Background
Duchenne muscular dystrophy (DMD) is a rare, progressive, life-limiting neuromuscular disorder [
1] occurring in 15.9 to 19.5 per 100,000 live male births [
2‐
4]. It is caused by mutations in the dystrophin gene [
2,
5]; lack of dystrophin compromises muscle structure and integrity, leading to progressive muscular degeneration [
6,
7]. Patients with DMD are typically identified in early childhood with symptoms including delays in motor milestones and frequent falls [
8]. Over time, these patients experience progressive functional impairments leading to loss of ambulation (LOA), pulmonary insufficiency, cardiomyopathy, and early mortality [
2,
5,
9].
Although there is presently no cure for DMD, advancements to the standard of care, including the introduction of systemic corticosteroids in the 1990s, have helped slow disease progression and improve survival [
10‐
12]. However, the impact of these changes in standard of care across the full range of clinically-relevant disease progression milestones experienced by those with DMD has not been fully characterized. In 2017, Ryder et al. published a systematic review examining the epidemiology, burden, and treatment of DMD; however this review focused only on studies published between 2011 and 2015 [
6]. Other reviews focused on the prevalence of DMD [
13] or the impact of surgery on pulmonary decline [
14]. While robust outcomes data are available from large cohort studies including the Cooperative International Neuromuscular Research Group (CINRG) [
15], Duchenne Registry [
16], and Centers for Disease Control and Prevention’s Muscular Dystrophy Surveillance, Tracking, and Research Network (MD STARnet) [
17], a synthesis of data from recent studies is lacking [
18]. The objective of this systematic review was to characterize the clinical course of DMD in the era of corticosteroid treatment in North America.
Methods
A comprehensive search of the Medline/Medline In-Process and EMBASE databases was performed (see Additional file
1: Table S1 for search strategy), the design of which was guided by the study-specific PECOS (Population, Exposures, Comparators, Outcomes, Study design) criteria (Table
1). Studies published in English between database inception (1946) and November 2018 that reported estimates of the age at occurrence of key clinical milestones occur among males with DMD were selected. To focus on more generalizable outcomes from a more homogeneous set of patients, the review targeted observational studies from North America (or international studies including North America patients) that aimed to estimate the frequency of key clinical events from large (n > 50) samples of DMD patients treated with corticosteroids. Animal studies, or studies that included patients with other muscular dystrophies, were excluded.
Table 1
PECOS criteria to outline the scope of the literature review
Exposures/comparators | Subgroup Corticosteroid treatment By age By disease status at baseline |
Outcomes | Clinical/functional measures measured over a minimum of 1 yeara Pulmonary function tests: Forced vital capacity, peak expiratory flow Assessment of cardiac function: Ejection fraction, left ventricular end diastolic dimension, shortening fraction Key clinical outcomes LOA Scoliosis Need for ventilatory support Pulmonary dysfunction Cardiac dysfunction/cardiomyopathy Mortality |
Study design | Prospective or retrospective studies Case series |
Outcomes of interest that describe the clinical course of DMD included LOA, scoliosis, need for ventilatory support (stratified by any ventilation/type unspecified, non-invasive ventilation [NIV] or invasive ventilation [IV]), pulmonary dysfunction, cardiomyopathy, and mortality. Relevant measures included the mean or median age at the outcome of interest, or the percentage experiencing the outcome over time or at a particular time (
t)
. Scores on assessments of ambulatory, pulmonary, or cardiac function over a minimum of one year of follow-up were also included (Table
1). Two reviewers independently screened abstracts and potentially eligible full-text articles for inclusion, and any discrepancies were resolved through discussion to achieve consensus.
Data were extracted by two researchers; study characteristics extracted included authors, year, study duration, objective(s) and design, sample size, and inclusion and exclusion criteria. Patient characteristics included details of corticosteroid treatment and baseline demographics. Cohorts were classified as ‘corticosteroid-treated’ if all patients were so treated, ‘mixed corticosteroid use’ if the sample represented a mix of corticosteroid-treated and -untreated patients, and ‘likely corticosteroid-treated’ if the study was published after 2005 and did not state the sample was untreated. Available data on use of cardioprotective medications, such as angiotensin-converting enzyme (ACE) inhibitors, were also extracted where available.
