Regenerative-system dysfunction in organ fibrosis: Maladaptive regeneration, fibrotic memory, and therapeutic reprogramming

Abstract

Organ fibrosis involves extracellular matrix accumulation and dysfunction of interacting regenerative, immune, and stromal compartments. This narrative review integrates evidence from seven organs within an injury-repair-memory framework while distinguishing bona fide stem/progenitor cells from differentiated parenchymal cells and niche components. Maladaptive regeneration includes impaired lineage progression, senescence, persistent injury-associated states, and pathological cell-niche signaling. Fibrotic memory is evaluated using explicit persistence, stimulus-withdrawal, re-exposure, and causal-maintenance criteria rather than equated with ongoing disease. Direct evidence is restricted to particular models, including cultured fibroblast mechanical memory, selected stromal epigenetic and activation states, conditioned immune responses, and post-obstructive renal epithelial senescence. Cross-organ generalization, metabolic-memory mechanisms, and cell-autonomous memory in transitional epithelia remain incompletely established. Therapeutic reprogramming denotes controlled, durable correction of pathological cell states, identities, or differentiation trajectories; senolysis, niche modulation, cell replacement, and organoid-guided screening are distinct complementary approaches. Conflicting findings underscore disease-stage and cell-type dependence. We propose a translational roadmap prioritizing biomarkers and human-relevant models before durable therapeutic state correction, with explicit assessment of function, safety, and persistence after treatment withdrawal.

Key Words: Fibrosis; Regeneration; Stem cells; Cellular senescence; Epigenesis; Genetic; Extracellular matrix; Cellular reprogramming; Fibrotic memory

Core Tip: Fibrosis can involve failed tissue regeneration as well as scar accumulation. This review compares organ-specific evidence for maladaptive repair and fibrotic memory without treating all repair-associated cells as stem cells. Direct memory evidence is separated from persistent abnormalities during ongoing disease. Therapeutic reprogramming requires controlled, durable cellular-state or lineage correction; extracellular matrix normalization, immune and metabolic modulation, cell replacement, and organoid-guided screening remain distinct. The proposed framework prioritizes testable mechanisms, context-dependent reversibility, and safety rather than assuming that preclinical state changes restore human tissue function.

INTRODUCTION

In idiopathic pulmonary fibrosis (IPF), pirfenidone and nintedanib slowed functional decline in pivotal trials, but those trials did not demonstrate restoration of normal tissue architecture[1,2]. More broadly, fibrosis involves interacting injury, repair, immune, and matrix pathways rather than excessive scar production alone[3]. These limitations motivate examination of regenerative-system dysfunction alongside conventional antifibrotic mechanisms.

During chronic injury, bona fide tissue stem/progenitor populations and injury-responsive differentiated parenchymal cells are continuously exposed to inflammatory mediators, mechanical stress, genomic damage, and metabolic disturbances. Rather than executing coordinated regeneration, these cells may enter dysfunctional states characterized by exhaustion, senescence, impaired differentiation, or persistence of transitional phenotypes that fail to restore mature tissue architecture. Fibrosis may therefore involve both failed regeneration and disrupted communication among parenchymal, immune, stromal, and extracellular matrix (ECM) compartments[4,5].

In this review, the term stem/progenitor cell is restricted to populations with experimental evidence of self-renewal and/or injury-responsive lineage reconstitution. Differentiated proximal tubular epithelial cells, mature hepatocytes, cardiomyocytes, fibroblasts, macrophages, mesenchymal stromal cells, and the ECM are not treated as a single stem-cell category; they are described as parenchymal, immune, stromal, or niche compartments. In the adult heart, where a broadly accepted resident cardiac stem-cell population has not been established, the discussion focuses on limited cardiomyocyte renewal and fibroblast-state dynamics rather than assuming a bona fide cardiac stem cell[6,7].

This review focuses on three interconnected concepts. Maladaptive regeneration describes injury-associated repair that fails to reconstruct functional tissue. Fibrotic memory is assessed using the operational criteria in “Definition and conceptual boundaries”. Therapeutic reprogramming is used narrowly for controlled, durable correction of a pathological cellular state, identity, or differentiation trajectory; niche or pathway modulation, cell replacement, and organoid modeling are considered separately[3,4].

Rather than discussing fibrosis as isolated organ-specific diseases, we examine lung, liver, kidney, heart, skin, intestine, and bone marrow fibrosis as representative systems that reveal common principles of regenerative failure. We first discuss fibrosis as a consequence of impaired regeneration, then summarize major mechanisms driving stem/progenitor dysfunction, examine the concept and biological basis of fibrotic memory, compare conserved and organ-specific mechanisms, and finally evaluate therapeutic strategies designed to restore regenerative capacity[3,5].

Review methodology and evidence appraisal

This narrative review used a structured search to inform conceptual synthesis rather than a systematic review or meta-analysis. PubMed/MEDLINE and Web of Science Core Collection were searched for English-language articles published from January 1, 2000, through June 1, 2026. The core query combined fibrosis-related terms with regeneration- and memory-related terms: (fibrosis OR fibrotic) AND (“stem cell” OR progenitor OR “failed repair” OR “maladaptive regeneration” OR “transitional state” OR senescence OR “epigenetic memory” OR “mechanical memory” OR “trained immunity” OR organoid) AND (lung OR liver OR kidney OR heart OR skin OR intestine OR “bone marrow”). Reference lists were also screened for relevant primary studies and major conceptual reviews.

Eligible reports addressed fibrosis-associated repair, cellular or niche persistence, or a directly relevant experimental method in at least one of the seven organs. Human tissue studies, lineage-resolved experiments, genetic or pharmacological perturbation, injury withdrawal, re-exposure, and reversibility tests were prioritized. Reviews were retained for background and comparison, not as substitutes for primary evidence supporting specific mechanistic grades. Reports lacking a relevant fibrosis or repair endpoint, duplicate reports, and unsupported cross-organ extrapolations were excluded from the evidence tables. Non-English full texts were outside the search scope; untranslated reports were not used to support evidence grades. Preprints and conference abstracts were not used as graded mechanistic evidence; an available peer-reviewed final article was preferred. The seven organs represent epithelial, mechanically stressed, and hematopoietic/stromal repair systems. Selection was purposive rather than exhaustive and may favor well-studied organs, accessible English-language publications, positive findings, and studies fitting the proposed framework. Contradictory and negative findings were therefore explicitly considered, but this does not eliminate selection or publication bias.

Evidence grades in Tables 1, 2, 3 and 4 apply only to the specified organ, model, cell type, and endpoint. Strong denotes convergent direct perturbational or persistence evidence with human support or independent replication; moderate denotes a direct but context-limited experiment, or convergent associations without complete causal testing; emerging denotes descriptive or insufficiently tested evidence for the particular claim. Study directness is stated separately: Lineage tracing establishes origin or fate, whereas causal contribution requires an appropriate perturbation; human association is not human therapeutic efficacy. A strong grade for a mechanism does not establish memory, reversibility, or clinical benefit. Memory grading follows “Definition and conceptual boundaries” and Table 2, with ongoing-injury associations separated from direct persistence tests.

Table 1 Mechanisms contributing to regenerative-system dysfunction in organ fibrosis.

