Lab-Grown Organs

Established The subject is manufacturing: building a transplantable organ outside a body, then putting it into a person. Four routes are competing for the same clinical slot and each of them fails at a different wall. Established Organoids — self-organising three-dimensional cultures grown from pluripotent or tissue stem cells — fail at vascularisation and at maturity. Established Three-dimensional bioprinting fails at cellularisation and function. Established Decellularised scaffolds, in which a donor organ is stripped to bare extracellular matrix and reseeded with the recipient's cells, fail at endothelialisation and thrombosis. Established Xenotransplantation fails at none of those, because it begins with a whole working organ, and fails instead at chronic immunology.

Established Two numbers govern the first three routes and belong at the top of any honest account. Oxygen and nutrients reach only about 200 to 400 µm from a perfused surface, which is why every organoid stalls at roughly the same size whatever tissue it is meant to become. Established And a typical organoid contains 10^4 to 10^6 cells, against 160 to 360 billion cells in a human liver and 2 to 3 billion cardiomyocytes in a human heart. Established One liver's worth of parenchyma is therefore thousands to tens of millions of organoids, and there is no published method for assembling them into an organ.

Established The route that is actually reaching patients is the pig. The longest a human being has lived with a transplanted animal organ is 271 days, on a gene-edited pig kidney, and there are now two cleared multi-patient clinical trials. Frontier No living human recipient has yet reached one year on a xenograft, and every graft to date has been lost or removed. Frontier The sharpest single result in the topic cuts against the industry's own design orthodoxy: a pig kidney carrying one gene edit, not ten, gave a human decedent 61 days of dialysis independence with an antibody-mediated rejection episode that was fully reversed.

Established The diffusion limit is the field's governing physical constant, and it is measured, not estimated. Oxygen and nutrients penetrate roughly 200 to 400 µm from a perfused surface, and in vivo diffusion is confined to within a few hundred microns of a capillary. Established In human brain organoids, more than 50 per cent of cells sitting 150 to 600 µm from the surface become hypoxic or necrotic by day 60 of culture. Established Models put the viability ceiling for a homogeneous sphere at about 1.4 mm in diameter; conventionally grown organoids reach 3 to 4 mm and struggle past 4 to 5 mm as the oxygen gradient across them steepens. Established This is why organoids of wildly different tissue identity — cerebral, intestinal, renal, hepatic — converge on the same size range: the limit is not developmental, it is transport, and it is the same transport limit for all of them. Established The standard review of organoid vascularisation states the position flatly: no method tried has achieved a vessel network in an organoid matching the in vivo standard.

Established The scale-up arithmetic is the second governing constant and it is worse than the first. A single human liver requires on the order of 160 to 360 billion cells; a heart holds roughly 2 to 3 billion cardiomyocytes, and an organoid contains 10^4 to 10^6 cells. Established One liver's worth of parenchyma is therefore thousands to tens of millions of organoids, at reproducible quality, under good manufacturing practice. Established The review that assembles these figures also names the reproducibility problem behind them: batch-to-batch variation is large, protocols vary widely between laboratories, and the animal-derived matrices most organoid work depends on show substantial lot-to-lot inconsistency in composition, stiffness and viscoelasticity.

Established Organoids are not small organs; they are persistently fetal tissue. Ergün and Wörsdörfer, reviewing the field in 2025, put it directly: even after prolonged culture, organoids derived from pluripotent stem cells generally remain in a fetal-like developmental state, containing immature cell types. Frontier The canonical critique of kidney organoids specifically — Little and Combes, asking whether they are accurate models or fortunate accidents — is the standard reference for off-target cell populations and nephron counts in these cultures; the full text was not obtained for this brief, so no percentage or count is printed here. Established What can be said without it is that maturation is a distinct unsolved problem from vascularisation, that it is not obviously downstream of it, and that a fetal cardiomyocyte with disorganised sarcomeres and no T-tubules will not do a heart's mechanical work however well perfused it is.

Established Decellularised scaffolds have had two decades and have never produced meaningful function in vivo. The approach is elegant: perfuse a donor heart or kidney with detergent until only the extracellular matrix remains, then reseed that matrix — which already carries the correct vascular architecture — with the recipient's own cells. Established Stoian and colleagues, reviewing the two decades in 2024, report that despite work on porcine and human hearts using cardiomyocytes derived from induced pluripotent stem cells, heterotopic implantations in small and large animal models have not shown considerable in vivo functionality. Established Incomplete recellularisation of parenchymal regions remains a challenge, and the review names the reason the inherited vascular tree does not save the approach: full revascularisation is an essential prerequisite because of the risk of thrombosis if vascular walls are bare. Frontier That is an unusually clean statement of the barrier — the scaffold hands you a perfect pipe network with no endothelium, and blood clots on bare matrix. Established It is also a review by people committed to the approach reporting that the approach has not worked.

Established Bioprinting has solved resolution and has not solved viability, and the tradeoff between them is the technology's defining constraint. The 2019 SLATE system — stereolithography apparatus for tissue engineering — uses digital light processing with blue light projected from below and food-dye photo-absorbers to confine cure depth, achieving pixel sizes of 10 to 50 µm and printing structures in a matter of minutes. Established Its demonstrations were real: an alveolar model in which red blood cells took up oxygen while flowing past an air sac during simulated breathing, and hepatocyte-loaded constructs implanted into mice with chronic liver injury for 14 days, with surviving liver cells recovered. Frontier The tradeoff between resolution and viability is structural rather than incidental: finer voxels mean longer print times, higher cumulative light dose and, in extrusion approaches, higher shear through narrower nozzles, all of which cost cell viability, while coarser voxels cannot resolve a capillary. Established A capillary is roughly 5 to 10 µm in bore, so a 10 to 50 µm voxel prints arterioles and not exchange vessels — precisely the length scale at which the diffusion limit binds. Established The lead author of the SLATE work, in the release announcing his own headline result, gave bioprinting a horizon of becoming a major component of medicine within two decades — restraint from the person with most to gain from a shorter number.

