Duchenne muscular dystrophy (DMD) is an X-linked condition caused by deficiency of functional dystrophin protein. Dystrophin deficiency causes increased susceptibility of the sarcolemma to mechanical stress and results in high intracellular calcium levels, mitochondrial dysfunction, and increased production of reactive oxygen species that lead to a cycle of loss of muscle fibres, chronic inflammation, and decreased muscle regenerative capacity. Ultimately, muscle fibres are replaced by connective and adipose tissue resulting in muscle dysfunction. Patients experience progressive weakness in skeletal muscles and cardiomyopathy. Several isoforms of dystrophins are expressed in the central nervous system, and a large proportion of patients have cognitive involvement ranging from mild attention deficit to severe autism (Doorenweerd, 2020). DMD patients lose ambulation and have a shortened life expectancy. For a detailed review of DMD pathophysiology, we refer readers to the following reviews (Duan et al., 2021; Starosta and Konieczny, 2021). Standards of care includes steroids, multidisciplinary approaches and depending on the country, ataluren for nonsense mutation or exon skipping therapy for patients with eligible mutations. Standards of care and steroids extend life expectancy and the quality of life for patients with DMD (Birnkrant et al., 2018a; Birnkrant et al., 2018b; Birnkrant et al., 2018c); however, there is an unmet need with regards to treatment.

In the past decades, several compounds have been tested in preclinical and clinical studies for their abilities to restore functional dystrophin levels in the muscle cells or to alter the fate of downstream molecular pathways involved in DMD pathophysiology. Breakthroughs in biomedical research have led to the development of ASOs and gene therapy treatments that increase the levels of dystrophin. Except for the US Food and Drug Administration (FDA)-approved steroid for use in DMD (deflazacort), four other antisense oligonucleotides (ASOs) for exon skipping (eteplirsen, golodirsen, casimersen, viltolarsen) have received approval by the FDA through the accelerated approval pathway; viltolarsen has additionally received approval in Japan [ ¹ (Aartsma-Rus and Krieg, 2017; Aartsma-Rus and Corey, 2020; Dhillon, 2020; Shirley, 2021)]. One more compound, the small molecule for stop codon readthrough namely ataluren, has received conditional approval by the European Medicines Association (EMA), but not the FDA (Ryan, 2014). However, the clinical development of several compounds with promising early-phase results has been halted due to disappointing late-phase results, tempering expectations in the DMD community.

The difficulty in translating results from animal models to humans, the heterogeneity of the DMD genotypes and symptoms such as cognitive involvement, the rarity of the disease, the lack of efficacious outcome measures or fully validated biomarkers are some of the most important reasons for the high attrition rate between early- and late-phase clinical development. In this article, we review data on the compounds that have failed in clinical development with published results within the last 5 years. Additionally, we present our perspective on the reasons for the high attrition rate in DMD clinical development and propose solutions.

To identify compounds for which clinical development in the DMD population has been terminated, we performed a comprehensive review of four databases (Ovid MEDLINE, Embase, Scopus, and Cochrane) in English and French, restricted for the last 5 years (2016–2021). For a full list of the keywords used for the different databases, please refer to Supplementary Material. We selected full-text peer-reviewed publications of clinical studies of DMD (+/− Becker muscular dystrophy) patients of pharmacological interventions only. We made a list of the different compounds identified in the selected papers of our research strategy and we checked the status of their development online and on ClinicalTrials.gov.

Additionally, we searched ClinicalTrials.gov using the term “Duchenne muscular dystrophy” and filters “terminated” OR “suspended” OR “withdrawn” with no time limitation. Studies with unknown status were not included and we selected only studies of a pharmacological intervention intended to treat DMD (+/− Becker muscular dystrophy). In total, we identified 10 compounds from the literature review and another six from the ClinicalTrials.gov search whose development has been discontinued.

Despite almost three decades of preclinical and clinical research, no treatment other than steroids has formally demonstrated clinical efficacy in DMD. The attrition rate between early-phase and late-phase development is high (Dovari et al., 2021). Several developments have been discontinued due to inefficacy despite early-phase positive results or signals of efficacy in secondary outcomes. The pressure by investors and by stakeholders to present and publish positive results has probably contributed to an inflation of confirmatory trials with compounds that are unlikely to gain approval. Currently, several compounds are in clinical development for DMD, with five of them having received approval subject to further proof of efficacy. There is a clear need for better clinical trial design and more reliable outcome measures. Reasons for failures and suggested solutions are summarised in Table 2.

More extensive preclinical studies could contribute to more effective clinical trials; however, current preclinical evaluation modes are not optimal. The commonly used mdx mouse model has significant differences from patients affected by DMD, mainly because an accelerated muscle regeneration rate in mice results in milder symptoms (Egorova et al., 2021). The pathophysiology of the disease may also differ. For example, myostatin levels are higher in mdx mice. This likely resulted in differences in the efficacy of myostatin inhibitors in mdx mice and DMD patients. Additionally, confounding factors encountered in clinical trials such as the use of steroids are not an issue in preclinical studies. The recently reported dystrophin-deficient rat could allow more rigorous preclinical studies in a more relevant model and a larger number of animals, which would help tackle the issue of inter-animal variability (Larcher et al., 2014; Nakamura et al., 2015).

