Ascorbic acid in Charcot–Marie–Tooth disease type 1A (CMT-TRIAAL and CMT-TRAUK): a double-blind randomised trial

Summary Background Ascorbic acid reduced the severity of neuropathy in transgenic mice overexpressing peripheral myelin protein 22 (PMP22), a model of Charcot–Marie–Tooth disease type 1A (CMT1A) associated with the PMP22 duplication. However, in three 1-year trials, ascorbic acid had no benefit in human beings. We did a multicentre 2-year trial to test the efficacy and tolerability of ascorbic acid in patients with CMT1A. Methods Adult patients (aged 18–70 years) with symptomatic CMT1A were enrolled from nine centres in Italy and the UK, and were randomly assigned (1:1 ratio) to receive 1·5 g/day oral ascorbic acid or matching placebo for 24 months. The randomisation sequence was computer generated by block randomisation, stratified by centre and disease severity, and patients were allocated to treatment by telephone. The primary outcome was change in the CMT neuropathy score (CMTNS) at 24 months. Secondary outcomes were timed 10 m walk test, nine-hole peg test, overall neuropathy limitations scale, distal maximal voluntary isometric contraction, visual analogue scales for pain and fatigue, 36-item short-form questionnaire, and electrophysiological measurements. Patients, treating physicians, and physicians assessing outcome measures were masked to treatment allocation. Analysis of the primary outcome was done on all randomised patients who received at least one dose of study drug. This study is registered, numbers ISRCTN61074476 (CMT-TRAUK) and EudraCT 2006-000032-27 (CMT-TRIAAL). Findings We enrolled and randomly assigned 277 patients, of whom six (four assigned to receive ascorbic acid) withdrew consent before receiving treatment; 138 receiving ascorbic acid and 133 receiving placebo were eligible for analysis. Treatment was well tolerated: 241 of 271 patients (89% in each group) completed the study; 20 patients (nine receiving ascorbic acid) dropped out because of adverse events. Mean CMTNS at baseline with missing data imputed was 14·7 (SD 4·8) in the ascorbic acid group and 13·9 (4·2) in the placebo group. Mean worsening of CMTNS was 0·2 (SD 2·8, 95% CI −0·3 to 0·7) in the ascorbic acid group and 0·2 (2·7, −0·2 to 0·7) in the placebo group (mean difference 0·0, 95% CI −0·6 to 0·7; p=0·93). We recorded no differences between the groups for the secondary outcomes at 24 months. 21 serious adverse events occurred in 20 patients, eight in the ascorbic acid group and 13 in the placebo group. Interpretation Ascorbic acid supplementation had no significant effect on neuropathy compared with placebo after 2 years, suggesting that no evidence is available to support treatment with ascorbic acid in adults with CMT1A. Funding Telethon-UILDM and AIFA (Italian Medicines Agency) for CMT-TRIAAL, and Muscular Dystrophy Campaign for CMT-TRAUK.


