Abstract
Objective: To evaluate the efficacy of dichloroacetate (DCA) in the treatment of mitochondrial myopathy, encephalopathy, lactic acidosis and stroke-like episodes (MELAS).
Background: High levels of ventricular lactate, the brain spectroscopic signature of MELAS, correlate with more severe neurologic impairment. The authors hypothesized that chronic cerebral lactic acidosis exacerbates neuronal injury in MELAS and therefore, investigated DCA, a potent lactate-lowering agent, as potential treatment for MELAS.
Methods: The authors conducted a double-blind, placebo-controlled, randomized, 3-year cross-over trial of DCA (25 mg/kg/day) in 30 patients (aged 10 to 60 years) with MELAS and the A3243G mutation. Primary outcome measure was a Global Assessment of Treatment Efficacy (GATE) score based on a health-related event inventory, and on neurologic, neuropsychological, and daily living functioning. Biologic outcome measures included venous, CSF, and 1H MRSI-estimated brain lactate. Blood tests and nerve conduction studies were performed to monitor safety.
Results: During the initial 24-month treatment period, 15 of 15 patients randomized to DCA were taken off study medication, compared to 4 of 15 patients randomized to placebo. Study medication was discontinued in 17 of 19 patients because of onset or worsening of peripheral neuropathy. The clinical trial was terminated early because of peripheral nerve toxicity. The mean GATE score was not significantly different between treatment arms.
Conclusion: DCA at 25 mg/kg/day is associated with peripheral nerve toxicity resulting in a high rate of medication discontinuation and early study termination. Under these experimental conditions, the authors were unable to detect any beneficial effect. The findings show that DCA-associated neuropathy overshadows the assessment of any potential benefit in MELAS.
Mitochondrial myopathy, encephalopathy, lactic acidosis, and stroke-like episodes (MELAS) is a devastating multisystem syndrome characterized by a progressive encephalopathy and stroke-like episodes leading to disability and early death. MELAS is most commonly associated with a mitochondrial DNA A-to-G point mutation at nucleotide 3243.1–3 The mechanisms by which the resulting impairment in mitochondrial respiratory chain function causes the MELAS phenotype are incompletely understood. There is no effective treatment for this devastating condition. Based on our observation that the degree of cerebral lactic acidosis correlates with neurologic impairment, we hypothesized that lowering lactic acid may mitigate the clinical phenotype in MELAS.4–6 Dichloroacetate (DCA) is a potent lactate lowering agent that has been used to treat congenital and acquired conditions associated with lactic acidosis.7,8 The lactate-lowering effect is based on the interaction of DCA with the pyruvate dehydrogenase enzyme complex, located in the mitochondria (figure 1). The complex catalyzes the irreversible decarboxylation of pyruvate to acetyl CoA, which is the rate-limiting step in the aerobic oxidation of glucose, pyruvate, and lactate. The complex undergoes rapid, post-translational modulation in activity, due in part to reversible phosphorylation of the E1 component. DCA inhibits the kinase involved in this phosphorylation, thus locking the enzyme complex in its unphosphorylated, active form. Several reports on the open-label use of DCA have suggested that it may be beneficial in mitochondrial disease,9,10 and specifically in MELAS.11–16 However, DCA also has been associated with peripheral nerve toxicity.17,18 This finding is of particular concern in the MELAS patient population because peripheral neuropathy can occur as part of the natural history alone and in the presence of diabetes mellitus, which is also importantly associated with MELAS. In a series of 32 patients with MELAS with the A3243G mutation, 22% fulfilled the electrodiagnostic criteria for polyneuropathy.19 To evaluate the safety and benefit of DCA in MELAS 3243 patients, we conducted a randomized, double-blind, placebo-controlled clinical trial.
Figure 1. Interaction of dichloroacetate (DCA) with the pyruvate dehydrogenase enzyme complex. ATP = adenosine triphosphate; PDH = pyruvate dehydrogenase; Pi = inorganic phosphate.
Methods.
Study population.
