Phenotypic Spectrum of Dystrophinopathy Due to Duchenne Muscular Dystrophy Exon 2 Duplications

Abstract

Background and Objectives

To describe the phenotypic spectrum of dystrophinopathy in a large cohort of individuals with DMD exon 2 duplications (Dup2), who may be particularly amenable to therapies directed at restoring expression of either full-length dystrophin or nearly full-length dystrophin through utilization of the DMD exon 5 internal ribosome entry site (IRES).

Methods

In this retrospective observational study, we analyzed data from large genotype–phenotype databases (the United Dystrophinopathy Project [UDP] and the Italian DMD network) and classified participants into Duchenne muscular dystrophy (DMD), intermediate muscular dystrophy (IMD), or Becker muscular dystrophy (BMD) phenotypes. Log-rank tests for time-to-event variables were used to compare age at loss of ambulation (LOA) in participants with Dup2 vs controls without Dup2 in the UDP database and for comparisons between steroid-treated vs steroid-naive participants with Dup2.

Results

Among 66 participants with Dup2 (UDP = 40, Italy = 26), 61% were classified as DMD, 9% as IMD, and 30% as BMD. Median age at last observation was 15.4 years (interquartile range 8.79–26.0) and 75% had been on corticosteroids for at least 6 months. Age at LOA differed significantly between participants with Dup2 DMD and historical controls without Dup2 DMD (p < 0.001). Valid spirometry was limited but suggested a delay in the typical age-related decline in forced vital capacity and 24 of 55 participants with adequate cardiac data had cardiomyopathy.

Discussion

Some patients with Dup2 display a milder disease course than controls without Dup2 DMD, and prolonged ambulation with corticosteroids suggests the potential of IRES activation as a molecular mechanism. As Dup2-targeted therapies reach clinical applications, this information is critical to aid in the interpretation of the efficacy of new treatments.

Duchenne muscular dystrophy (DMD) is an X-linked recessive neuromuscular disease characterized by progressive weakness of the skeletal, respiratory, and cardiac muscles. DMD affects 1 in 3,500–5,050^1-3 live male births and is the most severe phenotype caused by pathogenic variants in the DMD gene, which encodes for the dystrophin protein. Deletions are the most common molecular defect (68%–77% of cases),^4,5 followed by point mutations, duplications (11%–13% of cases),^4-7 and small rearrangements.^8-10

These pathogenic variants result in 1 of 5 different phenotypes based on clinical symptoms and severity: DMD (loss of ambulation [LOA] prior to age 13 years), intermediate muscular dystrophy (IMD), Becker muscular dystrophy (BMD, LOA after age 16 years), manifesting female carrier, and isolated dilated cardiomyopathy. Phenotype can largely be predicted based on the reading frame rule, and in ∼90% of cases, out-of-frame pathogenic variants that induce a frameshift in the protein-coding sequence lead to the absence of dystrophin expression and the DMD phenotype.^4,11

Duplication of exon 2 (Dup2) is the most common duplication in the DMD gene (6.3%–9.8% of all duplications and 1%–2% of total cases) and results in a frame shift and thus an expected severe DMD phenotype. The majority of studies cataloging known DMD pathogenic variants in large databases have listed those with Dup2 phenotypically as DMD.^4,5,7,12,13 However, in a large cohort of 1,111 patients with dystrophinopathy (overlapping with the data set we report here), 20% of 20 participants with Dup2 were clinically classified with milder IMD or BMD phenotypes.⁷ More recently, a discordance in clinical severity between BMD and DMD phenotypes has been reported in a pair of Dup2 half siblings, confirming variability within the same family.^11,14 A potential explanation for this variability is offered by the identification of a glucocorticoid-responsive internal ribosomal entry site (IRES) within exon 5 of the DMD gene, which can drive expression of an N-truncated but functional dystrophin from an AUG codon within exon 6.^15,16 Patients who express this isoform have a markedly attenuated phenotype.¹⁷ Notably, studies in reporter systems showed that IRES activation was robust in the presence of exon 2 deletions and was ablated in exon 2 duplications in vitro,¹⁵ but such in vivo studies may not reflect variability in utilization of the IRES in muscle fibers, offering a mechanism for variability of phenotype.