Established The largest bioprinting programme has a public specification and a checkable, missed deadline. United Therapeutics and 3D Systems announced a printed lung scaffold in June 2022 with a stated specification of 200 million alveoli, 4,000 km of pulmonary capillaries and 44 trillion voxels, printable in as little as three weeks. Established The same release stated that printed lung scaffolds were demonstrating gas exchange in animal models and that human clinical trials of three-dimensional-printed, cellularised lungs would follow within five years, with kidneys and livers listed as under development and no timeline attached; that five-year window closes in mid-2027 and no human trial had begun as of this assessment. Established Separately, the 2019 Tel Aviv construct universally reported as the first printed human heart was rabbit-sized, contained the right patient-derived cell types, vessels and chambers, and did not pump.

Established Xenotransplantation is the only one of the four routes with cleared multi-patient trials and living recipients measured in hundreds of days. The record is 271 days: Tim Andrews, transplanted at Massachusetts General Hospital on 25 January 2025 with an eGenesis EGEN-2784 kidney, the organ removed on 23 October 2025 for declining function, the patient returned to dialysis and still listed. Established EGEN-2784's genetics decompose into three classes — elimination of three glycan antigens, insertion of seven human transgenes regulating immunity, inflammation, coagulation and complement, and genome-wide inactivation of endogenous retroviruses — and the FDA cleared an investigational new drug application for a combined phase 1/2/3 study in dialysis-dependent waitlisted patients aged 50 and over, with a 24-week primary endpoint, on 8 September 2025. Established The competing programme, United Therapeutics UKidney, uses ten edits — six human genes added, four porcine genes inactivated — and performed the first transplant in its EXPAND trial at NYU Langone on 3 November 2025: an initial cohort of six transplants across two centres, a twelve-week interval between the first and second, 24-week endpoints covering survival, graft function, proteinuria and zoonotic infection, lifetime monitoring thereafter, and an independent data monitoring committee gating each expansion. Established The regulatory hinge was February 2025, when the FDA cleared the first pig-kidney trial at all, for six patients expandable to fifty; before that every human xenotransplant was a single compassionate-use or expanded-access case.

Established The rest of the case record is short and every entry ends in graft loss. Two pig hearts: David Bennett Sr, transplanted 7 January 2022, died 8 March 2022 at about 60 days, with latent porcine cytomegalovirus in the graft post-mortem, indirect evidence of antibody-mediated rejection, and a published hypothesis that immunoglobulin given twice in the second month contributed to myocyte damage; and Lawrence Faucette, 58, transplanted 20 September 2023 with a ten-modification Revivicor organ, died 30 October 2023 at about 41 days of rejection. Established Two pig livers: a six-edit graft observed for ten days in a brain-dead decedent, producing bile within two hours and 66.5 mL by postoperative day 10, whose authors concluded the likely role is bridge therapy rather than permanent replacement; and a ten-edit auxiliary graft in a living 71-year-old man with hepatitis B cirrhosis and hepatocellular carcinoma who survived 171 days, with a thrombotic microangiopathy on day 38 that resolved with complement inhibition and plasma exchange. Established The preclinical benchmark for kidneys is six baboons on ten-gene-edited organs: median survival 261 days, maximum 337. Frontier Note the comparison rarely drawn — baboon median 261 against a human record of 271. Frontier Primate survival is not clearly ahead of human survival here, which is an unusual preclinical-to-clinical relationship and says something either flattering about the clinical immunosuppression or unflattering about the model.

Established The most important single result in the topic used a pig with one edit. Montgomery and colleagues transplanted a kidney from a GGTA1-knockout pig — that knockout and nothing else, no CD40 blockade, no additional genetic modification — into a single brain-dead human recipient for a pre-planned 61-day observation, and the graft achieved haemodynamic and electrolyte stability and dialysis independence. Established On postoperative day 10 there was glomerular IgM and IgA deposition with mesangiolysis, a phenotype not seen in allotransplants; on day 33 creatinine rose abruptly with antibody-mediated rejection and rising donor-specific IgG, and it was completely reversed with plasma exchange, C3/C3b inhibition and anti-thymocyte globulin. Established The authors identify pre-existing xenoreactive T cell clones that expanded and acquired effector profiles, not complement, as the residual barrier. Frontier Against a field standardised on ten or more edits aimed largely at complement and coagulation, that is a serious empirical challenge to the design orthodoxy — and it comes from a group whose own prior work used more heavily edited animals.

Established What has actually worked in human beings is small, hollow, or not an organ at all. The only long-follow-up manufactured-organ success is the 2006 engineered bladder series: seven children with spina bifida, autologous muscle and urothelial cells seeded onto a biodegradable bladder-shaped scaffold, seven to eight weeks from biopsy to implantation, follow-up beyond seven years. Established The first-in-human organoid transplant happened on 5 July 2022 at Tokyo Medical and Dental University: colonic tissue from a patient with refractory ulcerative colitis, cultured about a month into organoids of 0.1 to 0.2 mm and returned by colonoscopy, n equals one with planned expansion to eight. Established And the clearest demonstration that organ function can be replaced by manufactured cells is not an organ at all: zimislecel, a stem-cell-derived islet therapy, achieved insulin independence in 10 of 12 patients at twelve months and a mean 92 per cent reduction in insulin dose, under chronic immunosuppression. Frontier The pattern across all three is that replacing a function is tractable and replacing a thick, mechanically loaded solid organ is not.

Established There is a third way to solve the immunological problem, and it edits the graft rather than the recipient. The universal-donor or hypoimmune approach deletes the molecules the recipient’s T cells read — beta-2 microglobulin, which abolishes surface class I human leukocyte antigen, and CIITA, which abolishes class II — and then adds back a signal that stops the innate system killing what it can no longer identify. Established That second half is not optional, and it is where the design gets interesting: natural killer cells are licensed to destroy exactly the cells that display no class I, a mechanism called missing-self recognition, so the class I knockout that defeats T cells hands the graft to the NK compartment. Established Two answers exist in the literature. Overexpress CD47, which engages SIRPα and suppresses both macrophage phagocytosis and NK killing; or express a single-chain HLA-E or HLA-G that engages the inhibitory receptor NKG2A while remaining invisible to conventional T cells. Frontier A third school retains HLA-C instead of deleting everything, on the argument that a licensed inhibitory signal is more robust than a synthetic one and costs only partial T-cell visibility.