Module	Relevant compartment	Organ/model and direct evidence	Scoped evidence grade	Therapeutic implication and limit
Exhaustion and senescence	ATII cells; differentiated renal epithelia; hepatic stellate cells	Mouse ATII perturbation with human IPF observations; renal G2/M perturbation and post-obstructive senolytic treatment[17,19,20]. Stellate-cell senescence can limit liver fibrosis[22]	Strong for specified epithelial mechanisms; exhaustion is not established by senescence markers alone. Direction differs by cell type	Target harmful states selectively. Depletion is not rejuvenation; a pilot trial does not establish fibrosis reversal
Persistent maladaptive repair states	Transitional alveolar cells; proximal tubular cells; hepatic ductular/progenitor-associated cells	Lung transitional-state perturbation; renal stress perturbation; mouse ductular-reaction/Hedgehog experiments[26,27,47]	Strong in selected lung models; moderate, model-specific renal/liver causal support. Human state maps remain associative	Test maturation and functional recovery separately from disappearance of a marker
ECM/mechanotransduction	Fibroblasts; ATII cells; extracellular niche	Cultured lung fibroblast stiffness-switch assay; YAP/TAZ perturbation; murine alveolar tension model; human decellularized lung ECM conditioning[13,32-34]	Direct culture memory; strong selected mechanistic effects. Human in vivo memory and whole-organ restoration are unproven	Normalize harmful cues without assuming all ECM is detrimental or that mechanics alone restores tissue
Inflammatory conditioning	Macrophages and immune-epithelial niches	Conditioned macrophage/rechallenge-transfer experiments in HClO-induced mouse fibrosis; postviral lung niche perturbation[39,42]	Moderate, direct within named models; conditioning effects vary by endpoint. Macrophage NASH epigenomics is not a withdrawal test[43]	Distinguish harmful training, protective tolerance, and continuing niche stimulation
Metabolic/mitochondrial dysfunction	Alveolar and renal epithelia	Mouse/human-associated CD38 epithelial-aging work and renal ferroptotic-stress perturbation[21,47]	Moderate, organ-specific mechanistic support; metabolic-memory erasure remains emerging/unproven	Measure cell fitness and fibrosis separately from durable post-withdrawal state correction
Epigenetic stabilization	Systemic-sclerosis dermal fibroblasts; liver macrophages as a contrasting context	TGFB2 enhancer persistence/perturbation in patient-derived fibroblasts; macrophage chromatin remodeling during ongoing steatohepatitis[14,43]	Strong for the specified enhancer mechanism ex vivo; macrophage memory remains inferred. No generalized progenitor-memory grade	Test durable locus- and cell-specific correction; avoid equating altered chromatin with erased memory
Cross-organ state convergence	Distinct epithelial, immune, and stromal populations	Lung, human liver, and kidney state/niche analyses; functional experiments apply only to the modules actually tested[8,10,36,37]	Moderate for recurrent associations; emerging for a universal causal sequence	Use shared signatures for target nomination, followed by organ-specific perturbation

Grades apply to the named mechanism, model, cell type, and endpoint, not to an entire organ or therapeutic class. Directness, memory, reversibility, and clinical efficacy are separate judgments (“Review methodology and evidence appraisal”). ATII: Alveolar type II; IPF: Idiopathic pulmonary fibrosis; YAP/TAZ: Yes-associated protein/transcriptional coactivator with PDZ-binding motif; ECM: Extracellular matrix; HClO: Hypochlorous acid; NASH: Nonalcoholic steatohepatitis.

Table 2 Operational assessment of proposed fibrotic-memory layers.

Layer/carrier	Organ/model and test actually performed	Directness and scoped grade	Reversibility and remaining test
Epigenetic/dermal fibroblasts	Systemic-sclerosis fibroblasts: Retained TGFB2 enhancer program and chromatin-directed perturbation[14]	Direct ex vivo; strong for this enhancer mechanism, not for all chromatin signatures	Molecular program is perturbable in the assay; durable human tissue reversal is not established
Mechanical/fibroblasts	Cultured lung myofibroblasts transferred from stiff to soft substrates[13]	Direct culture persistence; moderate for generalization beyond the model	Retention depends on exposure history and assay conditions. Clinical memory erasure is untested
Inflammatory/macrophages	Conditioning, subsequent challenge, and transfer in an HClO-induced mouse fibrosis model[39]	Direct, model-specific response modulation; moderate. BCG and low-dose LPS have different effects	Fibrosis-modifying responses are demonstrated; durable normalization of human immune states is not
Niche/ECM	Human decellularized IPF matrix conditions cultured fibroblasts; postviral mouse lung immune-epithelial niche is functionally perturbed[34,42]	Direct niche instruction/persistence in these settings; moderate. Continued niche exposure is not cell-autonomous memory	Require removal of residual stimuli, cross-transfer controls, and durable functional follow-up to establish autonomous retention
Metabolic/epithelial cells	Renal metabolic-state and ferroptotic-stress studies demonstrate injury-associated dysfunction, not the full withdrawal-retention-reversal sequence[46,47]	Inferred; emerging for metabolic memory	Direct metabolic-memory erasure is unproven; test after stress and treatment withdrawal
Epithelial state/renal or alveolar cells	Senescent renal epithelia persist after relief of obstruction; senolytic treatment improves fibrosis without epithelial-specific targeting[20]. KRT8-associated lung states are a distinct context[8,25]	Direct, moderate evidence for the post-obstructive senescent-state example; KRT8 cell-autonomous memory remains inferred	Depletion is not reprogramming of surviving cells. Durable maturation and organ recovery require separate demonstration
Stromal activation state/hepatic stellate cells	Lineage-traced stellate cells deactivate after experimental liver-injury withdrawal but retain enhanced reactivation potential[52]	Direct, moderate evidence for sensitized activation-state memory; its epigenetic mechanism is not established by this experiment alone	Partial deactivation is demonstrated; complete return to a naive state or durable therapeutic memory erasure is not

Operational criteria: Document antecedent exposure, a suitable resolution/rest interval, and separation from continuing injury. Direct support comes from withdrawal-associated persistence, altered re-exposure responses, or selective perturbation of an already retained state. Perturbation only during ongoing injury is not a memory test. Molecular correction, cell depletion, histological regression, and functional recovery are different outcomes. TGFB2: Transforming growth factor beta 2; HClO: Hypochlorous acid; BCG: Bacillus Calmette-Guérin; LPS: Lipopolysaccharide; ECM: Extracellular matrix; IPF: Idiopathic pulmonary fibrosis; KRT8: Keratin 8.

Table 3 Organ-specific repair compartments, mechanistic evidence, memory, and reversibility.

Organ/context and compartment	Mechanism: Model and evidence grade	Memory: Directness and boundary	Reversibility/translational limit
Lung/IPF and postviral injury. ATII and airway progenitors; fibroblasts and immune niche are separate	Strong in selected mouse ATII-senescence, transitional-state, and mechanical-tension experiments; human observations support relevance[17,27,33]	Direct cultured fibroblast mechanical retention[13]. Postviral niche persistence is distinct from cell-autonomous epithelial memory[42]	Preclinical effects are context-specific; the open-label senolytic pilot establishes feasibility, not antifibrotic efficacy[23]
Kidney/AKI-CKD. Differentiated proximal tubular epithelia	Direct G2/M and ferroptotic-stress perturbation in mouse injury models; strong/moderate within those mechanisms. Human failed-repair signatures are associative[9,19,47]	Direct, moderate post-obstructive persistence and senolytic-response evidence, not epithelial-specific ablation[20]; not every failed-repair signature meets memory criteria	Partial experimental benefit does not establish full renal reconstruction or reprogramming of surviving cells
Liver/steatohepatitis, cirrhosis, and experimental regression. Hepatocytes, ductular/progenitor-associated cells, macrophages, stellate cells	Moderate direct ductular/Hedgehog mechanism in mouse models; human cirrhotic niche interactions add distinct evidence[10,26]	Macrophage epigenetic adaptation during ongoing disease is inferred memory[43]. Direct sensitized stellate-cell reactivation follows experimental regression[52]	Partial stromal inactivation differs from memory erasure. Arid1a deletion supports murine fibrosis reduction/regression, not human efficacy[63]
Intestine/fibrostricturing Crohn’s disease. Crypt stem cells and fibroblasts are distinct	Moderate direct cadherin-11 fibroblast evidence from patient-derived cells and experimental colitis fibrosis[56]	Emerging/unproven for stem-cell memory or an autonomous epithelial cause of fibrostenosis	Stromal pathway modification is not demonstrated restoration of crypt function or established stricture reversal
Heart/early injury vs established scar. Cardiomyocytes and resident fibroblast lineages	Strong direct lineage/ablation evidence for periostin-positive myofibroblast contribution to mouse scar and necessary acute healing[57]	Post-injury deactivation is observed; retained pathogenic responsiveness and therapeutic memory erasure are not established by that observation alone[57]	Experimental cardiac lineage conversion[58] and negative chronic-scar cell therapy[7] address different stages; neither establishes human scar reversal
Skin/experimental scarring and systemic sclerosis. Dermal fibroblast and adiponectin-marked lineages	Strong lineage/functional support for selected dermal populations; lineage origin alone does not establish every population’s necessity[59,60]	Direct ex vivo TGFB2 enhancer-memory evidence in systemic-sclerosis fibroblasts[14]; not all progenitors	Enhancer perturbability and lineage plasticity do not establish complete or safe human tissue reversal
Bone marrow/myelofibrosis models. Hematopoietic progenitors interact with distinct stromal populations	Strong Gli1 lineage/ablation evidence in ThPO/JAK2-associated models; LepR/PDGFRα evidence in a separate marrow model; supportive human findings[11,61]	Emerging/inferred memory: These studies do not perform a definitive retained-state withdrawal/rechallenge test	Preclinical stromal targeting does not establish durable human memory erasure; Gli1 and LepR labels are not assumed equivalent

Evidence is specific to the named cell, organ, model, and endpoint; no overall cross-organ confidence score is assigned. IPF: Idiopathic pulmonary fibrosis; ATII: Alveolar type II; AKI-CKD: Acute kidney injury-chronic kidney disease; TGFB2: Transforming growth factor beta 2; ThPO: Thrombopoietin; LepR: Leptin receptor; PDGFRα: Platelet-derived growth factor receptor α.