At least 1 mutation-specific therapy for Dup2 mutations has reached clinical trial (ClinicalTrials.gov identifier NCT04240314) and others are under study. In anticipation of such trials, and in order to determine the range of distinctive disease trajectories, we describe the phenotypic spectrum of dystrophinopathy in a large cohort of individuals with Dup2 pathogenic variants.

Methods

Participant Identification

In this cross-sectional retrospective observational study, participants with a genetically confirmed duplication of exon 2 were identified through 2 different networks: the Italian DMD Network^18-21 and the United Dystrophinopathy Project (UDP) database, a large genotype–phenotype consortium.^11,22-24 Both networks are large databases of demographic, clinical, and genetic information of participants with dystrophinopathies. For the UDP, inclusion criteria include symptoms consistent with a dystrophinopathy along with an X-linked family history, absent or altered dystrophin expression on muscle biopsy, or clinical genetic analysis that demonstrates a pathogenic variant in the DMD gene. Participants enrolled without a defined mutation undergo genetic analysis as part of the project, and for all participants, the DMD mutation is determined by complete interrogation of copy number for all exons, and sequencing of all exons in patients without exon copy number changes. There are no exclusion criteria. For the Italian network, the enrollment included all the patients with a confirmed genetic diagnosis of dystrophinopathy due to an exon 2 duplication. Patients were identified from the 15 national neuromuscular centers, thus including the majority of all known living patients with Dup2 in Italy. The UDP began enrollment in 2003 and the Italian DMD Network began enrollment in 2018. Any participant present in the database as of August 2020 was included.

Standard Protocol Approvals, Registrations, and Patient Consents

The work completed in this article has been approved by the Institutional Review Board at Nationwide Children's Hospital (614 722 2708) under the Translational Research in the Dystrophinopathies protocol for the UDP cohort and the Ospedale San Raffaele Ethical Committee under the protocol Banca-Inspe for the Italian cohort. Consent has been obtained from each patient/participant or (for minor participants) from their parents or guardians in order to obtain and analyze participant data.

Clinical Characterization

Data collected included age, survival status, clinical phenotypic characterization (DMD/IMD/BMD), ambulation status, functional status, disease-related complications, and corticosteroid use. When applicable, other experimental therapeutic agents were recorded. Clinical and genetic data were anonymized and entered into a secured database. After review of the clinical information, a clinical classification of DMD, IMD, or BMD was assigned, based on previously published criteria: DMD, LOA at age ≤12; IMD, LOA between ages 13 and 15, inclusive; and BMD, LOA at age ≥16.^11,25 As these criteria are based upon age at LOA, classification in ambulant participants was assigned by clinicians based upon a combination of clinical judgment and functional scores at last assessment (e.g., 6-minute walk test, North Star Ambulatory Assessment). Both of these outcome measures have been well-characterized in DMD and one can determine if a certain time or score at a particular age is more characteristic of a DMD, IMD, or BMD phenotype.^26-29 Two participants were too young (<4 years of age) for expert clinicians to establish a clinical classification. Data sources included both clinical evaluation reports from which data were extracted and research evaluations. When available, respiratory function was recorded as forced vital capacity percent predicted (FVC%p) at last visit. Cardiomyopathy was assessed by review of the clinical echocardiographic reports and recorded as either present (defined as left ventricular ejection fraction [LVEF] <55%) or absent (LVEF >55%) from clinical echocardiogram reports in each database. When available, the first abnormal or last normal echocardiogram were also recorded. Lastly, participants were grouped based on corticosteroid (CS) use as “CS-never” (never treated or treated for <6 months, consistent with earlier usage^22,24) and “CS-ever” (received at least 6 months of corticosteroid treatment prior to LOA or last evaluation timepoint).

