Limb Regeneration

Established A limb is not a large organ. It is a patterned composite with a proximodistal axis, a joint series, an obligatory dependence on nerve supply, and a size-scaling problem that no amount of cell biology removes. The salamander rebuilds one. The mouse rebuilds the tip of its terminal phalanx and nothing more proximal. The human rebuilds the tip of the terminal phalanx, mostly in childhood. The distance between those three facts is the entire subject, and most of the public argument about it is conducted without either of the two numbers that actually govern the answer.

Established The first number is a decline curve. Repeatedly amputate a mouse digit tip and it regenerates through four cycles and then stops: excess bone volume falls 57, 43, 19, 13 per cent and then to zero, with 65.7 per cent fewer osteoblasts at the fifth attempt. Frontier The finding that matters more is that adjacent, un-amputated digits on the same animal were impaired too — the exhaustion is systemic, not a local stem-cell pool being drawn down. Established The second number is a growth rate. An axolotl limb regenerate grows at about 0.04 cm/day in its early phase and 0.02 cm/day in its late phase, so 25 cm of human forearm is 625 to 1,250 days of growth even if every patterning problem were already solved.

Speculative This brief argues that the field's two organising stories are both wrong in the same place. Handwave The enthusiast story is that mammals have a suppressed regenerative programme awaiting one released brake; Established the axolotl differs from us by a 32-gigabase genome, roughly ten times human, that is missing Pax3 outright, which is not a switch. Handwave The sceptic story is that humans lost the pattern information; Established human limb cells demonstrably retain positional memory and cannot act on it. Frontier What is missing is an actuator, and the binding scientific link is proximodistal axis specification — the exact place the field's best 2026 mammalian result fails.

Established Epimorphic limb regeneration runs through a blastema, and a blastema is not a ball of stem cells. It is a heterogeneous, lineage-restricted mass in which connective-tissue-derived cells carry the positional information and other lineages contribute their own kind back. Established The patterning hierarchy has now been mapped rather than asserted: dorsal and ventral tissues do not specify structure directly, they license posterior cells to express Shh — dorsal through WNT10B, ventral through FGF2 — and the resulting anterior–posterior signalling then drives the pattern. Established Cherubino and McCusker's 2026 synthesis in eLife states the comparison that organises everything downstream: humans retain positional memory in limb cells but cannot use it to recreate missing structures the way axolotl cells do. Established Laurencin and Nair, writing from inside the field a decade earlier, list the barriers as blastema heterogeneity and gene-expression complexity, loss of epimorphic capacity in adult mammals, the repair-not-regenerate default that resolves injury to scar, and incomplete knowledge of the signalling that triggers blastema formation at all. Frontier Their proposed remedy, Regenerative Engineering, is a conjunction of five things — biomaterials, morphogenesis-competent cells, applied electrical and mechanical forces, immune modulation, enhanced innervation — none of which is individually solved, and that structure is honest about where the field stands.

Established The axolotl genome is 32 gigabase-pairs, about ten times the human genome, and reading it took a purpose-built assembler. Nowoshilow and colleagues needed 110 million PacBio long reads at 32-fold coverage with an N50 read length of 14.2 kb, optical mapping, and an assembler written for the job (MARVEL) to produce the 2018 assembly. Established The assembly's headline biological finding is a negative one: the axolotl does not contain Pax3, and the Pax7 paralogue has assumed its functions. Established This is what the genome cost the field. Until 2018 the premier regenerating vertebrate had no reference sequence, which meant no routine reverse genetics, no clean transcriptomic mapping, and a comparative-genomics literature that could not ask whether a mammalian orthologue existed for anything the salamander was doing. Frontier The consequences persist: functional genetics in an animal with a 32 Gb genome and a long generation time still runs at a fraction of mouse throughput, which is one concrete reason the mechanistic gap between axolotl description and mammalian intervention has closed so slowly.

Established Positional identity along the proximodistal axis is set by retinoic-acid destruction, not retinoic-acid production, and that makes it a titratable variable. Duerr and colleagues showed in 2025 that CYP26B1, the enzyme that degrades retinoic acid, is graded along the axis and significantly higher distally, so signalling level is set by breakdown rate. Established Blocking CYP26B1 with talarozole at 1 µM caused 66.7 per cent of distally amputated axolotl limbs to regenerate full or partial stylopod duplications — a distal amputation growing proximal structures, which is positional memory being overwritten chemically. Established The same work identified Shox as a retinoic-acid-responsive gene differentially expressed proximally versus distally; ablating it shortened stylopod and zeugopod with failed endochondral ossification while leaving autopod length normal. Frontier The authors note that SHOX haploinsufficiency causes short stature in humans, which is suggestive of a conserved reading mechanism and is not evidence that a mammalian blastema would read a CYP26 gradient, because no mammalian blastema exists to test it on.

