Neuromuscular Remodeling in Knee Osteoarthritis Across Aging: An Evidence-Bounded Joint–Afferent–Motor Unit Framework

Abstract

Quadriceps weakness is a major determinant of disability in knee osteoarthritis (KOA), particularly in later life, yet weakness often exceeds the loss of muscle size. This discrepancy is clinically important because joint-derived activation failure occurs against a background of declining neuromuscular and regenerative reserve. We critically examine an evidence-bounded joint–afferent–motor unit framework in which abnormal afferent input from the diseased knee reduces effective neural drive through arthrogenic muscle inhibition (AMI). Persistent under-recruitment may be associated with altered motor-unit behavior and loading, impaired muscle quality, and greater reliance on compensatory reinnervation. We deliberately distinguish demonstrated human observations from provisional transitions and hypothesis-generating mechanistic extensions. Human evidence supports AMI and impaired muscle quality and provides preliminary cross-sectional evidence of altered motor-unit behavior; however, human studies have not established a temporal sequence from persistent AMI to motor-unit remodeling or denervation–reinnervation. Direct KOA-related neuromuscular junction (NMJ) evidence remains predominantly preclinical, and progressive NMJ degeneration has not been demonstrated in humans with KOA. Aging, sex, obesity, diabetes, systemic inflammation, pain sensitization, inactivity, and other comorbidities are treated as effect modifiers and competing explanations rather than proof of causation. We propose a testable, multimodal validation strategy combining joint imaging, standardized voluntary-activation testing, neurophysiology, high-density electromyography, motor-unit-number estimation, muscle imaging, and, when justified, tissue-level NMJ phenotyping. For older adults, the practical implication is mechanism-directed care: restore joint-related activation when inhibition dominates, use neural augmentation to enable adequate loading, address muscle-quality and metabolic constraints when present, and reserve NMJ-directed interventions for future human subgroups defined by convergent structural and physiological evidence.

Introduction

Osteoarthritis is a leading cause of disability and is projected to impose an increasing global burden as populations age.¹ The modern disease model is no longer cartilage-centered: synovium, subchondral bone and menisci contribute to symptoms and structural progression.^2,3 Ligaments, capsule, sensory nerves and periarticular muscles also participate in the joint-as-organ phenotype.^4,5 The clinically relevant problem in later life is not simply that KOA is common; it is that joint pathology may interact with declining neuromuscular reserve to amplify disability. Yet the neuromuscular system is still often positioned downstream of the joint, as though weakness were merely a consequence of pain and inactivity.

Quadriceps weakness is present across radiographic severities and is linked to impaired function and postural instability.^6–8 Knee buckling is also common and functionally limiting.^9,10 Longitudinal associations with incident disease, cartilage loss and joint-space narrowing are not uniform, but they show that strength is not simply a late marker of end-stage joint destruction.^11–13 The quadriceps regulate knee flexion during weight acceptance and contribute to attenuation of impulsive loads; weakness can therefore alter the mechanical environment of the diseased joint.^14–16 The important observation is not that KOA causes weakness. It is that the magnitude and persistence of weakness are often poorly explained by muscle size.

Voluntary activation failure accounts for a substantial fraction of this discrepancy.¹⁷ AMI describes a joint-related inability to fully activate periarticular muscle despite an anatomically intact motor pathway.^18–20 In symptomatic KOA, activation deficits coexist with weakness and can limit the response to strengthening.^21,22 The term is nevertheless used too loosely. AMI is sometimes treated as a unitary spinal reflex, sometimes as any reduction in surface electromyographic amplitude, and sometimes as a clinical synonym for quadriceps weakness. These usages collapse distinct levels of analysis and obscure the central question: what happens to the motor unit when effective neural drive remains depressed for months or years? In an older patient, that distinction matters because identical low torque may reflect activation failure, tissue-quality loss, generalized neuromuscular aging, or combinations of these processes.