Table 4 Classification and evidence status of therapeutic strategies.

Approach/category	Organ, model, and target	Directly supported effect/evidence stage	What is not established/main limitation
Progenitor-state rejuvenation/reprogramming only if durable and controlled	Partial-reprogramming experiments in aging/progeroid mice and tissue-repair settings[53]	Preclinical improvement in selected age-associated cellular and tissue phenotypes	Not a general fibrosis trial, a validated memory-erasure assay, or established human progenitor rejuvenation
Epigenetic state correction/candidate direct reprogramming	Systemic-sclerosis dermal fibroblast enhancer perturbation[14]; Arid1a deletion in murine liver-fibrosis/regression models[63]	Direct locus-level mechanism ex vivo; preclinical liver-fibrosis reduction, including intervention after injury cessation	No established human antifibrotic efficacy or complete memory erasure; delivery, identity stability, and oncogenic risk require separate testing
Direct lineage conversion/reprogramming	Murine cardiac fibroblasts converted toward induced cardiomyocytes after injury[58]	Animal proof of lineage conversion and selected functional effects	Not reversal of established chronic human scar; efficiency, maturation, integration, and lineage control remain limiting
Senomorphics/state modulation; senolytics/cell depletion	Mouse pulmonary fibrosis, post-obstructive renal senescence, and an open-label human IPF pilot[15,20,23]	Preclinical benefit in specified endpoints; human pilot feasibility	Senolysis is not reprogramming; pilot data do not establish antifibrotic efficacy. Beneficial senescent states may be lost
ECM/mechanotransduction modulation/niche intervention	Cultured lung fibroblast stiffness-switch and YAP/TAZ perturbation models[13,32]	Direct effects on mechanical retention or activation in the particular assay; primarily preclinical	Matrix instruction is not cell-intrinsic memory; clinical restoration and optimal timing are unproven
Immune-state modulation/complementary unless durable resetting is shown	Macrophage conditioning in mouse systemic-sclerosis-like fibrosis[39]; IL-22 liver progenitor studies[64]	Conditioning changes selected fibrosis outcomes; IL-22 promotes progenitor expansion experimentally, with human associations	Opposing conditioning effects; IL-22 proliferation is not demonstrated fibrosis reversal. Infection and impaired repair require assessment
Metabolic correction/state modulation	CD38-related alveolar aging and renal ferroptotic-stress models[21,47]	Direct preclinical modification of specified stress/repair pathways and fibrosis-associated outcomes	No complete withdrawal-based metabolic-memory erasure; systemic and cell-specific effects need evaluation
Cell replacement/separate regenerative strategy	Tested cell products in chronically scarred rodent hearts[7]	Negative primary evidence: No rejuvenation in that chronic-scar setting	No class-wide efficacy inference; engraftment, maturation, host response, and disease stage must be tested for each product
Organoid-guided screening/research platform, not itself therapy	Human pluripotent-stem-cell-derived nephron organoids and expandable adult liver cultures[50,51]	Direct platform evidence for development/injury modeling and culture-based experimentation	Memory washout/rechallenge protocols are a proposed extension here; no direct patient effect from screening alone. Incomplete niche and model fidelity remain limitations

A mechanistic or model-specific result is not a clinical evidence grade. Reprogramming requires controlled and durable intrinsic state, identity, or trajectory correction. Depletion, transient pathway modulation, matrix correction, cell delivery, and screening are different categories. IPF: Idiopathic pulmonary fibrosis; ECM: Extracellular matrix; YAP/TAZ: Yes-associated protein/transcriptional coactivator with PDZ-binding motif; IL-22: Interleukin-22.

Box 1: Concise glossary

Regenerative-system dysfunction: Failure of coordinated repair involving distinct stem/progenitor, differentiated parenchymal, immune, stromal, and extracellular-niche compartments; an integrative framework, not a single cell identity.

Maladaptive regeneration: An injury-associated repair response that fails to complete functional tissue restoration and can sustain pathological remodeling.

Fibrotic memory: A retained fibrosis-relevant state meeting the operational criteria in “Definition and conceptual boundaries” and Table 2; not a synonym for ongoing chronic disease.

Direct reprogramming: Controlled alteration of cellular identity or a durable intrinsic state; direct lineage conversion changes identity without an intended pluripotent intermediate.

State modulation: Modification of cellular activity or responsiveness without demonstrated durable identity correction; senolysis is cell depletion, not state reprogramming.

Niche modulation: Modification of ECM, mechanics, or intercellular signals without necessarily changing the intrinsic identity of resident cells.

FIBROSIS AS A DISEASE OF FAILED REGENERATION

From acute repair to maladaptive regeneration

Successful tissue repair requires a coordinated sequence of stem and progenitor cell activation, proliferation, lineage commitment, and niche-guided maturation. In acute injury, transient scar formation may provide structural support while regenerative programs restore tissue integrity. However, under persistent or repetitive injury conditions, this tightly regulated process becomes progressively disrupted[3,4].

Chronic exposure to DNA damage, oxidative stress, inflammatory cytokines, altered mechanical forces, and metabolic imbalance continuously activates injury-response pathways in regenerative cells. Instead of progressing toward tissue restoration, these cells may remain trapped in persistent wound-associated states. Thus, maladaptive regeneration does not represent a simple failure of repair; rather, it reflects an abnormal repair process in which injury-responsive cellular programs remain activated without successfully rebuilding normal tissue architecture[3,4]. This concept offers a possible explanation for continued fibrosis after the initial trigger has been reduced. Whether repair-associated states become autonomous or remain dependent on residual injury and niche signals must, however, be tested rather than assumed[4].

Regenerative-cell dysfunction as an active contributor to fibrosis

Stem and progenitor cell dysfunction has traditionally been interpreted as a secondary consequence of an already established fibrotic environment. However, emerging evidence suggests that dysfunctional regenerative cells may actively contribute to fibrosis progression[4,8,9]. Under chronic stress conditions, regenerative cells can promote fibrosis through several mechanisms, including secretion of inflammatory mediators, failure to complete appropriate differentiation programs, abnormal communication with fibroblasts and immune cells, and remodeling of the extracellular niche[3,5]. Epithelial failed-repair states are well characterized in alveolar and renal tubular systems[8,9]. Human liver niche analyses, marrow stromal perturbation studies, and the intestinal fibrosis literature illustrate distinct parenchymal, stromal, and immune contributions[10-12]. These examples should not be interpreted as equivalent evidence for stem-cell-driven fibrosis in every organ. Therefore, fibrosis may not simply occur when repair fails to proceed. In some circumstances, fibrosis may represent the consequence of repair continuing in an abnormal and unresolved cellular state[4].

An evidence-calibrated injury-repair-memory framework

Individual components of the injury-repair sequence are not new: Chronic injury can impair regenerative cells; unresolved repair can activate fibroblasts and remodel the niche; and persistent chromatin or mechanical states can maintain profibrotic behavior[3-5,13,14]. The contribution of the present review is the cross-organ integration of these components and the explicit separation of evidence-supported modules from proposed connections. We therefore do not present the entire sequence as a single universally established mechanism.