Frontier The result that makes this a live proposition rather than a mouse argument is one patient, one islet transplant, and no immunosuppression at all. Allogeneic primary islet cells carrying the beta-2 microglobulin and CIITA knockouts with CD47 overexpression were transplanted into a person with type 1 diabetes in 2025 with no immunosuppressive cover, and the sponsor reported surviving, functioning graft with detectable C-peptide and imaging evidence of the graft at six months. Established The cell dose was deliberately sub-therapeutic: this was an experiment about immune survival, not about insulin independence, and it should not be read as a diabetes result. Frontier It is nonetheless the first direct human evidence that editing the graft can substitute for suppressing the recipient, which is a different claim from anything else in this brief and, if it generalises, a larger one. Frontier One patient, one sponsor, one tissue, no independent replication, and no published peer-reviewed report at the time of writing.

Established The comparison that gives that result its weight is the encapsulation route, which tried to solve the same problem with a membrane and failed on this brief’s own physical constant. A device that walls transplanted islets off from the immune system also walls them off from the capillary bed, and the oxygen and nutrient diffusion limit that governs every construct in this subject governs encapsulated cells identically. Established The device-encapsulated islet programme that would have removed immunosuppression was discontinued for inadequate C-peptide, while the same sponsor’s unencapsulated product, given with standard immunosuppression, reached insulin independence in most recipients. Frontier So the field has now run the comparison directly: a physical barrier costs more function than it saves, and a genetic one is the only immune-evasion strategy with a surviving human graft behind it.

Frontier The vessels that carry blood in every published organoid success are the host's. Human cerebral organoids transplanted into mouse cortex stayed stable between two and four weeks while neural-progenitor grafts shrank, and host-derived CD31-positive vessels penetrated them significantly more than progenitor grafts at four weeks; stage-matched organoids kept in culture showed substantially higher apoptosis than the grafts. Frontier The best in-vitro answer, organoids co-cultured with human umbilical vein endothelial cells, matured to around 200 days with thicker neuroepithelia and significantly reduced apoptosis and hypoxia markers — but showed steady blood flow only after transplantation into mouse cortex. Frontier Both point the same way: engineering a lumen is not engineering a circulation, and no in-vitro method has closed that gap.

Frontier The strategy that is actually winning in vitro is to give up on beating the diffusion limit and get under it. Sliced organoids at an air–liquid interface, 200 to 500 µm thick, sustain culture eight to ten months with high viability and no necrotic cores; intact organoids at the same interface reach six months or longer. Frontier Every one of those gains comes from reducing thickness. Speculative Extend the logic and a transplantable liver might not be liver-shaped at all — a large-area, thin, folded or stacked construct with an engineered perfusion manifold. Speculative What would have to be true is that such a thing can be surgically anastomosed and that the fibrotic response to a very large foreign surface area is tolerable. Handwave Nobody is building it; it is named here because the alternative to a wall is sometimes a different shape rather than a bigger hammer.

Frontier Interspecies blastocyst complementation works, and its efficiency is the number nobody quotes. Human induced pluripotent stem cells injected into cloned pig embryos deficient in SIX1 and SALL1 produced metanephric tubules containing 40 to 60 per cent human cells, with minimal human contribution to non-target organs. Established The denominator: 1,820 embryos transferred yielded two chimeric fetuses at embryonic day 25 and three at day 28, an efficiency of roughly 0.27 per cent. Frontier A parallel in-vitro route mixed human iPSC-derived nephron progenitors with dissociated fetal pig kidney cells at a three-to-one ratio and produced organoids with glomeruli, tubules and ureteric bud — whose authors state plainly that maturation and filtration still require transplantation and vascular invasion. Speculative Three things would have to become true here: efficiency improving by orders of magnitude; human contribution to brain and germline excludable reliably enough to satisfy a regulator, which the commentary itself raises; and the vasculature problem not simply reappearing. Speculative The third is the killer and is badly under-discussed — a human kidney grown inside a pig has pig endothelium unless endothelium is complemented too, so the interspecies route ends up facing the xenotransplant immunology it was meant to avoid.

Frontier The near-term xenotransplant product may be extracorporeal rather than implanted. Four brain-dead decedents were supported by pig-liver cross-circulation: three for 72 to 84 hours with the native liver intact, and a fourth for 48 hours after hepatectomy, holding haemodynamics, pH, lactate, ammonia and clotting time on xenograft support alone with minimal immunosuppression. Speculative Both pig-liver studies conclude toward bridging rather than replacement. Speculative If the honest clinical product is days of extracorporeal support or a months-long bridge to an allograft, then trial designs built on 24-week endpoints and lifetime monitoring are measuring something the technology does not do. Speculative What would have to be true is that chronic rejection kinetics are irreducible over years; the record — every graft lost or removed — is consistent with that and does not establish it.

Frontier The failure modes are becoming named clinical entities, which is what a maturing field looks like. Xenotransplantation-associated thrombotic microangiopathy now has an onset window, a treatment and a documented resolution, and reviews list it with progressive interstitial fibrosis as appearing within months rather than years. Frontier Porcine endogenous retroviruses have been inactivated genome-wide in clinical donor herds with no transmission documented — against a total human exposure of a handful of patients over a few years, a denominator too small to license the word safe. Frontier Meanwhile the edit-count question raised by the one-edit kidney stays open, and nobody with a commercial programme has an incentive to run the experiment that would close it.

Frontier One route treats vascularisation by refusing to build any. Ectopic organogenesis in lymph nodes — hepatocytes seeded into an already-perfused niche, in a phase 2a trial registered as NCT04496479 — is the only programme here that resolves the standing wall by putting tissue where blood already goes. Speculative Generalise it and you get a testable prediction the literature has not stated: borrowed-vasculature approaches should succeed in rough order liver, then pancreas, then thymus, then kidney, and never heart. Speculative Liver function is largely mass-dependent and tolerates being in the wrong place; kidney function depends on the corticomedullary gradient and a connection to a collecting system; cardiac function is mechanical and topological, so an ectopic heart is not a heart. Frontier That ordering is falsifiable and cheap to check against trial registries over a decade.