Established Limb regeneration is nerve-dependent, and the dependency has been reduced to a molecule at least once. Kumar, Godwin, Gates, Garza-Garcia and Brockes identified nAG, an anterior-gradient family protein, as a secreted ligand for Prod1 and a growth factor for blastemal cells; it is expressed sequentially in the regenerating nerve and then the wound epidermis, and denervation abrogates that expression. Established The decisive part is the rescue: local nAG expression by electroporation is sufficient to sustain a denervated blastema and regenerate distal structures. Established But nerves do a second job that nAG does not do. Wells and colleagues showed in the axolotl accessory limb model that the number of nerves connected to the central nervous system sets the size of the resulting leg; isolated nerves without CNS connection do not scale growth, and non-neural extrinsic factors were excluded as direct regulators of allometric growth. Established That same paper supplies the rate. Growth proceeds in stages: an early tiny-limb phase at roughly 0.04 cm/day and a late tiny-limb phase at roughly 0.02 cm/day. Established A 25 cm forearm at those rates is 625 to 1,250 days — one and three-quarter to three and a half years; a 60 cm arm is 1,500 to 3,000 days. Frontier No published proposal exists for accelerating blastemal growth without losing pattern fidelity, and that absence is a finding rather than a gap in this brief's reading.

Established Mouse digit-tip regeneration is the only mammalian instance, and it is level-dependent, imperfect and finite. Regeneration occurs at the tip of the terminal phalanx; amputations at the level of P2 do not regenerate. Established Dolan and colleagues amputated repeatedly and measured the decline: excess bone volume regenerated fell 57 to 43 to 19 to 13 per cent across four cycles and then failed outright at the fifth, with 65.7 per cent fewer osteoblasts in fifth-cycle digits. Frontier Adjacent, un-amputated digits on the same animals also regenerated poorly, which implies a depletable systemic resource rather than a local progenitor pool — a single study, mechanism unidentified, and the most consequential unreplicated result in the mammalian half of this subject. Established Human digit-tip regeneration at the distal level is real and age-dependent, the clinical observation base being conservatively managed children's fingertip amputations reported from the 1970s onward. Frontier This brief deliberately prints no number for it: the primary series and the modern audit of the claim could not be obtained in preparing this page, and a fact that a correction elsewhere on this page depends on cannot be published from memory. Established What “level-dependent” means in practice is worth stating plainly, because it is where most translation arguments quietly cheat: the boundary is a line partway along the last bone of the digit, so the regenerating structure is a bone tip, a nail bed and soft tissue, with no joint, no second skeletal element and no proximodistal series to specify. Frontier A mammal that regenerates across that line has not demonstrated a small version of limb regeneration; it has demonstrated the one part of limb regeneration that does not require pattern.

Established Growth factors can force skeletal regeneration at a non-regenerative amputation level — and what comes out is growth without pattern. Yu and colleagues amputated mice at mid-P2, a level that does not regenerate, then delivered FGF2 at four days and BMP2 at nine days. Established The result was a P3-like phalangeal bone dorsally, a sesamoid-like bone distoventrally, a synovial joint complex with articular cartilage, and tendon and ligament. Established At 45 days bone volume and length were significantly increased but did not reach unamputated values, and the digits carried an average of 1.6 ectopic skeletal elements each across 44 animals; the authors describe the outcome as regeneration “albeit imperfectly”. Established Note precisely what formed: ectopic elements, not a correct P2–P3 series. Established The companion result is that BMP9 induces a joint where none should form, at measurable frequency: 51 per cent of neonatal digits (49 of 96 by micro-CT) formed a distal skeletal element and 61 per cent (60 of 98 by histology) formed a joint-like structure with a synovial cavity, against 0 of 32 in Prg4 knockouts, with sequential BMP2/BMP9 giving 58 to 70 per cent. Established Adult responses were dominated by chondrogenesis rather than joint formation, which is the age-dependence problem restated in a single assay.

Established The deeper regenerators are the ones that should be shaping expectations, and they cut against the appendage story rather than for it. Hydra reassembles an entire animal from a roughly 70,000-cell reaggregate, with several polyps emerging and able to feed by about day six and a basal disc detaching one to four weeks later; β-catenin RNAi abolishes axis formation entirely. Established Its head organiser is a self-limiting Wnt3/β-catenin/Sp5 feedback loop — Sp5 is induced by Wnt signalling and represses Wnt3 — and breaking it by RNAi produced ectopic heads in 50 per cent of uncut animals one day after the second round and 100 per cent two days after the third, the ectopic heads being fully functional, feeding and expressing neuronal markers. Frontier Flatworms are now undermining the textbook: in Stenostomum brevipharyngium, an earliest-branching catenulid, regeneration is supported by multiple division-competent populations resident inside differentiated tissues rather than a canonical neoblast compartment, and the authors argue neoblast-like cells may have evolved convergently. Established The one mammalian appendage that fully regenerates is the deer antler, at up to 2 cm/day — the fastest documented animal tissue growth — roughly 10 kg of tissue a year from three characterised stem-cell populations. Established It is paired with strong positive selection on p53 cofactor and regulator genes and highly efficient apoptosis in the reserve mesenchyme, and the animal throws the organ away annually. The moral is unflattering to the human case: the mammal that solved regrowth solved cancer suppression first and regenerates a disposable structure with no joints and no autopod.

