**Genotype–Phenotype Correlations in Duchenne and Becker Muscular Dystrophy Patients from the Canadian Neuromuscular Disease Registry**

### **Kenji Rowel Q. Lim 1,**†**, Quynh Nguyen 1,**† **and Toshifumi Yokota 1,2,\***


Received: 29 October 2020; Accepted: 21 November 2020; Published: 23 November 2020

**Abstract:** Duchenne muscular dystrophy (DMD) is a fatal neuromuscular disorder generally caused by out-of-frame mutations in the *DMD* gene. In contrast, in-frame mutations usually give rise to the milder Becker muscular dystrophy (BMD). However, this reading frame rule does not always hold true. Therefore, an understanding of the relationships between genotype and phenotype is important for informing diagnosis and disease management, as well as the development of genetic therapies. Here, we evaluated genotype–phenotype correlations in DMD and BMD patients enrolled in the Canadian Neuromuscular Disease Registry from 2012 to 2019. Data from 342 DMD and 60 BMD patients with genetic test results were analyzed. The majority of patients had deletions (71%), followed by small mutations (17%) and duplications (10%); 2% had negative results. Two deletion hotspots were identified, exons 3–20 and exons 45–55, harboring 86% of deletions. Exceptions to the reading frame rule were found in 13% of patients with deletions. Surprisingly, C-terminal domain mutations were associated with decreased wheelchair use and increased forced vital capacity. Dp116 and Dp71 mutations were also linked with decreased wheelchair use, while Dp140 mutations significantly predicted cardiomyopathy. Finally, we found that 12.3% and 7% of DMD patients in the registry could be treated with FDA-approved exon 51- and 53-skipping therapies, respectively.

**Keywords:** Duchenne muscular dystrophy; Becker muscular dystrophy; dystrophinopathy; genotype-phenotype correlations; Canadian Neuromuscular Disease Registry; reading frame rule; dystrophin; multiple logistic regression analysis; exon skipping therapy

#### **1. Introduction**

Duchenne muscular dystrophy (DMD) is the most common inherited neuromuscular disorder worldwide, affecting approximately 20 per 100,000 male births (1:5000) [1,2]. DMD is an X-linked recessive disorder that is characterized by progressive body-wide muscle degeneration, with proximal muscle weakness starting at 3–5 years and loss of ambulation during the early teens [3,4]. Cardiac and respiratory symptoms often appear during the third decade of life, which eventually lead to death. DMD is primarily caused by mutations in the *DMD* gene that lead to an absence of dystrophin. Dystrophin is a protein responsible for stabilizing muscle cell membranes during contraction–relaxation cycles; its loss increases the susceptibility of muscles to tear during use [5–7]. There is a milder form of the disease called Becker muscular dystrophy (BMD), which is caused by mutations in the same gene. However, mutations in BMD patients generally only reduce the amount or functionality of the dystrophin produced, as opposed to the complete absence of dystrophin seen in DMD [8–10].

DMD and BMD are part of a group of disorders called the dystrophinopathies, which are all characterized by mutations in the *DMD* gene. Stark differences between the fatal DMD and mild BMD prompt us to understand how differences in genotype (i.e., mutation) impact phenotype (i.e., clinical outcome). This is especially important since there is no cure for DMD at present. To study these genotype–phenotype correlations, among other purposes, dystrophinopathy patient registries were formed by local, national, and international initiatives to collect information on patient clinical outcomes and *DMD* mutations. Perhaps the most extensive of these would be the TREAT-NMD DMD Global Registry [11] and the Leiden Open Variation Database (LOVD) [12,13], each having data from more than 7000 dystrophinopathy patients across the world. Canada in particular has the Canadian Neuromuscular Disease Registry (CNDR), a national patient registry established in 2011 that also contributes to the TREAT-NMD database [14,15]. As of 1 December 2019, with 4310 registrants, dystrophinopathy patients make up the second-largest disease group in the CNDR at 13.3% [15]. Amyotrophic lateral sclerosis has the most number of registered patients at 36.1%; myotonic dystrophy, limb–girdle muscular dystrophy, and spinal muscular atrophy patients make up 10.5%, 5.9%, and 5.3% of CNDR registrants, respectively.

Here, we aimed to evaluate genotype–phenotype correlations specifically in the Canadian DMD/BMD population, using the information on 402 patients from the CNDR. Similar studies have been conducted previously [11,16–22]; however, most of these investigated a limited number of clinical phenotypes. There may also be correlations unique to the Canadian population that would otherwise not be observed from a global database. We particularly examined the relationships between patient genotype and clinical diagnosis (DMD/BMD), as well as between patient genotype and clinical outcomes (e.g., wheelchair use and cardiomyopathy status). We also determined the applicability of recent U.S. Food and Drug Administration (FDA)-approved exon skipping DMD therapeutics to the CNDR DMD patient population, given the increasing entry of this class of therapies into the clinic. Finally, this work provides the most recent characterization of the *DMD* mutation landscape in Canada.

#### **2. Materials and Methods**

#### *2.1. Study Population and Design*

This study was approved by the University of Alberta Health Research Ethics Board—Health Panel (reference Pro00092569). Participants in the CNDR provided informed consent and agreed to have their data shared for research purposes. For this study, the following information was used from CNDR patient records, which were provided directly from the clinic by neuromuscular specialists in the CNDR network: weight, height, clinical diagnosis, genetic data (test information, mutation type, mutation location), neuromuscular data (motor function, therapies received), cardiac history (presence of cardiomyopathy, left ventricle ejection fraction (LVEF), cardiac medications received), respiratory data (use of non-invasive/invasive ventilation, forced vital capacity (FVC)), and gastrointestinal data (feeding tube use, major nutritional route). Clinical diagnosis (DMD/BMD) was at the discretion of the neuromuscular specialist attending to the patient on the basis of clinical and genetic characteristics. All genetic data were derived from accredited testing laboratories across Canada as part of standard clinical practice. If a patient had information in the registry from more than one visit, data from the most recent visit was considered for analysis. All patient data were de-identified before provision to the study team.

The initial study population consisted of 508 dystrophinopathy patients in the CNDR from 1 January 2012 to 3 July 2019. This included 414 DMD patients, 78 BMD patients, 13 female *DMD* mutation carriers, 2 intermediate muscular dystrophy (IMD) patients, and 1 with an unknown diagnosis (Figure 1). We filtered out patients who did not have genetic testing data or a definite DMD/BMD diagnosis, leaving us with 420 patients (350 DMD patients, 61 BMD patients, 9 female carriers). Data from these patients were used for comparisons of clinical outcomes across groups. For correlational analysis between genotype and clinical diagnosis as phenotype, we focused only on

the 342 DMD and 60 BMD patients with non-negative genetic test results. On the other hand, for the analysis between genotype and clinical outcomes (wheelchair use, presence of cardiomyopathy, LVEF, FVC), we restricted our analysis to include only the 342 DMD patients.

**Figure 1.** Study population and design. Patient data from the Canadian Neuromuscular Disease Registry between January 2012 and July 2019 were used for this study. The number and groups of patients evaluated for the various analyses performed are shown. DMD, Duchenne muscular dystrophy; BMD, Becker muscular dystrophy; FC, female carrier; IMD, intermediate muscular dystrophy; G/P, genotype–phenotype.

