The recent clinical trial of losmapimod for the treatment of facioscapulohumeral muscular dystrophy

Abstract

Facioscapulohumeral muscular dystrophy is a progressive muscle disorder that is likely linked to aberrant DUX4 expression. Losmapimod, a p38 kinase inhibitor, was utilized to suppress DUX4 and its downstream effects. Phase I/II clinical trials showed promising functional improvements in muscle strength and reachable workspace, despite showing no significant reduction in DUX4-driven gene expression. Based on these findings, a larger phase III trial was conducted, prioritizing functional outcomes. However, the trial failed to demonstrate a clear clinical benefit, highlighting key challenges in drug development for the disease. Notably, the underlying mechanism by which p38K inhibition affects DUX4 in mature muscle remains poorly understood, and most preclinical studies with losmapimod were performed in vitro on immature muscle cells. This raises concerns about the relevance of in vitro models for drug testing. The failure of losmapimod underscores the need for better disease models, such as xenografts, and a deeper understanding of DUX4 regulation, before advancing future therapies to clinical trials.

1. Background and pathogenesis

Facioscapulohumeral muscular dystrophy (FSHD) is the third most common form of muscular dystrophy in the human population, affecting between one in 14,000 and one in 20,000 individuals worldwide [1]. FSHD typically begins with facial and shoulder weakness, followed by involvement of other muscle groups [2]. The disease generally has a slow but steady progression of weakness overall, with a loss of strength of 2–4 % per year [3], although some patients experience periods of stability followed by rapid deterioration [4]. FSHD is an autosomal dominant disease linked to mutations that result in epigenetic derepression of the DUX4 gene, encoded by the D4Z4 satellite repeats in the subtelomeric region of chromosome 4q [5–8]. DUX4 is a transcription factor that plays a key role in early embryogenesis but that is silenced in most mature tissues by CpG methylation [9,10]. The most common form of FSHD, termed FSHD1, is initiated by a deletion in the D4Z4 array that results in 10 or fewer D4Z4 repeats, which is associated with a hypomethylation and a greater probability of aberrant DUX4 expression in skeletal muscle [11–13]. When the nascent RNA is polyadenylated, in individuals with the permissive 4qA allele, DUX4 is made [14,15]. In the less common form, FSHD2, methylation is defective due to mutations in epigenetic regulators [16–18], leading to derepression and DUX4 synthesis – again, only in the presence of the permissive 4qA allele [19]. FSDH1 and FSHD2 are otherwise identical in their presentation and prognosis, with a 6-year risk of wheelchair use of 24 % [20,21].

As the pathogenic agent in FSHD, DUX4 has been the focus of many studies aimed at identifying possible treatments for the disease, but it is an elusive target: it is made only sporadically, in minute amounts that are difficult to measure, in a small number of myofibers per muscle and by only a small percentage of the myonuclei in each myofiber [15]. DUX4 protein levels are thus vanishingly low [22]. Together with the difficulty of identifying appropriate regions of affected muscles to biopsy, these factors make the expression of DUX4 mRNA and protein very hard to assay. As an alternative, many investigators have chosen to focus instead on several of DUX4’s downstream gene products, with the (so far unproven) assumption that the expression of their mRNAs accurately reports on the levels of DUX4. This approach may be further complicated by the fact that different studies analyze different sets of gene products, encompassing both primary DUX4 targets (i.e., those directly activated by DUX4) and secondary targets (indirectly regulated through downstream pathways) [23], which can lead to inconsistencies in interpreting DUX4 activity. Moreover, the amounts of the proteins encoded by these mRNAs have never been determined, and so protein biomarkers for DUX4 expression and FSHD must still be identified and validated.

Despite these drawbacks, researchers have been developing a number of experimental approaches to reduce DUX4 expression (for reviews see [1,24,25], including: (i) the design of antisense oligonucleotides and siRNAs to the DUX4 ORF and polyadenylation signal and introducing them into muscle tissue either via transduction by AAV or by linking them to antibodies or ligands that target receptors on the surface of skeletal myofibers [26–31]; (ii) oligonucleotides that include the DNA sequence recognized by DUX4, to act as surrogate binding sites [32,33]; (iii) the identification and targeting of proteins with which DUX4 interacts [34–36]; (iv) screening libraries of small molecules to identify compounds that selectively reduce DUX4 expression without altering muscle development [37–39]. Of these approaches, three have so far reached the clinic: (i) an oligonucleotide conjugated to antibody to the transferrin receptor, tested by Avidity Biosciences (NCT05747924), with promising results that were recently announced (https://www.aviditybiosciences.com/wp-content/uploads/2024/10/Final-NMSG-2024-FORTITUDE-Poster-16SEP24.pdf); (ii) an siRNA linked to a peptide ligand of αVβ6 integrin, developed by Arrowhead Pharmaceuticals (NCT06131983); and (iii) losmapimod, a small molecular inhibitor of the alpha and beta forms of p38 kinase advanced by Fulcrum Therapeutics (NCT05397470). Here we focus on the latter, as some of the results of the clinical trials have been published [40–42] (Fig. 1).