Frontier Retinoic-acid catabolism as a proximodistal dial is the most important new idea in the subject and it comes from one laboratory. The CYP26B1 result is unreplicated at the time of writing, and its significance is methodological rather than mechanistic: it converts positional memory from a descriptive concept into a chemical variable with a clinically known pharmacological handle. Frontier The open question is whether the gradient is instructive in a tissue that has never built a blastema, and no mammalian experiment addresses it.

Frontier Sequential rather than simultaneous growth-factor delivery is the best current mammalian result, and it is one year old with no replication. FGF2 then BMP2, five days apart, at a non-regenerative level, in 44 animals, produced ectopic elements and a joint. Frontier The frontier question it opens is whether ordering is the variable that matters — whether a longer, more finely staged sequence approaches correct pattern asymptotically, or whether ordering buys growth and nothing else and pattern requires an entirely different class of input.

Frontier The bioelectric route to limb induction rests on a single frog paper. A silicone BioDome cap loaded with a silk-protein gel delivered a five-drug cocktail for 24 hours to adult Xenopus laevis; over 18 months the animals grew a limb containing bone, neurons and vasculature, swam with it, and responded to touch at it. Several toes formed without underlying bone. Frontier The release describing the work does not state a sample size and the primary paper could not be obtained for this brief, so no n is printed here. Speculative A 24-hour intervention producing an 18-month morphological outcome is either the most important result in the field or an artefact, and no external replication was located; it is interesting because it is unreplicated, which is a reason to fund a replication rather than to cite it as a milestone.

Frontier Regeneration exhaustion being systemic is the finding nobody predicted and nobody has explained. If impairment spreads to digits that were never injured, the limiting quantity is not a local blastemal progenitor pool, and no candidate has been named — metabolic, immune, or progenitor-compartment. Speculative A measurable systemic variable whose depletion tracks the 57-43-19-13-zero decline would be a genuinely new object in regenerative biology, and identifying it is a cheaper experiment than any limb-scale programme.

Speculative Is nerve dependence reducible to a molecule, or is it two requirements wearing one coat? nAG rescues a denervated blastema, which argues for reducibility. Established Nerve number also sets regenerate size, which is a separate function that no secreted factor has been shown to substitute for. Frontier And Prod1, nAG's receptor, is a salamander-specific three-finger protein with no established mammalian orthologue — human AGR2 and AGR3 exist, the receptor does not, and this is a specific obstacle rather than a formality.

Speculative Two proposals follow directly from the published results and nobody is testing either. The first is to stop demanding a full-size regenerate. If nerve number sets regenerate size, a limb is not obliged to emerge at final dimensions: a correctly patterned miniature could in principle be induced first and grown out afterwards by increasing innervation and mechanical load, which decouples the patterning problem from the rate problem entirely. Handwave What would have to be true is that a scaled-down but correctly patterned appendage can be induced at all, and that allometric growth in an adult body follows the same rules it follows in a larval axolotl; neither has been tested. Speculative The second is modular assembly: BMP9 makes a joint, sequential FGF2 and BMP2 make bone plus a joint complex, so skeletal elements and joints could be induced as separately triggered modules rather than expected to fall out of one blastema. Handwave That requires soft tissue — nerve, vasculature, muscle, tendon routing — to follow a skeleton laid down module-wise rather than co-developing with it, which is assertion. Both are coherent, cheap relative to a limb programme, and unfunded.

Frontier The whole-body regenerators pose a frontier question the appendage literature mostly ignores: what stops a pattern? Hydra's answer is a self-limiting feedback loop in which the organiser induces its own repressor, so that a head is produced once and not repeatedly. Speculative Nothing in the mammalian digit work has an equivalent termination circuit, and the ectopic elements that the best mammalian protocol produces are exactly what an induction without a stop rule should look like. Handwave Whether termination is the missing half of mammalian patterning, rather than initiation, is an unasked question with a cheap first experiment attached to it.

Frontier Two live results were located for this brief and deliberately not used for numbers. A 2025 PNAS paper argues that embryonic mammalian limb regeneration is driven by neural-crest recruitment and reprogramming, which would name a cell source rather than a pathway and would be a different kind of target than anything above. Frontier A 2025 computational paper claims a transcriptomic definition of the regenerative versus non-regenerative boundary around the mouse digit wound bed, which is the molecular statement of the P2/P3 line. Speculative Both are recorded here as leads, unread, because a brief that prints a figure it could not open is worth less than one that says which door was locked.

Established The binding link is proximodistal axis specification, and the field's own best mammalian result fails exactly there. The workback chain runs: induce a blastema-like mass at a non-regenerative level; specify the correct positional identity within it; produce a joint series rather than one joint; recruit nerve supply adequate to sustain and scale growth; vascularise faster than the tissue outgrows the 200–400 µm diffusion limit it shares with every engineered tissue on this map; achieve a growth rate a patient will tolerate; keep a tumour ledger over the whole period; route muscle and tendon; define a regulatory endpoint; find a payer. Established Link two is the first hard scientific stop: FGF2 followed by BMP2 induces growth and yields 1.6 ectopic elements per digit, which is the definition of specification failing while proliferation succeeds.