The variability of DMD genotypes results in a wide range of phenotypes. For example, in patients with certain point mutations, alternate splicing can lead to some production of functional dystrophin and consequently less severe symptoms. Other genetic modifiers, such as LTBP4, may also contribute to phenotypic variability (Barp et al., 2015; Hightower and Alexander, 2018; Bello et al., 2020). Furthermore, differences in standards of care may result in significant differences in the disease trajectory. The age at which steroids are started, the type of steroids used (deflazacort vs. prednisone) as well as the steroid regimen (e.g., daily vs. alternate days vs. 10 days on -10 days off) also influence disease trajectory (Schreiber et al., 2016; de la Peña et al., 2017; Pradhan and Das, 2018). Additionally, the disease follows a non-linear trajectory. Loss of ambulation occurs between 6 and 18 years of age, which leads to significant variability in phenotype around the age of 7, the typical age of a patient included in a clinical trial. In children between 4 and 7 years, motor function outcomes used in clinical trials, such as the NSAA score, might show development-associated improvement before declining due to the natural course of the disease (Goemans et al., 2013; Mcdonald et al., 2013; Ciafaloni et al., 2016; Ricotti et al., 2016). For another example, respiratory outcomes may improve due to the normal growth of patients leading to increased lung volumes. Furthermore, it is highly important to normalise data for both age and height in this population, as some DMD patients have significantly stunted growth, mainly due to chronic steroid treatment (Hogrel et al., 2020). With regards to clinical trial design, the need for larger samples and longer duration is hampered by the rarity of the disease. Additionally, the regulatory pressure for the use of placebo arms contradicts the wishes of the community (Peay et al., 2018). The stratification of patients by disease trajectory (Muntoni et al., 2019b) to better define inclusion criteria, the use of natural history data to enrich placebo arms (Bello et al., 2016; Goemans et al., 2020a; Mercuri et al., 2020; Osorio et al., 2020), the sharing of placebo data via platforms (Bello et al., 2016; Goemans et al., 2020a), the use of innovative and sensitive outcome measures (Seferian et al., 2015; Lilien et al., 2019), or the use of Bayesian modelling could all result in more efficient and informative clinical trials (Fouarge et al., 2021).

There is a significant unmet need with regards to outcome measures used in clinical trials; different endpoints are meaningful depending on the age and disease stage of DMD patients. The use of surrogate endpoints, such as dystrophin expression for the compounds which induce dystrophin production, require further investigation, as the extent to which these measures predict efficacy is unknown (Aartsma-Rus and Arechavala-Gomeza, 2018). The optimisation of dystrophin quantification methods with the aim to improve accuracy and reproducibility will provide an undeniable pharmacodynamic endpoint for clinical trials. Towards this step, the harmonisation of protocols of the most used assays (e.g., immunochemistry, western blot) for the quantification and localisation of dystrophin will help minimise intra- and inter-lab variability. Several steps are currently being taken towards the optimisation of current assays and the use of new technologies which will help the reliable and accurate dystrophin quantification (Wells, 2019). However, there are different arguments for the clinical relevance of low levels of dystrophin; some patients who produce low dystrophin levels from birth show a milder phenotype, but it is still under investigation if low levels can have the same result after years of disease evolution (de Feraudy et al., 2021). Additionally, dystrophin levels can be variable in healthy individuals (Aartsma-Rus et al., 2019).

The FDA has granted accelerated approval based on truncated dystrophin expression as a surrogate endpoint, whereas the EMA has been consistently requiring functional data (Aartsma-Rus and Arechavala-Gomeza, 2018). This more conservative attitude translates into a clear controversy between the two agencies when it comes to drug approvals. Given the burden of the intravenous infusion, the risk of infections associated with the use of port-a-cath, the potential side effects as well as the cost of such drugs, the EMA position seems reasonable. Nevertheless, the difference in the position of the two agencies complicates clinical development, which must integrate both the possibility of accelerated approval and the need to prove efficacy by strong clinical data -sometimes even when this drug is commercially distributed in the US-. This is currently the case for exon skipping ASOs, like casimersen, golodirsen and viltolarsen, which are commercially distributed in the US, and still in large scale placebo-controlled trials. The qualification of dystrophin levels as a validated surrogate endpoint will be key in the next years; the translation of a limited re-expression of dystrophin into a sustainable clinical benefit could potentially not only support early approval but also help to compare compounds with a mechanism of action based on dystrophin re-expression. The minimum amount of dystrophin that is needed, the critical age or tissue quality at which re-expression must be initiated to be efficient and the time needed to observe clinically significant benefit is currently unknown. On the other side, functional outcome measures are very much dependent on both the observer’s opinion and the patient’s fatigue and motivation. The use of endpoints based on continuous monitoring with wearable technologies, such as the 95th percentile stride velocity, has the potential to dramatically decrease the number of patients needed as well as the study duration (Haberkamp et al., 2019; Poleur et al., 2021).

Now, breakthroughs in the fields of genetic therapies bring several opportunities in the preclinical and clinical pipeline of DMD, which need to be efficiently assessed in clinical trials. The rarity and variability of DMD render the design of clinical trials challenging. Undoubtedly, the DMD community needs to act to avoid repeating the mistakes of the past. The development of representative disease models for preclinical studies, the identification of sensitive biomarkers, the use of innovative technology, and the application of statistics will assist the roadmap to a more efficacious clinical trial design.