitive to deterioration over time, whereas the global cognitive assessments were evidently insensitive to deterioration. Not all patients completed all scheduled neuropsychological examinations: 46 participants, 24 in the lamotrigine group and 22 in the placebo group, completed four subsequent evaluations (table 3). The MMSE was completed by all on every occasion. The subsequent analyses report the statistics from the repeated-measures ANOVA including those 46 participants. Additional analyses using simple factorial ANOVA offered similar results.
Results of neuropsychological assessments
Tests of motor function (i.e., the Pegboard test and the Finger Tapping test) showed a significant loss of dexterity and slowing over time (F = 23.30, p < 0.001; and F = 10.90, p < 0.01, respectively), consistent with a deterioration in motor performance (table 3). The between-groups and group-by-time interactions were not significant.
Of the cognitive tests, the full scale IQ (WAIS-R), the MMSE, and the Paired Associates test did not show a time-dependent change (table 3), whereas the word fluency test results improved over time, probably representing learning. No medication effects were found. For the immediate Paragraph Recall of the WMS, there was a significant interaction between time and treatment (F = 8.96; p < 0.01), reflecting better learning for those on placebo.
For the tested subset of study participants, no differences between those on lamotrigine and those on placebo were seen on the Beck Depression Inventory at 12, 24, and 30 months.
Effects on cerebral metabolic rate.
A total of 26 participants were able to complete all four scheduled PET scans. In those 26, of whom 14 were on lamotrigine and 12 on placebo, a significant decrease in metabolic rate over time was found in the basal ganglia (F = 25.06, p < 0.001), the frontal cortex (F = 6.52, p < 0.02), the temporal cortex (F = 5.50; p < 0.05), and the thalamus (F = 6.13, p < 0.05), but no treatment effect was detected (table 4). In contrast, neither the parietal metabolic rate nor the cerebellar rate decreased over time (table 4). The most profound decrease in metabolic rates occurred in the basal ganglia, with a decrease of approximately 7% per year (figure 4). The most profound decrease occurred in the last 6 months (between 24 and 30 months), suggesting that the changes in metabolic rates of glucose in the basal ganglia may not be linear over the progression of the illness.
Results of positron emission tomography 18F-deoxyglucose measurements
Figure 4. Percentage changes in metabolic rates of glucose in different brain regions over time as assessed in 26 patients who had four different PET scan assessments. —▪— = basal ganglia; —□— = frontal lobe; - - ▵ - - = parietal lobe; - - ⋄ - - = occipital lobe; - - ▴ - - = temporal lobe; —*— = thalamus; —♦— = cerebellum.
Post hoc power calculations.
For post hoc power calculations, the crucial finding of the study was the large SD (2.20 units) of deterioration relative to the overall deterioration (i.e., 2.0 units) in TFC for the aggregate group of 55 participants over 30 months. Given an actual difference of 0.22 units (or 10% less deterioration) in those on medication versus those on placebo, in retrospect our trial had a power of 10% to detect a significant difference at a one-sided α of 0.05. Our initial assumption of 50% less deterioration over 30 months (or 1.0 point) would have required 46 patients per treatment arm for a power of 70%, or 60 patients per arm for a power of 80%. Instead, it turned out that we had achieved a power of just 51% to detect a 50% difference.
We then recalculated the number of patients required in a trial aimed at detecting 0.22 points with a 80% power. For such a trial, 1,230 patients per arm would be needed.
With the final cohort size of this trial (approximately 27 patients per arm), we would have had a power of 80% to detect a difference of 1.50 TFC units after 30 months, or 70% power to detect a difference of 1.30 points. The actual difference in TFC units in this study after 30 months was 0.22 units.
Attrition rate.
Nine of the 64 patients (14.1%), including 5 on treatment and 4 on placebo, did not complete the study (figure 1). This equals an attrition rate of 3.6 per 64 patients per year, or 5.6% per year. Those patients who withdrew from the treatment arm included two who had persistent nausea, even after a decrease in dose, and three who lost interest in the study. Of those on placebo, one man had a myocardial infarction, whereas three participants lost interest and terminated contact with the nurse coordinator.