Frontier The systemic-exhaustion result changes the shape of the whole bottleneck argument. If regenerative capacity were a suppressed programme, releasing the suppression would be a one-time cost. Frontier A four-cycle decline with 65.7 per cent fewer osteoblasts at failure, spreading to digits that were never cut, describes a consumable rather than a switch — and a consumable has to be measured, budgeted and possibly replenished before any multi-year growth protocol is credible. Speculative Nobody has named the consumable, so nobody can currently say whether a 625-day forearm protocol would exhaust it in month three.

Established The second-order bottleneck is rate against a comparator. At 0.02–0.04 cm/day, 25 cm of forearm is 625 to 1,250 days of continuous, supervised, tumour-surveilled growth. Speculative An osseointegrated prosthesis with a peripheral-nerve interface restores function on a timescale of months at a cost a health system already pays, which means that even a fully solved regenerate competes badly on every axis except being the patient's own tissue. Handwave The one honest escape is that no acceleration mechanism has been proposed because nobody has looked, rather than because the rate is a physical constant — blastemal growth rate has never been measured in a mammal at all, so the claim that it is fixed is itself unevidenced.

Established Three further links are real but not yet binding, because nothing has reached them. A joint series rather than a single joint has never been attempted at an adult non-regenerative level. Frontier Vascularisation keeping pace with a growing regenerate has never been measured in a mammal, and the 200–400 µm diffusion limit means a mass that outgrows its perfusion necroses from the centre regardless of how well it is patterned. Speculative And a prospective tumour ledger across a multi-year induced-proliferation protocol does not exist for any species, which matters because the only mammal that regrows an appendage did so alongside strong positive selection on p53 cofactors and regulators. Handwave Ordering these correctly is the point: a field that solves vascularisation before it solves specification will have built a well-perfused ectopic lump.

Established This brief waits on results nobody is currently producing, which is a different condition from waiting on a hard experiment. The first dependency is a mammalian blastema atlas: single-cell transcriptomic characterisation of an induced proliferative mass at a non-regenerative level, scored against the axolotl blastema signature, so that the field can say whether FGF2/BMP2 produces a blastema or merely a growth. Frontier Comparative genomics is the second: the axolotl assembly exists, but the orthologue map that would tell us which salamander components have mammalian counterparts is incomplete in the specific place it matters, Prod1.

Frontier The third dependency is immunological and is currently unreadable from this brief. Macrophage depletion has been reported to abolish salamander limb regeneration, and the strong form of the hypothesis holds that the adult mammalian deficit is substantially an immune phenotype and therefore pharmacologically modifiable within a lifetime. Speculative The primary could not be obtained in preparing this page and no number from it is printed; the honest statement is that the immune hypothesis is live, cheap to test relative to everything else here, and unadjudicated in this brief.

Established The fourth is shared infrastructure rather than a discovery. Vascularisation keeping pace with a growing mass is the same wall that bounds organoids and printed tissue in Lab-Grown Organs, and the fibrotic default that resolves mammalian injury to scar is the same one argued over in Whole Organ Regeneration. A limb programme does not need to solve either alone, and would be the first beneficiary if a neighbouring field did.

Frontier The fifth dependency is a human assay that does not exist. The claim that human limb cells retain positional memory is inferred from cellular behaviour, not from a test that reports where a given human fibroblast thinks it is. Speculative Until such an assay exists, the central optimistic premise of the subject — that the information is present and only the actuator is missing — cannot be checked, and every downstream argument inherits that uncertainty rather than resolving it.

Established The experiments below are ordered so that failing one makes the next unnecessary, which is the property a workback plan should have and most roadmaps in this field do not.

Frontier One: is it a blastema? Take the FGF2-then-BMP2 mid-P2 mass and run single-cell RNA sequencing against the axolotl blastema signature, reporting lineage composition and the fraction of connective-tissue-derived cells. This is achievable now with existing reagents and is the cheapest decisive experiment in the subject. Frontier Two: does it know where it is? Score induced skeletal elements by identity, not presence — morphology plus Shox, Meis and Hoxa13 expression — and report the fraction that are positionally correct rather than ectopic. Established The current answer is 1.6 ectopic elements per digit, so the measurement already exists; what is missing is any intervention that moves it.

Frontier Three: can a mammal make a joint series? BMP9 gives one joint at 51–61 per cent frequency in neonates and chondrogenesis in adults; the experiment is serial joints with correct articular surfaces at an adult non-regenerative level. Frontier Four: does the axolotl scaling law hold in mammals? Measure regenerate size as a function of axon count in a mouse digit, replicating the CNS-connection dependence. Established This has never been attempted, and it is the experiment that would tell us whether nerve supply is a ceiling or a dial in our own tissue.

