Clinical Outcomes in Duchenne Muscular Dystrophy: A Study of 5345 Patients from the TREAT-NMD DMD Global Database

LIST OF ABBREVIATIONS

INTRODUCTION

TREAT-NMD, an “EU funded network of excellence” founded in 2007, has pioneered the formation of national and global patient registries in the field of rare neuromuscular diseases. In Duchenne muscular dystrophy (DMD), TREAT-NMD has facilitated the establishment of standardised national patient registries in multiple countries worldwide via the use of a standardised mandatory data set collected by each registry [1]. These national registries have combined data in an effort to create the TREAT-NMD global database for DMD [2], a unique global resource containing mutational information for over 7000 DMD patients and clinical data from over 5000 DMD patients from 31 countries. Data derived from registries and natural history studies are complementary. Registries involve much wider population cohorts and may demonstrate country-specific differences usually not captured by natural history (observational) clinical studies. At the same time, natural history studies provide high quality longitudinal data, including clinical and multi-disciplinary assessment not captured by registries. Collectively, they lead to an increased understanding of the complexity, variability and progression of DMD [3, 4].

DMD is a progressive, muscle-wasting disease with an X-linked mode of inheritance, affecting between 1 in 3500 to 1 in 5000 live male births globally [5]. Affected boys become symptomatic at 3–5 years of age due to proximal muscle weakness. Untreated DMD boys lose independent ambulation at an average age of 9.5 years of age [6]. The use of corticosteroids in controlled clinical trials has been shown to delay this loss of ambulation by 1-2 years [7]. Life-expectancy is typically limited to the second or third decade due to respiratory failure or cardiomyopathy. However, the application of ventilatory support in addition to a multidisciplinary care approach has lead to a growing adult DMD population, with several patients surviving as far as the fourth or fifth decades [8, 9]. Currently there is no cure for DMD. The implementation of care recommendations (including corticosteroids, cardiac medications and assisted ventilation) improve outcomes and quality of life, but their effects on the underlying disease mechanisms are as yet unknown [10–12]. Many experimental therapies designed to repair the primary genetic defect are being pursued, including the mutation specific approach of ‘exon-skipping’ and stop codon read through therapies (e.g. Altaluren, Translarna^TM). However, a lack of comprehensive information on the natural history of DMD poses significant barriers to applying meaningful outcome measures to clinical trials, and subsequently to gaining regulatory approval. In order to gain a comprehensive understanding of clinical outcomes and the evolving heterogeneity of the disease, data from large numbers of patients from a variety of countries (where healthcare often differs) is required. In this study we determined clinical outcomes in relation to genotype and corticosteroid treatment in a large, multinational registry cohort of DMD patients.

MATERIALS AND METHODS

RESULTS

Corticosteroid use

Corticosteroid use within the registry was recorded in four main categories: “corticosteroids yes” (reflecting current corticosteroid use, dose unspecified), “corticosteroids no” (reflecting never used corticosteroids) and “past corticosteroid use”(dose, duration of past use and stopping reasons not specified). Almost half of the study population (49.7%: n = 2658) were currently using steroids, 37.7% (n = 2015) had never used corticosteroids and 9.8% (n = 522) reported past steroids used. For the remaining 2.8% (n = 150) patients the use of corticosteroids was unknown. Corticosteroid use varied with age; younger patients reported more corticosteroid use than older patients. Current care standards recommend commencing corticosteroid use at around 4–6 years of age [15]; and this was reflected within the database. Until the age of 14 years corticosteroid use within the database increased in an age dependent manner, with patients aged 6–8, 9–11 and 12–14 reporting the greatest corticosteroid use (57%, 74% and 64% respectively) (Fig. 1). Corticosteroid use notably decreased after the age of 14 as ambulation decreased. In the database, population 20 years and older, only 15.2% (n = 127) of patients reported current corticosteroid use. This was further reduced in the population aged 30 years and older where only 4.4% (n = 6 of 135) reported current corticosteroid use. The early use of corticosteroids has only been common practice since 2004, and DMD patients born before 1998 are less likely to have ever used corticosteroids. (Table S2. in the supplementary index provides an overview of corticosteroid use per age group within the database).

Fig.1

Overview of corticosteroid use within the global TREAT-NMD DMD database. Corticosteroid was reported as “corticosteroids yes” (current corticosteroid users/red bars), “corticosteroids never used ”(corticosteroids never used/green bars), “past corticosteroid use” (used corticosteroids in the past/blue bars) or “corticosteroid use unknown” (corticosteroid use unknown/orange bars”).