As a natural history group for comparison of age at LOA, the same data were extracted from 618 participant records in the UDP database with a known of date of LOA, a clinical diagnosis of DMD, and completely characterized steroid treatment status; this subset was considered our dystrophinopathy control group. Participants were excluded from this comparator group if they had duplications of exon 2 or other pathogenic variants known to be associated with an ameliorated phenotype, including deletion of exons 3–7 or deletion in the rod domain in which low-level endogenous skipping of exon 44 may occur.³⁰

Statistical Analysis

Group comparisons were evaluated using χ² or Fisher exact tests for categorical variables and Wilcoxon rank sum tests or Kruskal-Wallis tests for continuous variables. For time to event variables (age at LOA), Cox proportional hazards regression with log-rank tests were used. For those who were still ambulant, their time was censored at their last recorded visit. All analyses were conducted using R for Statistical Computing. Results were considered significant when p < 0.05.

Data Availability

Data not provided in the article because of space limitations may be shared (anonymized) at the request of any qualified investigator for purposes of replicating procedures and results.

Results

Cohort Description

All identified participants with Dup2 (n = 66) from both networks were included in the analysis, including 40 (61%) from the UDP database and 26 (39%) from the Italian network. There was no significant difference in age at diagnosis, age at last examination, or current age between the 2 cohorts (Table 1). The median (interquartile range [IQR]) age at diagnosis was 4.6 years (2.6–6.7) and median (IQR) age at last observation was 15.4 years (8.8–26.0). Sixty-three participants (95%) were alive at the time of analysis, 2 were deceased, and survival status was not available for 1 participant.

Table 1.

Summary of Dup2 Participants

Thirty-nine participants (61%) were classified as DMD, 19 (30%) as BMD, and 6 (9%) as IMD. Two very young participants (<4 years) at the time of this study were not phenotypically categorized. Participants in the Italian cohort were more likely to be classified as the milder BMD or IMD compared to the UDP cohort, which was predominantly DMD (p = 0.036). Forty-seven participants (75%) were classified as having steroid exposure (CS-ever), with similar rates between the Italian (76%) and UDP (74%) cohorts. Participants in the UDP cohort had a shorter total CS exposure time (2.95 vs 4.15 years).

Motor Function/LOA

In a survival analysis of all participants with information on ambulation status (an analysis that incudes age at last observation for participants who are still ambulant), there is no evidence of a difference in age at median LOA between the Italian (13.0 years) and US (12.7 years) groups (hazard ratio [HR] 0.79, 95% confidence interval [CI] 0.38–1.68; p = 0.54). However, as expected, there was evidence that age at LOA was different between the CS-ever (16.5 years) and CS-never (11.0 years) participants (HR 0.45, 95% CI 0.21-0.94; p = 0.03; Figure 1). Similar steroid effects were found within each cohort; in the Italian cohort, CS-never participants had a median age at LOA of 11.0 years vs 17.0 years for CS-ever participants, and in the UDP cohort, CS-never participants had a median age at LOA of 12.0 vs 16.5 years for the CS-ever.

Figure 1. Age at Loss of Ambulation in the Entire Dup2 Cohort.

Data were analyzed regardless of clinician-assigned clinical classification (e.g., Duchenne muscular dystrophy/intermediate muscular dystrophy/Becker muscular dystrophy). (A) Loss of ambulation (LOA) is similar between the United Dystrophinopathy Project (UDP) and Italian cohorts. Two participants from the UDP–duplication of exon 2 (Dup2) cohort are not included due to unknown ambulation status. Median age at estimated LOA was similar between groups: 13 years for the Italian cohort and 12.7 years for the UDP cohort (hazard ratio [HR] 0.79, 95% confidence interval [CI] 0.38–1.68; p = 0.54). (B) Loss of ambulation among entire Dup2 cohort suggests a significant corticosteroid treatment effect. Three participants are not included due to incomplete steroid information. Estimated LOA was significantly older for those who were exposed to >6 months of corticosteroids (CS-ever; median 16.5 years) vs those exposed to less than 6 months (CS-never; 11.0 years) (HR 0.45, 95% CI 0.21-0.94; p = 0.03).