Frontier Five: what is the mammalian growth rate? Nobody has published a blastemal growth rate for a mammal, so the 0.02–0.04 cm/day figure that drives every timeline on this page is an amphibian number being used as a proxy. Speculative Six: what is the tumour ledger? Neoplasia incidence across a multi-year induced-proliferation protocol in a large animal, prospectively counted — the deer's answer to this question was p53 amplification, and ours is unknown. Frontier Seven: does it work? Grip and pinch force, not histology; a regrown structure that cannot be used is a tumour with good manners.

Established Experiment two is the binding one, and experiments one and five are the cheapest. Nothing beyond link two is worth funding until a mammalian induced mass can be shown to produce the correct element for its amputation level rather than an additional one. Speculative The counter-argument from inside the field is that specification may be a downstream consequence of getting the blastema right, so experiment one is not merely diagnostic but potentially decisive — and that is a genuine disagreement about ordering rather than a hedge. Frontier Experiment eight, unlisted above because it is a measurement rather than a manipulation, is to identify the depletable systemic quantity that the four-cycle decline implies; it requires no limb, no primate and no device.

Established The only device in this subject that has produced a limb is a silicone cap. The BioDome is a wearable bioreactor: a sealed chamber holding a silk-protein hydrogel that presented a five-drug cocktail to an adult frog stump for 24 hours and was then removed. Frontier Its engineering claim is that the wound environment, not the drug, is the intervention — that a stump held in a controlled fluid-filled space behaves differently from one allowed to close. Speculative Scaling that to a human stump means a sterile, perfused, instrumented chamber worn for months rather than a day, and nothing of the kind has been built.

Established Sequential delivery is a device problem before it is a biology problem. The best mammalian protocol needs one factor at day four and another at day nine, spatially localised; a multi-year human protocol implies programmable, implantable, refillable release with dose verification in a growing tissue nobody can currently image at the resolution the pattern is specified at. Add continuous tumour surveillance over that period and the engineering object is closer to an implanted process plant than a dressing.

Speculative Two requirements are usually left out. The first is innervation engineering: if nerve number sets regenerate size, a human protocol needs a way to grow and route axons into a target at a rate matched to the tissue, which is an active-guidance problem rather than a permissive-scaffold problem. Speculative The second is load: limb skeletons are shaped in part by mechanical use, and a structure grown over years on a body that cannot yet use it has no obvious source of the forces that finish it. Handwave Both are stated in every roadmap as items to be addressed and appear in no experimental programme this brief could find.

Frontier The unglamorous requirement is measurement. Pattern is specified at the scale of individual skeletal condensations, and there is no clinical imaging modality that resolves an emerging element's identity in living tissue over months; the field currently reads its outcomes by micro-CT and histology, which means killing the animal. Speculative A human protocol needs a non-destructive readout that says at week twelve whether the structure forming is a phalanx or an ectopic nodule, because the alternative is a multi-year trial whose primary endpoint is only observable at the end of it.

Speculative The nearest technology is the competitor. Osseointegrated prostheses with peripheral-nerve interfaces restore usable function on a timescale of months, and a hypothetical multi-year regenerate has to beat them on something other than speed, cost or risk. Speculative The defensible remaining ground is congenital limb absence, growth in children, and contexts where device maintenance is impossible — and this brief could not source a formal statement of that market argument, so it is offered as a position to be tested rather than a finding.

Frontier Bioelectric medicine is the adjacent programme with the most to give and the most to prove. The claim that resting membrane potential distributions carry morphological information, and that gap-junction and ion-channel manipulation can rewrite them, is developed at length in Bioelectric Medicine; the limb-relevant exhibits are the frog result above and a flatworm head-shape literature this brief flags rather than quantifies. Established What it supplies here is a non-genomic control variable, which is exactly what a field stuck on a 32 Gb genome should want.

Established Three other neighbours matter concretely. Lab-Grown Organs owns the vascularisation wall and the manufacturing route, and is where a limb programme would get its perfusion engineering. Frontier Cellular Rejuvenation owns partial reprogramming, the one demonstrated method for returning adult mammalian tissue to a more plastic state without immediate loss of identity. Speculative Nanomedicine owns targeted delivery, which is what a spatially patterned multi-factor protocol would actually be built from. Speculative Comparative genomics, deer-antler biology and the developmental genetics in Genetic Engineering supply the rest.

Frontier One adjacency is usually mislabelled as a rival and is really an input. Peripheral-nerve interfacing, developed for prosthetic control and discussed in Neuroplasticity Engineering, is the same discipline a regenerate would need in order to acquire the innervation that sets its size and to be usable once grown. Speculative A limb programme that succeeded would still be shipping a device-adjacent product: a stump interface, a controlled growth chamber, and years of instrumented follow-up. The clean opposition between growing a limb and building one is rhetorical, not technical.