#### *2.2. Statistical Analysis*

All statistical analyses and plotting were performed using GraphPad Prism version 8.4.3 (GraphPad Software, San Diego, CA, USA). A two-sided Fisher's exact test was done to determine statistically significant differences between groups of categorical variables, while a two-tailed, unpaired Student's *t*-test was done for continuous variables. A multiple logistic or linear (least squares) regression analysis was used to construct inferential models studying the relationships between genotypes and clinical outcomes, with the latter serving as dependent variables. Patients with missing information were excluded from the multiple regression analyses by the software. A *p*-value of less than 0.05 was considered statistically significant.

#### **3. Results**

#### *3.1. Clinical Characteristics*

Table 1 summarizes the clinical characteristics of the three subgroups in our study population: DMD, BMD, and female carriers. The female carriers all appear to be healthy, at least based on the parameters reviewed. However, the low number of carriers in our cohort (*N* = 9) makes it difficult to accurately compare with other subgroups. Thus, we decided to perform a comparative analysis of clinical characteristics only between DMD and BMD patients.


**Table 1.** Summary of clinical characteristics for patients with genetic data in our study population.

<sup>1</sup> count data: frequency (%), continuous data: median (interquartile range), <sup>2</sup> DMD versus BMD, <sup>3</sup> vamorolone and testosterone counted as one group, <sup>4</sup> digoxin up to MRA counted as one group. DMD, Duchenne muscular dystrophy; BMD, Becker muscular dystrophy; FC, female carrier.

The DMD patients in our population were significantly younger by 7 years (*p* < 0.0001; mean ages of 10.5 versus 17.9 years old, respectively) and had lower body mass indices (BMIs) by 3 points (*p* < 0.005; mean BMIs of 18.1 versus 21.3, respectively) than the BMD patients. As expected, DMD patients used the wheelchair significantly more than BMD patients (*p* < 0.005), required more support for walking (*p* < 0.05) or sitting (*p* < 0.05), and were mostly on deflazacort therapy (*p* < 0.05). In terms of cardiac outcomes, no significant differences in cardiomyopathy status between DMD and BMD patients were observed in our population. However, the age of cardiomyopathy onset was significantly earlier for DMD at an average of 13.0 years than BMD at an average of 23.0 years (*p* < 0.05). Despite LVEF values being significantly lower in BMD than DMD patients (*p* < 0.05), both subgroups were well within the healthy LVEF range at >50%. These LVEF results likely reflect how patients from both groups also received standard cardiac medications in the form of angiotensin-converting enzyme inhibitors, angiotensin II-receptor blockers, and β-blockers, among others. FVC values were significantly reduced in DMD than in BMD patients (*p* < 0.005; 76.0% versus 88.0% on average, respectively). Perhaps due

to scarcity in the available data, no significant differences in other respiratory or gastrointestinal parameters were found between the two patient subgroups.

#### *3.2. Genetic Characteristics*

Genetic testing data was available for 350 of 414 DMD patients (85%) and 61 of 78 BMD patients (78%) (Figure 1). The majority of mutations were deletions of at least one exon in the *DMD* gene in 69% (241/350) of DMD patients and 80% (49/61) of BMD patients, or 71% (290/411) of patients in total (Figure 2a). This was followed by small mutations, i.e., point mutations and insertions/deletions within exons or splice sites, in 17% (71/411) of patients, and duplications of at least one exon in 10% (41/411) of patients. Negative results were found for 2% of patients, i.e., these patients were clinically diagnosed as having DMD/BMD, but genetic testing failed to identify a variant. However, as these patients were also not tested via gene sequence analysis, it remains possible that they could have deep intronic mutations in the *DMD* gene that were missed.

Mapping out all large deletions (>1 exon) revealed two mutation hotspots, one from exons 3 to 20 and another from exons 45 to 55 (Figure 2b). More than half of all patients with deletions at ~65% had mutations in the distal hotspot, whereas only ~21% were in the proximal hotspot. Moreover, most deletions in the proximal hotspot were represented by only one patient. The most common deletion was a deletion of exon 45, which was in 18 out of 290 patients (6%) with large deletion mutations (Figure 2c). Out of the 18 most common large deletion mutations, 17 were in the distal exons 45–55 mutation hotspot. Conversely, mapping out all large duplications (>1 exon) in our DMD and BMD patients revealed one hotspot from exons 3–10 (Figure 2d). However, note that most exon duplication patterns were represented by only one patient. The most common duplications were an exon 2 duplication and an extensive exons 5–65 duplication, which were each found in 3 out of 41 patients (7%) with large *DMD* duplication mutations (Figure 2e).

Small mutations were spread out across the entire gene, ultimately affecting all four major dystrophin protein regions: the N-terminal actin-binding domain (exons 2–8), the central rod domain (exons 8–61), the cysteine-rich domain (exons 63–69), and the C-terminal domain (exons 70–79) (Figure 3a,b). Exons were assigned to protein domains following information from the Leiden Muscular Dystrophy dystrophin page (https://www.dmd.nl/). Exon 18 harbored the greatest number of small mutations in our combined DMD and BMD population (Figure 3b). More than half (51%) of all identified small mutations were nonsense point mutations, followed by 27% being small insertions/deletions, 13% being splice site mutations, and 4% being missense mutations (Figure 3c). Interestingly, two DMD patients each carried two different small mutations—one with c.8729A>T and c.8734A>G (both missense mutations; reported in the LOVD to frequently co-segregate with each other and are classified as benign), and one with c.10127T > C (a missense mutation) and c.10133dup (a frameshifting insertion mutation). There was also one DMD patient who had both a duplication of exon 61 and a nonsense c.9100C > T point mutation; for purposes of this study, this patient was grouped with other duplication mutation carriers. A survey of nonsense point mutations in our population showed that 47% (17/36) involved a C-to-T transition (Figure 3d).

*J. Pers. Med.* **2020**, *10*, 241

**Figure 2.** Overview of genetic characteristics in the study population. (**a**) *DMD* mutations in patients grouped according to type (deletions, duplications, small mutations); (**b**) Map of large *DMD* deletions (>1 exon) in Duchenne and Becker muscular dystrophy patients (DMD, BMD), with their frequencies (# patients) on the right (*N* = 290); (**c**) Top 18 most common large *DMD* deletions in DMD and BMD patients; (**d**) Corresponding map of large *DMD* duplications (>1 exon) in DMD and BMD patients (*N* = 41); (**e**) Top 8 most common large *DMD* duplications in DMD and BMD patients.

**Figure 3.** Overview of small mutations in the study population. (**a**) The positions of small mutations identified in Duchenne and Becker muscular dystrophy (DMD, BMD) patients are shown according to the domain/region of the dystrophin protein they affect, with each dot representing a unique mutation. The color of the dots correspond to the legend in (**c**); (**b**) The positions of small mutations, shown according to the *DMD* exon they are located in; (**c**) Distribution of *DMD* small mutations according to type (small insertions/deletions, splice site mutations, nonsense mutations, missense mutations, others); (**d**) Frequency of point mutation types in DMD and BMD patients. (*N* = 71).