Further comparison of the 2 cohorts used a subset analysis of only those participants who had reached LOA, excluding still-ambulant participants. As expected, this subset analysis demonstrates slightly different values for median LOA from the survival analysis reported above, but still shows no significant differences between the Italian and UDP cohorts. For the subset of 32 participants (48%) who had a recorded age at LOA, the observed median age at LOA was 12.1 years (interquartile range [IQR], 10.8–14.2) and there was no significant difference between the Italian (12.0 years; IQR, 11.0–13.0) and UDP cohorts (12.2; IQR, 10.8–17.2 years) when all phenotypes were included (Table 1). When comparing steroid use, CS-ever participants had a later age at LOA of 12.6 years (IQR, 11.8–17 years) vs 11.0 years (IQR, 9.9–12.3 years) in CS-never participants, but this difference was not statistically significant (Table 2), likely due to the smaller sample size in this subset analysis.

Table 2.

Steroid Exposure Among the Dup2 Cohort

Under the assumption that distinct DMD and BMD phenotypes occurred in patients with Dup2, in order to determine whether Dup2-associated DMD was different from other DMD-associated phenotypes, we analyzed only those participants given a clinical diagnosis of DMD and compared them to a cohort of non-Dup2 UDP controls. As a group, even those participants with a clinical classification of DMD walked significantly longer: median age at LOA for the UDP controls was 10.5 years, compared to 12.0 years for the Dup2 cohort (HR Dup2 vs UDP: 0.29, CI 0.17–0.51; p < 0.001) (Figure 2).

Figure 2. Age at Loss of Ambulation in Patients Clinically Diagnosed With DMD Phenotypes.

(A) All participants with duplication of exon 2 (Dup2) with Duchenne muscular dystrophy (DMD) phenotype. Estimated loss of ambulation (LOA) was not significantly older for those who were exposed to ≥6 months of corticosteroids (CS-ever; median of 12.0 years) vs those exposed to <6 months of corticosteroids (CS-never; 11.0 years) (hazard ratio [HR] 0.50, 95% confidence interval [CI] 0.21–1.19; p = 0.12). (B) All participants with Dup2 with DMD phenotype compared to a natural history comparator group. Comparison of participants with Dup2 clinically characterized with DMD, regardless of steroid treatment status, compared to participants without Dup2 DMD from the United Dystrophinopathy Project (UDP) database shows prolonged ambulation. The median age at LOA for the UDP cohort was 10.5 years vs 12.0 years for the combined Italian and UDP Dup2 cohorts (HR 0.29, 95% CI 0.17–0.51; p < 0.001). (C) CS-never participants with Dup2 with DMD phenotype compared to a natural history comparator group. Comparison of only corticosteroid-untreated participants with Dup2 clinically characterized with DMD to participants without Dup2 DMD from the UDP database also shows prolonged ambulation. Among CS-never participants, the median age at LOA for the UDP cohort was 10.0 years vs 11.0 years for the Dup2 CS-never participants (HR 0.65, 95% CI 0.52–0.82; p < 0.001).

Respiratory and Cardiac Function

Spirometry was performed in 39 participants with Dup2 at a median (IQR) age of 14.9 years (9.65–22.25) (Table 3). Median (IQR) FVC%p was 90.0% (70.5–98.5). Although this is a limited dataset, all participants had FVC%p above 60% until at least 20 years of age. Information on cardiac status was available in 55 participants; 24 had been diagnosed with cardiomyopathy. Of those with an ejection fraction (EF) reported, the median (IQR) EF was 58% (50–63.4). One CS-never participant, who lost ambulation at 11 years, received an implantable cardioverter defibrillator at 14.9 years. Scatter plots of FVC%p and EF% by age and diagnosis are displayed in eFigure 1, links.lww.com/WNL/B729.

Table 3.