Frontier The institutional pull is toward organoids as models, not as organs. The FDA's roadmap for reducing animal testing targets making animal studies the exception rather than the norm within three to five years, and cutting primate antibody testing from six months to three where new-approach data supports it. Frontier That is a large and near-certain demand signal — and toxicology does not need a vascularised, mature, 10^11-cell organ. Frontier It needs a reproducible 10^5-cell one.

Established The central objection to hypoimmune cells is not that they will be rejected; it is that they will not be. The immune surveillance these edits defeat is the same surveillance that clears a cell which has acquired a transforming mutation, and CD47 overexpression is not a neutral engineering choice: it is the canonical don’t-eat-me signal that human tumours use for exactly this purpose. Frontier A pluripotent-derived product carries the residual-undifferentiated-cell problem already, and hypoimmune editing removes the backstop that would otherwise have caught a teratoma early. Speculative The honest framing is that the field has traded a managed, visible, treatable risk — rejection, which announces itself and responds to drugs — for an unmanaged, invisible one that declares itself late.

Frontier Kill switches are the stated answer and they are in direct mechanical conflict with the evasion package. The three routes in use are a suicide enzyme cleared by a prodrug, an inducible caspase that dimerises on a small molecule and kills better than 90% of carrying cells within about half an hour, and a surface tag that lets a clinical antibody clear the product. Established The third route is the one most often proposed for off-the-shelf products and it depends on antibody-dependent cellular cytotoxicity and complement — effector mechanisms delivered by natural killer cells and macrophages, which are precisely the cells CD47 overexpression was installed to inhibit. Frontier The inducible caspase avoids that conflict because it is cell-intrinsic, and inherits a different one: it must remain expressed for the lifetime of a graft intended to persist for decades, and transgene silencing in long-lived cells is well documented. Speculative Worse, a switch only has to fail in one cell. Any clone that silences or deletes the cassette is selected for by the very event the switch is meant to control, so a kill switch is a population-level control that behaves badly at exactly the moment it is needed.

Established Persistence is the metric that distinguishes a cell product from an organ, and the two applications want opposite answers. An allogeneic effector-cell therapy is useful if it survives weeks; a manufactured endocrine graft or a vascularised organ is useful only if it survives years, and hypoimmune editing is the only intervention that plausibly delivers the second without lifelong pharmacology. Frontier That is also why this module matters to an organ brief rather than only to a cell-therapy one: the surface a recipient’s antibodies actually see on a transplanted organ is the endothelium, and a manufactured organ lined with hypoimmune endothelium is the only version of this subject that escapes immunosuppression rather than merely reducing it. Speculative It also means the module cannot be evaluated until the endothelialisation problem this brief names as binding is solved, because there is no organ-scale endothelium to edit.

Established A workback plan for a manufactured, transplantable, human-sized heart has ten links and only one of them is currently binding. Established One: produce 2 to 3 × 10^9 functional cardiomyocytes per organ reproducibly — the cell number is known, the GMP yield and batch coefficient of variation are not published. Established Two: mature them past the fetal phenotype, measured as adult-like sarcomere organisation, T-tubules, a force–frequency relationship and upstroke velocity. Unsolved. Established Three: build a vascular tree that is patent, non-thrombogenic and perfusable at organ scale, measured as time-to-thrombosis of a recellularised or printed coronary tree under physiological flow and pressure. Unsolved, and there is no published benchmark of any kind. Established Four: deliver oxygen through a myocardial wall more than a centimetre thick, measured as oxygen-tension mapping under perfusion — which the diffusion constant makes entirely a vascular problem, so it is not an independent link at all. Established Five: achieve coordinated, load-bearing contraction, measured as developed pressure and ejection fraction in a working-heart rig. The 2019 print failed exactly here. Frontier Six: match electrically to the recipient without re-entrant arrhythmia over months — arrhythmia has been the recurring failure mode of pluripotent-stem-cell cardiomyocyte grafts in primates, and the large-animal literature on it was not obtained for this brief.

Established The remaining four links are institutional and industrial rather than scientific, and they are not yet doing any work. Established Seven: survive implantation immunologically — an autologous iPSC route removes the problem and adds six to twelve months of per-patient manufacturing; an allogeneic route reintroduces lifelong immunosuppression. The trade-off is unresolved because nobody has had to resolve it. Established Eight: manufacture within a clinically usable time, meaning weeks rather than months; the printed-scaffold claim is three weeks and the cellularisation time on top of it is unquantified. Established Nine: a regulatory route for a manufactured solid organ, for which there is no precedent and no defined endpoint. Established Ten: beat the comparator — which means beating a pig heart at 41 to 60 days and a mechanical ventricular assist device that routinely runs for years.

Frontier Link three is binding, and it is a scientific unknown with a crisp experimental definition. Links four, five and seven are all gated on it; the field's own review of the decellularisation route names bare vascular walls and thrombosis as the prerequisite; and there is no published time-to-thrombosis figure for a whole-organ vascular tree anywhere in the literature. Frontier Link two, maturation, is the second-order blocker and would become binding the moment link three moved. Established The asymmetry that should be stated explicitly: for a heart, link three binds. Frontier For a liver, link three can be bypassed entirely by placing tissue in an already-perfused niche — which is why the liver programmes are in phase 2a and the heart programmes are still printing scaffolds.

Established Manufacturing consistency is the quiet constraint on the universal-donor route, and it is a genomic-stability problem rather than a process one. The whole economic case is that one clonal master bank serves thousands of recipients, which concentrates risk in the same proportion that it spreads cost. Established Human pluripotent lines recurrently acquire culture adaptations, including dominant-negative TP53 mutations and characteristic chromosomal gains, and each additional simultaneous edit lowers efficiency multiplicatively while raising the chance of a translocation between cut sites. Frontier A hypoimmune line is a multiplex-edited clone, so it sits at the intersection of both hazards, and the release testing that would detect them is long-read whole-genome work on the bank rather than anything performed on a patient dose.