#### *3.3. Relationships between Genotype and DMD*/*BMD Diagnosis as Phenotype*

The reading frame rule predicts at least 90% of the time [11,22] if a given *DMD* mutation will lead to a DMD or BMD phenotype. Most out-of-frame mutations give rise to DMD, while most in-frame mutations give rise to BMD [8]. To determine how well this rule holds in our population, we examined the frequency of out-of-frame and in-frame deletions in our DMD and BMD patients from the CNDR (Figure S1a–c). Of the 238 DMD patients in our cohort with deletion mutations not involving either exon 1 or 79, 87% (208/238) had out-of-frame mutations and 13% (30/238) had in-frame mutations (Figure 4a). On the other hand, of the 49 BMD patients with corresponding deletions, 16% (8/49) had out-of-frame mutations and 84% (41/49) had in-frame mutations.

Considering the deletions themselves, 96% (208/216) of observed out-of-frame deletions led to DMD, with only 4% (8/216) leading to BMD (Figure 4b). The in-frame deletions displayed a less skewed behavior—with 42% (30/71) giving rise to DMD and 58% (41/71) to BMD. Since the in-frame deletions did not predominantly favor one phenotype over the other to the same extent as out-of-frame deletions, we decided to map them out across the *DMD* exons. This will allow us to see if the location of the in-frame deletion is a key determinant of whether a patient develops DMD or BMD. The majority of in-frame deletions leading to DMD were found to start within the N-terminal exons 3–20 hotspot (Figure 4 and Figure S1b). In particular, of the 19 in-frame deletions solely associated with DMD, 14 or 74% of them started in this region. DMD-associated N-terminal in-frame deletions also tended to partially or completely remove more functional domains on the resulting dystrophin protein than their BMD-associated counterparts (Table S1). On the other hand, 67% (10/15) of in-frame deletions located at the distal half of the gene past exon 43 led to a BMD phenotype or to a mix of either a DMD or BMD phenotype (Figure 4c).

**Figure 4.** Analysis of large deletions and duplications, and their effect on the *DMD* reading frame. (**a**) Distribution of in-frame and out-of-frame deletions in Duchenne and Becker muscular dystrophy (DMD, BMD) patients; (**b**) Distribution of phenotypes associated with in-frame and out-of-frame deletions; (**c**) Map of in-frame *DMD* deletions in DMD and BMD patients, black: DMD, white: BMD, striped: both; (**d**) Frequencies of DMD and BMD patients with hybrid or fractional repeat-forming in-frame deletions; (**e**–**g**) Corresponding plots of (**a**–**c**) for duplications in DMD and BMD patients.

As these distal in-frame deletions all occur within the central rod domain of the dystrophin protein, one could model in silico how well these preserve the filamentous, helical structure of the region. Depending on where the exon breakpoints are, an in-frame deletion can give rise to either a hybrid or a fractional repeat unit in the rod domain. Hybrid repeats maintain the filamentous structure of the rod domain, whereas fractional repeats disrupt it [23–25]. Using the eDystrophin database (http://edystrophin.genouest.org/) [25], we obtained modeling predictions for the repeat structures formed by the various distal in-frame deletions (Table 2). Although hybrid repeat-forming deletions were found in more BMD than DMD patients, no significant association was found between clinical phenotype (DMD/BMD) and the predicted repeat structure formed by an in-frame deletion in the exons 45–55 hotspot region (Figure 4d). Interestingly, despite giving rise to a predicted fractional repeat unit, the in-frame deletion of exons 45–47 led to BMD 91% of the time (10/11 patients) rather than DMD (Table 2).


**Table 2.** Repeat structure modeling of in-frame *DMD* deletions within the exons 45–55 hotspot.

<sup>1</sup> Information obtained from the online eDystrophin database.

We next examined the frequency of out-of-frame and in-frame duplications in our DMD and BMD patient population (Figure S2a,b). Of the 35 DMD patients in our cohort with duplication mutations, 83% (29/35) had out-of-frame mutations and 6% (6/35) had in-frame mutations (Figure 4e). Meanwhile, we only had five BMD patients with duplication mutations, one of which had an out-of-frame mutation, with the remaining four having in-frame mutations. In terms of the duplications themselves, out-of-frame duplications led to DMD 97% (29/30) of the time and to BMD 3% (1/30) of the time; in-frame duplications led to DMD in 60% (6/10) of cases and to BMD in 40% (4/10) of cases (Figure 4f). Similarly, as we did with the deletions, we mapped out all in-frame duplication patterns across the *DMD* exons (Figure 4g). Only nine unique in-frame duplications were found in our population, with those at the proximal end of the gene mostly associated with BMD and those at the distal end all associated with DMD.

Notably, less than 10% of small mutations (6/71) were associated with BMD in our study population. Due to the low representation of this mutation type among BMD patients, an analysis of genotype–phenotype correlations may be premature and therefore was not performed.

#### *3.4. Relationships between Genotype and Clinical Outcome as Phenotype*

We then proceeded to perform a series of multiple regression analyses to determine any relationships between patient genotypes and clinical outcomes, focusing on data from DMD patients (Figure 1). For genotype, we considered the location of the mutation according to which dystrophin protein domain/s or dystrophin isoform/s they affect. Exons were once again assigned to protein domains following information from the Leiden Muscular Dystrophy dystrophin page (https://www.dmd.nl/). For clinical outcomes, we looked at wheelchair use (combined permanent and intermittent use), cardiomyopathy status (presence or absence), LVEF, and FVC. In constructing these models, we also took into account the effect of other parameters such as age, BMI, steroid use (past or present), and use of cardiac medications, as appropriate. The results of these analyses are summarized in Table S2 and Table S3.

Multiple logistic regression analysis revealed that there is a 6.136 times increase in odds (95% confidence interval (CI): 1.44, 33.99; *p* < 0.05) that a DMD patient will require wheelchair use when they have mutations affecting the dystrophin rod domain (Table S2). Mutations affecting the C-terminal domain yielded an odds ratio of 0.0281 (95% CI: 0.001, 0.30; *p* < 0.005), indicating that their presence was associated with decreased wheelchair use in our DMD patient population. A similar relationship was found for mutations affecting the Dp116 and Dp71 isoforms (both *p* < 0.005). Across all models with wheelchair use as the selected outcome, age had an odds ratio greater than 1.75 (*p* < 0.0005), and BMI as well as steroid use were not significant predictors. All area under the receiving operator curve (AUC) values were at least 0.93. When cardiomyopathy status was used as an outcome, only mutations affecting the Dp140 isoform showed a significant relationship, with an odds ratio of 0.3662 (95% CI: 0.14, 0.92; *p* < 0.05) (Table S2). Age gave an odds ratio of at least 1.31 (*p* < 0.0005), with BMI and steroid use not being significant predictors of cardiomyopathy status; AUC values were at least 0.83. Unfortunately, models could not be generated for the other genotype categories, as these groups did not have any patients with cardiomyopathy.

Multiple linear regression analysis with LVEF as the outcome yielded no genotypes as significant predictors (Table S3). Age, steroid use, and use of cardiac medications all yielded significant estimates (β) in the produced regression models (individual R<sup>2</sup> > 0.3). Age and use of cardiac medications gave negative estimates (*p* < 0.0005 and *p* < 0.005, respectively), while steroid use gave positive estimates (*p* < 0.05). On the other hand, when FVC was used as an outcome, mutations in the C-terminal domain gave a significant β in the model at −19.24 (95% CI: −36.56, −1.91; *p* < 0.05). No other genotype categories yielded significant β values. Age and steroid use had significant estimates in all produced models for FVC (individual R<sup>2</sup> values >0.4), with age having negative β values (*p* < 0.0005) and steroid use having positive β values (*p* < 0.005).