Established The largest sustained funder of this work buys reconstruction, not regrowth. The Armed Forces Institute of Regenerative Medicine consortium led by the Wake Forest Institute for Regenerative Medicine received $40 million over five years in its 2023/24 award, across six named areas: craniofacial, extremity, genitourinary and lower abdomen, skin and wound healing, on-demand blood, and cellular therapies for trauma. Established Extremity regeneration is one of six. Established The two previous consortia since 2008 are characterised by the institution itself as having produced more than 20 clinical studies across burns, limb, genitourinary, facial and skull injuries — reconstruction and wound healing. Frontier No funded programme located for this brief has limb regrowth as its stated objective, which is a documented absence rather than an inference from silence.

Established There is no regulatory endpoint for a regrown limb, and this is further from being solved than it looks. “Limb present” is not an endpoint; function scales validated for a regenerated appendage do not exist; and the comparator arm would have to be a prosthesis, which converts the trial into a device-versus-biology superiority question at multi-year follow-up. Speculative Nothing has reached this constraint, so it is not currently binding — but it is the link where a successful animal programme would stall for a decade if nobody starts drafting now.

Frontier The binding institutional constraint is the mismatch between grant length and readout length. A 24-hour intervention with an 18-month endpoint, a five-cycle amputation series, and a growth process measured in hundreds of days do not fit a three-year award with annual milestones. Speculative What such an award does fit is descriptive sequencing, which is why the literature is rich in blastema transcriptomics and thin in multi-year pattern-formation outcomes. Handwave The optimistic reading is that this is the cheapest barrier on the page to remove, because removing it requires a longer cheque rather than a discovery.

Frontier A second institutional fact shapes what can be believed here: the key results come from very few groups. The proximodistal dial, the nerve-scaling law, the sequential-factor protocol, the joint-induction series and the frog limb each rest on a single laboratory, and several rest on the same laboratories as each other. That is not an accusation; it is the normal structure of a small field with expensive animals. Speculative But it means the replication budget, not the discovery budget, is what would most change the state of knowledge, and no funder in this area has an explicit replication line.

Established The consent problem here is unusual because the burden is measured in years. A protocol implying 625 to 1,250 days of induced proliferation, immobilisation, sequential drug delivery and tumour surveillance, with a realistic prospect of an ectopic and unusable outcome, is not comparable to consenting to a device fitting. Speculative A patient who declines a prosthesis to attempt a regenerate forgoes years of function for an uncertain result, and no framework exists for presenting that trade honestly.

Established The tumour question is not a footnote; it is the design constraint. The one mammal that regrows an appendage did so by acquiring strong positive selection on p53 cofactors and regulators and highly efficient apoptosis in its growth zone. Speculative A human protocol proposes sustained proliferation without that inheritance, over a period long enough for a neoplasm to become clinically relevant, in a patient who is otherwise well. Frontier No prospective neoplasia ledger has been published for any long-duration induced-regeneration protocol in a large animal.

Established Children are where the biology is most permissive and the ethics least. Digit-tip regeneration is age-dependent and best documented in conservatively managed children; joint induction by BMP9 works in neonatal mice and collapses to chondrogenesis in adults. Speculative Any serious translational programme therefore points at paediatric subjects for mechanistic reasons, which is precisely the population where a multi-year experimental protocol with an unquantified tumour risk is hardest to justify. Established A related hazard already has a documented instance: the 2008 pig-bladder-powder coverage generated public expectation of finger regrowth from a report whose own investigator called the procedure not a sure thing, and unregulated clinics have historically followed exactly that kind of coverage.

Speculative The distributive question is the one the field never states. If a regenerate costs years of supervised care and a prosthesis costs a fitting and a service interval, then a working limb-regeneration technology arrives as the most expensive option for the least urgent version of the problem, in health systems that already ration amputation care. Handwave The scenario in which it is genuinely liberating — congenital absence, or settings with no device supply chain — is the scenario in which nobody can pay for it, and that mismatch is an ethical fact about the technology rather than a detail of its financing.

Speculative The population that a working technology would serve is large and badly served, and it is not primarily the traumatic-amputation population. Congenital limb absence, dysvascular and diabetic amputation, and injury in settings where prosthesis fitting and maintenance are unavailable together define a need that devices address unevenly. Speculative A regenerate that took three years but required no fitting, no service interval and no supply chain would be worth more in exactly the places least able to fund it.

Handwave The larger implication is not about limbs at all. A mammal induced to rebuild a patterned, jointed, innervated composite on demand would have demonstrated control over positional information in adult tissue, which is the same capability that whole-organ regeneration, scar-free repair and much of engineered-tissue manufacture are separately trying to buy. Speculative The limb is a hard test case for a general capability rather than a specialised goal, and that is the strongest argument for funding it at a level its own market cannot justify.

Established The counter-argument deserves equal weight. The comparative record says regeneration has been repeatedly traded away in large, long-lived, cancer-prone animals, and the one mammalian exception pairs it with hypervigilant tumour suppression and annual disposal of the organ. Speculative A civilisation that acquired limb regeneration might find it had acquired a lifetime oncology surveillance obligation with it, and the honest position is that nobody has costed that.