Frontier And there is no accepted potency assay for immune evasion, which is a regulatory gap with no research answer. Loss of surface class I can be measured by flow cytometry in an afternoon, but that measures the edit, not the function: whether a given graft survives depends on the recipient’s natural killer repertoire and KIR genotype, which vary between people and are not part of any release specification. Speculative Until an assay exists that predicts evasion in a named recipient rather than confirming a genotype in a vial, every hypoimmune product is released against a proxy, and the first failure will be discovered in a patient.

Established This subject waits on results other fields are producing for their own reasons, and on one that nobody is producing at all. Established It depends on stem-cell manufacturing science — expansion, purification and cryopreservation of clinical-grade cells at 10^9 to 10^11 scale — which the cell-therapy industry is advancing because cell therapies pay for it. Established It depends on gene editing at herd scale, covered by Genetic Engineering, which supplies the edited animals the only working route currently uses. Frontier It depends on transplant immunology, and specifically on characterising the pre-existing xenoreactive T cell repertoire that the one-edit decedent study names as the residual barrier — a question about human immune memory rather than about pigs. Frontier And it depends on large-animal electrophysiology for any cardiac construct, because graft-induced arrhythmia rather than graft failure has been the recurring endpoint in primate work with stem-cell-derived cardiomyocytes.

Speculative The dependency nobody is discharging is endothelialisation. A method for confluent, non-thrombogenic endothelial coverage of a whole-organ vascular tree, measured as time-to-thrombosis under physiological flow, is a single well-defined unsolved measurement that gates the entire scaffold and printing programme. Frontier It is not being pursued as a named objective by any funded programme this assessment could identify, because it sits between vascular biology, which studies vessels in bodies, and tissue engineering, which studies constructs without flow. Frontier The neighbouring subject that would most change this one is in-situ regeneration, covered by Whole Organ Regeneration: if an organ can be induced to rebuild itself inside a patient, the manufacturing problem dissolves rather than being solved. Speculative The reverse is also true and less often said — a manufactured organ that worked would remove most of the clinical urgency from regeneration research. Frontier The dependency that is discharged, and worth noting because it used to be binding, is source-animal genetics: gene editing at herd scale is no longer the limiting step for xenotransplantation, which is why the argument has moved to how many edits are actually needed rather than whether they can be made.

Established Six experiments would move this field more than any amount of additional organoid characterisation, and four of them are cheap. Frontier First, and most important: a time-to-thrombosis benchmark. Perfuse a recellularised or printed organ-scale vascular tree with whole blood at physiological pressure and flow, and publish hours to occlusion against endothelial coverage fraction. Frontier No such number exists for any construct, which means the field's named binding barrier has never been given a scale, and no programme can currently say whether it is a factor of two or a factor of a thousand from clinical viability.

Frontier Second: a randomised comparison of edit counts. The one-edit decedent kidney gave 61 days of dialysis independence with reversible rejection; the ten-edit programmes give hundreds of days under heavier immunosuppression; nobody has run the two side by side in matched recipients under matched regimens. Frontier The experiment is a decedent study with two arms and it would settle whether edits two through ten are buying anything, which is a question with commercial consequences and therefore no commercial sponsor. Speculative Third: transplant a GGTA1-knockout-only organ into a living recipient with the plasma-exchange and complement-inhibition rescue protocol pre-specified. Speculative If reversible rejection is genuinely reversible in a living patient, the whole edit-escalation programme is optional.

Frontier Fourth: oxygen-tension mapping through a thick engineered wall under perfusion. Publish the depth profile through more than a centimetre of engineered myocardium and the diffusion argument stops being theoretical. Frontier Fifth: report the GMP economics honestly — cost and yield per 10^9 cardiomyocytes, with batch coefficient of variation on purity and maturity. Frontier No such figure is published by anyone, and a cost model for a manufactured organ without one is not a cost model. Speculative Sixth, and the one that tests a hypothesis rather than a device: track the ordering prediction that borrowed-vasculature approaches will succeed for liver, pancreas and thymus and fail for kidney and heart. Speculative That is a decade-long registry study costing nothing, and it distinguishes a field making progress from a field making announcements.

Established The engineering requirements are known precisely enough to be quoted as a specification, which is unusual and slightly misleading. The published lung-scaffold target is 200 million alveoli, 4,000 km of pulmonary capillaries and 44 trillion voxels, with a claimed print time of as little as three weeks. Established Printing resolution is 10 to 50 µm in the best-published photopolymer system, which resolves arterioles and does not resolve capillaries. Frontier The resolution-versus-viability tradeoff is the governing engineering constraint: at finer voxels, print time, cumulative light dose and, in extrusion systems, nozzle shear all rise, and each of those costs living cells, while at coarser voxels the exchange vasculature the tissue needs cannot be laid down at all.

Frontier Cellularisation, not printing, is where the engineering actually stops. A voxel count is a geometry problem and geometry has been demonstrated; putting the right cell in each voxel, keeping it alive through the build, and keeping the endothelial lining confluent under flow is a different discipline with no demonstrated organ-scale result. Established Beyond the construct itself, a manufactured organ needs bioreactor capacity for 10^9 to 10^11 cells per patient, a perfusion circuit that maintains the construct from print to implantation, a cold or normothermic chain to the operating theatre, and a release-testing regime for a product that is alive and cannot be terminally sterilised. Frontier None of that is exotic; all of it is absent. Speculative And an autologous route multiplies every one of those requirements by the number of patients, because each organ is a batch of one — which inverts the usual manufacturing economics, since there are no unit-cost gains from volume, only throughput gains from parallel capacity. Frontier The xenotransplant route faces the mirror image and it is a far more familiar problem: a designated-pathogen-free, gene-edited herd is a facility, a veterinary programme and a qualification dossier, all of which industry knows how to build.