#### *3.5. Applicability of Exon Skipping Therapy to DMD Patients in Canada*

A particularly promising approach to treat DMD is exon skipping using small single-stranded nucleic acid analogues called antisense oligonucleotides (AOs). In this strategy, AOs are designed to bind specific splicing enhancer sequences in out-of-frame *DMD* exons by base pairing. This results in the exclusion of targeted exons from the final mRNA transcript, restoring the reading frame and thereby allowing for the synthesis of shorter, partially functional dystrophin proteins [26,27]. With the increasing number of exon skipping therapies entering the clinic and receiving FDA approval, we sought to determine their applicability to DMD patients in Canada. We evaluated the applicability of the top 10 single exon skipping strategies that can treat the most number of patients registered in the global TREAT-NMD DMD database [11], and we also evaluated two multiple exon skipping strategies that target exons within the *DMD* mutation hotspots [18]. Exon 51 skipping treated the most number of DMD patients with deletions at 17%, as well as the most number of DMD patients overall (with deletions, duplications, and small mutations) at 12.3% in our cohort, which was similar to the trend observed worldwide in a previous TREAT-NMD study [11] (Table 3). This was followed by exon 45 skipping at 15.8% of DMD patients with deletions or 11.1% of all DMD patients and then by exon 44 skipping at 12.9% of DMD patients with deletions or 9.4% of all DMD patients. Exon 53 skipping is only the fourth most applicable single exon skipping therapy in our cohort, as opposed to being ranked second among TREAT-NMD DMD patients [11]. For the multiple exon skipping strategies, exons 45–55 skipping was applicable to 66.8% of DMD patients with deletions or 50.9% of all DMD patients in Canada (Table 3). Exons 3–9 skipping was less applicable, at 7.9% of all DMD patients with deletions or 9.1% of all DMD patients.


**Table 3.** Applicability of single and multiple exon skipping strategies to DMD patients in Canada.

<sup>1</sup> Rank information obtained from Bladen et al. (2015) [11].

#### **4. Discussion**

We characterized *DMD* mutation data from DMD/BMD patients registered in the CNDR between 2012 and 2019, with a subsequent analysis of genotype–phenotype correlations. This study partly builds on previous work done by the Canadian Pediatric Neuromuscular Group (CPNG) in 2011, who studied the spectrum of *DMD* mutations in 773 patients across Canada from 2000 to 2009 [16]. We observed a similar abundance of mutation types across patients as the CPNG, with deletions forming the largest group (71% here compared to 64% from the CPNG study), followed by small mutations and duplications (Figure 2a). We found similar *DMD* mutation hotspots, with the exception that the CPNG observed a more extensive duplication hotspot from exons 2–20. In terms of overall genetic characteristics, our findings were largely consistent with those from global database studies (TREAT-NMD, LOVD) [11,18], indicating underlying commonalities in *DMD* gene mutability between patients in Canada and the rest of the world.

Perhaps the most well-known genotype–phenotype correlation in the field concerns the reading frame rule [8]. As in other studies (e.g., [11,16,18,19]) we found exceptions to this rule, with only 87% of DMD patients in our population having out-of-frame deletions and 84% of BMD patients having in-frame deletions (Figure 4a), for a total exception rate of 13%, which was higher than what was observed in the TREAT-NMD and LOVD databases [11,18]. Examining the 36 in-frame deletion patterns in our cohort revealed that deletion location and size matter, particularly if it affects dystrophin protein-binding domains mostly concentrated at the N-terminal end of the protein (Figure 4c, Table S1). In-frame deletions within the rod domain-coding region past exon 45, which do not code for any known protein-binding domains, were mostly associated with BMD. However, the number of impacted binding domains does not completely predict the disease phenotype of in-frame deletions. Consider our in-frame deletions that start on exon 13: exons 13–44 and 13–53 deletions lead to BMD, while the sandwiched exons 13–47 deletion leads to DMD. All three affect the same dystrophin protein-binding domains (Table S1) and yet have varying clinical consequences.

It is possible that regions other than the currently known protein-binding domains may be more critical for dystrophin function. For instance, a previous study looked at 97 patients from the Universal Mutation Database (UMD)-DMD registry with in-frame deletions before exon 35 and suggested that certain protein-binding domains may be dispensable to dystrophin function [28]. Characterizing these other potential critical regions in the *DMD* gene would be essential to understanding patients with mutations not governed by the reading frame rule. These regions can be identified through a combination of extensive patient database study and in vitro validation with patient-derived cells or induced pluripotent stem cell-derived models [29] of patient mutations. The identification of such regions will also benefit the development of gene replacement or correction therapies for DMD [24] to ensure that the dystrophin protein variants used or produced by these approaches are as functionally close as possible to the full-length version.

One concern for in-frame deletions affecting the central rod domain is also whether or not they can preserve its repeating, filamentous structure. Intuitively, in-frame mutations that can maintain this structure would be more likely to lead to BMD. While we observed this to be somewhat true for hybrid repeat-forming deletions, the same surprisingly cannot be said for fractional repeat-forming deletions (Figure 4d). In fact, one study of LOVD patients with in-frame mutations between exons 42 and 57 even found that fractional repeat-forming deletions were more commonly associated with BMD (72% of the time) than DMD [24]. The same study showed that the position of in-frame mutations relative to hinge 3 (exons 50–51) better determines phenotype than the predicted repeat structure formed by the deletions, which is a finding corroborated by another report [30]. This suggests that other parameters should be considered when evaluating the consequences of in-frame mutations on dystrophin structure, such as effects on overall protein flexibility or intra-protein interactions between residues. However, it is important to point out that knowing this information would still not be sufficient to explain certain cases, such as why the same in-frame deletion leads to a mix of DMD and BMD patients (e.g., deletions of exons 45–47, 45–49, 48, 48–49, and 49–51; Figure 4c and Figure S1b). In these cases, genetic modifiers [31,32] or spontaneous exon skipping events (as discussed in the next paragraph) may play a role in determining patient phenotypes.

We also saw a few out-of-frame deletion patients in our cohort to be exceptions to the reading frame rule, particularly those with deletions in exons 3–6, 3–7, 3–21, 7–8, 42–43, and 43 (Figure S1b). Two mechanisms have been proposed to explain such exceptions. The first is the use of alternative translational start sites further downstream in the *DMD* transcript [33–35]. For instance, a series of immunofluorescence experiments performed on skeletal muscle biopsies from exons 3 to 7 deletion patients suggested that there was a potential alternative initiation codon in exon 8 [34]. Dystrophin was not detectable when antibodies recognizing the 5 end of exon 8 in the protein were used; however, dystrophin was detected using antibodies recognizing the 3 end of exon 8. This may explain why a deletion of exons 3–7 is typically associated with BMD or with milder DMD phenotypes [18,35,36]. The second mechanism is the occurrence of spontaneous exon skipping events that convert out-of-frame into in-frame mutations. A well-documented example is the spontaneous skipping of exon 44 that occurs when the exons flanking it are deleted [37,38]. In fact, exon 44-skippable deletions are usually associated with a higher number of dystrophin-revertant fibers and milder DMD phenotypes such as prolonged ambulation [36,39–42]. In addition, of the six out-of-frame deletions that we have listed as exceptions, five of them can be converted into in-frame deletions with the skipping of just one exon adjacent to the deletion. This spontaneous exon skipping may be tied to how the junction sequences formed by a deletion influences splicing, i.e., if it creates or destroys exon splicing silencer/enhancer sequences [37]. Further study into this phenomenon may also provide hints regarding the formation of dystrophin-revertant fibers.