Introduction
Charcot-Marie-Tooth disease (CMT), a disease for which no drug treatments are available, is the most common inherited neuromuscular disorder, and is characterised clinically by distal wasting, weakness, and sensory loss. The main subtype is type 1A (CMT1A), representing about half of all CMT cases. CMT1A is associated with an autosomal dominant duplication of the peripheral myelin protein 22 gene (PMP22), causing peripheral nerve demyelination and secondary axonal loss. 1 Ascorbic acid reduced the severity of neuropathy in transgenic mice overexpressing PMP22, an animal model of human CMT1A, compared with untreated mice. 2 Some clinical indicators improved during treatment, suggesting partial reversion of the phenotype. Ascorbic acid promotes myelination in vitro and possibly decreases PMP22 expression. 2,3 Evidence of effi cacy of ascorbic acid in the animal model prompted initiation of randomised controlled trials to test the effi cacy of ascorbic acid in patients with CMT1A.
Three trials assessing 1-year ascorbic acid treatment in patients with CMT1A have been completed. In a Dutch proof-of-concept trial, ascorbic acid had no eff ect in fi ve patients (aged 14-24 years) treated with 2 g/day compared with six patients on placebo (aged 13-24 years). 4 Findings of an Australian trial of 42 children treated with an equivalent dose (30 mg/kg per day) failed to show a signifi cant eff ect on disease course compared with 39 children on placebo. 5 Similarly, in a French trial of 179 adult patients treated with one of two doses of ascorbic acid (1 or 3 g/day) or placebo, ascorbic acid had no signifi cant benefi t. 6 In the fi rst two trials, motor conduction velocity of the median nerve was the primary outcome measure, whereas the CMT neuropathy score (CMTNS) was the primary endpoint in the French trial. Two trials reported positive results from post-hoc analyses: in the Australian trial, fi ve children treated with ascorbic acid had a large increase in motor conduction velocity in the median nerve; 5 and in the French trial, the CMT examination score (CMTES, the clinical component [symptoms, signs] of the CMTNS) was signifi cantly better in the 3 g/day group than in the other groups. 6 As acknowledged by the authors themselves, these trials were both underpowered and too short. 7 Moreover, fi ndings obtained from post-hoc analyses should be interpreted with caution. 8,9 Another crucial point is the choice of motor conduction velocity as the primary endpoint for the Dutch and Australian trials because, although axonal damage correlates with clinical progression, the correlation of motor conduction velocity with clinical progression is controversial. 1 The only validated CMT-specifi c outcome measure is the CMTNS, which is a composite of symptoms, signs, and motor and sensory action potential amplitudes of the ulnar or median nerve. 10 The scale ranges from 0 (normal) to 36 (worst). CMTNS progression rate was 0·68 points per year in a North American study of adults with untreated CMT1A 11 and 0·5 points per year in the placebo group of the French trial. 6 We did a phase 3, multicentre, placebo-controlled, double-blind randomised trial to assess the effi cacy and tolerability of chronic ascorbic acid treatment in patients with CMT1A in Italy (CMT-TRial Italian with Ascorbic Acid Long term [CMT-TRIAAL]) and the UK (CMT-TRial with Ascorbic acid United Kingdom [CMT-TRAUK]). In accordance with the consensus from the 136th European Neuromuscular Centre (ENMC) workshop that the CMTNS was the most appropriate primary endpoint for a trial of ascorbic acid in CMT1A, we chose the CMTNS as our primary outcome. 7 Our secondary outcomes were also those suggested at this workshop. 7,12

Patients
Detailed methods are reported in the published protocol. 12 Patients with clinical and genetic diagnosis of CMT1A were enrolled from eight Italian hospitals and one UK hospital. Eligible patients were aged 18-70 years and had a CMTNS between 1 (excluding the electrophysiological component) and 35 (including the electrophysiological component). Exclusion criteria were contraindications to ascorbic acid (nephrolithiasis, glucose-6-phosphate dehydrogenase defi ciency, or iron overload), extra-dietary ascorbic acid supplementation in the 3 months before screening, limb surgery in the 6 months before screening or planned for before fi nal assessment, other causes of neuropathy (appropriate blood tests to exclude other causes of neuropathy were done at screening), or other clinically signifi cant neurological or systemic diseases. At screening, women who were pregnant (patient's statement, no pregnancy test required), breastfeeding, or planning to become pregnant during the study were not included. Women of childbearing age were advised to adopt contraception for the entire duration of the study. Women who became pregnant during the study were advised to stop treatment and notify the treating neurologist immediately. They were followed up regularly for the duration of the study to comply with the intention-to-treat design. The protocol was approved by the institutional review board at every site, and patients gave written informed consent before any study-related procedures.