Established The comparators this technology has to beat are not other manufactured organs; they are machines and drugs that already work. A mechanical ventricular assist device supports circulation for years and is routine; dialysis supports renal function indefinitely at a cost in quality of life; normothermic machine perfusion is expanding the usable donor pool without any manufacturing at all. Established Islet cell therapy has reached insulin independence in most treated patients, which shows that replacing a function with manufactured cells is a solved problem class in at least one organ. Speculative The nearest adjacent failure is instructive and is missing from most accounts: the bioartificial liver devices ELAD, HepatAssist and MARS carry a human trial record generally null or equivocal on survival, and the review documenting it could not be obtained for this brief — an absence named here rather than papered over, because a field that does not read its own failed programmes repeats them. Frontier A manufactured heart competing against a device that already runs for years has a much harder case to make than a manufactured liver competing against nothing. Speculative The comparator that most threatens the whole enterprise is the one nobody in it likes to name: a durable, fully implantable, non-thrombogenic mechanical pump would end the manufactured-heart programme regardless of how the biology turned out, and it faces the same materials problem — blood on a foreign surface — from an engineering direction rather than a biological one.

Established Within this map the near neighbours are close enough that the boundaries need stating. Whole Organ Regeneration owns rebuilding an organ in place from the patient's own tissue; Limb Regeneration owns appendages with joints and a proximodistal axis; Artificial Wombs owns extracorporeal gestation; Nanomedicine owns sub-micron agents and inherits the same diffusion limit from the other side. Established This brief owns everything manufactured — organoids, printing, scaffolds, blastocyst complementation and xenotransplantation — and it owns the vascularisation wall. Frontier Genetic Engineering and Designer Organisms supply the edited source animals; Precision Medicine supplies the matching and monitoring; Xenobiology is where the question of how far a chimeric organism can be pushed lives.

Established There is no regulatory precedent for a manufactured solid organ and the nearest analogue is a poor fit. The engineered bladders of 2006 went through a route designed for devices and cell products; the xenotransplant programmes are proceeding as biologics with combined phase 1/2/3 designs intended to support a licence application; a printed, cellularised heart is neither. Established What the xenotransplant trials have established is an architecture other manufactured organs will inherit: small initial cohorts, mandated intervals between transplants, independent data monitoring committees gating each expansion, 24-week functional endpoints, and lifetime monitoring for zoonotic and opportunistic infection. Frontier That last requirement is genuinely novel — it commits a health system to indefinite surveillance of a cohort, and possibly of their close contacts, as a condition of the therapy existing.

Frontier The money is being pointed somewhere else, and this explains funding patterns better than technical difficulty does. The FDA's animal-testing roadmap creates a large, near-term, regulator-backed market for organoids as toxicology models — and that market wants reproducibility at 10^5 cells, not vascularisation at 10^11. Frontier Simultaneously, the cleared xenotransplant trials point transplant capital at pigs, which are a nearer-term biological substitute. Speculative The manufactured-organ route may therefore end up squeezed between a cheaper model market above it and a nearer-term biological substitute below it, serving neither. Speculative That is a market-structure hypothesis rather than a scientific one, and it deserves stating precisely because it predicts the observed funding pattern without needing to claim the science is harder than it looks.

Established Two institutional facts are missing and neither is a research problem. There is no designated allocation rule for the first scarce manufactured or xenogeneic organs — the existing allocation system is built around deceased-donor scarcity and clinical urgency, and a manufactured organ inverts several of its assumptions, including that supply is exogenous. Frontier And there is no established supply chain of clinical-grade, designated-pathogen-free, gene-edited source animals at any scale beyond single-programme herds. Speculative Both are constraints a regulator, a payer and an industry could choose to supply, and neither has.

Established The consent problem in this field is unusually sharp because the record is unusually clear. Every human xenograft to date has been lost or removed. Established Recipients enrolled under expanded access have generally been people ineligible for conventional transplant, which is the standard justification and also the standard critique: the population with least to lose is being asked to bear the highest uncertainty. Frontier The decedent studies — the 61-day kidney, the ten-day liver, the cross-circulation series — are ethically the cleanest experiments in the field and depend entirely on family consent for research on a brain-dead body, a consent regime that is thinly codified and varies by jurisdiction.

Frontier Chimera research raises a question that regulators have not answered and researchers have raised themselves. The blastocyst-complementation commentary explicitly flags that brain and germline knockouts in the donor cells may be needed, which is a way of saying that the acceptability of the technique depends on being able to prove a negative about where human cells ended up. Speculative No agreed threshold exists for how much human contribution to a pig brain is too much, and no assay is agreed for measuring it. Established Animal welfare is a live and under-argued cost: the working route requires purpose-bred, gene-edited, pathogen-controlled pigs killed at organ retrieval, at eventual scale.

Speculative The allocation problem arrives before the technology does. When the first few manufactured or xenogeneic organs exist, they will be scarce, expensive, and of uncertain durability — which means the allocation criteria used for deceased-donor organs, built on urgency and waiting time, will mis-assign them. Frontier An organ expected to fail within a year is not appropriately given to the sickest patient; it is appropriately given to the patient for whom a year is decisive, which is a different person and a criterion no existing allocation system encodes. Speculative And if manufactured organs are autologous, they are inherently unequal: the price is per patient rather than per unit, and the queue becomes a market.

Established The demand is measured and it is the reason all of this is funded. As of mid-2026 there were 109,817 people on the United States national transplant waiting list, 12 people died each day waiting, more than 49,000 transplants were performed in 2025, and a new candidate was added roughly every seven minutes; kidneys are the organ most needed. Established That is one country's list, and it counts only people considered eligible to be listed at all, which understates true demand by an unknown factor.

Frontier Removing organ scarcity would change more than transplant medicine. If organs became a manufactured good, the clinical criteria for offering transplantation would loosen, because listing criteria exist largely to ration; the age and comorbidity thresholds that currently exclude patients are downstream of scarcity rather than of biology. Speculative That produces a demand expansion nobody has modelled: the number of people who would benefit from a new kidney is far larger than the number currently listed for one. Speculative It also removes the moral architecture of donation — the gift, the family decision, the reciprocity norms that organ procurement systems are built on — and replaces it with a supply chain and a price.