A total of 16 of 28 (57.1%) patients on active drug had documented side effects. These included nausea, dizziness, headache, skin rash, depression, insomnia, and night sweats. Three patients had mild increases in liver enzymes that resolved without any change in dosage. One patient had seizures at 8 months. This, together with depression requiring hospitalization, represented the only severe side effects. By contrast, 8 of 27 (29.6%) patients on placebo had minor side effects, including nausea, dizziness, insomnia, confusion, and mild increase in liver enzymes (in 2 patients).
Discussion.
We found no clear evidence that lamotrigine has a significant effect on disease progression in HD, in terms of the primary response or the secondary variables. However, an important finding of this study was the determination of which measures are sensitive to disease-related deterioration over a period of 30 months. These measures will be useful in future intervention studies.
An important issue in this study is the power to detect clinically relevant effects, given the sample size and the SD of the primary response variable, the TFC scale. The trial design was based on prior estimates of TFC deterioration in the baclofen trial,25 assuming 50% less decline in the treated group. The final SD of TFC decline turned out to be higher than anticipated, thus diminishing the statistical power of the study. Our post hoc power analysis showed that we had achieved a power of only 51% to detect a 50% change, whereas the actual difference in disease progression (10%) would have provided only a 10% chance of obtaining statistically significant results. Therefore, the results of the trial do not exclude the possibility that lamotrigine does indeed have an effect on disease progression in early HD.
Of patients on lamotrigine, 53.6% (n = 15/28) reported symptomatic improvement, compared with only 14.8% of patients (n = 4/17) on placebo (p = 0.006). The chorea subscale of the QNE showed less deterioration of choreic movements in those on lamotrigine than in those on placebo, suggesting an antichoreic effect of the drug. Riluzole, a drug that also inhibits glutamate release, has been shown to reduce dyskinesias in HD animal models, including primates.26 The potential antichoreic effect of lamotrigine warrants further assessment. The symptomatic relief in patients on lamotrigine is unexplained but may be due to the antichoreic effect seen in this study, or may reflect mild mood-elevating properties of the drug.
The differences in the rate of TFC decline between the participants in the baclofen study (overall, 1.70 ± 1.43 units) compared with the current lamotrigine study (overall, 2.0 ± 2.2 units) may be related to the different study populations.25 The ages at entry in the baclofen study (treated group, 38.0 years; placebo group, 40.8 years) were clearly lower than in this study (46.1 and 43.2 years, respectively). The mean TFC scores at enrollment in the baclofen study (10.75 units for the treatment group and 10.00 units for the placebo group) were also lower than in the lamotrigine study (11.32 and 11.33 units, respectively).25 This shows that the group of patients enrolled in the baclofen study was younger and more severely affected than those in the lamotrigine study. It is conceivable that younger patients with more progressive disease deteriorate at a more predictable rate over the mid-course of their disease than older and less severely affected patients. This would have implications for the design of trials aimed at neuroprotection in asymptomatic patients: in these patients, SDs of decline may be even larger, and consequently larger study populations would be required.
One important methodologic issue is whether the large SD in the TFC score represents true biologic variation in disease progression, an instrument-related assessment artifact, or an observer-related artifact. In the first case, attempts to increase the statistical power of future intervention trials in HD should be directed toward increasing the number of patients enrolled. In addition, better assessment methods would allow relatively smaller trials with an increased power. A key issue in using instruments is the coefficient of variation (the SD divided by the mean values). This study suggests that the coefficient of variation was smaller for the PET measurements than for the clinical and cognitive variables. Because the PET measurements of caudate glucose metabolism were clearly sensitive to decline, it could be argued on methodologic and statistical grounds that an effort should be made to include PET measurements in future intervention studies in HD. Ideally, new instruments could be developed and assessed that display a smaller coefficient of variation over time than does the TFC.