Randomisation and masking
Patients were randomly assigned (1:1 ratio) to receive either 1·5 g ascorbic acid per day as three 0·5 g tablets (one at breakfast and two at dinner) or matching placebo, for 24 months. Ascorbic acid and placebo tablets had identical appearance, taste, and smell, were centrally blistered, packed, and labelled before being sent to the pharmacy of each study centre. The randomisation sequence was computer generated (by a pseudo-random number generator) by the randomisation unit (Neuroepidemiology Unit, Carlo Besta Neurological Institute, Milan, Italy), which was independent of the study. The sequence was known only by staff at the randomisation unit and the drug dispenser. Treatment was allocated centrally by telephone, and was stratifi ed by centre and disease severity (CMTNS ≤10 vs >10, or CMTES ≤8 vs >8 if electrophysiological assessment was not done in the previous 3 months), with a block size of four (unknown to investigators) within each centre. The sequence was available in opaque sealed envelopes at every centre for emergency unmasking. Patients, treating physicians, and physicians assessing outcomes with clinical scales at baseline and during follow-up were masked to treatment allocation.

Procedures
We decided to give 1·5 g/day ascorbic acid, subdivided into two doses, on the basis of available pharmacokinetic data showing cellular saturation at doses of 100 mg/day, 13 the US Institute of Medicine recommendations of a tolerable upper intake level of 2 g/day, 14 and the dose given to mice. 2,12 After providing informed consent, patients underwent physical examination and laboratory testing at screening. Eligible patients returned at visit 1 (baseline) and were randomly assigned to treatment groups. Study visits were done at 6, 12, 18, and 24 months. Drug compliance, concomitant drug use, clinical scales, and electrophysiology were assessed at baseline and every 6 months. Clinical scales assessed were CMTNS, maximal voluntary isometric contraction with a handheld myometer (Cit Technics, Haren, Netherlands) for distal arm movements (hand-grip and large three-point pinch) and leg movements (foot dorsifl exion and plantar fl exion, by use of a stabilising device), 18 overall neuropathy limitations scale, 19 timed 10 m walk test, and nine-hole peg test. 12 The 36-item short-form questionnaire (SF-36) 15,16 and pain and fatigue visual analogue scales 12,17 were used at baseline and at 12 and 24 months. 12 Safety was assessed by physical examination, adverse events, and serious adverse events every 6 months, and by laboratory tests at baseline and at 6, 12, and 24 months.
At every visit, we took electrophysiological recordings from the motor ulnar, median, and peroneal nerves and the sensory ulnar nerve (non-dominant side, antidromic technique), and diff erences between baseline and negative peak amplitudes of compound motor action potentials and sensory action potentials were assessed. Foot and hand skin temperature was checked and kept at 32-34°C. Patients had laboratory tests of blood cell count, urine and its sediment, serum iron, total iron binding capacity, ferritin, and uric acid. Adherence was assessed by counting tablet returns every 3 months (including visits or contacts for drug delivery at 3, 9, 15, and 21 months), and measurement of serum ascorbic acid concentrations by spectrophotometry 20 every 6 months. Patients taking less than 80% of tablets were judged to be non-compliant.
The primary outcome was change in CMTNS at 2 years; for CMTNS, CMTES, and CMTNS subitems, data are presented at 12 and 24 months. Secondary outcomes were timed 10 m walk test, nine-hole peg test, overall neuropathy limitations scale, distal maximal voluntary isometric contraction, visual analogue scales for pain and fatigue, SF-36, and electrophysiological measurements; for these outcomes, data are presented at 24 months. An independent data and safety monitoring committee supervised the safety and data quality, and did a masked interim analysis of the primary outcome measure at 1 year, to confi rm or disconfi rm study continuation.