Fig. 1.

Proposed DUX4 activation pathway, with p38 MAPK inhibition.

Loss of epigenetic silencing results in the aberrant expression of the DUX4 gene. The p38 MAPK (α and β isoforms) regulates DUX4 expression, likely through phosphorylation of chromatin regulators. P38 inhibition by losmapimod reduces differentiation-induced DUX4 during early myogenesis, but in late myogenesis residual DUX4 expression persists despite p38 blockade.

2. Targeting DUX4 expression by inhibiting p38 kinase with losmapimod

The p38 MAPKs are serine-threonine kinases p38α (MAPK14), p38β (MAPK11), p38γ (MAPK12), and p38δ (MAPK13), that modulate cellular responses to environmental stimuli. Activation of p38K is generally associated with processes that promote cell differentiation and inhibit cell proliferation, which are affected directly or via other pathways [43–45]. Its effects have been delineated in several cell types, including liver cancer, which proliferate more when p38K has been suppressed [44]. These differences are due at least in part to hyperactivation of the JNK/c-Jun pathway, the activity of which increases in human tumors when the p38α activity is inhibited [44,51]. In skeletal muscle, activation of the p38K pathway promotes myogenesis [46–50]. When satellite cells undergo asymmetric cell division, p38α and p38β are only activated in the daughter cells, which subsequently proliferate and differentiate to form myotubes and myofibers [48]. p38K activity stimulates these later events as well, by promoting transcription of MyoD and MEF2C [46,52,53]. Consistent with this, p38α-deficient myoblasts exhibit impaired differentiation, and cell proliferation continues in skeletal muscles of p38α^Δ/Δ newborn mice [43]. As in some cancers [44,51], increased proliferation is due at least in part to the activation of the JNK/c-Jun pathway [43].

p38 Kinases α and β are inhibited by losmapimod, a small molecule that was first clinically tested for treating chronic obstructive pulmonary disease (COPD) and was deemed safe and well tolerated [54]. It was initially identified as a potential FSHD therapeutic through screens of small molecules that inhibit β2 adrenergic signaling, which increases DUX4 expression in cellular models of FSHD [38]. The p38 MAPKs are activated by β2 adrenergic signaling, raising the possibility that p38K activation mediates the repression of DUX4 by β2 agonists. This was not the case, however. Although the mechanism is not known, losmapimod reduces the expression of DUX4 and some of its downstream target genes in FSHD myotubes in vitro [55,56], with some weaker but corroborative results in vivo, in mice engrafted with FSHD cells from patients [55] (Fig. 1). The results of phase I/II trials of losmapimod have been published [40–42], and a preliminary report of a more extensive phase III trial has recently been released (https://ir.fulcrumtx.com/news-releases/news-release-details/fulcrum-therapeutics-announces-topline-results-phase-3-reach/). The phase I/II trials were remarkably thorough, well planned, and meticulously carried out, and they yielded some promising results. The results of the phase III trial were therefore awaited with enthusiasm, which made its failure all the more devastating to investigators and patients alike.

3. Outcomes of the clinical trials

Briefly, the phase I/II trial studied a cohort of 39 FSHD patients given 15 mg losmapimod with food twice daily, and 38 given a placebo similarly (Table 1). All completed the double-blind study. Patients were examined for 24–48 weeks, with periodic evaluation of serum drug levels, inhibition of p38K activity in blood, muscle dynamometry, time-up-and-go measurements, self-assessment by patients of their overall health and the status of their disease, assessment of “reachable work space” (RWS: a 3 dimensional motion sensor-based assessment of upper arm mobility, which is significantly reduced in FSHD patients [57]), MRI to evaluate muscle integrity and fatty infiltration, and biopsies to assay mRNA levels for DUX4 and several of the downstream genes activated by DUX4 [58,59]. The study was powered to learn if statistically significant changes occurred at the mRNA level, which, based on in vitro studies of immature FSHD muscle cells (myotubes), was anticipated to decrease by at least 2-fold. Although no significant differences in DUX4-driven mRNA levels were found (CCNA1, KHDC1L, MBD3L2, PRAMEF6, SLC34A2 and ZSCAN4 were assayed, with TBP, HMBS and CDKN1B as reference genes), significant differences in patients’ well-being, RWS, and in the strength of particular muscles were identified in the cohort receiving losmapimod [42]. As these results were promising but far from definitive, a phase III trial was redesigned to provide sufficient statistical power to determine if the drug reliably improved these and perhaps other negative consequences of FSHD.