We synthesize a context-dependent sequence: Chronic or repetitive injury → regenerative-system stress → persistent maladaptive repair → failed restoration and niche remodeling → fibrosis. Fibrotic memory is proposed as a persistence layer that can stabilize particular cellular or niche states after substantial stimulus reduction. Failed repair alone does not establish such memory. In Figure 1, solid arrows identify locally supported interactions, whereas dashed arrows identify proposed cross-organ connections or incompletely validated memory feedback. These modules are not a universally established causal chain. Interventions directed at repair dysfunction or pathological persistence may complement antifibrotic therapy, but their effects require organ-specific testing[3,4] (Figure 1).

Figure 1 Schematic framework linking chronic injury, maladaptive regeneration, fibrotic memory, and progressive fibrosis. Chronic or repetitive injury, including inflammatory, mechanical, genomic, and metabolic stress, disrupts regenerative cell function and promotes senescence, transitional arrest, impaired differentiation, metabolic dysfunction, and epigenetic locking. These maladaptive cellular states alter inflammatory signaling and interact with immune cells and fibroblasts, promoting fibroblast activation, extracellular matrix (ECM) deposition, matrix stiffening, and Yes-associated protein/transcriptional coactivator with PDZ-binding motif-dependent mechanotransduction. Reciprocal signaling among regenerative cells, immune cells, fibroblasts, and the ECM can reinforce failed repair, maladaptive regeneration, and progressive fibrosis. Epigenetic, niche, inflammatory, and metabolic memory are shown as candidate mechanisms that may sustain pathological responses after the initiating injury has been reduced. Green solid arrows indicate supported activation or feedback relationships, blue solid arrows indicate pathological progression, red dashed lines indicate inhibitory relationships, and gray dashed arrows indicate proposed or less well-established connections. The strength and persistence of these mechanisms are context-, cell type-, and organ-dependent, and the different memory layers should not be interpreted as having equivalent levels of evidence. ECM: Extracellular matrix; YAP/TAZ: Yes-associated protein/transcriptional coactivator with PDZ-binding motif.

MECHANISMS OF REGENERATIVE-SYSTEM DYSFUNCTION IN FIBROTIC REPAIR

Exhaustion and senescence

Stem/progenitor cell exhaustion and cellular senescence represent two related but distinct forms of regenerative impairment. Exhausted progenitor cells exhibit reduced proliferative potential and impaired differentiation capacity after prolonged stress exposure, whereas senescent cells undergo relatively stable cell-cycle arrest while maintaining active transcriptional programs[5,15,16].

In lung models, alveolar type II (ATII) cell senescence and age-associated progenitor dysfunction can impair epithelial renewal and promote fibrosis[17,18]. In kidney injury models, proximal tubular G2/M arrest contributes causally to profibrotic signaling[9,19]. Senescent renal epithelia can also persist after relief of obstruction, and senolytic treatment reduces fibrosis in that setting, although the drug intervention is not epithelial-specific[20]. These findings support particular dysfunctional states, not universal progenitor exhaustion across organs.

CD38-dependent alveolar epithelial aging provides an additional lung-specific link between metabolic dysfunction and fibrosis[21]. The direction of senescence effects nevertheless differs by compartment: Senescence of activated hepatic stellate cells limits experimental liver fibrosis and facilitates its resolution[22]. Thus, epithelial senescence associated with failed repair should not be equated with the transient senescence of fibrogenic stromal cells that restricts scarring. Timing and cell identity are essential to interpreting apparently conflicting findings. From a therapeutic perspective, senescence-prone regenerative states represent attractive targets, although selective intervention remains challenging because senescence is biologically heterogeneous and not universally pathological[15,23,24].

Impaired lineage progression and persistent maladaptive repair states

Effective regeneration requires appropriate lineage progression, but chronic injury can arrest injury-responsive cells in persistent transitional or failed-repair states. We use the term impaired lineage progression rather than lineage infidelity unless direct fate-mapping demonstrates adoption of an inappropriate lineage[4,8,9].

Keratin 8 (KRT8)-positive transitional alveolar states normally participate in ATII-to-ATI differentiation but may persist in fibrotic lungs[8,25]. Proximal tubular epithelial cells can adopt inflammatory failed-repair states[9], whereas progenitor-associated ductular reactions can promote stromal signaling in experimental liver injury[26]. These are different repair contexts; proximal tubular cells are injury-responsive differentiated parenchymal cells, not a bona fide stem-cell population. These states should not be interpreted as direct epithelial-to-fibroblast conversion without lineage-tracing evidence. Their fibrogenic contribution is more plausibly mediated by prolonged inflammatory and stromal signaling and continued niche remodeling[8,9,27].

State identification alone remains descriptive. Lineage tracing resolves ancestry, while targeted perturbation tests functional contribution[28,29]. In the lung, regulated Notch and interleukin (IL)-1β signaling can support airway progenitor plasticity during regeneration[30], whereas persistent transitional-state signaling can promote fibrosis[27]. A repair-associated state is therefore not intrinsically pathological; its duration, maturation trajectory, and signaling context determine its interpretation.

Niche remodeling and ECM-driven dysfunction

Stem and progenitor cell activity is fundamentally dependent on a supportive tissue niche. Fibrosis progressively disrupts this environment by altering ECM composition, increasing matrix stiffness, modifying basement membrane properties, activating fibroblast populations, and impairing vascular support[3,5,31].

Mechanical signaling provides distinct kinds of evidence. Cultured lung myofibroblasts can retain activated behavior after transfer from stiff to soft substrates, directly supporting mechanical memory within that assay[13]. Yes-associated protein/transcriptional coactivator with PDZ-binding motif (YAP/TAZ) perturbation establishes a role in fibroblast mechanosignaling[32], and increased tension on alveolar stem cells promotes experimental lung fibrosis[33]. The latter studies support mechanistic effects, but do not by themselves establish retained memory after withdrawal.

Decellularized human fibrotic lung matrix can drive profibrotic behavior in cultured fibroblasts, directly demonstrating matrix instruction rather than epithelial memory[34]. Human lung, liver, and kidney atlases identify associated pathological cell populations and niches[10,35-37]. These maps do not alone prove that disease-associated ECM causes progenitor dysfunction; any functional validation must be attributed to the specific interaction actually tested.

Niche effects also include cellular signals: Basal-cell-derived WNT7A promotes fibroblast activation in experimental studies of the IPF niche[38]. Together, these findings support testing both matrix and cell-cell interactions. They do not establish that matrix normalization alone restores regenerative competence, or that every matrix component is detrimental[13,32].

Inflammatory imprinting and immune-regenerative miscommunication

Inflammation is an essential component of normal tissue repair. However, fibrosis frequently develops when inflammatory responses fail to transition from injury containment toward regenerative resolution[3,39,40]. Previous inflammatory exposure can alter subsequent responsiveness. Conditioning and transfer experiments in a mouse systemic-sclerosis-like fibrosis model provide direct evidence for fibrosis-modifying trained or tolerized macrophage responses[39]. Separately, severe acute respiratory syndrome coronavirus 2-associated paracrine senescence can sustain inflammatory signaling, but this is not by itself an operational demonstration of fibrotic memory[41].

After viral pneumonia, persistent immune-epithelial niche interactions can impede resolution of abnormal lung repair[42]. During ongoing diet-induced steatohepatitis, macrophages undergo niche-specific epigenetic remodeling[43]; this demonstrates disease-associated state adaptation, not memory after diet withdrawal or fibrotic-memory erasure. In bone marrow, Gli1-positive stromal cells respond to inflammatory niche signals and contribute causally to experimental fibrosis[11]. These organ-specific findings should not be combined into a single established immune-memory mechanism.

Direction and autonomy also matter. In the same fibrosis-conditioning study, Bacillus Calmette-Guérin (BCG) training aggravated selected fibrotic outcomes whereas low-dose lipopolysaccharide conditioning attenuated them[39]. In aging-lung experiments, macrophage responses were strongly shaped by the recipient microenvironment[44]. These results argue against assuming that all persistent immune phenotypes are cell-autonomous or profibrotic, and favor selective modulation over indiscriminate inflammatory suppression.