A rehabilitation-centered literature has established that AMI can be modified, at least transiently, by interventions that alter afferent input or augment muscle activation.^23,24 Evidence from related knee injury and reconstruction literature also supports transient reversibility with afferent- or activation-targeted interventions.^25,26 It has not established whether persistent AMI changes motor-unit number, recruitment territory, discharge behavior or NMJ integrity. Conversely, the neuromuscular aging literature describes motor-neuron loss, collateral reinnervation and eventual failure of reinnervation reserve.^27–29 The intersection is particularly important in later life, where chronic joint-derived reductions in effective drive may be superimposed on a smaller reserve for compensation. The primary gap is whether objectively measured, sustained joint-related activation failure precedes motor-unit or tissue change within individuals, and whether reduced compensatory and regenerative reserve modifies that relationship. Joint-related underloading in an older muscle may have different consequences from the same activation deficit in a muscle with greater reserve; this interaction remains to be tested.

We therefore propose an aging-modified joint–afferent–motor unit framework with a conditional NMJ extension. Its contribution is to connect rehabilitation and neuromuscular aging evidence through explicitly bounded, falsifiable transitions, rather than to infer progression from coexisting abnormalities. The framework contains five overlapping domains: (1) abnormal joint afference; (2) reduced effective neural drive expressed as AMI; (3) altered motor-unit recruitment and discharge; (4) impaired muscle quality and regenerative reserve; and (5) conditional NMJ remodeling. These domains are not a validated sequence or staging system. Aging, sex and metabolic comorbidity may modify susceptibility and recovery, while generalized neuromuscular aging and neuropathy remain competing explanations.

The distinction between remodeling and degeneration is central. Remodeling may remain adaptive or reversible; degeneration requires persistent structural loss or transmission failure that exceeds compensatory capacity. Human KOA studies document activation failure, altered motor-unit behavior and impaired muscle quality. They do not establish that persistent AMI causes denervation–reinnervation or NMJ degeneration. Direct KOA-related NMJ observations are predominantly preclinical. The novelty of the framework is therefore its separation of observed phenotypes from proposed temporal links, with repeated measurements, anatomical controls and explicit falsification criteria for each link.

Review Approach and Evidence Classification

This critical narrative review used an iterative, concept-driven search of PubMed/MEDLINE, publisher records and reference lists, with citation-chain checking. The original search covered literature through 31 July 2026; a targeted revision update through 16 September 2026 examined sex, hormonal context, AMI duration, adipokines, diabetes and NMJ injury using PubMed-indexed records retrieved through Europe PMC and publisher sources. Search blocks combined KOA, joint effusion or synovitis with AMI, voluntary activation, spinal reflexes, transcranial magnetic stimulation, motor units, electromyography, denervation, reinnervation, muscle quality, fibrosis, adipose infiltration, satellite cells, NMJ, agrin, LRP4, MuSK, acetylcholine receptors and arthroplasty. Additional terms included sex, estrogen, menopause, persistence, longitudinal, insulin resistance, leptin and adiponectin. Human mechanistic studies, longitudinal cohorts, tissue studies, systematic reviews and experiments perturbing joint afference were prioritized. Aging and metabolic studies were included to examine otherwise unsupported links. This was not a systematic review with a prespecified protocol or an exhaustive screening flow.

Evidence was classified according to both source and inferential distance. “Direct human evidence” refers to observations in KOA or closely defined peri-arthroplasty cohorts. “Translational evidence” refers to human experimental joint perturbation or animal KOA models that inform mechanism. “Mechanistic analogy” refers to findings from aging, disuse, or neuromuscular disease that establish biological plausibility but cannot establish a KOA-specific causal pathway. The labels are not formal risk-of-bias grades; they define how far a conclusion may reasonably be generalized. Mixed-age cohorts were not treated as evidence specific to adults older than 65 years unless age-stratified analyses supported that inference.

No formal tool-based risk-of-bias assessment, pooled effect estimate or certainty-of-evidence score was performed. Interpretation instead considered design, sample size and selection, sex and age coverage, comparator choice, measurement specificity, confounding and temporal ordering. Null or inconsistent findings were retained when they challenged a proposed link; differences by muscle, task or outcome were not combined into a uniform deterioration narrative. Examples include heterogeneous cortical responses, the absence of some between-severity differences in motor-unit and muscle-quality measures, and sex-dependent functional recovery. A null finding in a small study was not treated as proof of equivalence. This qualitative approach cannot exclude selection or confirmation bias, and the proposed transitions require prospective testing.