Statistical analysis
On the basis of longitudinal data, 11 we defi ned a signifi cant diff erence as an improvement in CMTNS by 0·5 points or more in the ascorbic acid group compared with a worsening by 1 point in the placebo group at 24 months. Therefore, we expected a CMTNS of 13·0 in the ascorbic acid group and 14·5 in the placebo group at 24 months, with a common SD of 6·5. 10 With these values, a twosided α of 0·05, and the assumption that 20% of patients would be lost to follow-up or discontinue treatment, we calculated that a sample of 272 patients (136 per group) would provide 90% statistical power to detect a signifi cant diff erence between the ascorbic acid and placebo groups.
The main analysis of the primary outcome included all randomised patients who took at least one dose of study treatment. We compared the study groups for the primary outcome by use of a linear mixed model for longitudinal data. To keep imputed values to a minimum, patients who stopped treatment were asked to attend scheduled followup visits. For patients who were unavailable, missing data were imputed according to Rubin's multiple imputation approach. A per-protocol analysis was done for the primary outcome and all secondary outcomes, and included patients who accomplished the outcome measures and were compliant with treatment (≥80% of tablets). For analysis of the primary outcome, we used the covariates baseline CMTNS, centre, age, and sex, and we also tested for fi rst-order interaction terms (centre per treatment and sex per treatment). Between-group comparisons of proportions were done with Pearson's χ² test or Fisher's exact test, and comparisons of continuous variables were done with the Wilcoxon rank-sum test. Two-sided p values of less than 0·05 were judged to be signifi cant; p values were not adjusted for multiple comparisons. Analyses were done with Stata (version 10.0) and SAS (version 9.2).
This study is registered, numbers ISRCTN61074476 (CMT-TRAUK) and EudraCT 2006-000032-27 (CMT-TRIAAL). CMT-TRIAAL and CMT-TRAUK shared a common protocol but have diff erent names and registration numbers because funding was separate in Italy and the UK.

Ascorbic acid Placebo
Patients Higher CMTNS, CMTES, or overall neuropathy limitations scale score indicates more severe disease; higher visual analogue scale score indicates more severe symptom (pain or fatigue); higher SF-36 score indicates better quality of life. CMTNS=Charcot-Marie-Tooth neuropathy score. CMTES=Charcot-Marie-Tooth examination score. N=Newtons. SF-36=36-item short-form questionnaire. CMAP summatory=sum of compound motor action potentials of the three motor nerves (ulnar, median, and peroneal nerves). MCV=motor conduction velocity. *Mean from median and ulnar nerves.

Role of the funding source
A representative of the main sponsor (Telethon Italy) attended meetings of the steering committee as nonvoting external participant. The study database is held by the Neuroepidemiology Unit, Carlo Besta Neurological Institute, Milan, Italy. The funding sources had no role in the study design, data collection, data analysis, data interpretation, preparation of the report, or the decision to submit for publication. The corresponding author and all coauthors had access to all data and had fi nal responsibility for the decision to submit for publication.