Metabolic and mitochondrial dysfunction

Metabolic adaptation is increasingly recognized as an important component of fibrotic remodeling. Fibrotic tissues commonly exhibit mitochondrial dysfunction, altered lipid metabolism, increased glycolytic activity, and disturbed redox homeostasis[3,45]. In renal injury models, adaptive and fibrotic repair trajectories differ in metabolic pathway activity[46], and ferroptotic stress promotes pro-inflammatory proximal tubular states with functional evidence from experimental perturbation[47]. These findings support a role for metabolic stress in maladaptive repair; they do not demonstrate metabolic memory after stress withdrawal. However, the causal relationship between metabolic dysfunction and fibrosis remains incompletely defined. Metabolic changes may represent both consequences of chronic injury and active mechanisms that stabilize pathological cellular states[45]. Despite these uncertainties, metabolic pathways remain attractive therapeutic targets because they are potentially pharmacologically modifiable and may provide opportunities to improve cellular fitness without directly manipulating cell identity[45].

Multi-omic evidence: State discovery rather than causal proof

Single-cell and spatial approaches identify injury-associated epithelial states, stromal subsets, and fibrotic niches across organs[8,10,36,37]. The atlas component is associative; where these studies include lineage analysis or functional perturbation, stronger inference applies only to the tested population or interaction, not to every inferred communication network. For example, multiomic modeling with functional follow-up nominates NFAT5 as a regulator of proximal tubular failed-repair programs[48]. Accordingly, we distinguish cell-state discovery, lineage assignment, and causal perturbation[28,29]. Organoids and related culture systems provide experimentally tractable models, but their existence does not establish therapeutic efficacy or a validated memory assay[49-51] (Table 1).

FIBROTIC MEMORY: SUSTAINING FIBROSIS BEYOND THE INITIAL INJURY

Definition and conceptual boundaries

In this review, fibrotic memory denotes a retained fibrosis-promoting or fibrosis-sensitizing cellular or niche state that outlasts the expected acute response to the initiating exposure. Operational assessment requires documentation of the antecedent exposure, an appropriate resolution interval, and separation from continuing injury. Direct evidence may comprise: (1) Persistence after substantial stimulus reduction or withdrawal; (2) An altered response to re-exposure following a defined recovery or rest period; or (3) Selective perturbation of an already retained state that changes later fibrotic behavior. A perturbation performed only during continuing injury establishes pathway involvement, not memory. Residual injury and ongoing niche stimulation must be excluded or explicitly accounted for; matrix-conditioned behavior while cells remain on diseased ECM does not establish cell-autonomous memory. Findings observed only during ongoing chronic disease are classified as persistent disease states or inferred memory. Directness and strength are assessed separately, as summarized in Tables 2 and 3; the terms epigenetic, mechanical, inflammatory, and metabolic specify candidate mechanisms rather than interchangeable proof of memory[13,14,39].

Cellular carriers of fibrotic memory

Candidate carriers include stromal cells, selected epithelial states, conditioned immune cells, and retained extracellular niches. Direct support differs substantially among these compartments[13,14,20,39]. They should therefore be evaluated by the operational tests above rather than grouped solely because they remain abnormal.

Stromal examples include cultured lung fibroblast mechanical memory and retained enhancer activity in systemic-sclerosis fibroblasts[13,14]. In experimental liver-fibrosis regression, lineage-traced hepatic stellate cells can enter an inactive but sensitized state and reactivate more readily on subsequent stimulation[52]. This supports activation-state memory, but does not by itself identify the responsible epigenetic mechanism.

Epithelial evidence is not uniformly indirect. Post-obstructive renal epithelial senescence has withdrawal-associated persistence and senolytic-response support, although the intervention is not epithelial-specific[20], whereas many KRT8-associated lung states are identified during persistent disease and lack equivalent cell-autonomous memory tests[8,25]. A postviral immune-epithelial niche can remain pathological after infection, but this does not establish that epithelial cells retain the state independently of immune signals[42].

Memory formation is therefore context-dependent and must be distinguished from persistent stimulation within an abnormal niche. A retained cellular state, a maintained extracellular environment, and their reciprocal interaction are experimentally separable explanations.

Epigenetic memory

Selected stromal models provide direct epigenetic evidence. Systemic-sclerosis dermal fibroblasts show persistent activation of a transforming growth factor beta 2 (TGFB2) enhancer, with mechanistic evidence from chromatin-directed perturbation[14]. This is a defined ex vivo fibroblast context, not proof that all fibrosis-associated chromatin changes encode durable memory. By contrast, niche-specific macrophage epigenetic remodeling in diet-induced steatohepatitis was characterized during disease[43]. Without comparable withdrawal, rest/rechallenge, or retained-state perturbation experiments, it supports disease-associated remodeling rather than direct inflammatory or epigenetic fibrotic memory. Accordingly, strong mechanistic support for a particular enhancer program should not be extended to macrophages or regenerative progenitors merely because their chromatin profiles differ in fibrosis. The latter populations require their own persistence and functional tests[14,43]. Although epigenetic interventions offer the possibility of directly modifying pathological cellular states, their therapeutic application remains challenging because restoring normal identity without inducing unwanted plasticity requires precise control[53,54].

Inflammatory memory

Inflammatory memory is best assessed through conditioning, a defined interval, and subsequent challenge or transfer, rather than persistent inflammatory gene expression alone[39]. In a hypochlorous acid-induced systemic-sclerosis-like mouse model, BCG training and low-dose lipopolysaccharide conditioning produced different macrophage response programs and opposing effects on selected skin and lung fibrotic outcomes[39]. This is direct, fibrosis-relevant evidence within a specific conditioning model, not a broadly validated mechanism across organs or patients. The effects were endpoint-dependent and should not be represented as uniform improvement or worsening of every fibrosis measure. The contrasting conditioning effects support testing which immune states are harmful, protective, or niche-dependent before attempting durable immune-state correction[39,44].

Mechanical and niche memory

Cultured lung myofibroblasts can retain activation after transfer from a stiff to a soft substrate, providing direct mechanical-memory evidence within an experimentally controlled system[13]. YAP/TAZ-dependent mechanosignaling contributes to fibroblast activation, but pathway activation or inhibition alone is not equivalent to a withdrawal-based memory test[32].

ECM effects are not uniformly pathological. Increased mechanical tension on alveolar stem cells promotes experimental pulmonary fibrosis[33], whereas hyaluronan-Toll-like receptor 4 signaling supports surfactant-protein-C-positive alveolar progenitor renewal and protects against severe fibrosis in mice[55]. Decellularized fibrotic ECM also instructs fibroblast behavior[34]. These findings distinguish harmful mechanical or compositional changes from protective matrix signals; none alone demonstrates cell-autonomous epithelial memory.

Direct evidence for mechanical retention is strongest in controlled culture-transfer experiments; diseased-matrix experiments establish niche instruction rather than quantitative human in vivo memory[13,34]. Combined matrix and cellular interventions are therefore a testable strategy, not an established requirement for restoring function.

Metabolic memory

Metabolic and mitochondrial abnormalities can accompany persistent repair defects and interact with other regulatory programs, but this provides a rationale for a metabolic-memory hypothesis rather than its demonstration[45-47]. A direct test would require persistence after metabolic-stressor withdrawal and selective correction of the retained state with an effect on later fibrotic behavior. The cited renal stress and repair studies do not establish this sequence, so durable metabolic-memory erasure remains unproven in these examples[46,47].

Evidence strength, reversibility, and unresolved questions

The strongest direct memory evidence is context-specific: Cultured fibroblast mechanical retention, a selected systemic-sclerosis enhancer program, and sensitized hepatic stellate cells after fibrosis regression[13,14,52]. Conditioned macrophage responses and post-obstructive renal epithelial senescence provide additional model-specific functional evidence[20,39]. Most claims about cell-autonomous KRT8-state memory or metabolic-memory erasure require further testing; retained postviral niche signaling should be distinguished from autonomous epithelial memory[8,42,47]. These distinctions, including direct vs inferred evidence and the degree of reversibility, are summarized in Table 2.

Several important questions remain unanswered: (1) Are different memory layers sequential events or parallel processes? (2) Which cell populations dominate memory formation in different organs? (3) Can pathological memory be selectively erased without impairing tissue identity? and (4) Is memory reversal required for functional regeneration, or is partial attenuation sufficient? Addressing these questions will be essential for converting a persistence framework into measurable therapeutic targets, rather than treating a descriptive signature as proof of reversibility[3,5].