Conceptual Model, Evidence Boundaries and the Aging Context

Conceptual Model vs Demonstrated Transitions

The proposed relationships should not be read as an established sequence of disease progression (Figure 1). Table 1 defines the constructs and the limits of their measurement. Experimental human joint perturbation demonstrates that altered afference can suppress quadriceps activation, and clinical KOA studies document voluntary activation failure. The proposed link from persistent AMI to motor-unit remodeling rests on between-person observations rather than demonstrated temporal ordering. Links to muscle quality and NMJ integrity may also reflect parallel effects of aging, metabolic disease or inactivity. Table 2 distinguishes demonstrated perturbation effects, provisional associations and hypotheses; none of the distal links establishes progressive NMJ degeneration in human KOA.

Figure 1.

The solid arrow represents the experimentally supported effect of joint-afferent perturbation on activation failure. Dashed arrows indicate proposed relationships without confirmed temporal ordering or causal interpretation. The question mark denotes the current evidence gap regarding the relationship between impaired muscle quality and neuromuscular junction (NMJ) alterations in knee osteoarthritis (KOA), for which direct human evidence remains insufficient. Human KOA studies support activation failure, altered motor-unit behavior, and impaired muscle quality, whereas KOA-related NMJ evidence is mainly preclinical. Illustrations are schematic.

Abbreviations: AMI, arthrogenic muscle inhibition; KOA, knee osteoarthritis; NMJ, neuromuscular junction.

Table 1.

Operational Definitions and Claims That Can Legitimately Be Made

Construct	Operational Definition	Measures that Support the Claim	What the Measure Cannot Establish
AMI	Incomplete voluntary activation attributable to joint-related neural mechanisms	Interpolated twitch, superimposed burst, central activation ratio	Anatomical site of inhibition; denervation; NMJ failure
Persistent AMI (proposed research definition)	Reduced activation at two or more standardized visits spanning at least 3 months; 6-month sensitivity definition	Evoked activation relative to method-specific limits and measurement error; repeated context measures	Unvalidated sampling rule; not an established damage threshold or duration inferred from symptoms
Motor-unit remodeling	Change in recruitment, discharge, territory or number of functional units	HD-sEMG decomposition, needle EMG, MUNE, force steadiness	A cross-sectional enlarged potential cannot prove prior denervation
Muscle quality	Force capacity and contractile composition relative to muscle size	Specific strength, MRI fat fraction, ultrasound echo intensity, biopsy fibrosis	A single accepted KOA threshold or the direction of causation
NMJ remodeling	Structural or molecular adaptation at the nerve–muscle synapse	Pre/post-synaptic apposition, terminal Schwann cells, AChR topology, transmission measures	Progressive degeneration without longitudinal functional loss
NMJ degeneration	Persistent synaptic/axonal loss or transmission failure exceeding reinnervation reserve	Convergent morphology, electrophysiology and longitudinal decline	Not yet directly demonstrated in human KOA

Abbreviations: AChR, acetylcholine receptor; AMI, arthrogenic muscle inhibition; HD-sEMG, high-density surface electromyography; KOA, knee osteoarthritis; MRI, magnetic resonance imaging; MUNE, motor-unit-number estimation; NMJ, neuromuscular junction.

Table 2.

Evidence Status of the Proposed Transitions: Demonstrated, Provisional and Hypothetical