A stimulus to initiate this study, and most relevant to HD, were the findings of McGeer and Zhu,10 who demonstrated that lamotrigine blocks kainic acid–induced striatal degeneration in rats in a dose-dependent way, with doses of 16 mg/kg yielding most protection. Lamotrigine was also able to attenuate striatal lesions in rats induced by a variety of neurotoxins, such as malonate, 3-acetylpyridine, and 3-nitropropionic acid, always at doses of 12 to 16 mg/kg.11
The presumed neuroprotective effects of lamotrigine are based on its ability to inhibit corticostriatal glutamate release by blocking presynaptic voltage-dependent sodium channels,14,27 and perhaps both presynaptic and postsynaptic calcium channels.28,29 In rat brain slices, lamotrigine inhibits glutamate release induced by the sodium channel opener veratridine, with IC50 values between 23 and 150 μM, which matches therapeutic plasma concentrations. However, the electrically stimulated glutamate release (more similar to the physiologic situation) is affected to a much lesser degree.27
Neuroprotection by lamotrigine has been demonstrated in various models of acute neuronal degeneration, such as the global ischemia model in gerbils,30 motor neuron cell death in the rat facial motor nucleus after axotomy,31 and malonate-induced degeneration of basal forebrain cholinergic neurons.32 Not only is the time course of acute neuronal damage in these animal models different from the slow chronic degeneration in human HD, but the protective dosages and concentrations used in these acute models substantially exceed those tolerated by humans. For example, in the malonate-induced basal forebrain cholinergic neuron degeneration, rats received 16 mg/kg intraperitoneally.32 Because we used significantly lower doses of lamotrigine in this clinical trial than those that had protected from excitotoxicity in animal models, our results do not exclude a role for excitotoxic neuronal death in HD.
We did not measure serum or CSF concentrations of lamotrigine. Therefore, it could be argued that the possibility of insufficient bioavailability of the drug has not been excluded, either because of noncompliance or insufficient perfusion of the blood–brain barrier. However, this explanation is very unlikely. The nurse coordinator was able to make an assessment of compliance, and the doses used were those that have demonstrated efficacy in epilepsy. In the various animal models demonstrating effects of lamotrigine, the doses on a per-kilogram basis were higher than those used in our study. Assuming a mean weight of 70 kg in the participants of our study, their maximum dose was 5 to 6 mg/kg/day. Higher doses would probably be accompanied by more side effects and thus decrease the tolerability of the drug. In many participants (n = 15), 400 mg/day had to be reduced because of side effects. Thus, an increased dosage of lamotrigine does not seem a feasible option for further intervention studies, and other candidate drugs need to be assessed. One such candidate might be riluzole, for which a modest benefit in ALS has been demonstrated in large trials.33,34 Riluzole is also a more potent blocker of sodium and calcium currents than lamotrigine.35
Despite the absence of a clear drug effect, the study yielded valuable information on measurements that are sensitive to clinical decline in patients with HD. In general, tests of motor function and TFC were more sensitive than instruments assessing cognitive measures. The TFC is easy to administer, it has a good inter-rater reliability,25 and it reflects the effects of the disease on a patient’s daily life. However, the TFC, in spite of these features, may not be sensitive to more subtle factors influencing progression of illness. For example, patients with strong family support may remain at home longer than others without such support. Furthermore, people who are self-employed may be less sensitive to mild deterioration in functional capacity than those working for others. It has been found that cognitive and mood changes figure prominently as determinants of changes in functional capacity.36 Mood plays a role in the performance on some of the items of the TFC, as well as in most of the other assessment scales. The positive effects of lamotrigine on mood may in fact have masked some of the underlying deterioration in functional capacity. Notwithstanding these caveats and recognizing its limitations, the TFC still represents a reliable but insensitive measure of progression of illness.
Measures of global cognitive performance in this type of intervention such as the MMSE, or the much more complex full scale IQ on the WAIS-R, were unable adequately to reflect disease progression. Tests of memory like the WMS are equally insensitive to deterioration because they are subject to learning improvement over repeated testing. Current methods for assessing cognitive deterioration that focus on frontal lobe dysfunction17 may be more appropriate for longitudinal assessment.
The yearly deterioration of total functional capacity (i.e., 0.72 points/year in the treated group and 0.84 in the placebo group) is remarkably similar in magnitude to that described in other studies.15,16,25 In a previous intervention study in HD that compared baclofen with placebo, the mean yearly deterioration in TFC was 0.68 ± 0.57 points per year, with 0.53 ± 0.46 for the placebo group and 0.85 ± 0.64 for the baclofen group.25 In an initial analysis of data from the large prospective database on the natural course of the disease that was set up by the multicenter Huntington Study Group, a yearly deterioration of 1.0 points was found.17 Therefore, averaged yearly rates of deterioration between 0.5 and 1.0 points must be considered characteristic of disease progression in early symptomatic HD. However, individual variations in functional decline may be large. We identified 13 people, or 23.6% of the total sample, who did not deteriorate on this measure over the 30-month period using this scale. On the other hand, a dramatic decline in functional capacity of seven or eight points in two patients was also noted. Clearly, in a small proportion of people, a less predictable rate of decline may be evident. Because all assessments of TFC score in this study were made by one person, the nurse coordinator (MAH), interobserver variance cannot explain these results.