To minimize variation between subjects, we limited inclusion to individuals harboring the A3243G mtDNA point mutation, who have the MELAS phenotype, i.e., who have a history of stroke-like episodes, focal seizures, or both. Additional inclusion criteria were 1) evidence of cerebral lactic acidosis (CSF lactate > 2.75 mM/L [normal 0.6–2.2 mM/L] and 1H MRSI estimated brain lactate > 5 IU) and 2) normal transaminases or less than fourfold elevated above the upper limit of normal.
Study design.
We conducted a double-blind, placebo-controlled, randomized, cross-over study of DCA in MELAS. Thirty-six patients were screened to randomize 30 patients to receive either oral DCA at 25 mg/kg/day by mouth divided in two daily doses or placebo for 2 years followed by crossover to the alternative treatment arm for a third year. Placebo was identically supplied and formulated except that it contained no DCA. Participants were assigned on an individual basis to a given treatment sequence beginning with DCA or placebo treatment. Based on a computer-generated randomization list the research pharmacy at the University of Florida mailed study medication for each participant to the clinical investigators at Columbia University Medical Center. Participants and clinical researchers remained blinded to treatment assignment throughout the trial. Lactate measures were anticipated to decrease as a result of DCA treatment, and the clinical investigators were therefore blinded to all lactate results throughout the study. Participants remained on the same allocation throughout the initial 24 months and were then crossed over to the alternative treatment if they continued in the study. The code was revealed to the clinical investigators once the study had been terminated. In addition to the study drug, subjects were given a combination of vitamins and nutrients providing thiamine at 10 mg/kg/day, CoQ10 at 5 mg/kg/day, l-carnitine at 50 mg/kg/day, and alpha-lipoic acid at 10 mg/kg/day to standardize their supplement intake.
We tested the hypothesis that oral DCA at 25 mg/kg/day would improve global clinical outcome in subjects with MELAS and the A3243G mutation. Subjects were evaluated every 3 months for a total study duration of up to 36 months (13 visits). Neuroimaging outcome measures were assessed semiannually at 0, 6, 12, 18, 24, 30, and 36 months only. CSF lactate was studied at 0, 12, 24, and 36 months only.
The primary outcome measure was the Global Assessment of Treatment Efficacy (GATE) score. The GATE score was determined by clinical investigator consensus at the end of each 6-month follow-up period, and was based on the following elements: 1) neurologic examination as semi-quantitatively rated with the Columbia Neurologic Score, 2) neuropsychological performance score, 3) health-related event inventory (HREI, frequency of seizures, strokes, migraine headaches, hospitalizations, health care encounters, and medication changes), and 4) Karnofsky score (assessing daily living functional abilities). The GATE score consisted of a 5-point scale with 0 representing no change, 1 = improved slightly, 2 = improved, –1 = worsened slightly, and –2 = worsened.
Secondary outcome measures included 1) 1H MR spectroscopic imaging (MRSI) to estimate degree of cerebral lactic acidosis, 2) venous lactate, 3) CSF lactate, and 4) MRI to evaluate focal lesions and global atrophy. At the quarterly visits (3, 9, 15, 21, 27, and 33 months), the investigators scored clinical global outcome based on Columbia Neurologic Score, Karnofsky, and HREI data.
Assuming a combined mortality and dropout rate of 30% for the duration of the trial, the anticipated study completion by 20 subjects was predicted to give us 80% power to detect a ½ unit difference in GATE score, the primary outcome measure between placebo and DCA treatment (see supplementary data for a detailed description of the outcome and safety measures [available on the Neurology Web site at www.neurology.org]).
Safety evaluations.
Adverse events were ascertained by questioning participants at each visit and during interval telephone contact. In addition, we studied motor and sensory nerve conductions and safety blood tests. All adverse events were documented and reported to a data safety monitoring board (DSMB) consisting of two physicians with expertise in mitochondrial disease, one lay member, and one biostatistician. The DSMB reviewed all adverse events and made recommendations as to dose reduction or discontinuation of study medication, as well as early study termination.
Setting.
All evaluations took place at a single site, the Columbia University Medical Center in New York City. Participants traveled to the study site from locations across the United States. The study team made travel arrangements for patients and caregivers and travel expenses were covered.
Data analysis.