Speculative The likeliest outcome over the next two decades is not abundance but a two-track system. Xenotransplantation supplies kidneys and bridging livers under intensive immunosuppression and lifetime surveillance; manufactured cell therapies replace discrete endocrine and hepatic functions; and thick, mechanically loaded solid organs — hearts above all — continue to come from human donors and machines. Frontier A civilisation that solved the kidney and not the heart would look a lot like this one, only with 90,000 fewer people on a list. Speculative That is a large humanitarian gain and a small civilisational one, and the distinction is worth keeping straight: this technology, at its plausible best, removes a specific and terrible scarcity without changing anything structural about how long people live.

These horizons track results that would change what can honestly be said, not predictions of clinical availability.

• 10 yr: Frontier A living human recipient passing one year on a xenograft is the single most likely field-defining result, and its absence a decade from now would say more than any announcement. Frontier The two cleared kidney trials will report 24-week endpoints on tens of patients, which is enough to establish whether hundreds of days generalises beyond selected expanded-access cases. Frontier The United Therapeutics printed-lung prediction expires in mid-2027 and can simply be checked. Speculative A published time-to-thrombosis benchmark for an organ-scale vascular tree is achievable within this window at modest cost and would, for the first time, put a number on the binding barrier. Speculative Organoids will be widely adopted as regulatory toxicology models; this is close to certain and is not progress toward a transplantable organ.
• 25 yr: Speculative Ectopic and borrowed-vasculature approaches deliver a licensed hepatic therapy, or they do not, and the ordering prediction — liver, pancreas, thymus, then kidney — becomes checkable rather than rhetorical. Speculative Blastocyst complementation either improves efficiency by two to three orders of magnitude from 0.27 per cent and solves the porcine-endothelium problem, or it settles into a research tool. Speculative If minimal-edit xenografts prove sufficient, source-animal supply becomes an agricultural problem rather than a genetic-engineering one, which changes the cost structure of the whole field.
• 50 yr: Speculative A vascularised, matured, thick engineered tissue at clinically useful mass is plausible on this horizon and requires an endothelialisation result that does not currently exist. Handwave A manufactured heart supporting circulation for years requires all ten workback links, of which five are unsolved scientific unknowns; nothing in physics forbids it and no demonstrated trend delivers it. Speculative The more probable fifty-year picture is that solid-organ manufacture is superseded — by in-situ regeneration, by durable mechanical devices, or by cell therapies that make whole-organ replacement unnecessary for most indications.
• 100 / 250+ yr: Handwave On-demand manufacture of any organ from a patient's own cells within a clinically usable time is a coherent endpoint with no component demonstrated at organ scale. Handwave Growing bodies without brains as organ sources, periodically proposed, requires both a developmental-biology capability nobody has and a moral settlement nobody has attempted. Speculative The honest statement at this range is that the constraint is a transport problem in tissue, that transport problems in engineering are usually solved rather than evaded, and that the field has not yet produced the measurement that would show progress.
• Universal-donor cells: Frontier Within ten years the hypoimmune islet result either replicates in a second sponsor at a therapeutic cell dose or it does not, and that is the whole question for the module. Speculative A hypoimmune endothelium lining a manufactured vascular tree is the only route by which this subject reaches a transplant without lifelong immunosuppression, and it is downstream of the endothelialisation problem rather than an alternative to it. Handwave Whether a graft engineered to be invisible can be made safely removable is unresolved in principle, not merely unengineered.

• Depends on This brief sits downstream of results other subjects are producing for their own reasons. It waits on gene editing at herd scale, which supplies the source animals the only clinically effective route uses, and on the stem-cell manufacturing science that cell therapies are advancing because cell therapies pay for it. It waits, less obviously, on transplant immunology's characterisation of the pre-existing xenoreactive T cell repertoire, because the best-controlled study in the field names that repertoire rather than complement as the residual barrier, while every gene-edit strategy in commercial use is aimed at the other target. No typed dependency edge on another brief is claimed, because none of those neighbours has produced a result this subject is presently blocked on: the binding barrier is endothelialisation of an organ-scale vascular tree, which nobody has as a stated objective.
• Requires (not on this map) Two constraints sit outside this map and neither is a scientific result. The first is a supply of clinical-grade, designated-pathogen-free, gene-edited source animals beyond the single-programme herds that exist today: the only route with living recipients measured in hundreds of days depends entirely on pigs bred, edited, housed and qualified to a standard with no industry behind it, no second supplier and no published cost per animal — and if the one-edit result holds, that herd specification changes fundamentally. The second is an allocation rule for the first scarce organs, which arrives before the technology does: an organ expected to last a year is not appropriately assigned to the sickest patient but to the patient for whom a year is decisive, a criterion no existing allocation system encodes, and one a manufactured supply inverts further because supply stops being exogenous and price becomes the queue. A third arrives with the universal-donor module and is regulatory rather than scientific: a release standard for immune evasion. There is today no accepted potency assay that predicts whether an edited graft will survive in a particular recipient, only assays that confirm the edit was made, and because survival depends on the recipient's natural killer repertoire rather than on the product alone, no amount of characterisation of the vial supplies it. Somebody has to write that standard before an off-the-shelf hypoimmune product can be released against anything but a proxy, and no brief on this map writes standards.
• Enables What this subject would enable, if it worked, is the removal of scarcity from transplantation, with effects running well past transplant medicine, because listing criteria exist to ration and would loosen if supply became manufactured. What it enables today is narrower and worth stating exactly: a validated comparator. Any in-situ regeneration programme, any long-horizon repair proposal, any argument that a damaged organ can be rebuilt in place now has to beat 271 days on a pig kidney and years on a mechanical assist device, both moving. That is a benchmarking contribution rather than a technological one, so no typed enabling edge is recorded.
• Adjacent Adjacent to Whole Organ Regeneration, which attacks the same clinical problem from inside the patient and shares the diffusion and fibrosis walls; to Limb Regeneration, which owns the pattern-axis problem this subject does not have; to Nanomedicine, which inherits the same 200 to 400 µm constant from the delivery side; to Genetic Engineering and Designer Organisms, which build the source animals; to Precision Medicine for matching and surveillance; to Xenobiology, which owns how far a chimeric organism can be taken; and to Artificial Wombs, the same engineering problem — living tissue sustained outside a body on an artificial circulation — at whole-organism scale.

