Follistatin vs myostatin antibodies — different strategies, similar limits

What happened instead is a case study in how a single well-validated target can absorb the best efforts of multiple well-resourced pharmaceutical companies and still not yield the clinical outcomes everyone expected. And it is a story best told by looking at the two main pharmacological strategies side by side — because their differences illuminate what the pathway is actually doing, and their shared failures illuminate something broader about muscle biology.

The first strategy is follistatin-based. Follistatin is the body's own natural neutralizer of myostatin — an endogenous binding protein that wraps around myostatin and related TGF-beta family members and prevents them from signaling. Follistatin-344, the predominant circulating splice variant, was developed as a research tool and eventually as a gene therapy payload. The most significant human-relevant work came from Lee Sweeney's laboratory at the University of Pennsylvania and Jerry Mendell's group at Ohio State, who used adeno-associated viral vectors carrying the FS-344 gene to deliver follistatin overexpression directly to dystrophic muscle in mouse models and then, under compassionate use and early clinical protocols, in small numbers of human patients with Becker muscular dystrophy and inclusion body myositis. These were not large randomized trials. They were proof-of-concept studies in small cohorts. The muscle effects were measurable. The safety signals were manageable. The functional improvements were modest. Acceleron Pharma built a broader program around the follistatin-based approach using engineered fusion proteins before pivoting to different targets.

The critical pharmacological feature of follistatin is that it is broad. Follistatin binds myostatin, but it also binds activin A, activin B, and other TGF-beta superfamily ligands that share the same structural binding interface. This breadth is both its appeal — you suppress multiple negative regulators of muscle simultaneously — and a source of complexity. Activin A is not just a muscle growth inhibitor. It is involved in inflammation, in reproductive hormone regulation, in the differentiation of red blood cell precursors, in bone metabolism. Blocking activin alongside myostatin means you are pulling on multiple threads in a system that was built with those threads integrated.

The second strategy is the monoclonal antibody approach: drugs engineered to bind a single target with high selectivity, blocking it before it can signal. Several large pharmaceutical companies ran programs with different specificities within the pathway.

Eli Lilly, in collaboration with Novartis (and later MorphoSys acquiring the program), developed bimagrumab — not an anti-myostatin antibody but an antibody that blocks the ACVR2B receptor itself, the receptor that both myostatin and activin bind to. Blocking the receptor rather than the ligand means you catch everything that uses that receptor: myostatin, activin A, activin B, GDF-11. Pharmacologically, bimagrumab and follistatin-based approaches end up with overlapping target coverage. Bimagrumab in trials produced real increases in lean mass and real decreases in fat mass — effects striking enough that programs exploring bimagrumab for obesity and metabolic disease have continued even as its muscular dystrophy applications paused. A 2021 trial published in JAMA Network Open found significant body composition changes in type 2 diabetic patients with obesity. The pathway is doing something metabolically interesting beyond pure muscle hypertrophy.

Pfizer's program targeted myostatin more selectively. Domagrozumab, also known as PF-06252616, is a monoclonal antibody that binds circulating myostatin directly, preventing it from reaching its receptor, without the broad activin blockade of follistatin or bimagrumab. It was studied primarily in Duchenne muscular dystrophy. Phase 2 results were mixed — there were trends in some functional measures, but the primary endpoints were not met, and the program did not advance to phase 3 for Duchenne. Regeneron developed REGN1033, another anti-myostatin antibody, which showed encouraging results in mouse aging models and was studied in healthy older adults, again finding muscle mass increases that did not consistently translate into proportional functional improvements.

The pharmacological contrast between these strategies matters for understanding what the trials found. More selective myostatin blockade produced lean mass gains but limited functional benefit. Broader blockade of the ACVR2B pathway produced larger lean mass and body composition effects and may have more metabolic interest — but it also hit more systems and created a more complex safety profile to manage.

The safety signals in the ACVR2B pathway blockade trials were not trivial. Epistaxis — nosebleeds — was a common side effect reported across multiple programs, attributable to the role of the pathway in vascular biology. Telangiectasias, small dilated blood vessels visible in the skin, appeared in some trials. Cardiac concerns were monitored across programs, as myostatin and activin signaling have roles in cardiac muscle as well as skeletal muscle. These were not uniformly disqualifying safety issues — bimagrumab's metabolic program continued, and the nosebleeds were generally manageable — but they illustrated that broadly blocking a pathway present throughout the body is not the same as specifically fixing what is wrong with a diseased muscle.

The functional gap — muscle grew but function did not follow proportionally — has multiple proposed explanations, and the honest answer is probably that several are operating simultaneously. Connective tissue and tendon compliance do not scale rapidly with muscle hypertrophy. Motor neuron organization and the fine-tuned recruitment of motor units require neural adaptation that happens on a different timescale than fiber growth. In a disease context like Duchenne, the muscle tissue available to respond to a growth signal is already compromised by cycles of degeneration and fibrosis. You can send the signal. The tissue available to receive and act on it may not be in a position to translate muscle mass into functional performance.

There is also the question of what physical therapy and resistance exercise contribute. In the animal models that showed the most compelling results, the follistatin or antibody intervention was often combined with exercise or was in healthy young tissue. Clinical trial populations in muscular dystrophy are, by definition, patients whose disease has been progressing for years. Whether combining myostatin pathway inhibition with structured physical rehabilitation would yield different functional outcomes is a hypothesis that some researchers continue to explore.

The lesson the field is drawing — slowly, through an expensive series of trials — is that muscle mass and muscle function are related but not identical. Pharmacologically increasing muscle mass in a diseased or aging human being does not automatically produce proportional functional benefit. The muscle is real. The function requires more than mass. Both follistatin-based approaches and the more selective myostatin antibodies appear to hit this same ceiling from different directions. Their shared limits are informative. They suggest the ceiling is less about the choice of pharmacological strategy and more about the underlying biology of what mass-building alone can and cannot do for a muscle that needs to perform.

That does not close the question. It narrows the important questions. Combination approaches — pairing myostatin pathway inhibition with resistance training, or with complementary interventions that support connective tissue and neural adaptation — remain active areas of investigation. Patient selection matters: who in the muscular dystrophy or sarcopenia spectrum has enough functional muscle remaining to benefit from a growth signal is a question that registry data and biomarker work are beginning to address. The target is not wrong. The strategy for realizing its potential in patients is still being refined, a process that has been slower and more complicated than the original animal data suggested it would be.