Table 1.

Comparative overview of losmapimod clinical trials in FSHD.

Aspect	Phase 1 [40]	Phase 2 (ReDUX4) [42] NCT04003974	Phase 3 (REACH) NCT05397470
Patients	30 total (10 healthy volunteers, 20 FSHD subjects)	77 FSHD subjects (39 losmapimod, 38 placebo)	260 FSHD subjects (1:1 randomization)
Selection Criteria	Genetic confirmation of FSHD1; muscle biopsy eligibility (MRI-normal or STIR+ muscles)	Genetically confirmed FSHD1; Ricci score 2–4.	Adults with FSHD1/2; functional impairment in reachable workspace (RWS)
Dosage	7.5 mg or 15 mg twice daily	15 mg twice daily	15 mg twice daily
Administration	Oral, 3.5 h postdose muscle biopsies in FSHD subjects	Oral, 48-week double-blind period followed by open-label extension (OLE)	Oral, 48-week double-blind treatment
Primary Outcomes	Safety, pharmacokinetics (PK), target engagement (pHSP27:total HSP27 ratio)	Reduction in DUX4 activity in muscle biopsies	Absolute change in RWS (reachable workspace)
Secondary Outcomes	Muscle and plasma drug concentrations	MRI-based Muscle Fat Infiltration (MFI), Muscle Fat Fraction (MFF) and Lean Muscle Volume (LMV). Patient Global Impression of Change (PGIC). Reachable Work Space (RWS).	MFI, PGIC, Neuro-QoL Upper Extremity, static muscle strength.
Results	Dose-dependent PK; target engagement in blood/muscle; no serious adverse events	No DUX4 reduction; Reduction in MFI and less fat accumulation. Stabilization in RWS (−0.77 %/yr vs. −9.96 % placebo)	No improvement in RWS or secondary endpoints; trial halted
Key Challenges	Establishing muscle penetration and biomarker reliability	DUX4 activity failed to correlate with clinical benefits; shifted focus to functional outcomes (RWS)	RWS lacked sensitivity; need for patient-centric/composite endpoints

The decision to focus on functional outcomes, like muscle strength and RWS, in the absence of measurable changes in the expression of DUX4 and its downstream gene products, could be justified by the fact that DUX4 is devilishly hard to assay quantitatively, even at the mRNA level, with changes in RT-qPCR results typically varying over a 4-fold or greater range even within a cohort. This would make it extremely difficult to detect differences between the populations of treated vs control individuals when only a ~2-fold change in DUX4 levels was anticipated. This difficulty persisted even when individual patients were biopsied before and after treatment, perhaps because the regions that were biopsied, although nearby and guided by MRI, were not identical. Although the measurable increase in strength in some muscles in losmapimod-treated patients may not be functionally significant, and might be due to the anti-inflammatory effects of the drug [55], a preservation or increase in RWS might well be. Based on the results of the phase I/II trial, the possibility that losmapimod improved RWS in FSHD patients and patient-reported global impression of change compared with placebo [42], and as the FDA prioritizes outcomes that demonstrate meaningful improvements in how patients feel, function, or survive [60], there was good reason to hope for a positive outcome in phase III. It was not to be, however: although the participants who received losmapimod, per the regimen shown in Table 1, appeared to accumulate less fat in their muscles, achieve greater RWS, and gain more strength in their shoulder abductor muscles than those who were treated with placebo, these outcomes failed to reach statistical significance (https://ir.fulcrumtx.com/news-releases/news-release-details/fulcrum-therapeutics-announces-topline-results-phase-3-reach).

4. Shortcomings of the preclinical models

In retrospect, there is one key fact that should have tempered our optimism: the molecular mechanism underlying FSHD – the upregulation of DUX4 in mature human muscle tissue and the potential role of p38 kinase – is not known. Indeed, recent reports point to a number of factors other than p38 kinase that are likely to contribute to FSHD pathogenesis, including a lncRNA and possible interactors [61,62], the inhibition of PAX7 activity by DUX4 [63,64], aberrant splicing [65], nonsense-mediated RNA decay [66], modifiers of disease severity [67–70] and changes in the population of fibroadipogenic precursor cells in FSHD muscle [71], in addition to other relevant systemic factors, such as inflammation and muscle usage. Targeting the inhibition of p38 kinase to the exclusion of these other contributing factors may be too focused an approach to yield significant functional improvements or changes in DUX4 expression.