CROSS-ORGAN EVIDENCE SUPPORTING AN EVIDENCE-CALIBRATED FRAMEWORK

Standardized organ-by-organ comparison

Lung: ATII cells and airway progenitors participate in epithelial repair. ATII senescence, persistent transitional-state signaling, and excessive mechanical tension have functional support in mouse fibrosis models with complementary human observations[17,27,33]. Direct mechanical memory is demonstrated in cultured fibroblasts[13]. Persistent postviral immune-epithelial niche signaling represents a different mechanism and does not establish cell-autonomous epithelial memory[42]. Preclinical state correction and the feasibility of an early senolytic pilot must not be equated with reversal of chronic human fibrosis[23].

Kidney: Proximal tubular epithelial cells are differentiated injury-responsive cells. Perturbational studies support roles for G2/M arrest and ferroptotic stress in maladaptive repair[19,47], whereas human transcriptomic signatures establish association and biomarker potential[9]. Renal epithelial senescence after relief of obstruction provides a selected persistence example with a senolytic treatment response, not epithelial-specific ablation evidence[20]. This exception should not be generalized to every failed-repair state or interpreted as complete restoration of renal function.

Liver: Mature hepatocytes are principal parenchymal regenerative cells, while ductular/progenitor responses are context-dependent. Experimental progenitor-associated ductular reactions can promote Hedgehog-mediated stromal signaling[26]; human cirrhotic-niche data provide complementary interaction evidence[10]. Macrophage epigenetic changes during steatohepatitis do not independently prove post-withdrawal memory[43]. In contrast, lineage-traced hepatic stellate cells can retain enhanced reactivation potential after fibrosis regression[52]. Thus, partial stromal inactivation, complete memory erasure, and hepatocyte regeneration are distinct outcomes.

Intestine: Crypt stem cells function within an inflammatory and stromal niche, but direct evidence that their retained state causes fibrostenosis is limited. In fibrostricturing Crohn’s disease, patient-derived fibroblasts and experimental colitis models support a functional role for cadherin-11 in fibrogenic remodeling[56]. This is stromal evidence, not a direct intestinal stem-cell or fibrotic-memory test.

Heart: Adult repair involves limited cardiomyocyte renewal and fibroblast-lineage responses rather than an assumed universal resident cardiac stem-cell population[6]. Lineage tracing and ablation demonstrate that periostin-positive myofibroblasts contribute to scar formation and are required for adaptive healing after myocardial infarction; some cells become less activated after injury cessation[57]. Fibroblast-to-cardiomyocyte conversion is experimental[58], and tested cell therapies did not rejuvenate chronically scarred rodent hearts[7]. These findings distinguish early repair from established scar and do not establish durable human memory erasure.

Skin: Lineage-resolved studies identify intrinsically fibrogenic dermal populations and adiponectin-lineage contributions to myofibroblasts in selected mouse models[59,60]. These findings should not be generalized to all epidermal or adipocyte progenitors. A TGFB2 enhancer program in systemic-sclerosis fibroblasts supplies a distinct, direct ex vivo epigenetic-memory example[14]. Molecular reversibility in this assay does not establish complete reversal of human skin fibrosis.

Bone marrow: Hematopoietic stem/progenitor cells interact with fibrosis-driving stromal populations. Gli1-lineage tracing and ablation in thrombopoietin- and JAK2-associated models establish a causal stromal contribution, with supportive human observations[11]. A separate study identifies leptin-receptor-expressing stromal cells and platelet-derived growth factor receptor α signaling in experimental myelofibrosis[61]. These marrow studies do not establish post-withdrawal stromal memory or equivalent behavior of all Gli1- and leptin-receptor-marked populations.

Conserved vs organ-specific mechanisms

Across organs, conserved processes include persistent injury-response states, fibroblast activation, ECM remodeling, and disrupted communication among regenerative, immune, and stromal compartments[3,5]. These shared processes support comparative analysis but do not imply that one cellular lineage or one causal sequence operates identically in every organ.

The principal repair compartment and evidentiary basis differ: Epithelial-state dysfunction is functionally supported in selected lung and kidney models[17,19,27,47], while stromal contributions are directly supported in skin and bone marrow[11,59,61]. Human liver niche maps and intestinal fibroblast experiments address different levels of mechanism[10,56]. The framework is therefore a scaffold for organ-specific hypotheses, not evidence for one conserved lineage or a uniform causal sequence.

Therapeutic implications of cross-organ comparison

Cross-organ comparison supports selection of targets by pathological cellular state, but any intervention must be matched to organ-specific cell identity, evidence grade, and reversibility[3,5]. “Therapeutic reprogramming and complementary regenerative strategies” therefore distinguishes direct cellular reprogramming from complementary niche or pathway modulation, cell replacement, and experimental platforms. This classification avoids treating every intervention that influences repair as reprogramming (Table 4).

THERAPEUTIC REPROGRAMMING AND COMPLEMENTARY REGENERATIVE STRATEGIES

Why conventional antifibrotic therapy is insufficient

The pivotal pirfenidone and nintedanib trials in IPF established slowing of functional decline, not reconstruction of normal tissue architecture[1,2]. This distinction motivates investigation of repair-directed approaches alongside established antifibrotic mechanisms[3]. If fibrosis persistence results from both maladaptive regeneration and retained pathological memory, then reducing scar formation alone may leave the underlying regenerative system dysfunctional[3,4]. A broader research objective is to test whether correction of dysfunctional cellular states, pathological niches, and intercompartmental communication improves durable tissue repair[3,4]. Figure 2 presents these outcomes as conditional experimental goals, not established effects of the illustrated approaches.

Figure 2 Therapeutic strategies targeting regenerative dysfunction and their conditional outcomes in organ fibrosis. In this review, therapeutic reprogramming is reserved for interventions intended to produce a controlled and durable correction of a pathological cellular state, identity, or differentiation trajectory. Direct epigenetic reprogramming, lineage conversion, and durable progenitor-state rejuvenation meet this definition. Senolytic depletion, conventional pathway inhibition, extracellular matrix normalization, immunomodulation, cell replacement, and organoid testing are biologically distinct; they are classified as complementary state-modifying, niche-modifying, replacement, or research strategies rather than automatically labeled as reprogramming. Fibrotic regenerative dysfunction is sustained by pathological cellular states and altered epigenetic, inflammatory, mechanical/niche, and metabolic signals. Candidate interventions are grouped into three conceptually distinct categories: (1) Direct cellular reprogramming, including progenitor-state rejuvenation and epigenetic resetting; (2) State and niche modulation, including extracellular matrix-Yes-associated protein/transcriptional coactivator with PDZ-binding motif normalization, immune-state regulation, and metabolic restoration; and (3) Regenerative replacement and organoid-based approaches, including cell replacement and organoid-guided modeling or repair. These strategies are proposed to restore regenerative competence, promote resolution of maladaptive transitional states, recover functional differentiation, support tissue restoration, and reduce fibrosis. Dashed connections indicate hypothesized or experimentally testable relationships rather than established therapeutic effects. Durable benefit requires organ- and disease-stage-specific validation, persistence after treatment withdrawal, functional assessment, and safety evaluation. ECM: Extracellular matrix; YAP/TAZ: Yes-associated protein/transcriptional coactivator with PDZ-binding motif.

Cell-state rejuvenation and senescence targeting

Progenitor-state rejuvenation aims to restore proliferative and differentiation competence within viable regenerative cells. Senomorphic approaches modulate senescence-associated signaling and may qualify as state modulation when effects are durable; senolytics eliminate selected cells and should be described as cell depletion rather than reprogramming[15,23,62]. Early clinical investigation of senolytics in IPF established feasibility but did not establish antifibrotic efficacy. Experimental studies support selective targeting of senescent or exhaustion-prone states in selected models[15,23,24]. Selected models support improvement in particular repair-associated phenotypes, but the IPF pilot did not establish durable correction of fibrotic memory[15,23]. The post-obstructive kidney example demonstrates that depletion can improve fibrosis after injury relief, not that surviving cells have been reprogrammed or that memory has been completely erased[20]. Major limitations include the heterogeneity of senescent states and the possibility that indiscriminate elimination of senescent cells may interfere with physiological repair processes[15]. Potential risks and clinical uncertainties depend on the targeted senescent population, intervention, and exposure schedule; they should not be inferred indiscriminately from the tumor risks of lineage reprogramming[62].