Cardiac and Respiratory Function Among Participants With Dup2

Discussion

A duplication of DMD exon 2—the most common duplication pathogenic variant in the dystrophinopathies—results in a shift in reading frame and is predicted to result in an absence of dystrophin expression and a DMD phenotype. Consistent with this prediction, the majority of large databases of dystrophinopathy genotypes have characterized Dup2 individuals as DMD,^4,5,7,12,13 although these databases do not typically describe how phenotypic classifications were derived. In a previous analysis of the UDP database where prospective classification criteria were defined for 1,111 participants, 20% of patients with Dup2 were defined as having milder phenotypes.¹¹ In the present cohort, containing more than triple that number of participants with Dup2, 39% have a milder phenotype, with 24% of these milder cases clinically classified as IMD and 76% classified as BMD using either clinical examination (in still-ambulant patients) or age at LOA criteria. These data suggest that in at least some patients with Dup2, molecular mechanisms of disease amelioration may be present. This hypothesis is supported by analysis of only those patients with Dup2 clinically classified as DMD (LOA <13 years) in comparison to a large cohort of controls from the UDP database, where participants with Dup2 DMD walk 1.5 years longer than those with non-Dup2 DMD genotypes.

One possible mechanism for this attenuated phenotype is the presence of a functional IRES within exon 5, first identified in the laboratory of one of the current investigators (K.M.F.). This IRES drives cap-independent translation that is initiated at an AUG codon within exon 6. The resulting dystrophin isoform lacks the first calponin homology (CH1) domain within the canonical dystrophin actin binding domain 1 (ABD1). Despite predictions that disruption of the ABD1 domain obligatorily results in significant loss of dystrophin function, patients who express dystrophin lacking the CH1 domain (the ΔCH1 isoform) demonstrate abundant dystrophin expression, properly localized at the sarcolemmal membrane.¹⁶ More importantly, they show minimal functional defects, with the most prominent symptoms being myalgias and elevations in serum creatine kinase.¹⁷ Consistent with the observation of a predominance of DMD phenotypes among patients with Dup2 pathogenic variants, in vitro experiments showed the IRES to be inactive in the presence of an exon 2 duplication within the mRNA.¹⁵ However, IRES utilization is controlled by a variety of IRES trans-activation factors, which play a role in activation in mammalian cells by a variety of mechanisms.^31,32 Expression or regulation of these auxiliary factors may well vary in the population, allowing a possible pathway for variability of expression of the highly functional ΔCH1 isoform in muscle tissue, leading to variable phenotypes.

Because the DMD exon 5 IRES has been shown to be steroid-responsive in both cell culture and in the Dup2 mouse model,¹⁵ the observation of an apparent CS treatment effect in our Dup2 cohort is of particular interest. Treatment of participants with Dup2 with CS seemed to provide a larger treatment effect, as judged by LOA, in comparison to UDP control participants without Dup2 similarly treated with CS. This apparent differential effect of CS therapy could be consistent with IRES activation, suggesting that treatment of patients with Dup2 with CS results in more benefit for a given dose. We note that all but 2 participants with a milder phenotype were exposed to corticosteroids for at least 6 months prior to losing ambulation, perhaps supporting the hypothesis that corticosteroids may provide an additional benefit in participants with Dup2. If CS-induced IRES activation is responsible for a more robust CS effect on age at LOA, the increased use of CS in modern clinical practice may contribute to the higher reported fraction of patients with Dup2 identified with IMD or BMD in this analysis in comparison to previous studies,¹¹ as our clinical definitions are based on LOA.

It is worth noting that if either of these IRES-based explanations—variable endogenous IRES activation or increased IRES-driven expression due to CS therapy—is active, they are not sufficiently robust to result in maximal IRES activation. None of the participants with Dup2 we describe presented with the exceedingly mild phenotypes associated with the presence of the p.Trp3X founder allele, in which expression at levels estimated at 15% resulted in ambulation into the 7th decade.^16,17 In contrast, the oldest ambulant participant with BMD we describe is currently 22.3 years old, but significantly and clearly affected by muscle weakness.

An alternate hypothesis for what appears to be a mildly attenuated phenotype is that alteration of native exon splicing results in exclusion of one copy of the duplicated exon 2. Low levels of alternative splicing leading to an mRNA with an open reading frame have been invoked as a cause of an attenuated phenotype in the presence of central rod domain out-of-frame deletion pathogenic variants, particularly in the case of exon 44.^33,34 In the presence of an exon 2 duplication, native splicing events resulting in exclusion of a single exon 2 would restore a normal DMD transcript. Expression of even very low levels of a normal protein might be expected to result in amelioration of the phenotype, as levels of dystrophin expression around 3% of normal in a nearly full-length dystrophin due to altered central rod region splicing have been shown to do.³⁵