Under the intent-to-treat principle we compared GATE score distributions over the initial 24-month treatment period among treatment groups (including all periods for which patients were scheduled to be on DCA at the beginning of the period, regardless of whether study medication was discontinued). For patients who did not return for study visits due to inability or unwillingness to participate, the worst possible GATE score was imputed. This conservative approach was chosen to prevent the error of falsely assuming a stable disease course in those who missed visits when, in fact, a worsening occurred. We had to guard against this error because missed visits occurred in the DCA group only and were associated with clinical worsening preventing participants from traveling. For patients who did not complete the 24-month follow-up period due to early study termination, no such differential missingness was observed and therefore, the last available score was carried forward when imputing data. In a post hoc analysis, we compared GATE score distributions at months 3 and 6 because many subjects were still taking study medication at these time points.
Given the small sample size and ordinal nature of the GATE score, we used Fisher exact test to assess differences between treatment groups.
We used type 3 analysis of variance (ANOVA) to test for treatment effects on secondary outcomes (Columbia Neurologic Score, Karnofsky, HREI, venous lactate, NP testing, 1H MRSI ventricular lactate, and MRI). Mixed linear models (with logarithmic transformation as needed to achieve normality) were generated to assess changes from baseline at 6, 12, 18, and 24 months. We report the least square means for quantitative assessments of outcome differences. To accommodate for the cross-over design and early study termination, we analyzed adverse events by treatment group over the initial 24 study months only using the significance test for proportions (Bonferroni correction was adopted to adjust for multiple comparisons). In a post hoc analysis, we compared the number of strokes and seizures (recorded as part of the HREI) between treatment groups.
Electrophysiologic properties of the peripheral nerves were compared to a reference population of normal control subjects established at the Columbia University EMG Laboratories. Values were considered abnormal when they were beyond two standard deviations from the mean. We defined worsening by either ≥30% deterioration in any single value or <30% deterioration in two or more values. Data were analyzed for treatment effect by ANOVA test over the initial 24 months of follow-up. Repeated measures of relative difference from baseline were modeled.
Results.
Thirty-six patients were screened to enroll 30 patients. Screen failures were due to 1) A3243G mutation not confirmed (n = 1), 2) full MELAS phenotype not present (n = 1), 3) absence of cerebral lactic acidosis (n = 1), and 4) withdrawal of consent (n = 3). The first patient was enrolled in August 2000, the last patient in November 2003. Randomization resulted in two groups with similar baseline characteristics (table 1).
Table 1 Baseline characteristics of randomized subjects
Among those randomized to DCA, 1 subject completed the entire 36-month study period, 2 subjects died, 2 withdrew consent, and 10 subjects completed less than the 36-month follow-up due to early termination of the study. In the placebo group, 4 completed the 36-month follow-up period, 1 died, 1 withdrew consent, and 9 subjects completed less than the 36-month follow-up due to early termination of the study (figure 2).
Figure 2. Flow of participants through each stage reporting the number of subjects screened for eligibility (I), the number of participants randomly assigned to dichloroacetate or placebo (II), the number of subjects completing study protocol (III), and the number of those analyzed for the primary outcome (IV).
The study was prematurely terminated at the recommendation of the safety monitoring board because of peripheral nervous system toxicity. All 15 patients randomized to DCA for the initial 24-month study period had been taken off study medication. Thirteen of the 15 patients were taken off medication because of peripheral neuropathy; two were taken off medication during an acute illness requiring hospitalization. In the placebo group, 4 of 15 patients were taken off medication because of peripheral neuropathy. However, it was later realized that one of these 4 patients had been exposed to DCA in error over a 3-month period. Four patients in the DCA arm missed visits due to inability to travel compared to none in the placebo arm (figure 3).
Figure 3. Diagram of follow-up for all participants indicating periods of exposure to allocated treatment, extent of study completion, and reasons for not receiving study medication or not completing the study. The top panel shows subjects assigned to dichloroacetate (DCA); the bottom panel shows subjects assigned to placebo. Each row represents one subject. The timeline is shown in columns. Dark shaded areas represent periods of DCA exposure, light shaded areas represent periods on placebo. Dotted areas represent periods of treatment interruption (participants were taken off assigned study medication). + = death; AE = study medication discontinued due to adverse event (mostly neuropathy related); T = study terminated; * = drug dispensing error; w/d = withdrew consent; M = missed visit due to inability to travel.
