Regenerative medicine has long carried the aura of the mythic, from Prometheus’s eternally renewing liver to Wolverine’s capacity for accelerated repair. For three millennia these stories have both entertained, and expressed, a deep intuition; that damage doesn’t need be permanent, and that our bodies possess the latent ability to restore form and function. Until recently, however, no widely accepted medicine has turned that intuition into routine clinical reality. We’ve become extraordinarily good at slowing decline and suppressing symptoms but we remain poor at eliciting adaptive repair.
In the companion article we reviewed the barriers that have prevented us from developing regenerative medicines at scale. Technological barriers have contributed, but they aren’t the whole story; progress in stem cells, organoids, gene editing and scaffolds has been substantial, but clinical translation has remained limited. At the core is the fundamental biological reality that repair and regeneration are not just the absence of damage or the arrest of its progression. They are active, highly orchestrated processes that unfold temporally and hierarchically until the initiating insult is resolved. By definition, successful repair requires some of the signals we’ve become expert at extinguishing, e.g. inflammatory signalling, to remain active, but only transiently and under tight spatial and temporal control. As a consequence, clinical endpoints and trial designs developed to evaluate therapies that suppress disease activity (such as symptom composites in autoimmune disease, forced vital capacity in interstitial lung disease, or relapse rates in multiple sclerosis) are inherently poorly suited to detect repair, regeneration and structural restoration.
Regulators and payers, quite rationally, demand evidence on endpoints they recognise. Venture capital, equally rationally, funds what has a plausible path to those endpoints. The result is a self-reinforcing loop in which true regenerative candidates are either never tested appropriately or are judged by criteria that guarantee they will appear merely incremental.
Our own regenerative medicines, mitochondrial complex I modulators (MCMs), offer evidence that another path is possible. MCMs don’t broadly suppress inflammation or extinguish proliferative responses. Instead, they modulate cellular energy sensing and signalling in a way that augments the body’s own repair programmes. In muscle they restore quality beyond baseline in preclinical aging and inflammation models, and in wider testing they repair multiple tissues and damage types, operating above conventional signalling nodes to promote resolution. A Phase 2 study of our lead evaluating musculoskeletal repair in secondary sarcopenia has recently started. The protocol includes muscle biopsies as an explicit move towards multi-dimensional measurement of repair rather than surrogate suppression. Of course, leramistat is not yet a finished medicine, but it is the right kind of starting material: systemic, oral, convenient, and mechanistically aligned with endogenous repair rather than suppression.
So how do we systematically make more medicines like this, or deliver the first? Our roadmap has six interdependent elements which need to be pursued in parallel and iteratively. The first step is to select indications where repair can actually be observed and measured with reasonable rigour within the constraints of clinical development.
Not every indication is equally tractable for demonstrating repair. The ideal starting point combines three features: a tissue compartment that can be meaningfully sampled or imaged at cellular resolution within feasible trial timeframes; a clear clinical unmet need where current suppression therapies leave structural damage and functional decline unaddressed; and systemic implications that allow the candidate to reveal broader programmed disease-resolving activity.
These criteria naturally limit the potential avenues that can be meaningfully investigated. Peripheral compartments, eg, blood, urine and saliva obviously lend themselves to repeated, minimally invasive sampling. However, at best they provide a proxy for the complex cellular, molecular and matrix processes occurring in the target tissue. Deep organs (heart, lung, kidney, brain) are difficult to sample repeatedly without unacceptable burden.
Imaging (MRI, DXA, CT, ultrasound) is improving rapidly, yet resolution, variability and the need for large datasets or long intervals between scans still limit sensitivity to true regenerative change.
Practically, this directs us toward muscle, skin and accessible vascular beds as leading compartments. Muscle, for example, permits safe, repeatable biopsies to interrogate cellular and molecular repair directly (progenitor activation, resolution of maladaptive inflammation, extracellular matrix remodelling), while advanced imaging quantifies architecture, quality and fatty infiltration, and functional composites (strength, performance, fatigue, exercise tolerance) capture patient-relevant benefit. Sarcopenia and related musculoskeletal decline in chronic disease offer both high unmet need and systemic readouts. These are not the only possible starting points, but they are among the most practical for generating unambiguous signals of repair.
With suitable indications identified, the next requirement is to design development around the biology of repair itself rather than the logic of suppression. Repair is a hierarchical, multi-phase process. However, current tools and endpoints were largely developed to quantify what existing medicines alter: inflammation, fibrosis or functional decline. They aren’t designed to detect restoration of normal tissue architecture and function. In IPF, high-resolution CT tracks fibrotic burden effectively but lacks resolution to confirm alveolar regeneration; IPF trials therefore use forced vital capacity as their primary measure of disease modification. In RA, ACR20/DAS28 and similar composites were selected historically because they detected responses to the treatments of the early 1990s rather than because they best reflect structural burden or repair potential. Even where mucosal healing is acknowledged as a key outcome in IBD, composite scores still emphasise suppression.
Preclinical models compound the problem. Rodent systems poorly replicate human immune dynamics, tissue microenvironments and long-term remodelling; promising effects frequently fail to translate. Organoids and organ-on-chip platforms reduce but do not eliminate this gap as they typically lack vascularisation, mechanical forces and complex immune interactions. Human data on repair therefore remain essential, but as of today we lack validated biomarkers that specifically track active restoration rather than injury or inflammation. Circulating markers of damage exist; signatures of organised resolution, progenitor mobilisation and matrix remodelling do not.