Our analysis has concentrated on LOA as an outcome, as it is relatively uniformly recorded in multiple record sources. However, our data preliminarily suggest that respiratory dysfunction may be milder in the Dup2 population. Recent publications suggest that the median age at reaching an FVC%p value below 50% is 16 years in patients with DMD.^36-38 Acknowledging the small sample size for available FVC data in our cohorts, none had an FVC%p below 60% prior to age 16, regardless of clinical phenotype or CS use. The cardiac data are similarly limited in this cohort and although cardiac function may be slightly milder in the BMD phenotypes, overall cardiac function did not appear to be significantly different than typically seen.³⁹

We acknowledge several limitations to our study. First, the sample size is relatively small, although this is the largest collection of phenotypic Dup2 data published to date. Another is its retrospective nature, with a reliance on data collected by a widespread group of clinicians. A third is the inherent risk in assigning clinical classifications to younger participants. We attempted to minimize this by utilizing functional outcomes data whenever possible for participants ages 4–9 years. For this group, outcomes on measures such as the 6-minute walk test and North Star Ambulatory Assessment have been established, allowing comparison of results to the range expected in DMD, and results far exceeding those expected performances could aid the expert clinicians in assigning a clinical classification of IMD or BMD. As a further conservative measure, phenotypic classification was not assigned to the 2 participants under the age of 4. Finally, we acknowledge that a variety of other genetic modifiers may influence the severity of DMD symptoms,⁴⁰ but our cohort size is too small to anticipate detection of effects from such known modifiers.

The largest limitation of our study is the absence of molecular data to correlate with these clinical findings. This is due in large part to the decreasing incidence of muscle biopsies in the diagnostic pathway for dystrophinopathies and the increasing availability of definitive copy number tests from genomic DNA. Such muscle studies would be informative. We note that our alternative hypothesis of amelioration (of mRNA splicing and IRES translation) are not mutually exclusive, and both may be active. Determination of whether one or both are depends on further studies of RNA and protein from muscle biopsies, which are beyond the scope of this article. Available tissues were limited, with wide variation in handling and time of storage; further prospective studies are planned, which should help clarify the underlying mechanisms. Nevertheless, our data describe the heterogeneity and range of dystrophinopathy phenotypes associated with duplications of exon 2, and may provide guidance for the enrollment in and interpretation of data from trials of therapies directed specifically to the Dup2 genotype.

Glossary

BMD

Becker muscular dystrophy

CI

confidence interval

CS

corticosteroids

DMD

Duchenne muscular dystrophy

Dup2

duplication of exon 2

EF

ejection fraction

FVC%p

forced vital capacity percent predicted

HR

hazard ratio

IMD

intermediate muscular dystrophy

IMD

intermediate muscular dystrophy

IQR

interquartile range

IRES

internal ribosomal entry site

ITAF

IRES transactivation factor

LOA

loss of ambulation

LVEF

left ventricular ejection fraction

UDP

United Dystrophinopathy Project

Appendix 1. Authors

Appendix 2. Coinvestigators

Study Funding

This work was supported by the NIH (NINDS NS085238) to K.M.F. and R.B.W. S.P. was supported by Fondazione Regionale per la Ricerca Biomedica FRRB CP2_10/2018, EJPRD19-118, and Telethon GGP17009.

Disclosure

A.A. Zambon, R. Alles, S. Conroy, M. Moore-Clingenpeel, and R.B. Weiss report no disclosures relevant to the manuscript. M.A. Waldrop has served on advisory boards for Avevis, Inc., and Sarepta Therapeutics, and is a site PI for a clinical trial sponsored by Astellas Gene Therapy. S.C. Previtali has served on advisory boards for Sarepta Therapeutics and Esperare Foundation and is site PI for a clinical trial sponsored by FibroGen Inc. K.M. Flanigan serves on advisory boards for Apic Bio, 4DMT, and Encoded Therapeutics, and receives royalties from Astellas Gene Therapies (formerly Audentes Therapeutics). Go to Neurology.org/N for full disclosures.