Epigenetic and lineage reprogramming

A demonstrated retained regulatory program provides a more direct target than a chromatin signature observed only during ongoing disease. The TGFB2 enhancer study is a specific example of such mechanistic targeting in systemic-sclerosis fibroblasts[14]. Potential approaches include enhancer-directed regulation, partial cellular reprogramming, and direct lineage conversion. Importantly, Arid1a suppression was tested in murine liver-fibrosis models: Genetic deletion reduced fibrosis, including in an experiment initiated after cessation of chronic CCl4 injury[63]. This is preclinical liver-fibrosis evidence, not merely a regeneration proof of principle; it does not establish human efficacy or operational memory erasure. Cardiac fibroblast-to-cardiomyocyte conversion likewise demonstrates experimental lineage reprogramming rather than reversal of established human cardiac scar[58]. The principal advantage of these approaches is their potential to modify persistent identity programs rather than only downstream fibrotic outputs. Their effects on established fibrosis and operationally defined memory nevertheless remain predominantly preclinical[14,53,58,63]. The cited studies do not establish clinical antifibrotic efficacy for these reprogramming approaches. Achieving sufficient state correction while preserving lineage stability remains a central translational challenge[53,54]. Potential risks include tumorigenesis, abnormal dedifferentiation, and unintended activation of developmental programs[53,54].

Niche and ECM/mechanotransduction modulation

If the extracellular niche actively regulates regenerative behavior, restoring niche integrity is an important complementary therapeutic objective rather than direct cellular reprogramming[13,32]. Potential strategies include ECM normalization, inhibition of pathological crosslinking, modulation of integrin or YAP/TAZ mechanotransduction, and correction of activated fibroblast states[13,32]. Mechanical-memory experiments justify testing whether changing pathological matrix signals improves the durability of a cellular response[13]. Conversely, the failure of tested cell therapies in chronically scarred rodent hearts cautions against assuming that cell delivery alone repairs an established niche[7]. Neither result proves that combined matrix normalization and reprogramming is necessary or sufficient for human tissue restoration. The cited mechanotransduction studies are preclinical. Their readouts support effects on selected fibroblast programs, but precise correction of diseased ECM and durable functional benefit require separate in vivo validation[13,32]. Safety considerations include impaired tissue stability, delayed wound healing, and disruption of normal repair processes[3].

Immune-state modulation to restore pro-regenerative inflammation

Immune intervention should not simply suppress inflammation but should restore the transition from injury containment toward regenerative resolution. It should be termed immune-state reprogramming only when a durable change in cellular responsiveness is demonstrated[3,39]. Candidate strategies include altering conditioning-dependent macrophage responses and correcting pathological cytokine environments. The opposing effects of trained and tolerized responses argue for state- and disease-specific testing rather than routine suppression of a presumed common immune-memory program[39]. IL-22 promoted liver stem/progenitor-cell proliferation in experimental models, with supportive associations in samples from patients with chronic hepatitis B[64]. This finding supports progenitor expansion, not an established antifibrotic treatment in those patients; effects on fibrosis regression or memory remain unproven. However, immune regulation remains highly context-dependent. The same inflammatory pathways may contribute to both pathological fibrosis and beneficial regeneration[3,39]. Targeted immune-state modulation should therefore be distinguished from durable immune reprogramming and evaluated for both pathological and beneficial effects on repair[39].

Metabolic and mitochondrial state correction

Metabolic interventions aim to restore cellular energy balance, mitochondrial function, and redox homeostasis. They should be considered reprogramming only if they produce a durable correction of pathological cell state rather than transient pathway inhibition[45]. Metabolic correction can influence cellular fitness without changing lineage identity. Lung CD38 and renal ferroptotic-stress studies provide organ-specific mechanistic examples[21,47]; they do not show that metabolism alone accounts for, or erases, retained fibrotic information. The cited intervention evidence is primarily preclinical and endpoint-specific. Changes in epithelial aging, stress responses, or fibrosis during treatment should be separated from durable correction after treatment withdrawal[21,46,47]. Important challenges include identifying reliable biomarkers and ensuring that systemic metabolic effects do not interfere with tissue-specific regenerative processes[45].

Cell replacement and organoid platforms: Distinct from reprogramming

Organoids are primarily platforms for modeling development, injury, and patient-derived cellular behavior. Human nephron organoids and expandable liver cultures illustrate these capabilities[49-51]. An organoid model is not itself a treatment; transplantation of a derived product would require a separate therapeutic evidence and safety assessment. An extension proposed here is to add controlled injury, washout, recovery, and re-exposure to such platforms, with persistent-state and functional readouts. The cited organoid studies establish platform feasibility, not validation of this complete fibrosis-memory assay[50,51]. Cell replacement supplies new cells rather than correcting the state of resident cells. Benefit must be demonstrated for the specified product, organ, and disease stage; tested cardiac cell therapies failed to rejuvenate chronically scarred rodent hearts[7]. Engraftment or cell survival alone should not be treated as evidence of tissue reconstruction. The possibility that unresolved niche dysfunction limits replacement is a research hypothesis, not proof that combining replacement with memory-directed treatment will restore tissue function.

Safety risks: Tumorigenesis, aberrant plasticity, and immune dysregulation

The biological plasticity that makes therapeutic reprogramming attractive also creates substantial risks[53,54]. Partial reprogramming improved selected age-associated phenotypes in experimental models[53], whereas in vivo pluripotency-factor reprogramming produced teratomas in another study[54]. These protocols are not interchangeable: The latter demonstrates a safety hazard of excessive identity destabilization, not that every partial-reprogramming protocol has the same risk. Senolysis and immune modulation require distinct assessments of cell loss, repair competence, and host defense[39,62]. Lineage-changing or cell-product approaches additionally require assessment of product heterogeneity, maturation, ectopic growth, and tissue integration; platform establishment alone cannot resolve these safety questions[49,54]. Future direct reprogramming strategies should prioritize controlled, reversible, and tissue-specific correction of pathological states. Complementary modulation, cell replacement, and research platforms should be evaluated using their own evidence and safety criteria[53,54] (Table 4).

CURRENT LIMITATIONS AND FUTURE DIRECTIONS

Distinguishing cause from consequence

A major unresolved issue in the field is whether stem/progenitor cell dysfunction represents an initiating event in fibrosis or primarily reflects adaptation to an already established pathological niche[3,4]. The available evidence suggests that both possibilities may occur depending on tissue context, injury type, and disease stage. In some settings, intrinsic defects in regenerative populations may predispose tissues to maladaptive repair. In others, ECM remodeling, inflammation, or stromal activation may occur first and subsequently impair regenerative capacity[3,4]. Resolving this causal relationship requires experimental systems capable of capturing disease progression over time. Longitudinal lineage tracing, inducible genetic models, repeated-injury systems, and human-based platforms will be essential to determine temporal relationships between regenerative dysfunction and fibrotic remodeling[28,29,49].

Quantifying fibrotic memory

Quantification should distinguish the strength and duration of retained states from the severity of ongoing injury, rather than repeat the operational definition in “Definition and conceptual boundaries”. Candidate readouts include retained enhancer activity[14], post-transfer mechanical responses[13], altered conditioned immune responsiveness[39], and spatially organized disease niches[10,36]. Spatial transcriptomics and its subsequent methodological development can localize these features[65,66], but spatial persistence alone does not establish causal memory or distinguish cell-intrinsic retention from continuing niche stimulation. Rather than treating fibrotic memory as a single binary phenomenon, these components should be measured independently and integrated into a multidimensional framework[13,14]. Establishing reliable biomarkers of memory burden will be essential for evaluating whether therapeutic interventions truly erase pathological states or merely reduce their downstream consequences[13,14].

Functional validation in human-relevant models

Cell and niche maps nominate candidate states, while functional inference depends on the specific perturbation performed. Human liver and kidney studies may combine mapping with functional experiments, but not every inferred interaction is thereby causally validated[10,36]. Priority experiments include longitudinal lineage tracing, inducible genetic or pharmacological perturbation, stimulus withdrawal and re-exposure, repeated-injury organoids or tubuloids, and transfer of cells between normal and diseased ECM environments. These designs distinguish association, causality, memory, and reversibility[13,28,29,49,50].