Repair is also temporally and biologically heterogeneous. It unfolds over months or years, varies between patients according to age, comorbidities and progenitor reserve, and may proceed intermittently rather than linearly. Standard registration trial durations (commonly 12–52 weeks) are therefore often mismatched to the biology. Background therapies can further complicate interpretation: anti-inflammatory or anti-fibrotic agents may inadvertently blunt the very pro-resolving signals required for repair.
The solution is to employ multi-dimensional endpoint packages from the earliest phases of development. Functional outcomes (strength, performance, patient-reported fatigue and function), structural repair readouts (histology from biopsies where ethical and feasible, quantitative imaging of tissue quality and architecture, spatial analyses), and mechanistic biomarkers collected over suitable timeframes and in samples that reflect repair kinetics. Adaptive designs with staged observation periods then allow us to capture the full arc of repair without making every study prohibitively long.
This measurement framework in turn requires more flexible trial architectures than those traditionally optimised for conventional symptom-modifying or disease-slowing agents. Even with the right mechanistic approach and measurement framework, progress is constrained by trial architectures optimised for conventional agents. Rigid, pre-specified, short-duration randomised designs leave little room for iterative exploration of dosing, timing, combinations or context-dependent repair biology.
The solution lies in deliberate redesign. Adaptive, seamless or platform designs such as those already used successfully in oncology basket trials and certain rare-disease or pandemic settings permit systematic learning while maintaining rigour. Early consultation with agencies on what constitutes acceptable evidence of repair will be essential. The objective is not to lower standards but to align standards with the biology we now seek to harness.
Even the most thoughtful trial designs will ultimately be limited, however, unless regulators evolve their expectations in parallel with the science. Historically, and with good reason, clinical development and regulatory review has followed the approach of reducing that which we can observe as quickly as possible. The endpoints defined as a result have become the compass for therapeutic development, set to detect benefit in the form of symptom reduction, slowed decline, or binary remission. They have not been set, however, to detect the restoration of structure, function or health. Or, indeed, to detect benefit relating to systemic aspects of disease that fall outside of direct symptom reduction or decline.
Established endpoints and guidance thus create a Catch-22 situation: without validated repair endpoints, we can’t prove regeneration, but without proof, we can’t qualify new endpoints. Regulators have shown willingness to evolve when presented with compelling biological rationale together with robust measurement strategies. The flexibility shown in rare diseases, or in oncology (pathological complete response, novel surrogates), and the availability of advanced therapy pathways (RMAT, PRIME) all point to this. Analogous operational flexibility needs to be extended to oral agents capable of repair and resolution. We can (and should) leverage scientific advice meetings, dedicated workshops on “repair” or “resolution” endpoints, and proactive sharing of mechanistic and real-world literature to build the case. The goal is to enhance, or create shared standards that recognise restoration, not merely slowed decline or symptom control.
Regulatory openness is only part of the picture. A viable commercial model is also essential if these medicines are to reach patients at meaningful scale. Regenerative medicines, particularly scaleable oral ones, won’t necessarily fit into conventional treatment algorithms. The GLP-1 receptor agonist class offers good and instructive example of this. These agents didn’t just slot into existing diabetes or obesity treatment algorithms. Instead, they helped redefine obesity as a chronic, biologically rooted condition amenable to medical intervention, generated large-scale outcomes data beyond glycaemic control, built awareness and adherence through sustained education and convenient dosing, and commanded premium pricing justified by transformative efficacy and convenience.
A systemic oral regenerative medicine could follow analogous logic. Positioning as a foundational therapy that restores resilience and resolves underlying drivers rather than managing downstream symptoms indefinitely creates a clear differentiation. Early health economic modelling of reduced long-term progression, lower polypharmacy burden, preserved independence in aging populations, and productivity gains support the value and commercial proposition. Notably, these are all outcomes that conventional suppression data rarely capture cleanly. The value case could be further strengthened by post-approval real-world evidence generation that sophisticated payors would likely demand.
The GLP1 agonist example could also be followed to enable pipeline expansion around the same mechanistic core (different tissues, optimised follow-on molecules), creating franchise value and diversifying risk.
Finally, patients themselves must become active partners in shaping what successful repair means in practice. The world is changing. Patients are increasingly informed, proactive and equipped with wearables, smartphones and AI-driven insights. Repair and related outcomes that matter to a patient may not map exactly onto a scientist’s or clinician’s definition. To address this, we believe in engaging patient advisory input early, to ensure we fully capture how patients feel and function (and survive). This means taking input on endpoint selection, integration of patient-reported measures of function, energy and independence, and digital capture of real-world performance, all of which can build scientific relevance and commercial value. It also enables us to make medicines that align with the broader shift towards proactive, personalised health management.
Summary
None of this is easy. It requires capital willing to back the longer arc, clinical investigators comfortable with new endpoint languages, regulators open to co-creating standards, and commercial teams prepared to create a new category rather than selling a new molecule in an old category. Developing scaleable oral regenerative medicine is a true blue ocean opportunity; it addresses unmet needs that suppression alone can’t resolve, and creates markets around restoration rather than perpetual management.
The pieces are in place to do this. We have a mechanistic class that augments rather than suppresses endogenous repair mechanisms. We have clinical-stage assets with the right systemic, oral profile. And imaging, biopsy and analytical capabilities that can make repair visible. The precedent of the GLP-1 agonists shows that an entire therapeutic paradigm can be rewritten within a decade when biology, measurement, regulation, and commerce are aligned around a genuinely new proposition.
The companion article in this pair closed with a call to reform measurements, trials, regulations, and paradigms so that human repair can be tested directly. This roadmap is our operationalisation of that call. If we get the development route, regulatory partnership, and commercial model right, the next chapter of medicine will not be about slowing the inevitable; it will be about actively restoring what disease and time have taken.