Speculative There is a third possibility that neither camp likes. The programme may be genuinely lost rather than suppressed — not a brake to release, and not information to recover, but a set of developmental capacities that large long-lived mammals dismantled because building them was expensive and the payoff was small. Handwave If that is right, the route to a human limb is not reactivation but reconstruction: assembling in engineering terms something evolution deleted, which is a far larger project than any current roadmap describes and is also the only version of the goal that does not depend on a comparative analogy holding. This brief declares the tie: the evidence supports neither the reactivation story nor the loss story cleanly, and saying so is the terminal position rather than a failure to reach one.

These horizons track the specific results that would have to land, not general optimism about regenerative medicine. The dates assume the current funding shape persists.

• 10 yr: Frontier Replication or failure of FGF2-then-BMP2, an answer to whether the induced mid-P2 mass is a blastema by single-cell criteria, a CYP26 inhibitor tested on a mammalian digit, and an attempt at the frog result outside the laboratory that produced it. Frontier An identified candidate for the systemic quantity that the four-cycle decline consumes is also a ten-year object, and a cheaper one than any of the above. Speculative The plausible clinical target in this window is a digit tip or a single joint, not a limb, and probably in a paediatric indication where the underlying biology is most permissive.
• 25 yr: Speculative A positionally correct skeletal element induced at a non-regenerative level in a large mammal, with element identity scored by marker rather than presence, and the first published mammalian blastemal growth rate — which would replace the amphibian proxy that every timeline including this one currently runs on. Speculative A regrown human distal phalanx with a functioning joint is the optimistic end of this horizon. Handwave An arm is not on it, and a brief that put one here would be describing a mood rather than a result.
• 50 yr: Speculative A patterned, small-scale forearm in a non-human primate would be the milestone that makes the human question real, and at that point the binding constraint switches from pattern to rate and to the tumour ledger. Handwave Multi-year supervised growth protocols with continuous neoplasia surveillance become the object of the field rather than an afterthought, and the regulatory endpoint problem — currently dormant because nothing has reached it — becomes the pacing item for a decade.
• 100 / 250+ yr: Handwave A human arm regrown to functional use requires either an acceleration of blastemal growth that nobody has proposed or a medical culture that accepts a two-to-eight-year growth period as ordinary. Handwave The comparator will also have moved: a prosthesis a century from now is not the prosthesis this brief is measuring against, and the honest terminal position is that limb regeneration may be achieved and still lose.

• Depends on No typed depends-on edge is claimed, and the reason is specific rather than modest. The inputs this subject needs — a mammalian blastema characterised against the axolotl signature, a positional-identity score for induced skeletal elements, an axon-count-versus-size measurement in a mammal, a mammalian blastemal growth rate, and a prospective tumour ledger over a multi-year protocol — are experiments nobody is running rather than results another brief on this map will deliver. It shares two walls with its neighbours: the 200–400 µm diffusion limit that bounds any growing tissue mass, and the fibrotic default that resolves adult mammalian injury to scar. Both are argued in Whole Organ Regeneration and Lab-Grown Organs, and neither is this brief's to solve.
• Requires (not on this map) The constraint is not a technology and not a regulation: it is that the instruments funding this work are shorter than the experiments that would settle it. The frog result was a 24-hour intervention with an 18-month endpoint. The finite-regeneration result required five sequential amputations with healing intervals between them. A mammalian growth-rate measurement, at the amphibian proxy rate of 0.02 to 0.04 cm/day, is a multi-hundred-day assay before analysis. None of those fit a three-year award with annual milestones, and what does fit is descriptive transcriptomics — which is why the literature has a great deal of blastema sequencing and almost no multi-year pattern outcomes. The largest sustained pot, $40 million over five years across six areas of which extremity is one, is structured for reconstruction deliverables. This is the cheapest binding constraint on the page, because relieving it requires a funder willing to write an eight-year cheque against a single morphological endpoint, not a discovery.
• Enables No typed enabling edge is claimed either. What this subject currently supplies the rest of the map is not a capability but a calibration: the finite-cycle result is the strongest available evidence that mammalian regenerative capacity is a depletable systemic resource rather than a suppressed programme, and the axolotl growth rate is the number that converts every regeneration timeline elsewhere from a biology question into an arithmetic one. If proximodistal specification were ever solved here, it would be the general result — control over positional information in adult tissue — that several neighbouring programmes are separately trying to buy.
• Adjacent Comparative genomics and long-read assembly, which is what made the axolotl legible at all; developmental genetics of the limb bud; bioelectric pattern control, which supplies a non-genomic control variable; immunology, which holds the most testable minority hypothesis in the subject; prosthetics and osseointegration, which set the comparator any regenerate must beat; oncology, which supplies the ledger; and within this map Whole Organ Regeneration, Lab-Grown Organs, Bioelectric Medicine, Cellular Rejuvenation, Nanomedicine and Biological Enhancement.