Ambulation

The proportion of ambulant and non-ambulant patients correlated with age and also independently with the use of corticosteroids. In the non-steroid treated population, 71% (n = 633) of patients under the age of 10 years remained ambulant (Fig. 2 and the corresponding Table S3. in the supplementary index). Beyond this age range, ambulation decreased significantly, with 52% (n = 464) of the non-steroid treated patients remaining ambulant at the age of 10 years. By comparison 79% (n = 1375) of patients currently taking steroids, remained ambulant at the age of 10 years. Median age at loss of ambulation in non-steroid treated patients was 10 years old, compared with 13 years in the steroid treated patients (p < 0.001).

Fig.2

Turnbull analysis for loss of ambulation. The red line indicates the “corticosteroid yes” group, the green line the “corticosteroid no (never)” group and the blue line the “past corticosteroid use” group. P-value indicates the difference between the steroid: yes and steroid: no groups.

Ventilation

Overall 10.3% (n = 549) of registry patients used assisted ventilation. Assisted ventilation fell into two categories, non-invasive (nasal/mouth mask) or invasive (tracheostomy). Non-invasive ventilation accounted for 74.0% (n = 406) of the ventilated population (average age 25.5 years) and invasive ventilation for 25.0% (n = 137) (average age 28.1 years). For 1.1% (n = 6) of the patients type of ventilation was unknown. The use of assisted ventilation increased with age (Fig. 3) with 2.9% (n = 129) of the patients under the age of 20 years reporting the use of assisted ventilation, compared with 50.4% (n = 420) of patients aged 20 years and older. As shown by Turnbull analysis corticosteroids had a positive effect on ventilation (p < 0.001) (Figure S1B. in the Supplementary Appendix).

Fig.3

Assisted ventilation within the global TREAT-NMD DMD database.

Corticosteroids and clinical milestones in patients aged 20 years and older

Whilst the majority of patients in the DMD global database were less than 20 years old, there was a sizable cohort of patients aged 20 and over 15.6% (n = 836).

33.9% (n = 283) of patients aged 20 years and older underwent scoliosis surgery. As observed in the younger patients group, the frequency of scoliosis surgery was lower in the corticosteroid treated population (χ² 4.4 p < 0.0001) (Fig. 4A). Assisted ventilation was reported in 50.4% (n = 420) of patients over the age of 20 years. 4.8% (n = 20) of these patients reported using corticosteroids, 16.9% (n = 70) reported past use, 75.5% (n = 318) reported never having used corticosteroids, and for 2.6% (n = 11) corticosteroid use was unknown. Chi-square testing of this group showed that corticosteroid use is associated with significantly less use of assisted ventilation (χ² 7.2 p < 0.0001) (Fig. 4B). Generally, non-invasive ventilation was utilised more than invasive ventilation except for patients over the age of 35. Patients aged 30 or older (n = 152) showed 75.5% (n = 115) dependence on assisted ventilation (36.5%: n = 42 invasive, 50%: n = 76 non-invasive).

Fig.4

Chi-square testing for scoliosis surgery (A), ventilation (B) and cardiomyopathy (C) in the patient group aged 20 years and older.

In contrast to the younger age group, further analysis of patients aged 20 years and older showed a cardioprotective effect of corticosteroids. Chi-square testing showed that in this age group cardiomyopathy was less frequent in the cohort using steroids (42%: n = 31) than in the past treated (62%: n = 71) or never steroid treated (60%:n = 229) cohorts (χ² 2.9, p = 0.0035) (Fig. 4C).

Genotype analysis in older age groups

We examined different age ranges of the older cohort of patients for most frequent mutations to assess whether any mutations in particular predispose to more severe or less severe course of disease. Whilst distribution of mutation type within the population aged 20 years and older was similar to the genotype distribution in the patient group under the age of 20 years (with deletions being most frequent followed by small rearrangements then duplications), detailed analysis showed that deletion of exon 45 was the most frequent mutation in the cohort aged 20 years and older accounting for 6.7% (n = 56) of the population. In contrast, this mutation was present in only 4.3% (n = 196) of patients under the age of 20 years. (Fig. S2A. and B. in the Supplementary Appendix). In addition, patients under the age of 20 years, never treated with corticosteroids and with a single exon deletion of exon 45 lost ambulation significantly later (mean age 15 years old) than patients in this age group with other most common single exon deletions (deletion exon 51, 44, 52 and 50) (mean age 11 years) (p = 0.027) (Fig. 5A). When treated with corticosteroids no significant difference in loss of ambulation was observed (p = 0.52) (Fig. 5B). The data was then further stratified to compare the abovementioned group of patients with single exon deletion 45 with mutations that would be rescued by current exon skipping strategies, specifically “skip exon 44” (excluding single exon deletion 45) and “skip exon 51” (excluding single exon deletion 50). An advantage was observed in terms of loss of ambulation for both single exon deletion 45 and the skip exon 44 group (Fig. 6A), but not for deletions including exon 45 that cannot be brought back into frame by exon 44 skipping. In the corticosteroid treated population median age of loss of ambulation for single exon deletion 45, skip exon 44, as well as the single exon deletion 50 and the skip exon 51, were in the same median age range (14 to15 years), (Fig. 6B).