In primary analysis, there were no significant differences between treatment groups for overall GATE distribution and for GATE comparisons at 3, 6, and 24 months (table 2).
Table 2 Global Assessment of Therapeutic Efficacy (GATE) scores at 3, 6, and 24 months by treatment group
For secondary outcomes, there were no significant differences between treatment groups for Columbia Neurologic Score scores, Karnofsky scores, NP scores, and venous, CSF, and brain lactate values in the overall treatment effect over 24 months (see table 2). Given the frequent early discontinuation of study drug, we also compared secondary outcomes at 6, 12, 18, and 24 months and found no significant differences between treatment groups (table 3). Figure E-1 (available on the Neurology Web site at www.neurology.org) illustrates the mean Columbia Neurologic Score, NP, Karnofsky, 1H MRSI lactate, and MRI scores over the initial 24-month study period. The frequency of strokes and seizures that occurred during the study period was similar in the DCA and placebo groups with 11 strokes and 1,064 seizures in the DCA group compared to 7 strokes and 860 seizures in the placebo group.
Table 3 Score changes by treatment group for secondary outcome measures at 6, 12, 18, and 24 months, and overall changes from baseline of secondary outcome measures over the initial 24-month treatment period
Clinical or electrophysiologic changes indicating peripheral neuropathy were the most frequent adverse events, occurring in 17 of 30 patients during the initial 24-month treatment period (4 in the placebo and 13 in the DCA group).
Clinical symptoms suggestive of peripheral neuropathy occurred in 19 of 22 patients treated with DCA (this includes the 7 patients initially randomized to placebo who were crossed over to receive DCA). Seventy-nine percent had presented by 3 months, 84% by 6 months, 95% by 12 months, and 100% by 18 months. Presenting symptoms were distal limb paresthesias (n = 14), pain (n = 5), distal numbness (n = 8), falls (n = 7), or subacute gait disturbance (n = 16), alone or in combination. In three patients with symptoms suggestive of neuropathy we could not document changes in nerve conductions. Conversely, an additional 3 patients presented with asymptomatic deterioration of nerve conductions. Seven patients in whom DCA had been held for safety concerns were restarted on DCA after their peripheral nerve symptoms had resolved. Three were restarted on full and four on half dose DCA. Four patients showed signs of peripheral neuropathy and DCA was again discontinued within 3 months of restarting.
DCA-related neuropathic symptoms resolved partially in 4 of 19 and completely in 13 of 19 patients (two patients withdrew consent after stopping DCA so that their follow-up information could not be included). The time interval between stopping DCA and reporting symptom resolution varied: complete resolution by 3 months had occurred in 26%, by 6 months in 57%, and by 9 months in 68%. The remaining 4 patients with known follow-up status have experienced partial symptom resolution up to the time of this report (<8 months of follow-up for 3 participants, and 19 months for the fourth subject).
Nerve conduction studies showed significant differences between treatment groups in the relative change from baseline over the 24-month treatment period in both sural SNAP (−38% in DCA group) and peroneal CMAP amplitudes (−55% in DCA group, p < 0.05). In contrast, the sural SNAP and peroneal CMAP amplitudes in the placebo group increased by 1% and 9% from baseline (table E-1 and figure E-2). In the DCA group, the main decline in nerve potential amplitudes occurred during the initial 6 months of treatment, when most patients were taking the study medication at full dose. Sixty-eight percent of DCA treated patients had a deterioration of nerve conductions by 6 months. However, nerve conductions also worsened in 20% of placebo subjects during the first 6 study months. The nerve conduction changes affected predominantly the motor and sensory amplitudes and were more notable in the legs than in the arms. Distal latencies and conduction velocities were largely unchanged.
Recovery to baseline levels occurred in 2 of 11 patients with DCA-related nerve conduction changes. Of the remaining 9 patients, 8 had partial resolution of changes (80% by 6 months, 100% by 18 months). One patient had no improvement during the period of observation. We have mathematically modeled the predicted pattern of recovery for motor and sensory nerve amplitude changes in the 8 patients with partial improvement during the observation period (figure E-3). The median time for recovery to baseline levels for the sural SNAP amplitude is predicted to be 23 months, and for the peroneal CMAP amplitude, 18 months, with considerable variation between subjects.