As for correlations between genotypes and clinical outcomes, it is important to emphasize that the regression analysis performed here produces an inferential model, i.e., a model that best describes the study population at its current state. There were a number of limitations with the study population as it is now that may have affected the analysis, mostly concerning low sample sizes for each mutation pattern observed and incomplete availability of clinical outcome data for all patients. The majority of DMD patients analyzed were within the younger range as well (Table 1), and so there may be some bias in the observed phenotypes. For practical reasons, we also limited our analysis to genotypes classified according to the protein domain or the dystrophin isoform affected by the respective patient mutations. We acknowledge that use of other stratification procedures may lead to differing conclusions.

With these in mind, we saw an increased likelihood of wheelchair use associated with mutations affecting the rod domain and, conversely, a decreased likelihood with mutations affecting the C-terminal domain and Dp116/71 isoforms in our DMD patient population (Table S2). It is interesting that a positive association with rod domain mutations was observed. Previous reports have shown that certain rod domain-coding mutations are associated with prolonged ambulation in DMD patients, e.g., exon 44-skippable deletions [36,40–42]. Once a sufficient number of patients are available, it may be worthwhile to further stratify rod domain mutations to pinpoint the importance of specific sub-regions. The finding regarding the C-terminal domain is striking, since one would expect it to be critical in localizing dystrophin to the muscle membrane [7]; note that C-terminal domain mutations were also significantly, positively correlated with FVC in our DMD patient cohort (Table S3). Interestingly, there has been a previous case of an 8-year-old boy reported to be asymptomatic despite having a nonsense mutation truncating the C-terminal domain [43]. Microdystrophins lacking most or all of the C-terminal domain have also been promising in *mdx* mice with improvements in skeletal and cardiac muscle phenotypes [44–46]. Our results complement such findings, inviting closer investigation into the importance of the C-terminal domain for dystrophin function in muscle. However, it is also important to note that our result is based on a small number of patients with C-terminal domain mutations (*n* = 10), and so further validation by conducting a regression analysis with a larger sample size is recommended.

The association of Dp116 and Dp71 with motor function was likewise unexpected, as these isoforms are not normally expressed in differentiated skeletal muscle. Dp116 is exclusively expressed in Schwann cells [47], and Dp71 displays mostly ubiquitous expression but is difficult to detect in differentiated skeletal muscle [48,49]. While some reports are now claiming otherwise [50,51], i.e., that these isoforms are in fact expressed in muscle (one study is described in the next section for Dp116), their functional significance in muscle remains unknown. As for other factors included in the model for wheelchair use, it was surprising that steroid use did not have a significant impact, contrary to a previous TREAT-NMD DMD registry report [17]. However, this observation may be restricted to the particular demographic of the population under study.

Mutations affecting Dp140 was the only genotype group determined to be a significant predictor of cardiomyopathy (Table S2); no significant genotypes were found as predictors for LVEF (Table S3). Dp140 is a non-muscle dystrophin isoform typically expressed in the central nervous system and the kidneys [52]; its expression in the heart (or skeletal muscles) has not yet been demonstrated. Based on our analysis, Dp140 mutations are apparently associated with the lack of cardiomyopathy. One group previously studied the relationship between cardiac dysfunction (LVEF <53%)-free survival and dystrophin isoform mutations, but they did not find any significant association with respect to Dp140 [51]. Instead, the authors observed that Dp116 mutations were significantly linked to better rates of cardiac dysfunction-free survival, which we did not see in our analysis. Note that Dp116 was thought to be a non-muscle dystrophin isoform; however, this study demonstrated that Dp116 mRNA expression was detectable in both human cardiac and skeletal muscle samples. Therefore, it remains possible that Dp140 may have a role in the heart, but this will have to be supported first by in vivo validation of cardiac Dp140 expression similar to what was done for Dp116 in the study above, and then by further confirmation of our result in other patient registries. Considering other factors in our model, steroid use was not a significant predictor for cardiomyopathy, but it was significantly, positively correlated with LVEF. This may be explained in part by the fact that our DMD patient cohort is relatively young and not well-suited for observing cardiac symptoms that manifest relatively late in the disease. Cardiac medications were significantly, negatively correlated with LVEF, but they may reflect the bias that patients with reduced LVEF are typically the ones receiving such treatments—the factor was included more as a control for other predictors.

There are other reports of genotype–phenotype correlations with respect to cardiac outcomes in the literature, with proximal/N-terminal mutations generally associated with worse cardiac symptoms than distal/C-terminal mutations [21,53–55]. Still, some studies demonstrated a lack of correlation altogether [21,56,57]. This issue of non-agreement across genotype–phenotype correlation studies is not only true for cardiac outcomes but also for skeletal muscle phenotypes. This clearly indicates the need for further work in this area, starting perhaps by standardizing data collection procedures to maximize comparability across patient registries as well as the amount of information obtained from each patient.

Within the last five years, we have seen the approval of three exon skipping AOs for DMD therapy by the FDA: eteplirsen (brand name Exondys 51, Sarepta) for skipping exon 51 in 2016 [26], and golodirsen (Vyondys 53, Sarepta) in 2019 [58] as well as viltolarsen (Viltepso, NS Pharma) in 2020 [59] for skipping exon 53; another AO, the exon 45-skipping casimersen (SRP-4045, Sarepta) is currently under FDA review. These FDA-approved therapies can treat a combined 26.5% of DMD patients with deletions or 19.3% of all DMD patients in Canada (Table 3), which is incredibly encouraging. Notably, the applicability of single exon skipping strategies was different for patients in Canada compared to global estimates from the TREAT-NMD DMD database [11], suggesting potential implications for future clinical trials. These findings highlight one of the major limitations associated with personalized therapies such as exon skipping, i.e., low patient applicability. One way to overcome this would be to develop multi-exon skipping strategies such as exons 45–55 skipping, which could treat more than half of all DMD patients (Table 3). Our data and those from other patient registries [18,60] also show that exons 45–55 deletions are commonly associated with mild BMD or asymptomatic phenotypes (Figure 4c and Figure S1b), confirming the viability of the approach as a treatment for DMD.

This last point raises a concern for other exon skipping strategies, i.e., if the in-frame-skipped dystrophin proteins they produce are indeed functional or associated with mild phenotypes. We have seen how some deletions lead to a DMD phenotype despite being in-frame, e.g., in our population, 42% of in-frame deletions were in DMD patients (Figure 4b). Encouragingly, the majority of patients with deletions equivalent to exon 51-skipped transcripts showed mild phenotypes [61], bearing well for eteplirsen. Therefore, consulting patient registries such as the CNDR when designing exon skipping strategies is recommended. Finally, despite the promise of exon skipping therapy, it cannot correct all mutations, and there remain concerns regarding its efficacy in patients. The continued development of other therapeutic approaches such as gene replacement with mini/microdystrophins or gene correction with genome editing strategies, as informed by genotype–phenotype correlation studies from patient registries, remains critically important.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2075-4426/10/4/241/s1, Figure S1: Summary of large *DMD* gene deletions, Figure S2: Summary of large *DMD* gene duplications, Table S1: In-frame deletions and their effects on dystrophin protein-binding domains, Table S2: Multiple logistic regression analysis for wheelchair use and cardiomyopathy status, Table S3: Multiple linear regression analysis for left ventricle ejection fraction (LVEF) and forced vital capacity (FVC).

