• ICN

Effect of mexiletine on muscle stiffness in nondystrophic myotonia: aggregated N-of-1 trials


Full title: Effect of mexiletine on muscle stiffness in patients with nondystrophic myotonia evaluated using aggregated N-of-1 trials: a clinical and validation study


Authors: Dr Bas Stunnenberg


On the 11th of December 2018, our research group published the results from an aggregated N-of-1 trial series in patients with inherited muscle stiffness in the prestigious high-impact Journal of the American Medical Association (JAMA).


Background

Building a reliable evidence-base for the treatment of rare diseases with large group randomized controlled trials (RCTs) is commonly limited by their small and often heterogeneous patient populations.Considering the large number of rare diseases with a lack of treatment evidence, the validation of innovative trial designs is urgently needed. However, it remains unclear what concessions should be accepted towards the level 1 evidence that is needed for drug registration and coverage decisions.


The case of mexiletine in nondystrophic myotonia

Nondystrophic myotonia (NDM) is a rare skeletal muscle ion channel disorder (‘channelopathy’). It is caused by inherited mutations in the skeletal muscle sodium or chloride channel genes. Patients with NDM experience a delayed skeletal muscle relaxation after voluntary contraction, resulting in functional limitation due to muscle stiffness, pain, weakness and fatigue.


Originally marketed as an anti-arrhythmic drug, the sodium channel blocker mexiletine used to be prescribed off-label to patients based on extensive clinical experience and the publication of several case reports, but coverage was discontinued in 2007 by the Dutch healthcare organization. This decision was based on the Cochrane review on drug treatment inmyotonia that concluded that there was no level 1 evidence available for the effectiveness of the drug (note that there was also no evidence for the ineffectiveness of the drug!).


Unique opportunity: direct comparison between the RCT and aggregated N-of-1 trials

During the design phase of our aggregated N-of-1 trial, an international cross-over RCT was published in the JAMA (Statland et al. 2011) that showed the effectiveness of mexiletine in NDM on a group level. Through collaboration with this international research group (The Clinical Investigation of Neurologic Channelopathies, or CINCH) we were able to match important study protocol components (such as outcome measures, mexiletine drug dosages and patient selection criteria) for direct comparison between the two trial designs.


Aggregated N-of-1 trials methods

Each of the 27 individual N-of-1 trials exposed a patient to between 1 and 4 treatment pairs or sets, with each set randomizing the order of mexiletine and placebo. A 1-week washout period was implemented between therapies. After each treatment set, pre-specified criteria (relating to the posterior probability of reaching a minimal clinically important difference) were used to determine whether the patient should continue to the next treatment set or discontinue, either for evidence of benefit of mexiletine, evidence of no benefit, or for reaching the maximum allowed number of treatment sets. Subsequently, a Bayesian hierarchical model (BHM) was used to integrate data from all available N-of-1 trials performed in all the patients. This produced estimates of treatment effects for each patient individually, as well as across the 2 genetic subtypes of the disease.


Main study findings

In this study, we show that aggregating individual N-of-1 trials through a Bayesian hierarchical model produces equally valid treatment efficacy results on the group level to a larger, previously conducted cross-over RCT (that served as the current gold standard design). (Figure 1)



Figure 1. Method validation (unpublished) Meta-analysis of results from the present aggregated N-of-1 trials and previous cross-over RCT on NDM group level. Mean treatment effects with 95% confidence intervals (95%-CI) of both trials on the primary outcome (IVR-stiffness) are represented.

In favor of the aggregated N-of-1 trial design; (1) less patients were needed (11 vs. 57 patients, see Figure 2 ), (2) Bayesian outcomes were better for interpretation (instead of the usual frequentistic p-value expressing the level of significance, posterior probabilities of reaching a minimally clinical important difference were calculated) and (3) the innovative design simultaneously allowed for estimation of individual treatment effects for the trial subjects, thus enabling personalized healthcare. These findings support the value of aggregated N-of-1 trials for assessing interventions in chronic rare diseases.



About the author

Bas Stunnenberg is a medical neurobiologist and medical doctor currently working at the Departments of Neurology and Health Evidence of the Radboud University Medical Center, Nijmegen, the Netherlands. In 2019, he will finish his neurology residency and PhD-thesis entitled “Personalized treatment in skeletal muscle channelopathies using N-of-1 trials”.

Afterwards, he will be appointed as a fellow in neuromuscular medicine and will continue his work on N-of-1 trials, Bayesian statistics and clinical reasoning in his field of clinical interest, i.e. neurological channelopathies (such as epilepsy, migraine, myotonia and periodic paralysis).


Contact details: bas.stunnenberg@radboudumc.nl