There was a significant difference between the frequency of neuromuscular adverse events between treatment groups. Additional adverse events affecting other systems were not significantly different between groups (table E-2). The severity of adverse events was rated as mild for 124 adverse events, moderate for 45 adverse events, and severe for 4 adverse events. The occurrence of serious adverse events was similar between treatment groups. Nine of 15 (60%) subjects in the DCA group had at least one serious adverse event, compared to 7 of 15 (47%) in the placebo group (Fisher exact test, p = 0.7). The most common serious adverse events were hospitalizations for seizures or strokes. Death occurred once in the placebo group, and twice in the DCA group. Changes in liver function enzymes occurred in four patients and were all classified as mild.
Discussion.
Based on the results of this clinical trial in 30 patients with MELAS and the A3243G mutation, we conclude that DCA at 25 mg/kg/day is not beneficial in the treatment of MELAS. Our findings show that peripheral nerve toxicity overshadows any potential benefit from DCA.
DCA treatment had been suggested for chronic cerebral lactic acidosis over 25 years ago.20,21 Since that time, it has been used in the treatment of congenital lactic acidosis and acquired lactic acidosis associated with heart failure.7,8 Because mitochondrial diseases are frequently associated with lactic acidosis, many patients with these diseases have been treated on a compassionate basis with DCA over the past 20 years. Several reports of open-label DCA treatment have been published, including a report on 37 patients with mitochondrial disease treated with 12.5 to 25 mg/kg/twice daily for an average of 3.25 years15: symptoms of peripheral neuropathy were described in four patients leading to the discontinuation of DCA in one. Subjective improvement occurred in 49%, worsening in 21%. Six patients withdrew consent and eight died during the study. In two patients with Leigh syndrome and one with abnormal brain myelination, the serum, CSF, and 1H MRSI lactate were lower and MRI lesions were improved after DCA treatment.12 In two Japanese siblings with MELAS, myoclonic seizures, abdominal pain, and headaches resolved with combined DCA and thiamine treatment.22 Three children with MELAS had decreased lactate levels and clinical improvement while treated with DCA.11 In a patient with MELAS with visual and auditory hallucinations, DCA at 12.5 to 100 mg/kg/day normalized lactate levels and stopped the hallucinations.14 DCA treatment in a 16-year-old patient with MELAS with advanced neurologic impairment resulted in reduced levels of serum and CSF lactic acid, neurologic improvement, and increased blood flow to the left frontal lobe on SPECT.16 Chronic DCA treatment for 5.3 years on average in four MELAS 3243 patients reportedly had symptomatic benefit, but nerve conduction changes were seen in one of three patients.13
Despite this extensive open label experience, it remained uncertain whether DCA was effective and safe. Our study is the first double-blind, placebo-controlled study to assess DCA efficacy in this patient population. Our experience suggests that any therapeutic benefit implied in previous reports is overshadowed by the consequences of peripheral nerve toxicity. None of those randomized to DCA tolerated study medication for the entire 24-month period. Nearly all patients developed peripheral neuropathy, evidenced either by symptoms and signs of neuropathy, or by electrophysiologic evidence of a length-dependent, axonal, sensorimotor polyneuropathy without a significant demyelinating component. These same characteristics are common in toxic neuropathies.23 The neuropathy is at least partially reversible, based on our observations that clinical symptoms had largely subsided within 6 months following DCA exposure. The electrophysiologic changes showed gradual improvement without full recovery within this same period. Our data predict a median return of peripheral nerve function to baseline within 2 years. We continue to follow these patients to validate the pattern of recovery. Based on our observation of worsening nerve conductions over time in the placebo group, we anticipate incomplete recovery in some patients, likely due to the underlying mitochondrial impairment compounded in some cases by impaired glucose metabolism.