Fig.5

Turnbull analysis of loss of ambulation in patients under the age of 20 years, never (A) and ever (B) treated with corticosteroids and with a single exon deletion of exon 45 (red line), exon 51 (green line), exon 44 (purple line), exon 52 (black line) and exon 50 (blue line).

Fig.6

Turnbull analysis of loss of ambulation in patients under the age of 20 years in mutations rescued by exon skip 44 and 51. (A) represents patients never treated with corticosteroids and (B) represents patients ever treated with corticosteroids. In both figures the red line indicates patients with deletion of exon 45 mutation, the green line indicates patients with mutations amenable to exon skipping with exon 44 (but not deletion of exon 45 mutations), the blue line indicates patients with deletion of exon 50 mutation and the black line indicates patients with mutations amenable to exon skipping with exon 51 (but not deletion of exon 50 mutations).

DISCUSSION

DMD is a severe, progressive condition affecting 1 in 3600–6000 live male births [15–16]. A range of therapeutic approaches targeted at causative mutations of the dystrophin gene have been proposed including exon-skipping, which aims to restore the correct reading frame and result in expression of at least partly functional dystrophin protein [17–21], and techniques to enable readthrough of premature stop codons. Despite promising results in early stage studies, failure of recent phase 3 trials to meet primary endpoints [22] suggests the clinical variability of DMD is greater than anticipated and that this may mask treatment benefit. Variability may partly be due to the disease-modifying effect of long-term corticosteroid use, increasingly common amongst younger patients, as recommended in international consensus guidelines [15, 16, 23]. Data from smaller natural history studies and national registries has also suggested that changes in care and particular mutations in the dystrophin gene may be correlated with clinical outcomes [24, 25].

The current evidence for corticosteroid treatment in DMD is largely from short-term data from randomised control trials; long-term data, particularly on prolongation of ambulation, are limited [26]. Our study shows that corticosteroids are advantageous for all major clinical outcomes. Patients taking corticosteroids retain ambulation status for longer, have scoliosis surgery less frequently, require ventilatory support later in life, and have a lower incidence of cardiomyopathy (in patients ≥20 years). While the mechanism underlying this advantage is still not completely understood, it is thought that corticosteroids may act to reduce the rate of muscle breakdown [27], and may reduce muscle necrosis and inflammation [28]. However, not all immunosuppressive or anti-inflammatory drugs have an effect on DMD, suggesting immunosuppression is not the only function of corticosteroids [29]. Alternative possible mechanisms of action are through increasing the number of muscle precursor cells [30, 31].

While advantages of corticosteroids are clear in terms of ambulation, their impact on skeletal defects, ventilation or cardiomyopathy has been controversial so far. Retrospective studies in Japanese DMD patients using the Remudy patient registry have shown patients using prednisolone lose ambulation on average 11 months later than their non-steroid treated counterparts [32]. Our results confirm the profound effect of corticosteroid treatment on ambulation, scoliosis surgery, ventilatory support and, to a lesser extent, cardiomyopathy [33]. Patients not treated with corticosteroids have a 90% probability of developing a clinically significant progressive scoliosis [34]. Daily corticosteroid use has been shown to reduce the incidence of scoliosis [35, 36]. Our data show that patients currently taking corticosteroids are less likely to have had scoliosis surgery. Furthermore, our data show that patients currently taking corticosteroids are less likely to require ventilatory support.

There have been previous concerns that, due to preserved skeletal muscle function and thus increased stress on a vulnerable myocardium, corticosteroids may worsen cardiomyopathy in DMD [37]. In our cohort, no negative effects of the use of corticosteroids on the heart were observed. Moreover, in the cohort of patients aged ≥20 years a cardioprotective effect was observed.

Corticosteroids tend to be initially prescribed in patients around age 4–6 years and discontinued at or shortly after permanent wheelchair dependence. We acknowledge this as a limitation of our study analysis since some patients will have stopped taking corticosteroids once they have permanently lost ambulation and were therefore considered “past steroid users” in our study. In this group, patients with unfavourable genotypes, lack of response to corticosteroids or intolerable side effects may be overrepresented. In general, our analysis in the older DMD population may be biased towards favourable outcomes, as –based on voluntary registration of living individuals – deceased patients are not included. This may be reflected in a lower than expected percentage of scoliosis surgery, cardiomyopathy and ventilation in patients aged ≥20 years.