Transition	Current Evidence	Evidence Class	Interpretation Permitted	Principal Limitation/Falsifier
Diseased joint/altered afference → reduced effective neural drive (AMI)	Experimental effusion suppresses activation and alters cortical responses.30,31	Experimental human evidence (translational)	Joint-derived afferent perturbation can reduce quadriceps output.	Acute distension is not chronic KOA; mechanism varies with pain, swelling and task.
Persistent AMI → motor-unit remodeling	KOA comparisons show altered motor-unit potentials and muscle- or task-dependent firing and strength relationships.32,33	Preliminary direct human evidence	Altered behavior is observed; denervation and reinnervation are not established.	No temporal ordering or validated exposure threshold; female-only findings do not establish sex differences.
Motor-unit/loading change → deterioration of muscle quality	Human biopsy and imaging show fibrosis, altered regenerative features and non-contractile tissue.34,35	Direct human coexistence; causal link provisional	Neural and tissue constraints can coexist and may reduce force per unit muscle size.	Direction remains unresolved; sex, adiposity, diabetes, activity and systemic inflammation may influence findings.
Motor-unit/muscle remodeling → NMJ remodeling	Rat KOA demonstrates endplate and MuSK/AChR remodeling.36 Aging biology provides a reinnervation-reserve analogy.27–29	Preclinical KOA + mechanistic analogy	KOA can coincide with NMJ remodeling; failure of compensation is biologically plausible.	AMI has not been shown to cause NMJ remodeling; rodent morphology cannot be assumed to represent human transmission failure.
NMJ remodeling → progressive NMJ degeneration	No direct human KOA demonstration; aging/NMJ studies define structural and translational benchmarks.37–39	Hypothetical in human KOA	Degeneration remains a testable human KOA hypothesis, not an established consequence of remodeling.	Requires longitudinal human structural plus physiological evidence; absence of such pathology would materially weaken the model.

Abbreviations: AChR, acetylcholine receptor; AMI, arthrogenic muscle inhibition; KOA, knee osteoarthritis; MuSK, muscle-specific kinase; NMJ, neuromuscular junction.

Sensorimotor and Neuromuscular Boundaries

The healthy knee contributes to motor control through a mixed afferent population. Low-threshold receptors encode position, movement and tissue deformation; group III and IV afferents signal mechanical and chemical threat. Classical experiments show that knee-joint afferents can alter quadriceps motor-unit activity.^40,41 Segmental reflex responses provide complementary evidence that joint input can change spinal output.^42,43 Joint input also interacts with muscle-spindle pathways and gamma-motoneuron control.^44–46 Group IV input can suppress group II-driven gamma effects, emphasizing state dependence.⁴⁷ The resulting motor effect cannot be reduced to a single inhibitory wire. It depends on receptor class, stimulus intensity, spinal state, task and concurrent input from muscle and skin.

Proprioceptive impairment in KOA is therefore better described as degraded sensorimotor information than as simple sensory loss.^48,49 A swollen joint may generate excessive pressure-sensitive input; damaged capsuloligamentous structures may remove reliable low-threshold information; nociceptor sensitization may increase protective drive; and reduced movement may diminish active sensory sampling. These processes can coexist and push motor output in different directions.

The motor unit comprises an alpha-motoneuron, its axon and terminal branches, the NMJs, and the innervated muscle fibers. NMJ maintenance depends on precise pre- and postsynaptic apposition and on signaling through agrin–LRP4–MuSK and related pathways.^50–52 LRP4 also acts as an agrin receptor and forms a complex with MuSK.⁵³ Activity is not merely an output of this system; it helps maintain it. Chronic changes in recruitment and discharge can alter fiber phenotype, trophic support and the demand for reinnervation.

Aging Modifies Reserve, Not Diagnosis

Aging provides a relevant but non-specific template for interpreting the framework. Motor-neuron loss is initially buffered by collateral sprouting, which enlarges surviving motor units.²⁷ Severe decline appears when reinnervation fails or surviving units cannot expand sufficiently.^28,29 This process contributes to age-related losses of muscle fibers, mass and function.^54,55 In the present framework, aging is therefore an effect modifier and competing explanation rather than a surrogate diagnosis: KOA may expose a vulnerable neuromuscular state, but age-related motor-unit remodeling cannot be relabeled as KOA-induced neurodegeneration without longitudinal, anatomically specific evidence. The clinically relevant aging question is whether reduced reserve changes the reversibility of a joint-specific neural disturbance, not whether chronological age itself constitutes a lesion.