**Author Contributions:** Conceptualization, K.R.Q.L. and Q.N.; methodology, K.R.Q.L., Q.N., and T.Y.; investigation, K.R.Q.L. and Q.N.; writing—original draft preparation, K.R.Q.L. and Q.N.; writing—review and editing, K.R.Q.L., Q.N., and T.Y.; supervision, T.Y.; project administration, K.R.Q.L., T.Y.; funding acquisition, T.Y. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Friends of Garrett Cumming Research & Muscular Dystrophy Canada HM Toupin Neurological Science Research Chair, Canadian Institutes of Health Research (CIHR) FDN 143251, 169193, and the Women and Children's Health Research Institute (WCHRI) IG 2874.

**Acknowledgments:** We would like to thank the CNDR Investigator Network for collecting, compiling, and providing the patient data used in this study. We would also like to thank Matthew Pietrosanu, a statistical consultant at the Training and Consulting Centre (TCC) in the Department of Mathematical and Statistical Sciences, University of Alberta, for his advice on the statistical analysis performed in this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


61. Waldrop, M.A.; Ben Yaou, R.; Lucas, K.K.; Martin, A.S.; O'Rourke, E.; Ferlini, A.; Muntoni, F.; Leturcq, F.; Tuffery-Giraud, S.; Weiss, R.B.; et al. Clinical Phenotypes of DMD Exon 51 Skip Equivalent Deletions: A Systematic Review. *J. Neuromuscul. Dis.* **2020**, *7*, 217–229.

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Review*

## **Advances in Genetic Characterization and Genotype–Phenotype Correlation of Duchenne and Becker Muscular Dystrophy in the Personalized Medicine Era**

### **Omar Sheikh <sup>1</sup> and Toshifumi Yokota 1,2,\***


Received: 10 August 2020; Accepted: 1 September 2020; Published: 3 September 2020

**Abstract:** Currently, Duchenne muscular dystrophy (DMD) and the related condition Becker muscular dystrophy (BMD) can be usually diagnosed using physical examination and genetic testing. While BMD features partially functional dystrophin protein due to in-frame mutations, DMD largely features no dystrophin production because of out-of-frame mutations. However, BMD can feature a range of phenotypes from mild to borderline DMD, indicating a complex genotype–phenotype relationship. Despite two mutational hot spots in dystrophin, mutations can arise across the gene. The use of multiplex ligation amplification (MLPA) can easily assess the copy number of all exons, while next-generation sequencing (NGS) can uncover novel or confirm hard-to-detect mutations. Exon-skipping therapy, which targets specific regions of the dystrophin gene based on a patient's mutation, is an especially prominent example of personalized medicine for DMD. To maximize the benefit of exon-skipping therapies, accurate genetic diagnosis and characterization including genotype–phenotype correlation studies are becoming increasingly important. In this article, we present the recent progress in the collection of mutational data and optimization of exon-skipping therapy for DMD/BMD.

**Keywords:** Duchenne muscular dystrophy (DMD); exon-skipping therapies; next-generation sequencing (NGS); Sanger sequencing; multiplex ligation probe amplification (MLPA); multiplex polymerase chain reaction (PCR); comparative genomic hybridization array (CGH); viltolarsen; eteplirsen; golodirsen

#### **1. Introduction**

Duchenne muscular dystrophy (DMD), a severe neuromuscular disorder, affects the skeletal and cardiac muscle of 1 in 5000 newborn boys [1], with very few treatment options available [2]. Frame-shifting mutations in the dystrophin gene [3] cause DMD by removing production of the 427 kDa protein dystrophin [4]. Without dystrophin, progressive muscle wasting occurs [5]. By contrast, in-frame dystrophin deletion mutations lead to the related condition Becker muscular dystrophy (BMD), which ranges in phenotype from subclinical to borderline DMD [6]. For this reason, the term DBMD is used to indicate the range of conditions that arise from dystrophin mutations. Though most mutations reported fit the dystrophin reading frame rule stated above, there are a significant number of exceptions, highlighting an intricate genotype–phenotype relationship [7]. Given the range of mutations underlying DBMD, precise genetic diagnosis and genotype–phenotype correlation analysis are crucial to design mutation-specific therapeutics like exon skipping [8,9]. Box 1 describes the

keywords used in this article. A better understanding of genotype–phenotype relationships in DBMD patients may lead to better design of exon-skipping therapies [9]. In this review, we describe the recent advances in molecular diagnostic approaches for DMD/BMD and discuss how exon-skipping therapy can be optimized.

**Box 1.** Definitions of keywords used in this article.

**Genotype**: genes that encode physical characteristics of an organism.

**Phenotype**: the observed characteristics resulting from the expression of those genes.

**Intron**: non-coding region of DNA that is removed by splicing prior to translation.

**Exon**: coding region of gene that appears in the mature RNA transcript.

**In-frame mutation**: a mutation that does not disrupt the reading frame of a gene during the transcription, likely not interfering with protein production.

**Out-of-frame mutation** (also known as frameshift mutation): a mutation that disrupts the reading frame, likely destroying protein production.

#### **2. Sequencing and Genetic Diagnosis Methodologies Relevant to DBMD**

Sequencing and mutation detection strategies are intertwined in DBMD. The key strategies used to study this disorder are listed below in Table 1.

**Table 1.** Sequencing and genetic diagnosis methodologies relevant to Duchenne/Becker muscular dystrophy (DBMD).


Sequencing methodologies provide precise genetic testing that can clarify the mutations seen in patients. Sanger sequencing is performed using nucleotides that lack a 3 -hydroxyl group, preventing the DNA polymerase from continuing the DNA chain at that position [17]. Though low throughput, it can complete partial sequencing cheaply.

Southern blot was originally used to examine DMD mutations before other techniques replaced it. Southern blot analysis using cDNA probes, which were established earlier [18], has been used to detect deletions and duplications of the dystrophin gene [19–21]. Southern blotting, however, is no longer commonly employed for DMD since it is time-consuming and requires several hybridization steps [11].

The use of multiplex polymerase chain reaction (PCR) for mutation detection has played a more prominent role in DMD genetic diagnosis [13]. Multiplex PCR, which allows for rapid detection of mutations using small or suboptimal samples of genomic DNA, is more efficient than Southern blotting [13]. One study indicated that the majority of the deletions detected by use of cDNA probes and Southern blot in the study could have been also characterized by multiplex PCR [20]. In 2006, Stockley et al. established the use of quantitative multiplex PCR to screen all 79 exons for deletions and duplications [15], strengthening the technique's applicability in DMD.