Historically the clinical progression of DMD was considered relatively homogeneous with predictable outcomes for muscle function and survival. We show that there is greater variability in clinical progression and outcomes than previously reported, with increasing evidence for the role of genetic modifiers [24, 38, 39]. Our analysis revealed that specific exons were associated with either positive or negative effects on disease severity. Such mutation specific responses have been suggested in case reports and smaller clinical studies [25, 40]. Our data highlights that single deletions of exon 45 may infer and advantage to survival when compared to other common single exon deletions. This could be due to exon 45 deletions having a higher frequency of spontaneous exon 44 skipping, giving rise to small amounts of functional dystrophin protein and resulting in long-term clinical benefit, as suggested by data from a smaller cohort [41]. This supports the use of exon-skipping antisense oligonucleotides to restore the reading frame of dystrophin mRNA and the further exploration of its potential therapeutic applications in DMD [17–21].

The global DMD platform continues its development by recruiting new members and national registries under the TREAT-NMD framework. Active research in the treatment of DMD introduces the need for post-marketing surveillance in the long-term assessment of safety and efficacy of therapeutic approaches. The global DMD registry is a good starting point in disease specific long-term data collection, which will allow the collection of specific post-marketing data and comparison of different therapeutic approaches. This work requires multicentre, multinational collaboration, an agreed additional standardized data set (e.g. adverse events forms) and agreed user roles (e.g. patient reported/clinician reported data) inbuilt in national registries to allow comprehensive, Good Clinical Practice (GCP) compliant data collection. Data collected in this way can contribute to better understanding of the course of the disease, disease management, and safety and efficacy profile of new therapies, and ultimately may lead to new policies and disease management protocols.

The infrastructure provided by TREAT-NMD and the standardized, common data set collected by the TREAT-NMD national registries has ultimately facilitated the formation of the TREAT-NMD global database for DMD [2]. This is a collection of more than 7000 genetically confirmed DMD patients worldwide of which we have clinical data, which passed quality control for 5345 patients. To our knowledge this is the single largest cohort of genetically confirmed DMD patients to date and has allowed us to highlight important DMD clinical milestones and outcomes. To facilitate the inclusion of patients, a flexible reporting methodology has been applied that makes use of both patient self-report and professional report depending on differences in medical and cultural conditions as well as practical considerations such as cost effectiveness. Great care has been taken in training, standardization, and curation, but some variability in reporting between different sites and countries is inherent to the registry approach. The trends uncovered will need to be taken into consideration for novel DMD treatments like exon-skipping, but will also have impact on DMD prognosis, management and care standards.

In addition to allowing access to large amounts of data, registries provide a mechanism for collecting longitudinal data at relatively low cost and often include information on patients who would be excluded from natural history studies or clinical trials. Unlike in randomized clinical trials, causality cannot be firmly established from registry data and we accept this as a limitation of our study. However, our data may help to power future clinical trials by highlighting the need for trial design to take into account the heterogeneity of the DMD population including patient age, mutation type and corticosteroid background. The data collected via registries contribute to better understanding of the cost-effectiveness of new therapies, allows assessment of the burden of illness, and helps in planning health economic actions and model healthcare system for a specific rare condition in the most effective way. It is also very reassuring that the association of corticosteroid treatment with improved clinical outcomes can be observed in the “real world” across many different countries with different healthcare settings and ethnic backgrounds and regardless of a specific corticosteroid regime.

CONCLUSIONS

Clinical trials in rare disorders such as DMD are challenging, expensive and liable to low recruitment numbers, lack of validation of outcomes in larger cohorts, and variability of assessment across multiple centres or countries. Power calculations are often based on assumptions and on data on a limited number of participants in single centres, and recent trials to prove the efficacy of novel drugs in DMD have not reached their primary endpoint in measures like the 6-minute-walk test [22]. While well-controlled clinical studies including natural history studies are expected to improve this knowledge and to de-risk future trials, complementary and supporting information can be obtained from pooling a limited, but well standardized set of data across multiple registries and countries. This cross-sectional approach allows correcting for country- or site-specific differences and for long-term changes in “real-world” standards of care and treatment. We suggest that through this approach, better assumptions of the variability in outcomes can be obtained, and power calculations and stratification strategies improved. In addition, with the development of new genetic therapies, comes the need for post-marketing surveillance to assess the long-term safety and efficacy of novel therapies. The global DMD registry will serve as an important initial source this information, allowing the generation of disease specific long-term data and facilitating the comparison of different therapeutic approaches, using standardized methods of data collection.

DECLARATIONS