Sex and Hormonal Context as Potential Effect Modifiers

Sex should be examined across the framework rather than treated solely as a demographic adjustment. Watabe et al studied 30 women, so their motor-unit findings cannot establish a sex interaction or be generalized directly to men.³³ The older-adult intermuscular-adipose study also enrolled only women.⁵⁶ Experimental estrogen-receptor work supports a role for hormonal context in satellite-cell maintenance and regeneration, with preliminary human menopausal-transition observations; these findings concern regenerative reserve, not a demonstrated KOA-specific neural pathway.⁵⁷ More recent mouse transplantation experiments show that estrogen-receptor dependence can vary with the host hormonal environment.⁵⁸

The reviewed human evidence does not establish that female hormonal status changes KOA-related gamma-loop gain, reinnervation reserve or transition probabilities between these domains. Nor should all outcomes be assumed to deteriorate faster in women: after total knee arthroplasty, women showed better relative preservation of quadriceps strength but greater early decline in physical function in one pooled cohort.⁵⁹ Sex differences may therefore depend on the outcome and treatment context. Future studies should prespecify sex-by-exposure interactions and, where relevant and feasible, record menopausal status, hormonal treatment and fat distribution. Table 3 incorporates these variables without treating sex as a surrogate for hormonal exposure.

Table 3.

Multimodal Phenotyping Framework

Domain	Minimum Dataset	Advanced Dataset	Interpretive Purpose	Confounders and Modifiers
Joint	Pain phenotype, effusion, radiographs, ultrasound synovitis	Contrast MRI, inflammatory phenotyping, gait loading	Define the afferent and mechanical exposure	Analgesics, injections, alignment, activity exposure
Neural activation	Repeated evoked activation with visit dates; proprioception	Quadriceps-relevant reflex testing, TMS, pain modulation	Separate activation failure from low force; quantify persistence and exposure	Attention, medication, contralateral OA, age/frailty
Motor unit	EMG and force steadiness at repeated visits	HD-sEMG decomposition, MUNE, needle EMG	Detect recruitment/rate-coding change and possible reinnervation	Adiposity, electrode geometry, neuropathy, age, activity
Muscle/NMJ	Strength, MRI/ultrasound composition	Biopsy, spatial/single-nucleus profiling, 3D NMJ imaging	Separate quantity, quality, regeneration and synaptic integrity	Nutrition, diabetes, renal function, sarcopenia, frailty
Cross-domain modifiers	Age, sex, adiposity, diabetes, medication and activity history	Menopausal/hormonal context, fat distribution, metabolic markers and neuropathy assessment as indicated	Prespecify interactions; distinguish local effects from systemic disease	Covariate roles depend on the causal question; avoid automatic adjustment for mediators

Abbreviations: EMG, electromyography; HD-sEMG, high-density surface electromyography; KOA, knee osteoarthritis; MRI, magnetic resonance imaging; MUNE, motor-unit-number estimation; NMJ, neuromuscular junction; OA, osteoarthritis; TMS, transcranial magnetic stimulation. Cross-domain variables apply across all rows. Sex and age are candidate effect modifiers; menopausal status and hormonal treatment should be recorded where relevant. Confounders, mediators and interaction terms must be prespecified for each causal question. The proposed persistence definitions are unvalidated research rules.

Competing Explanations and Anatomical Specificity

Any apparent downstream neuromuscular phenotype can arise from at least four non-exclusive explanations: reversible task adaptation to pain or instability; compensatory reinnervation after partial denervation; generalized aging or comorbidity, including diabetes, lumbar disease and sarcopenia; or deconditioning and altered activity exposure. Cross-sectional case-control designs cannot separate these pathways. A joint-related mechanism becomes more credible when the effect is anatomically greater in the affected quadriceps than in remote muscles, tracks objective activation failure, and changes appropriately after perturbation of the joint signal. These criteria are used throughout the review to prevent age, inactivity or systemic disease from being mistaken for evidence of a KOA-specific neurodegenerative process.

Domain I: How Joint Pathology Distorts Afferent Control

Effusion is a Mechanistic Perturbation, Not a Proxy for Chronic KOA

Joint effusion provides the cleanest demonstration that altered afference can suppress quadriceps output. Experimental distension reduces voluntary activation and can do so at volumes that produce little pain. Cryotherapy partly reverses the inhibition, supporting an afferent mechanism rather than irreversible muscle failure.³⁰ In a separate effusion experiment, quadriceps inhibition was accompanied by increased corticospinal excitability and a shorter cortical silent period.³¹ The result is important because it contradicts a common simplification: the cortex was not globally suppressed. Increased cortical output may have compensated for inhibition at a spinal or subcortical level.