Human-derived organoids, ex vivo tissues, and spatial assays should complement animal models and incorporate immune interactions, mechanical cues, and longitudinal measurements where feasible[49,65]. In vitro stiffness-switch assays control exposure but simplify the niche; ex vivo samples retain disease features but have limited longevity and may retain residual stimuli. Acute or self-resolving animal injury differs from long-standing human disease, and an early reduction in scar can compromise necessary repair. For example, acute cardiac myofibroblast depletion impairs adaptive healing, whereas tested cell therapies fail in established rodent scar[7,57]. Evidence of molecular normalization, histological collagen reduction, organ function, and patient benefit must therefore be reported separately. Table 5 summarizes what each platform can and cannot test.

Table 5 Experimental platforms for distinguishing association, causality, and memory.

Platform	Primary contribution/causal inference	Memory or reversibility requirements	Main limitation
Single-cell/single-nucleus RNA sequencing	Identifies cell states and heterogeneity; not causal by itself[8,37]	Longitudinal sampling can suggest persistence; exposure history and a functional test are needed	Sampling bias, dissociation effects, and lack of spatial context
Spatial transcriptomics	Maps tissue architecture and candidate interactions; not causal alone[65,66]	Can localize retained niches, but cannot alone exclude continuing stimulus	Resolution, analytical inference, and tissue-sampling limitations
Multiome/chromatin profiling	Links regulatory and transcriptional states; stronger inference requires perturbation[48]	Test retained programs after withdrawal and selectively perturb the candidate state	A chromatin signature is not itself proof of memory
Lineage tracing/barcoding	Establishes origin and fate, not causal necessity without an additional intervention[28]	Combine with withdrawal/re-exposure and appropriate functional perturbation[52]	Label specificity, recombination efficiency, and mostly experimental-animal use
Inducible perturbation/Perturb-seq	Tests candidate regulatory contributions when controls and endpoints are appropriate[29]	Perturb an already retained state after injury reduction; assess later function and re-exposure	Perturbation during ongoing injury may identify an injury pathway rather than memory
Organoids/expandable epithelial cultures	Experimentally tractable human-derived models[50,51]	Add and validate injury, washout, recovery, and rechallenge; this is a proposed extension, not automatic platform capability	Incomplete immune, vascular, and mechanical niches; variable maturation
Ex vivo diseased ECM/tissue systems	Tests matrix-conditioned behavior and niche instruction[34]	Transfer away from diseased ECM or remove residual stimuli to test cell-autonomous retention	Continued ECM exposure is continuing niche instruction, not proof of intrinsic memory
Longitudinal human sampling	Establishes clinical relevance and candidate biomarker trajectories[9,20]	Document injury reduction, recovery interval, and confounders; couple associations to functional models	Limited repeat sampling and no routine direct genetic causal test

Functional tissue recovery must be measured separately from cell-state or molecular normalization. Platform availability does not guarantee a valid memory test. Direct evidence depends on the exposure, resolution, transfer/rechallenge, and perturbation design described in “Definition and conceptual boundaries”. ECM: Extracellular matrix.

Designing safe and reversible reprogramming therapies

The ultimate challenge for therapeutic reprogramming is achieving sufficient correction of pathological states without disrupting normal tissue identity[53,54]. Because fibrosis involves complex interactions among regenerative cells, immune populations, fibroblasts, and ECM, single-target interventions may be insufficient. Future approaches will likely require precise, context-dependent modulation of multiple regulatory layers[3,5]. An ideal reprogramming strategy should possess several characteristics[53,54]: Tissue specificity; reversibility; controllable intensity; preservation of lineage identity; minimal risk of uncontrolled proliferation. The future of fibrosis therapy may therefore depend not only on discovering new molecular targets but also on developing strategies capable of safely resetting pathological cellular states[53,54].

Framework comparison and translational roadmap

The present framework builds on, rather than replaces, major syntheses of fibrosis mechanisms, maladaptive regeneration, and trained immunity[3-5,67]. Its proposed contribution is to align compartment identity, operational memory tests, organ-specific directness, and therapeutic categories within one appraisal structure (Table 6).

Table 6 Relationship of the proposed framework to selected major reviews.

Ref.	Principal scope	Overlap with this review	Additional emphasis of the present framework
Henderson et al[3], 2020	Fibrosis mechanisms and therapeutic development across organs	Immune-stromal interactions, ECM remodeling, and antifibrotic targets	Aligns repair-compartment identity with operational persistence tests and separates mechanism, memory, and clinical benefit
Zhu et al[4], 2021	Maladaptive regeneration and developmental pathways in NASH/fibrosis	Repair can become fibrogenic rather than simply insufficient	Compares seven organs without assuming identical progenitor biology; distinguishes persistent repair from directly tested memory
Distler et al[5], 2019	Shared and organ-specific fibrosis mechanisms	Conserved signaling alongside disease- and organ-specific biology	Uses model- and cell-specific directness to constrain cross-organ grades and therapeutic classification
Netea et al[67], 2020	Trained immunity in health and disease	Durable response changes and epigenetic/metabolic conditioning	Requires fibrosis-relevant endpoints and separates beneficial tolerance, ongoing niche stimulation, and autonomous retained states
Present review	Evidence-calibrated injury-repair-memory synthesis	Integrates established concepts rather than claiming a wholly new mechanism	Proposes testable links, explicit memory boundaries, a reprogramming/modulation/platform distinction, and a staged translational roadmap

This comparison describes differences in emphasis, not absence of related concepts from prior reviews or proof that the present framework is mechanistically superior. ECM: Extracellular matrix; NASH: Nonalcoholic steatohepatitis.

Near-term priorities are biomarker and model development. Candidate epithelial, stromal, and niche signatures should be evaluated in longitudinal cohorts with documented injury status and organ-specific functional outcomes. Assays should prespecify the recovery interval, distinguish residual stimulus from retained state, and report analytical reproducibility. Patient-derived cultures, organoids, and ex vivo tissues can then be used to test candidate-state reversibility rather than assumed to deliver clinical repair[9,37,50,51].

An intermediate step is causal validation: Selectively perturb the candidate state after injury reduction, measure its persistence after treatment washout, and test re-exposure, matrix transfer, and functional restoration where appropriate. Replication across complementary models is needed before interpreting reduced collagen or a normalized transcriptome as durable regenerative recovery. Existing withdrawal, transfer, and sensitized-reactivation experiments provide design examples, not interchangeable evidence across organs[13,20,52].

Longer-term therapeutic reprogramming requires targeted delivery, controlled intensity and duration, preservation of lineage identity, and surveillance for tumorigenesis, abnormal plasticity, and immune dysfunction. Candidate therapies should advance only when durable functional benefit and acceptable safety are shown in the relevant disease stage, with established antifibrotic care as an appropriate comparator[23,53,54].

CONCLUSION

Organ fibrosis involves more than ECM accumulation and fibroblast activation. It can also reflect dysfunction of a regenerative system composed of bona fide stem/progenitor populations, injury-responsive differentiated parenchymal cells, immune and stromal effectors, and the extracellular niche[3-5]. Evidence varies by organ and endpoint. Selected lung and kidney epithelial mechanisms have functional support[17,19,20,27], and stromal drivers are directly demonstrated in skin and marrow[11,59,61]. Mechanical and selected epigenetic or activation-state memory experiments are informative but context-restricted[13,14,52]. Cross-organ generalization and metabolic-memory erasure remain hypotheses rather than established therapeutic mechanisms. The injury-repair-memory framework is an evidence-calibrated, testable synthesis rather than a universally established mechanism. Its use should follow the operational boundaries in “Definition and conceptual boundaries” and Table 2, without treating persistent disease signatures as proof of memory. Therapeutic reprogramming should likewise be reserved for controlled and durable correction of pathological cellular state or lineage trajectory. ECM normalization, immunomodulation, metabolic therapy, cell replacement, and organoid-guided screening may be valuable complements, but they are biologically distinct strategies. Progress will require organ-specific causal validation, quantitative memory assays, and safe, reversible interventions that restore functional tissue repair.