Two Decades of Primary Mitochondrial Disease Trials: What Failed, What Worked, and What Comes Next

Key Takeaways

• A structured scoping review and horizon scan in Therapeutic Advances in Rare Disease identified 71 interventional trials in primary mitochondrial disease (PMD) from 2000 to November 2025, across 25 countries.
• The authors say many early trials of broad antioxidant and metabolic therapies failed because the drugs didn't directly correct the underlying defect or the trial designs didn't fit the biology. Recent progress has come from precision approaches such as nucleoside and gene therapy.
• The FDA approved elamipretide (Forzinity) for Barth syndrome in September 2025 (accelerated approval) and doxecitine/doxribtimine (Kygevvi) for thymidine kinase 2 deficiency in November 2025, using tailored evidence packages rather than large randomized trials.

A group of clinicians and patient-advocacy leaders, led by Amel Karaa of Massachusetts General Hospital and Harvard Medical School, report a review of clinical development in PMD, a group of ultra-rare, genetically diverse disorders of cellular energy production. The project began with a trial list compiled by Australia's Mito Foundation, and the authors describe it as both a scoping review and a horizon scan.

The team searched eight databases and registries, plus conference proceedings, sponsor press releases, and FDA and EMA sources, for human interventional studies from January 2000 to November 5, 2025. From 1,482 database records and 214 grey-literature and regulatory records, 1,384 were screened, 327 were assessed in full text, and 71 met the criteria. The authors excluded supplement and nutritional studies, healthy-volunteer studies, secondary mitochondrial dysfunction, and trials focused only on Friedreich's ataxia. Because the studies were so varied, they did not do a formal risk-of-bias assessment or meta-analysis and instead used a narrative synthesis.

Of the 71 trials, 12 were pediatric-only, 26 were adult-only, and 33 enrolled both. Most were Phase 2 (22), Phase 3 (10), or Phase 1/2 (9). Open-label designs were the most common (37), followed by 28 randomized, double-blind, placebo-controlled trials. Industry sponsored 47 and investigators led 21. The United States accounted for 36 trials, activity was concentrated in North America and Western Europe, and Latin America, Africa, and much of the Middle East had little. Small-molecule programs made up 53 trials, gene therapy 14 (almost all in LHON), and mitochondrial-augmentation cell therapies 7. At least 10 trials were terminated and at least 3 withdrawn. LHON had the most trials (20), followed by m.3243A>G disorders (15) and primary mitochondrial myopathy (8).

The failures, the authors argue, taught lessons about endpoints and heterogeneity. The 218-patient MMPOWER-3 trial of elamipretide in primary mitochondrial myopathy, for example, missed both co-primary endpoints, a walking test and a fatigue score. The authors say broad enrollment of mixed genotypes likely diluted any signal, and a later post hoc analysis led to a more targeted trial, now complete with results pending. They report similar challenges in other functional-endpoint trials. In contrast, they describe success in genetically defined diseases with more homogeneous features. Kygevvi received full FDA approval for TK2 deficiency with symptom onset at age 12 or younger, based on a matched external-control comparison of 78 treated and 78 untreated patients that showed mortality of 4% versus 36%. Forzinity's accelerated approval for Barth syndrome (patients weighing at least 30 kg) rested on improved knee-extensor strength, and the FDA requires a randomized confirmatory trial. Idebenone has been approved in Europe for LHON since 2015, but the FDA asked for additional controlled data.

Looking ahead, the authors propose matching trial populations to molecular biology, building natural-history studies and registries, validating outcomes and biomarkers, using adaptive and Bayesian designs, and engaging patients throughout. They also call for broader global access to trials and for publication of negative results. They note biomarkers such as GDF-15 and FGF-21 don't yet qualify as regulatory surrogate endpoints.

The authors acknowledge that they did not grade evidence quality, that unpublished or proprietary programs may be missing, that publication bias may favor positive findings, and that English-only sources may have left out data. They conclude that PMD drug development has changed markedly and is at a pivotal moment, one that depends on sustained investment in natural-history data, better endpoints, innovative trial designs, and equitable global access.