Results
Between March, 2006, and September, 2007, 277 of 354 patients (78%) were deemed eligible and were randomly assigned, of whom six withdrew consent before receiving treatment; 138 assigned to receive ascorbic acid and 133 assigned to receive placebo were eligible for the main analysis (fi gure 1). Demographic and clinical characteristics were balanced across treatment groups at baseline (table 1). Only one patient in the ascorbic acid group had a CMTNS of more than 30, indicating severe disease, fi ve patients (three in the ascorbic acid group) had a CMTNS of more than 25, and two patients (one in each group) had a CMTNS of less than 5.
In the prespecifi ed multivariate analysis, ascorbic acid treatment had no signifi cant eff ect on the primary outcome (p=0·491), and had no interaction with sex (p=0·891) or centre (p=0·529; webappendix p 1). Greater CMTNS at baseline (p<0·0001), older age (p=0·005), and longer time from start of study (p=0·004) were associated with worse prognosis. Centre had a signifi cant eff ect (p=0·042): in a subgroup analysis, patients treated at the centres in Milan and the UK had a slightly better outcome, irrespective of treatment group, than did those treated at the remaining seven centres, but this diff erence was signifi cant only at 12 months (p=0·04). This marginal diff erence, which was not signifi cant at 24 months, does not aff ect interpretation of treatment eff ect (webappendix p 1).
Findings of the per-protocol analysis of secondary outcomes showed no signifi cant diff erence between treatment groups for change in timed 10 m walk test, nine-hole peg test, overall neuropathy limitations scale, maximal voluntary isometric contraction, visual analogue scale scores for pain and fatigue, SF-36 scores, or electrophysiological measurements (table 3). Furthermore, changes from baseline to 24 months were negligible in both groups. Foot dorsifl exion seemed to be the most responsive measure with a mean worsening of more than 10% (tables 1 and 3).
Overall, ascorbic acid treatment was well tolerated. 15 patients (11%) in each treatment group discontinued treatment early, of whom six in the ascorbic acid group and fi ve in the placebo group discontinued because of oral or gastrointestinal symptoms (fi gure 1). 134 patients (97%) on ascorbic acid and 127 (95%) on placebo had drug adherence of more than 80%. 21 serious adverse events occurred in 20 patients, eight events in seven  Higher overall neuropathy limitations scale score indicates more severe disease; higher visual analogue scale score indicates more severe symptom (pain or fatigue); higher SF-36 score indicates better quality of life. N=Newtons. SF-36=36-item short-form questionnaire. CMAP summatory=sum of compound motor action potentials of the three motor nerves (ulnar, median, and peroneal nerves). MCV=motor conduction velocity. *Data are median change (95% CI). †Mean from median and ulnar nerves. Higher CMTNS or CMTES indicates more severe disease. CMTNS=Charcot-Marie-Tooth neuropathy score. CMTES=Charcot-Marie-Tooth examination score.

Discussion
In this 2-year randomised trial of high-dose ascorbic acid in the largest cohort of adults with CMT1A so far, we identifi ed no evidence that ascorbic acid is effi cacious for the primary outcome, CMTNS, or for several prespecifi ed secondary outcomes, which included impairment, activity limitation, and quality-of-life measures (panel).
Our fi ndings confi rm and expand the results of the Dutch, 4 Australian, 5 and French 6 1-year studies of ascorbic acid in CMT1A. These studies were limited by their statistical power and, in two studies, by the use of motor conduction velocity as the primary outcome. 4,5 We addressed these issues by choosing the CMTNS as our primary outcome, 7 and by powering our trial in accordance with the predicted change in CMTNS from Shy and colleagues' study 11 (1·4 points in 24 months). The most likely interpretation of our results is that ascorbic acid is not effi cacious in CMT1A. Three less likely interpretations are that the study was not long enough, the dose of ascorbic acid was insuffi cient, or the eff ect was too small to be detected by the outcome measures used.
With respect to study length, an unexpected fi nding was that disease progression was slower than was predicted from previous retrospective observational studies. 11,22 Mean CMTNS worsened by 0·46 points in 24 months (ie, 0·23 points per year) in the placebo group (per-protocol analysis), which is three times lower than that reported in the longitudinal study by Shy and colleagues (0·68 points per year), 11 lower than that in the placebo group of the French trial (0·5 points per year), 6 and slightly lower than that reported in a longitudinal study by Verhamme and colleagues (1·5 points worsening of an adapted CMTNS in 5 years). 22 However, the studies by Shy 11 and Verhamme 22 and their colleagues were partly retrospective, and therefore might not be directly comparable with our study. We feel that our large series is highly representative of the adult population with CMT1A treated at neurological centres.
The slow progression of CMTNS in adults with CMT1A raises the possibility that we might have missed a small eff ect of ascorbic acid in a 2-year study. Increasing the trial length in adults would be very diffi cult and perhaps impractical, but improved outcome measures could help. Participants of an ENMC workshop in 2009 agreed that some CMTNS components are not sensitive, and a modifi ed CMTNS is in development. 21 Furthermore, in our study, results indicated that development of myometry (especially foot dorsifl exion) as an outcome measure might be useful. The disease starts in childhood, so treatment eff ects might be more easily detected in children than in adults in future trials. However, existing evidence suggests that use of CMTNS is not appropriate in children younger than 10 years, and a paediatric CMTNS is in development. 21 With respect to the dose of ascorbic acid, we chose a dose that was about twice or thrice the equivalent dose given to mice (which synthesise ascorbic acid) 2 and was expected to achieve maximum and steady plasma concentrations with a low risk of side-eff ects. 12 This dose was well tolerated in our study. Although we feel that the dose we used should have been suffi cient to detect effi cacy, Micallef and colleagues 6 recorded weak evidence of an eff ect of 3 g/day ascorbic acid in the French trial, only because CMTES was signifi cantly better in the 3 g/day group than in the other group. This post-hoc fi nding should be interpreted with caution, and, notably, the most substantial contribution to the diff erence was from the CMTNS sensory component, which is the least reliable component. 6,10 Moreover, fi ndings of pharmacokinetic studies suggest that doses of higher than 500 mg/day do not substantially increase cellular concentrations because of saturation and tight metabolic control, 13 and, in the French trial, 6 dose increase from 1 to 3 g/day produced only a slight  , and the trial registries ClinicalTrials.gov and Current Controlled Trials with the search terms "peripheral neuropathy", "hereditary neuropathy", "Charcot-Marie-Tooth", "HMSN", "ascorbic acid", and "vitamin C". We also obtained information from handsearching of references in selected articles and conference proceedings, and from contacts with other investigators. We identifi ed seven studies assessing the effi cacy of ascorbic acid in patients with CMT1A. 21 Three are published randomised controlled trials assessing 1-year treatment with ascorbic acid treatment versus placebo, and ascorbic acid did not have a treatment eff ect in any of the trials. 4-6