For continuous variables, the mean, median, standard deviation (SD), confidence interval (CI), interquartile ranges (IQR), and range was extracted whenever available. For dichotomous and categorical variables, the number of patients and proportion was extracted. For studies reporting on the mean or median age at the outcome, the range of estimates was tabulated. The percentage of the sample who experienced the outcome at time of reporting was also described (where available). Data on the percentage experiencing the outcome at specific time points or over time were described using Kaplan–Meier (KM) curves, as well as presented as point estimates at time t by the original authors. Where available, scores on functional and clinical measures of interest over time were plotted using line graphs.
The strength of the available evidence was assessed using the STrengthening the Reporting of Observational studies in Epidemiology (STROBE) Statement for observational studies and non-randomized clinical trials [
19].
Discussion
A comprehensive systematic review was conducted to identify estimates of the age at key clinical milestones, and trajectories on relevant functional measures over time, among studies including North American patients with DMD. Age at LOA was the most widely reported with estimates available from many large studies; these tended to range from 10 to 14 years of age [
27,
34]. However, robust data on the timing of the onset of scoliosis-, cardiac-, pulmonary- and ventilation-related outcomes were less frequently presented, particularly from large longitudinal studies. While reported estimates of the mean age at diagnosis of scoliosis were fairly consistent across studies (at 14–15 years of age), how scoliosis was classified differed widely [
31,
35,
38]. Pulmonary function in DMD patients declines with age from the mid-teens [
30,
41], and while most have severe pulmonary dysfunction by 25 years [
30], the mean age at initiation of ventilatory support ranged from 15 to 22 years depending on the type of ventilation considered and treatment center [
33,
39]. Data on age at mortality in DMD were also variable, and estimates were impacted by the inclusion criteria of the individual studies; for example estimates of mortality among those with cardiomyopathy or on ventilation were drawn from populations surviving to adulthood [
40,
46]. In addition to selection criteria, factors impacting the timing of key clinical milestones include corticosteroid regimen [
31] and disease genotype [
26]. The findings of this review help summarize the likely timing of disease progression milestones for North American patients with DMD, and also highlight potential heterogeneity in timing observed both within and across study populations.
Estimates of time to key clinical milestones in this review included data from studies from the large North American registries (e.g. CINRG and MD STARnet), and findings are consistent with those from large observational studies and registries from outside of North America. The Translational Research in Europe—Assessment and Treatment of Neuromuscular Diseases (TREAT-NMD) network of DMD registries have published studies documenting the clinical course of patients with DMD [
49‐
52]. In a large survey of over 1500 DMD patients that characterized the impact of corticosteroid use, mean estimates of age at LOA ranged from 10.1 (non-corticosteroid-treated patients) to 11.4 (corticosteroid-treated) years [
49]. An analysis of over 5000 patients also from TREAT-NMD reported age at LOA of 13 years among corticosteroid-treated patients, and that up to 50% of patients required ventilation by 20 years of age [
50]
That longitudinal data describing survival specifically among North American DMD patients are few, was one of the major gaps identified in this review. However, mortality rates from included studies were consistent with findings of two important studies on mortality in dystrophin gene-related muscular dystrophy, which did not meet the inclusion criteria for the current review as they also included patients with Becker muscular dystrophy (BMD). The first study, which was based on vital statistics, estimated that 71% of mortality among those with BMD/DMD occurred between the ages of 15 and 29 years; the authors assumed it was most likely related to DMD [
53]. The second study, from MD STARnet, estimated mortality in almost 60% of that cohort by age 25 years, with most deaths occurring among those aged 20 to 25 years [
54]. Further follow-up from existing large DMD cohorts will help improve contemporary estimates of the timing of key clinical milestones.