Multiplex ligation-dependent probe amplification (MLPA) acts well as a first-pass assessment of DMD due to its speed and cheap cost [22]. This technique detects exon deletions and duplications. An MLPA probe consists of two probe oligonucleotides that hybridize to adjacent sites of the target sequence, followed by probe ligation. Probes hybridized are amplified by PCR and quantified, providing amplification products of unique size. The MLPA approach then provides the relative copy number of target sequences [22], which can detect most of the deletions and duplications in the DMD gene.

One rising alternative to MLPA is comparative genomic hybridization (CGH) array [16]. Since 2004, this approach has marked a new milestone for genetic diagnosis [23]. CGH is performed by using probes covering dystrophin exons and introns conjugated to a glass slide. Control and patient DNA is fragmented and hybridized to the probes, allowing for the detection of the relative abundance of each exon. However, unlike MLPA, it can pinpoint the location of breakpoints within introns [16]. This method can be applied to screen the genome both at the whole-gene level and the individual exon level for many disease genes including DMD [24]. The CGH platform can detect precise intron breakpoints in high resolution and sensitivity [25] while also being completely scalable [26]. Through the use of CGH, the ability to capture intronic mutations is notably improved [27]. Due to the high resolution of CGH [28], this technique has been used to probe intronic mutations in dystrophin using patient data [29,30]. Therefore, CGH is also a recommended technique used first to look at DBMD genetics.

Next-generation sequencing (NGS), which refers to sequencing strategies featuring a much greater sequencing volume than Sanger sequencing [10], is another prominent strategy relevant to DBMD [11] and can be used alongside other strategies such as MLPA to provide a reliable genetic diagnosis. Overall, targeted NGS can bolster a more precise understanding of ambiguous mutations [12] in contrast to MLPA which cannot identify some dystrophin mutations [11].

NGS features several potential diagnostic uses. For instance, NGS can accurately identify pathogenic small mutations in DBMD patients without a large deletion/duplication, especially in non-coding regions [31]. Therefore, this technique has great potential to improve the molecular diagnosis of DBMD. Lastly, whole-exome sequencing, which solely concentrates on the coding exon regions of the genome, is useful for the quick examination of exonic mutations [32]. Though this technique is not widely used, it has been used to identify small mutations giving rise to DBMD [33–35]. The broad range of NGS methodologies available supports precise genetic diagnoses [12].

#### **3. Exon-Skipping Therapies for DMD**

Exon-skipping therapy is based on the observation that not all of the 79 dystrophin exons are essential for functional protein [36]. Patients with in-frame deletions typically feature a milder BMD phenotype, despite not having all exons, which forms the basis for the approach of exon skipping [37]. Synthetic antisense oligonucleotides (AONs), which are engineered to resist nuclease degradation, are typically used to target mRNA of the dystrophin gene for removal, thereby restoring the reading frame and promoting the production of partially functional protein [36]. This truncated protein then compensates for the function of the full-length protein. Currently, many exon-skipping therapies are in clinical testing [38]. Thus far, exon skipping has shown effectiveness in delaying DMD progression [36]. Eteplirsen, which is designed to skip exon 51 [39], and golodirsen, which is designed to skip exon 53 [40], gained conditional approval in the US in 2016 and 2019, respectively. A newly approved AON, viltolarsen, has been especially promising. Based on compelling evidence of efficacy, viltolarsen received approval in Japan for the skipping of exon 53 [41] and was conditionally approved

by the U.S. Food and Drug Administration (FDA) in August 2020 [42]. A Phase II trial of viltolarsen demonstrated that the low dose group (40 mg/kg) rose from an average dystrophin production baseline of 0.3% to 5.7% of normal while the high dose group rose from an average dystrophin production baseline of 0.6% to 5.9% of normal in Western blots [43]. In parallel, the trial strengthened the evidence that viltolarsen can stabilize or improve muscle strength and functionality based on timed tests. Of the three approved therapies, which are compared in Table 2, viltolarsen has produced the highest observed increases in dystrophin production.

**Table 2.** Comparison of FDA-approved exon-skipping therapies for Duchenne muscular dystrophy (DMD). Mean dystrophin protein production (as a percentage), relative to healthy controls, is presented based on Western blot data. Baseline values are included for reference.


For this therapy to effectively treat patients, it must produce a stable dystrophin protein. In one study, researchers examined the stability of edited in-frame dystrophins lacking exons 45–53, exons 46–54, and exon 47–55, respectively; the edited protein lacking exons 46–54 featured the greatest stability [47]. Though this study provides biochemical and computational prediction of exon-skipping therapies, it does not demonstrate these results in vivo [9]. Nevertheless, exon-skipping schemes can cause a myriad of consequences at the protein structure level, which could influence therapeutic effectiveness.

In DMD, exon skipping is still challenged by its mutation-specific nature. Such therapies could be spread too thinly across many different mutations even though it can potentially treat many patients in total. For example, though 47% and 90% of nonsense mutations could be treated using single and double exon-skipping, respectively, this therapy development could necessitate targeting 68 of dystrophin's 79 exons [48]. Although technically more challenging, double exon skipping substantially raising the applicability of exon-skipping therapies compared to single exon skipping highlights the power of skipping more than one exon. In a dystrophic dog model, double exon skipping of DMD exons six and eight induced by cocktail AONs resulted in the systemic correction of the reading frame and truncated dystrophin expression in skeletal muscles accompanied by improved running speed [49]. The potential of multi-exon skipping is supported by the milder BMD phenotypes observed with the absence of exons 45–55 [50]. In particular, these patients largely featured no mortality and delayed loss of ambulation [51]. Multi-exon skipping of exons 45–55 is expected to benefit 47% of DMD patients [51]. In a DMD mouse model with a deletion mutation in exon 52, exons 45–55 skipping was induced by cocktail AONs, leading to systemic dystrophin expression and functional rescue [52]. Overall, successful development of multi-exon skipping will significantly expand the applicability and optimize the function and stability of truncated dystrophin.

#### **4. Patient Registries and the Personalization of Exon Skipping**

To better understand which patients are amenable to mutation-specific therapies, including exon-skipping, patient data must be collected broadly through studies and registries. In a foundational study, Baumbach et al. observed that 56% of DMD patients have detectable deletions, 29% of which mapped to a region proximal to the 5' end of the gene whereas 69% mapped to a region located centrally [53]. The Leiden patient registry reflects one major collection of data on the genetics of DBMD [7]. A large-scale study on the UMD-DMD registry from 2008 was performed on 2405 French patients with DBMD [54]. DMD patients featured 61% large deletions and 13% duplications whereas BMD patients featured 81% large deletions and 6% duplications. Comparatively, this indicates a similar deletion rate to Baumbach et al. Furthermore, this database study indicated that 24% of mutations are de novo events, reinforcing the relatively frequent occurrence of mutations in the dystrophin gene. Finally, this large-scale approach to genotype–phenotype in analysis coincides with the development of other international DMD patients' registries [54].

TREAT-NMD, an EU-funded multinational network, aims to establish comprehensive information on the natural history of DMD by acquiring data from a large number of patients from a variety of countries not limited to Europe [55]. Currently, the TREAT-NMD database contains a lot of mutational data [56], though as of 2015 15% and 57% of mutations submitted to the registry were from the Americas and Europe, respectively. In parallel, researchers across many countries are collecting mutational data on DBMD patients across the world. These efforts supplement consolidation of patient data into a global registry like TREAT-NMD [57–67].