Interpretation
We identifi ed no evidence of effi cacy of 1·5 g/day ascorbic acid in a large cohort of patients with CMT1A across a treatment period of 2 years. We feel that this study adds strong weight to existing evidence that ascorbic acid is not effi cacious in CMT1A.
increase in plasma concentrations. Finally, the US Institute of Medicine has set a tolerable upper intake level of 2 g/day, above which the risk of gastrointestinal and renal side-eff ects increases. We judged that our chosen dose was a good compromise by resulting in near saturation of plasma and cellular concentrations without risk of frequent side-eff ects, which could also unmask treatment. We await with interest the results of the continuing US study of 4 g/day ascorbic acid (NCT00484510) 21 and also fi ndings of tolerance to treatment because 5 g/day was poorly tolerated in a small study. 23 With respect to the size of the eff ect of ascorbic acid, if the eff ect is so small that only a biomarker can detect it, it would be unlikely to be clinically signifi cant in a shortterm study but might be important in the long term. To address this issue, we are studying PMP22 expression in skin biopsy samples 24 before and after treatment in a subgroup of patients from our study.
Findings of this study suggest that ascorbic acid is not effi cacious in adults with CMT1A. We have shown that a rigorous international randomised trial can be done in CMT. The use of the CMTNS should also provide the opportunity for meta-analysis of the effi cacy of ascorbic acid in CMT1A, when data from other trials that use this endpoint become available. Moreover, the data obtained from all these trials will provide unique information about the natural history of CMT1A.

Contributors
DP was the principal investigator. ASo was co-principal investigator, trial coordinator, and a member of the independent data and safety monitoring committee; to safeguard the committee's independence, ASo attended only open sessions of the committee's meetings. DP, ASc, LS, FV, RACH, and ASo contributed to the study design. DP, MMR, ASc, GMF, LS, GV, AQ, LP, and FG were local principal investigators at study centres. DP, MMR, ASc, GMF, TC, LS, GV, AQ, LP, FG, and ML participated in data collection and, with ASo, data assessment. DC was the clinical monitor of the study. DR did the statistical analysis. DP, MMR, LS, FV, DR, RACH, and ASo participated in writing and editing of the report. All authors interpreted the data and approved the fi nal report.