Accurately estimating the time of onset of gradually progressive manifestations of DMD can be difficult, and this along with changes in practice patterns and symptom detection, contribute to observed variability in estimates. For example, many studies reporting on scoliosis classify outcomes based on surgery, however with changing treatment patterns [
55] the utility of surgery as a proxy for clinically-significant scoliosis will decrease. Similarly, recommended strategies for ventilation vary among clinical centers [
39,
56,
57], and practice is changing (in particular for how IV is used) [
58], which will impact the comparability of estimates of the timing of respiratory decline across studies from different periods. Finally for cardiomyopathy, with advancements in screening tools [
59,
60] as well as evidence of benefits to early treatment [
61], it is likely that initial signs will now be detected earlier, which would result in an apparent decrease in the mean age at cardiomyopathy over the coming years.
There are several additional factors impacting the timing of key clinical milestones that require consideration. To capture the impact of corticosteroids in the management of DMD, only studies including patients from the corticosteroid treatment era were included. While details of corticosteroid treatment regimens were extracted and reviewed, there were important limitations that precluded analyzing outcomes according to regimen. First, details on the timing of initiation, duration, type, and dose varied within and between studies. Only a small number of studies reporting on LOA presented results according to agent; but the remainder of the studies for that outcome, and all of the studies for other outcomes of interest, did not stratify by corticosteroid regimen. However, variations in corticosteroid treatment patterns (in terms of duration and dosing) may have affected the timing when patients reached LOA [
28,
31,
62,
63], and other important clinical milestones [
20,
31,
46,
62‐
65]. Evidence on the impact of early initiation of corticosteroids (e.g. before age 6 years) remains mixed [
31,
66]; more work is needed to disentangle the potential confounding effect of disease severity and the potential risk for adverse effects of corticosteroid treatment on outcomes in real-world studies. Treatment with ACE inhibitors has also been shown to impact the clinical course of DMD by delaying the onset of cardiomyopathy; however, the use of ACE inhibitors remains variable [
20,
67]. While it might be anticipated that studies describing later cohorts would show delayed onset of milestones that define the clinical course of DMD, the interplay between treatment advances and the impact of earlier diagnostics would make these relationships less apparent. Finally, other genetic modifiers may also play a role in the timing of DMD progression [
26,
29,
50]; however, outcomes according to genotype are infrequently reported outside of treatment trials [
68,
69]. The move from biomarkers to precise genetic diagnosis may also impact the apparent clinical course [
70].
Variability in methodology and data sources may also have affected estimates. Data from the CINRG and MD STARnet registries, both large well-documented US cohorts that comprehensively collect longitudinal data on the clinical course of DMD, were used in ten studies within this review. Outside of those, most observational studies and treatment trials do not follow patients for a sufficient time to describe changes across the range of key clinical milestones [
21,
30]. Other challenges for studying disease progression in rare diseases include small sample sizes which can amplify the impact of heterogeneity in diseases with varied clinical courses; data presented from convenience samples and case series may not be generalizable, and the impact of selection biases on outcomes (particularly for diseases with high early fatality among more severe cases) can be substantial [
71,
72]. The numerous outcome measures used to assess progression in DMD also make comparisons difficult, a limitation recently acknowledged in a workshop held by the DMD research community [
73]. Finally, there are useful measures for characterizing DMD progression that were infrequently reported in the studies of this review, such as the North Star Ambulatory Assessment or upper arm function, which are important in understanding patient functional status and ability to participate in activities of daily living.
Some limitations to the published data warrant mention. First, while time to event data using KM curves were presented in some studies, many reported the mean age at an occurrence where the entire sample had not experienced the event at the time of study reporting. As such, these values can be interpreted as the lower limit for when key clinical milestones will occur in DMD. Second, some measures may only be administered to individuals who still have some functional capacity (e.g. tests of ambulation), and patients unable to complete the test would have been excluded. This type of survival bias would result in an inflation of apparent functional status for cohorts as a whole. Third, mean scores on functional tests may reflect the inclusion criteria of each study, rather than the underlying distribution of scores on that functional test among the DMD population. Fourth, because of heterogeneity in designs employed, measures selected, and populations included across studies, meta-analysis was judged to be infeasible [
74,
75]; as a result, overall summary estimates of the time to key clinical milestones were not calculable.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.