In 2015, TREAT-NMD's global database was used to assess more than 7000 dystrophin mutations [56]. Among large mutations, which comprise 80% of total mutations, 86% are deletions and 14% are duplications. This study, beyond providing an overview of mutations observed in a global group of DMD patients, also concludes that the skipping of exons 51 (14% of patients), 45 (9% of patients), 53 (8.1% of patients), and 44 (7.6% of patients) could apply to significant minorities of the registry's patients.

Inspired by the TREAT-NMD global registry, Japan established its own registry called Remudy. In a 2013 study examining 688 DBMD patients, the deletion of exons was most frequent followed by point mutations and duplications [68,69]. The most recent published analysis of Remudy concluded, based on a set of 1197 Japanese DMD patients, that 107 patients could benefit from exon 51 skipping while 111 could benefit from exon 53 skipping [70].

#### **5. Genotype–Phenotype Correlation Studies to Predict the Likely Outcomes of Exon-Skipping Therapies**

Through documenting the genotype–phenotype relationship, researchers may better design mutation-specific therapies such as exon-skipping. A greater understanding of genotype–phenotype relationships has been supported by data from clinical studies. Although the reading-frame rule holds in approximately 90% of DBMD cases [7], there are important exceptions. A 2007 review pooled DBMD patient data, based on MLPA, Southern blotting, or PCR analysis, concluded that in-frame deletion patterns result in a mixture of DMD and BMD phenotypes [71]. The deletion of exons 45–47, for instance, featured a 15% occurrence of DMD whereas the deletion of exons 45–51 featured a 48% occurrence of DMD (13 out of 27 patients).

Assessing the genotype–phenotype relationship in a subset of DMD patients might more directly indicate the merits of potential exon-skipping therapies. The 5' region of the gene, which includes exons 3–9, may be associated with complex genotype–phenotype correlations [72]. In one case study, a patient with an in-frame deletion of exon five featured a more severe than expected BMD phenotype despite the continued recognition of exon six [73]. By contrast, an in-frame deletion of exons 3–9, according to one study, mostly leads to a BMD phenotype [74,75]. The two closely examined patients featured especially mild BMD with only mild heart impairment. In addition, Nakamura et al. reported a patient with this deletion showing only a slight decrease in cardiac function but without muscle involvement at the age of 27 years. By examining this deletion in vivo, the researchers concluded that the removal of exons 3–9 via multi-exon skipping likely generates a mild BMD phenotype. Based on these observations, removal of exons 3–9 is a promising treatment for DMD patients with mutations in this region.

Findlay et al. examined 41 patients enrolled in the United Dystrophinopathy Project focusing on in-frame deletions around exon 45 [8]. All patients with Δ45–46 deletions (n = 4) carried a diagnosis of DMD whereas most patients with Δ45–47 deletions (n = 17) and Δ45–48 deletions (n = 19) were diagnosed with BMD. Based on these findings, the skipping of exon 46 for patients missing exon 45 may not rescue the DMD phenotype. Instead, the study illustrates how the skipping of exons 46–47 or 46–48 for these patients has a greater likelihood of producing a BMD phenotype. As a result of this cohort study, a clinical case can be made for multi-exon skipping, which remains in preclinical testing [76]. From this example, we can see how genotype–phenotype correlations can support the design of exon-skipping therapies, improving their personalization.

A systematic review of dystrophinopathy data from the published literature and unpublished databases examined 135 DBMD patients with in-frame deletions equivalent to the skipping of exon 51 [77]. Of these patients, the majority (n = 81) had BMD whereas 16 patients had more severe phenotypes and 6 had no definitive phenotype. The authors conclude that exon 51 skipping therapy, overall, is likely to produce milder BMD phenotypes in many patients.

To understand the genotype–phenotype relationships of in-frame deletions within the exons 45–55 mutational hot spot, 43 patients with DBMD patients were examined using MLPA, Southern blotting, and multiplex PCR [51]. The deletions examined are as follows: Δ45–55 (n = 7), Δ45–51 (n = 6), Δ45–48 (n = 5), Δ45–57 (n = 3). Researchers subdivided these groups into two groups based on truncated dystrophin conformation: hybrid type (Δ45–55, Δ45–58, Δ45–51) and fractional type (Δ45–57 and Δ45–49). Hybrid type conformation (n = 18) at large features a lower proportion of wheelchair-bound patients than the fractional type conformation (n = 6). Log-rank tests revealed a statistically significant difference between the hybrid and fractional groups (*p* < 0.05) of the age at which patients became wheelchair-bound. In other words, the fractional type appears to more consistently lead to an earlier loss of ambulation. This study provides another manner of predicting the viability of dystrophin protein produced by exon-skipping.

Larger studies of in-frame deletions can more strongly guide exon-skipping development [71]. Looking at in-frame deletions within the hotspot region, researchers determined that some mutations were unexpectedly severe, leading to a DMD phenotype rather than the expected BMD phenotype. For example, in-frame deletions starting from exon 49 and exon 50 featured 92% DMD and 90% DMD proportions, respectively, reinforcing the fact that not all potential exon-skipping strategies will resolve a severe phenotype.

Genotype–phenotype correlations of in-frame deletions also support multi-exon-skipping therapies, especially removing exons 45–55. In three patients each featuring in-frame deletion of the region, two developed heart failure while featuring no overt skeletal pathology whereas a third patient featured muscle atrophy and weakness [78]. The condition of all remained stable with treatment. A separate study examined nine patients with the same mutation and indicated that all nine patients had quadriceps and calf hypertrophy and no respiratory involvement. Meanwhile, two patients featured dilated cardiomyopathy [79]. These results suggest, like with the previous study, that the deletion of exons 45–55 is associated with a milder condition compared to smaller in-frame deletions in this region. A multi-exon-skipping strategy can recapture this phenotype by removing several exons, rather than simply skipping every exon in the region individually, and potentially treat over 65% of DMD patients featuring deletions [80].

#### **6. Conclusions**

Through comprehensive registries of patient data such as TREAT-NMD with the support of newly available genetic diagnosis tools, DBMD patients can be classified based on mutations, which will further help optimize therapy design while offering higher power for clinical trials [55]. Concurrently, the emergence of multi-exon skipping raises the overall applicability of this treatment strategy, although it is technically more challenging. For exon-skipping therapies to be as effective as possible, cohort studies of genotype–phenotype relationships in DBMD patients with the same resulting

mutation would support their design [9]. Because BMD can feature a plethora of truncated dystrophins, exon skipping resulting in truncated dystrophins linked to a milder BMD phenotype might be more beneficial. However, caution should be taken in interpreting these data as other factors, such as the variability of exon skipping efficacy among different exons, also need to be taken into account. Nevertheless, larger cohort studies utilizing patient registry data on genotype–phenotype correlation would greatly contribute to the rational design of mutation-specific therapies including exon skipping in the personalized medicine era.

**Author Contributions:** Literature review and writing—original draft preparation, O.S.; writing—review and editing, O.S., and T.Y.; supervision and funding acquisition, T.Y. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Friends of Garrett Cumming Research & Muscular Dystrophy Canada HM Toupin Neurological Science Research Chair, Canadian Institutes of Health Research (CIHR) FDN 143251, 169193, Fulbright Canada, and the Women and Children's Health Research Institute (WCHRI) IG 2874.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