A relatively low attrition rate of 5.6% per year reflects the importance of available support for patients in such trials. This number was kept low despite repeated PET measurements in large part owing to a person working solely with this trial.
In terms of biologic determinants, we were unable to relate CAG size to the rate of functional decline as measured by the TFC score over the 30-month follow-up period (figure 3). Others, using the same instrument, have been equally unsuccessful in relating TFC change to CAG size or onset age (a correlate of CAG size).37,38 In contrast, other measures have suggested a relationship between the rate of clinical deterioration and CAG size. In one study, deterioration in neurologic status as measured in the QNE and mental status as seen in the MMSE were related to CAG size.39 However, in this study, assessments were made at a single point in time and the duration of disease was derived from an estimated age at disease onset. Thus, this did not constitute a prospective longitudinal study of disease progression. A prospective volumetric MRI study with interimage intervals of at least 10 months did show a significant relationship between CAG size and volume loss.40 One explanation for these differences may be the greater time of patient follow-up in the latter study.
Deterioration in function was well reflected in the deterioration of cerebral glucose metabolism as measured by FDG PET, particularly in the basal ganglia. Decreased metabolic rates in HD brains has been documented in previous studies,41-43 but this study constitutes the largest follow-up study of FDG PET measurements in patients with HD. Another strength of the study was that all measurements were made with a single scanner, thus offering more robust data than in studies where the investigators switched to another scanner during the study.44 Consistent with what is known from neuropathology, the cerebral metabolic rate of glucose declined most in the caudate nucleus, and to a lesser extent in the frontal and temporal cortex and the thalamus, whereas a decline in the parietal and occipital regions and the cerebellum failed to reach statistical significance. Whether the metabolic decline, particularly in the basal ganglia, may reflect neuronal loss, loss of metabolic activity in remaining neurons, or simply atrophy with concomitant loss in recovery coefficients45 cannot be resolved from our data. In this study, the average yearly decline in cerebral glucose metabolism in the basal ganglia was 7%. Clearly, any medication that slows down this change in the metabolic rate of glucose or restores it close to baseline will provide in vivo biologic validation of efficacy. From a pragmatic point of view, this study clearly demonstrates that PET metabolic measurements can be used to assess disease progression to monitor effects of drug therapy.
Acknowledgments
Supported in part by a grant from the Huntington Society of Canada, Glaxo-Wellcome Canada, MRC Canada, and MRC Sweden (E.W.A.). M.R.H. is supported by the Canadian Networks of Centres of Excellence (NCE-Genetics), and is an established investigator of the BC Children’s Hospital.
Acknowledgment
The authors thank all physicians of the patients for their help and support of this study—in particular, Dr. M.H.K. Shokeir and Ms. Sharon Cardwell at Royal University Hospital, Saskatoon, Saskatchewan, Ms. Carol Demong at University of Calgary Medical Clinic, Calgary, Alberta, Dr. V. Ilivitsky at Royal Ottawa Hospital, Ottawa, Ontario, Dr. A.J. Stoessl at St. Joseph’s Health Centre, London, Ontario, and Dr. C. Greenberg at Children’s Hospital, Winnipeg, Manitoba.
This manuscript is dedicated to Dr. Jack B. Penney, Jr., who was committed to the development of effective therapeutics for HD.
- Received January 30, 1999.
- Accepted April 10, 1999.
References
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Hastening the Diagnosis of Amyotrophic Lateral Sclerosis
Dr. Brian Callaghan and Dr. Kellen Quigg
► Watch
Alert Me
Recommended articles
The CREST-E study of creatine for Huntington disease
A randomized, placebo-controlled trial of coenzyme Q10 and remacemide in Huntington’s disease
A randomized, double-blind, placebo-controlled trial of coenzyme Q10 in Huntington disease
A prospective evaluation of cognitive decline in early Huntington's disease