Notably, nearly all of the studies of losmapimod and other p38K inhibitors have been performed in tissue culture, under conditions in which mature human muscle fibers never develop. The only published experiment in animals was conducted using a xenograft model, in which DUX4 expression and its target genes were assessed 4 days post-engrafting [55], long before mature myofibers would have formed in the graft. A recent study reports that the expression of DUX4 in maturing FSHD myotubes in culture may go through two phases, an early phase in which DUX4 expression is up-regulated as myogenesis is initiated and that can be inhibited by losmapimod, and a later phase in which DUX4 expression is maintained in a p38K-independent manner [72] (Fig. 1), perhaps because DUX4 itself promotes the expression of p38K [49,50]. One might therefore expect that losmapimod would have an even smaller effect on fully mature human muscle tissue. (Inhibition of p38 kinase may also suppress myogenesis [46–48], which would likely exacerbate any beneficial effect the drug might have, especially in the context of pediatric FSHD). This study [72] also raised the possibility that DUX4’s downstream gene products can be expressed at significant levels when DUX4 levels are undetectable, suggesting that the downstream products may not be a reliable readout of DUX4 itself, even at the myotube stage. If so, the underlying mechanisms will be particularly challenging to understand.

This raises the question of the relevance of assessing the efficacy of drugs to treat FSHD in tissue culture. If, in fact, studies in culture are of only limited relevance to clinical outcome, and if one wishes to avoid going directly to clinical studies, there is only one alternative if one wishes to study the control of DUX4 expression in human cells: using mature xenografts prepared in mice with muscles or muscle cells from FSHD patients [73–77]. (NB: Culture systems as well as transgenic models [78–81] may be sufficient to determine the efficacy of reagents that target the ORF of DUX4, however). Although some studies of losmapimod have used xenografts [42,72], the stability and state of maturation of the fibers in these preparations has never been established. In other experiments, full maturation of grafts was shown to require electrical stimulation [73,74,76], which was not used in these studies. Additional research will therefore be needed to learn if inhibition of p38K indeed fails to suppress DUX4 and the DUX4 program in mature human FSHD fibers, and, if so, whether other, compensatory signaling pathways, perhaps including downstream pathways initially activated by DUX4 itself, can maintain its activity.

Future directions

Clinical trials of rare neuromuscular diseases, like FSHD, require meticulous planning, a large team of trained researchers, considerable resources, and the determination to participate by a limited population of patients dedicated to advancing therapies. Recent trials of some therapies, for diseases such as Spinal Muscular Atrophy and subsets of patients with Duchenne Muscular Dystrophy, have been successful, with approval of some or all of the therapies by European and American regulators. In these cases, extensive studies were performed in vitro and in murine models of the human diseases before initiating clinical trials. The use of well-established biomarkers, to assess the efficacy of the experimental treatments, and selection and stratification of the individuals to be tested, to minimize variability, were essential to the success of these trials.

Even with such preparatory studies, however, some clinical trials, including studies of virally-mediated gene therapy of Duchenne Muscular Dystrophy that were successful in mice, have failed in humans [82,83]. The lessons learned from both the successful and failed clinical trials have been shared [84–86] and should inform future studies. FSHD is a rare disease characterized by significant heterogeneity among patients, an often non-linear progression, and the possibility that the clinical phenotypes may vary significantly (e.g., phenotypes presenting with facial and scapular involvement may be more severe and progress more rapidly [87]), making the design of clinical trials and the establishment of reliable outcome measures particularly challenging. Future success of clinical trials of treatments for FSHD will likely require close matching of patients in the experimental and placebo groups for their disability, their genetics and epigenetics. It will also need to apply valid biomarkers [58,88], that, with robust functional assessments, can track disease progression. Current approaches to biomarker discovery include MRI to monitor muscle integrity and fat deposition [89], myostatin and other proteins in serum [90–94] (Mariot et al., personal observations), and the presence on the surface of FSHD myofibers of SLC34A2 [74] (Traficante et al., personal observations), a protein that is upregulated by DUX4. Validation of these or potentially other biomarkers of FSHD for use in clinical trials will be an important goal of future studies.

An additional lesson from the clinical trial of losmapimod that we might draw is to beware of basing a clinical trial for FSHD on results that were obtained largely in studies of immature muscle cells in vitro, no matter how promising they might be. Studies in mature muscle tissue, and if possible in mature human muscle tissue generated by xenografting, should facilitate the development of new therapies that avoid this potential pitfall.