Handwave “Scientists have three-dimensionally printed a working human heart.” The 2019 Tel Aviv construct was rabbit-sized, held the right patient-derived cells, vessels and chambers, and did not pump; the team's own statement was that the cells can currently contract, but they need them to work together. Any argument running from “a heart has been printed” to “organ printing is now an engineering problem” does all its work at this step, and it is the most over-reported result in the subject.

Frontier “Organoids are mini-organs, and vascularisation is being solved in the dish.” Organoids hold 10^4 to 10^6 cells against a liver's 160 to 360 billion, stay fetal-like indefinitely, and have no vasculature; the error is reinforced by the word itself. Established In every published success the vessels that actually carry blood are the host's: co-cultured organoids show flow only after transplantation into mouse cortex, and grafts are penetrated by host vessels. The best purely in-vitro answer is to make the tissue thinner: sliced organoids at an air–liquid interface survive eight to ten months without necrotic cores at 200 to 500 µm. Frontier That is an accommodation to the diffusion limit rather than a defeat of it.

Frontier “Pig organs need about ten gene edits to work.” A GGTA1-knockout-only pig kidney — one edit, no CD40 blockade, no other modification — supported dialysis independence for 61 days in a human decedent, and its day-33 antibody-mediated rejection was completely reversed. Frontier The authors identify pre-existing xenoreactive T cell clones, not complement, as the residual barrier — so the complement and coagulation transgenes dominating the ten-edit designs may address a problem that is no longer rate-limiting. Speculative The marginal contribution of edits two through ten is open, and this is the strongest steelman available for a minority position in the field. Frontier The mainstream reply is fair: 61 days in a brain-dead recipient is not 271 in a living one, the decedent model cannot show chronic rejection, and n equals one.

Established “The 271-day record belongs to the NYU Langone team, and the pig had 69 gene edits.” This is wrong four ways, and it is wrong in the peer-reviewed literature, not in a newspaper. A 2025 review in a peer-reviewed immunology journal tabulates the 271-day patient as NYU Langone, Montgomery group, 69 gene-edited. Established The 271-day patient is Tim Andrews, transplanted at Massachusetts General Hospital by the Riella and Kawai team — not NYU Langone. The NYU Langone case is Towana Looney, a different person, who received a United Therapeutics ten-gene-edited kidney and reached 130 days. Established And the 69-edit figure conflates functional edits with retroviral inactivation sites: eGenesis describes EGEN-2784 as three glycan-antigen knockouts plus seven human transgenes plus genome-wide inactivation of endogenous retroviruses — about ten functional edits plus dozens of retroviral-locus disruptions, which are different kinds of thing and should not be summed.

Established The fourth error is a date, and it is in a Nature-family journal. A 2025 Nature Biotechnology news item on the same case gives the transplant date as February 2024 and the removal as the end of October 2025 — an interval of about twenty months, contradicting the same article's own reference to eight months and the 271-day figure reported everywhere else. Established The correct date is 25 January 2025, given in eGenesis's own release, and the arithmetic settles it without adjudicating between sources: 23 October 2025 minus 271 days is 25 January 2025. Speculative This is an object lesson rather than a gotcha — a company release and a subtraction beat a high-prestige journal's news desk on a simple fact, while a peer-reviewed review propagated a separate set of errors about the same case. Frontier That is how a wrong case table enters the literature and stays there.

Frontier “Xenotransplantation's main risk is a pig virus jumping to humans.” Porcine endogenous retroviruses A, B and C are knocked out in clinical donor pigs and no transmission has been documented, while every human xenograft has been lost or removed to antibody-mediated rejection, thrombotic microangiopathy or declining function. Established The realised risk is immunological and realised in 100 per cent of cases; the zoonotic risk is real, is taken seriously by regulators, and is so far hypothetical against a denominator of a handful of patients — too small to license the word safe, and too small to make zoonosis the principal barrier when the principal barrier has failed every time.

Frontier “Decellularised scaffolds are obviously the answer — nature already built the vasculature.” The scaffold's vascular tree is bare matrix, and bare walls thrombose; two decades have produced no meaningful in vivo function in any recellularised solid organ, on the testimony of the field's own review. Frontier A widely repeated figure attributes about 2 per cent of adult function to the founding 2008 rat-heart result; the primary paper could not be obtained here, so that number is not printed as established. Note what the twenty-year review does not contain either: no recellularisation-efficiency figure, no function percentage, no human-scale cell number.

Frontier “The 2006 lab-grown bladders showed we can grow organs.” A bladder is hollow, thin-walled and low in metabolic demand, so host tissue perfuses it by ingrowth from outside — the one geometry in which the diffusion limit does not bind, which is why the result has stood for two decades without generalising to a single solid organ. Speculative The same reasoning explains the tracheal tissue-engineering failure of 2008 to 2014: a cell-seeded scaffold in a loaded, contaminated, poorly vascularised site does not remodel into an airway. That lesson was buried under the misconduct scandal that followed, and the documenting sources were unobtainable here, so it is offered as a lesson rather than a case history.

Frontier “Organoid research is aimed at growing transplantable organs, and a printed lung is about to enter trials.” The dominant institutional pull on organoids is toxicology, and a toxicology model needs no vascularisation, maturation or scale; the 2022 first-in-human organoid transplant was framed by its own investigators as establishing safety and efficacy rather than having established them. Frontier The printed-lung five-year timeline was stated in June 2022 and is checkable in mid-2027.

Speculative The steelman the mainstream rejects: the field may be building the wrong object. If borrowed vasculature works for organs whose function is mass-dependent, if minimal editing suffices, and if the honest product is a bridge, then the field is optimising a target — the organ-shaped, fully vascularised, matched construct — that no clinical pathway requires. Frontier The mainstream reply is strong: hearts cannot be bridged indefinitely, ectopic kidneys do not concentrate urine, every borrowed-vasculature result is small or early, and a field that abandons the hard target will still have no heart in fifty years. Frontier Serious people hold both positions, the evidence does not separate them, and the measurement that would — a time-to-thrombosis benchmark for an organ-scale vascular tree — has never been published by anyone.