Handwave “A soldier regrew his finger with pig-bladder powder.” Established The US Army's own 2008 account of the extracellular-matrix trial reports, for the first assessable patient, “an increase in length of the finger, but the wound closed before further growth could occur”; the second case was too early to assess; and the investigator is quoted saying the procedure “is not a sure thing” and had never been tested in humans for limb regrowth. Established The famous civilian case that seeded the coverage was a distal fingertip injury — a level at which human tissue regenerates conservatively without any treatment. Frontier The intervention was applied to an injury class that already self-repairs, which is the single most important sentence about it; this brief prints no figure for the spontaneous rate because the clinical series could not be obtained.

Handwave “Axolotls regenerate perfectly, so perfect regeneration is possible.” Established Salamander regeneration is slow — 0.02 to 0.04 cm/day in the tiny-limb phases — and it is not perfect. Speculative A 2022 report that regenerated salamander skeletal elements are bulkier than the originals is recorded in this brief's reading list without a figure, because it could not be opened; the direction of the correction is nonetheless the right one to hold.

Handwave “Humans lost the genetic information for limb pattern.” Established Human limb cells retain positional memory; what is absent is the ability to act on it during injury. Frontier This misconception is unusual because it is held equally by sceptics, who say we lost it, and by enthusiasts, who say we need only switch it back on. Speculative Both are wrong in the same place: the information is present and the actuator is missing, which makes this an engineering problem in a strictly limited sense — nobody has shown that a supplied blastema-like environment would read human positional memory, and that step is currently pure assertion.

Handwave “One suppressed switch stands between mammals and regeneration.” Established The axolotl genome is 32 gigabase-pairs, about ten times human, and lacks Pax3 entirely with Pax7 having assumed its functions. Established The regenerating vertebrate is not a mammal with a brake released; it is a differently built animal whose genome required a bespoke assembler to read. Frontier The finite-cycle result points the same way from the mammalian side: capacity that depletes over four rounds, with 65.7 per cent fewer osteoblasts at failure and impairment spreading to un-amputated neighbours, is a consumable, and a consumable is not a switch. Speculative Those two facts together are the strongest available evidence against the framing that both the field's enthusiasts and its sceptics lean on.

Handwave “Nerves are just a growth-factor source; supply the factors and the requirement goes away.” Established nAG rescue of a denervated blastema is real and clean. Established But nerve number also sets regenerate size, which is a second and separate function, and Prod1 — nAG's receptor — is a salamander three-finger protein with no established mammalian orthologue. Frontier The pathway does not obviously transfer, and this is an enthusiast's error rather than a popular one.

Handwave “Mouse digit regeneration proves mammals can regenerate limbs.” Established It occurs only at the tip of the terminal phalanx, not at P2; it survives four cycles and fails at the fifth; and the best induced result at a non-regenerative level produced an average of 1.6 ectopic skeletal elements per digit with bone volume and length below unamputated controls. Established What has been demonstrated in mammals is the induction of growth and of individual structures, not the restoration of a pattern.

Handwave “The frog result means mammals are close.” Frontier It is one laboratory, one paper, a 24-hour intervention, an 18-month endpoint, toes that formed without underlying bone, a sample size the public release does not state, and no external replication located. Speculative It is the most interesting unreplicated result in this cluster and should be described that way rather than as a milestone.

Handwave “The military is running a limb-regeneration programme.” Established The current consortium is $40 million over five years across six areas, of which extremity regeneration is one, and the historical output is characterised by the lead institution as more than 20 clinical studies in burns, limb, genitourinary, facial and skull injury — reconstruction and wound healing. Frontier That is a real and substantial programme and it is not an attempt to regrow an arm.

Handwave “The blastema is a ball of stem cells.” Established It is heterogeneous and lineage-restricted, with connective-tissue-derived cells carrying the positional information, and the patterning hierarchy runs dorsal/ventral to posterior Shh to anterior–posterior signalling. Established The field's own barrier list names blastema heterogeneity first, which is the opposite of treating it as a homogeneous pool.

Speculative A dismissed programme worth naming, and not worth asserting. From the 1940s to the 1970s a body of work claimed that mechanical trauma, repeated needling or hypertonic treatment could induce partial regeneration in normally non-regenerating limbs, including mammalian ones. Handwave No primary source for those claims was obtained in preparing this brief, so nothing is asserted about what was done or found; the reason for naming it at all is that the modern frog result is mechanistically a much better-controlled descendant of the same idea, and a field that abandoned a research line without a published post-mortem should expect it to return under a new name.

Handwave “If the axolotl can, we can — it is only a scaling problem.” Established Scaling is the part we can already compute, and it is unfavourable: 625 to 1,250 days for 25 cm of forearm at measured amphibian rates, 1,500 to 3,000 days for a whole arm. Frontier Pattern is the part we cannot compute, and it is where the best mammalian result fails. Speculative The correct summary is that a human limb is not one problem but three — a patterning problem we cannot solve, a rate problem we have not tried to solve, and a market problem that ensures nobody is paying to try — and that the third is why the first two look permanent.