The rate and the severity of DCA-related peripheral neuropathy was unexpected, given that two previous studies in a total of 97 patients with congenital lactic acidosis (including 10 patients with MELAS) did not report any significant peripheral neuropathy in relation to DCA exposure.8,24 The frequency of electrophysiologic abnormalities of nerve conductions following DCA treatment in our study also exceeds previous reports in the literature.18 Older age may be a risk factor for neuropathy, because the previously reported DCA exposed patients were younger on average than subjects in our study.8,18 Differences in DCA dose between adult and pediatric patients do not account for this finding, because the dose was calculated per kilogram body weight. To address the possibility of variations in DCA metabolism between age groups, we will determine plasma levels and finalize pharmacokinetic studies on stored samples from our clinical trial. It may be that MELAS A3243G patients are more vulnerable to DCA toxicity than patients with other diseases causing lactic acidosis.18 Diabetes mellitus is likely an additional contributing factor given that it commonly causes symptomatic or subclinical neuropathy with both axonal and demyelinating features.25,26 Twelve of 30 patients in our study (six in the DCA and six in the placebo group) had evidence of impaired glucose homeostasis compared to 1 of 27 with reported diabetes in the previously reported study.18
Due to DCA discontinuation and early study termination, our data do not permit conclusions regarding the possible effect of chronic DCA treatment on brain function in MELAS 3243, and any such effect on the brain would be overshadowed by peripheral nerve toxicity. Even though the differences were not statistically significant, we noted decreasing mean 1H MRSI brain lactate levels in the DCA group for the first 18 months of the study, compared to gradually increasing levels in the placebo group. The lactate-lowering effect of DCA on brain, venous, and CSF lactate was less than anticipated7,8,21 probably due to early medication discontinuation in our study.
We conclude that DCA at 25 mg/kg/day is not efficacious in the treatment of patients with MELAS because of the documented peripheral nerve toxicity and the lack of proof that the drug is effective. Based on our results, all MELAS A3243G subjects are at risk for this toxic effect, and oral thiamine supplementation at 10 mg/kg/day does not prevent this complication. Our data show that the therapeutic benefit of DCA at 25 mg/kg/day, under the experimental conditions of our study design, is negligible, whereas the toxicity is significant, resulting in an unfavorable therapeutic index. Should a DCA-related compound with lactate lowering properties and reduced peripheral neurotoxicity become available in the future, further studies would be needed to establish its efficacy in MELAS. We have shown that a clinical trial for this rare, clinically heterogeneous disease is feasible and we have developed a battery of reliable outcome measures. We demonstrated that our primary outcome measure, the GATE score, had sufficient sensitivity to detect a difference between treatment groups, with worsening in the DCA group (p = 0.16 at 3 months). Our experience underscores the importance of randomized, controlled trials in evaluating the efficacy of new treatments for MELAS.
Acknowledgment
The authors thank Dr. Stephen Cederbaum for his diligent service as chair of the Data Safety Monitoring Board (DSMB), and Dr. Ingrid Tein, Dr. Libby Wright, and Joseph Valenzano for their service as members of the DSMB. They thank Dr. Alan Hutson for his contributions to the statistical design of the trial, and Janethe C. Regus for administrative support. They thank the GCRCs and their staff at Columbia University and the University of Florida for their support. They thank the Tishcon Corporation for supplying the vitamin and nutrient supplement used during the clinical trial. They thank all the patients and families who contributed their time to this research effort.
Footnotes
-
Additional material related to this article can be found on the Neurology Web site. Go to www.neurology.org and scroll down the Table of Contents for the February 14 issue to find the title link for this article.
Editorial, see page 302
Supported by NICHD grant PO1-HD32062 (D.C.D. and S.D.), RR-00645 and RR-00082, K12 RR017648 (P.K.), Irving Research Scholar Award (P.K.), and the Colleen Giblin Foundation (D.C.D.).
Disclosure: The authors report no conflicts of interest.
Received April 18, 2005. Accepted in final form October 19, 2005.
References
Disputes & Debates: Rapid online correspondence
- Dichloroacetate causes toxic neuropathy in MELAS: A randomized, controlled clinical trial
- Reply from the Authors
- Dichloroacetate causes toxic neuropathy in MELAS: A randomized, controlled clinical trial
- Reply from the authors
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
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.