The experiment does not establish the chronic KOA mechanism. Acute saline distension is temporally precise, whereas chronic synovitis changes receptor sensitivity, movement behavior and central pain processing over years. It also cannot reproduce the mixed loss and gain of afferent function produced by meniscal, capsular and ligamentous pathology. Effusion studies should therefore be used to prove that joint afference can alter motor output, not to claim that distension explains all AMI.

Nociception Modifies Output, but Pain Intensity is Not the Mechanism

Experimental knee pain reduces maximal force, and clinical KOA is frequently accompanied by peripheral and central sensitization.^60,61 Heightened flexor-withdrawal responses indicate that spinal nociceptive circuitry is altered in at least a subset of patients.⁶² Pain can reduce voluntary effort, redistribute activity among synergists and increase antagonist co-contraction. Yet pain scores correlate imperfectly with activation. Analgesia may improve effort without normalizing the afferent state, and severe activation failure can occur with modest pain. Treating pain intensity as a surrogate for AMI therefore confounds symptom severity with sensorimotor mechanism.

Proprioceptive Noise and the Gamma Loop

KOA-related mechanoreceptor loss is often invoked to explain proprioceptive deficits, but the functional problem may be unreliable weighting of competing signals. Loss of capsuloligamentous input can coexist with increased nociceptor traffic and altered spindle sensitivity. The gamma loop is a plausible point of convergence because gamma-motoneuron activity regulates spindle gain and thereby influences alpha-motoneuron recruitment. The human evidence is indirect. Prolonged tendon vibration, which normally perturbs spindle-mediated facilitation, has a smaller effect on torque and activation in KOA than in controls, consistent with impaired gamma-loop contribution.⁶³ The finding identifies altered spindle-related control; it does not locate the lesion in joint receptors, gamma motoneurons or central integration.

Domain I should therefore be defined functionally: the diseased joint changes the reliability and salience of sensory input used to construct motor commands. Effusion, nociception and proprioceptive degradation are candidate drivers, not mutually exclusive phenotypes. A study that measures only pain or only radiographic severity cannot determine which afferent abnormality is dominant.

Domain II: AMI is a State of Reduced Effective Neural Drive

Measurement Defines the Claim

AMI cannot be diagnosed from weakness alone. Interpolated twitch and superimposed-burst methods infer incomplete activation when an external stimulus delivered during a maximal contraction produces additional force. Estimates vary with stimulus type and calculation method.^64,65 Joint angle, stimulus number and other protocol details further affect the estimate.^66–68 Surface electromyography is valuable for timing and coordination but is not a direct measure of central drive; amplitude is shaped by subcutaneous tissue, electrode geometry, motor-unit cancellation and action-potential propagation. A recent systematic review confirms consistent deficits in strength and voluntary activation across KOA cohorts while also exposing sparse evidence for spinal and cortical mechanisms.⁶⁹ Methodological heterogeneity is therefore part of the biological uncertainty, not a technical footnote.

The term AMI should be reserved for demonstrable activation failure attributable to joint-related neural mechanisms. Low force caused by fibrotic muscle, peripheral neuropathy, poor comprehension or fear of pain is clinically important but mechanistically different. Studies that do not separate these contributors risk assigning all unexplained weakness to AMI.

Spinal Mechanisms are Plausible but Incompletely Mapped

The prevailing model places altered joint afference upstream of spinal interneuronal networks that regulate the alpha-motoneuron pool. Candidate mechanisms include group II afferent effects and changes in presynaptic inhibition.^18,40,41 Segmental reflex studies support joint-afferent access to spinal circuits.^42,43 Altered flexion-reflex excitability and impaired spindle-mediated facilitation provide additional KOA-relevant evidence.^62,63 Human gamma-loop physiology provides an additional mechanistic template.⁴⁴ Joint-afferent studies show gamma-motoneuron modulation through group II and IV pathways.^45–47 The model is physiologically coherent, but KOA studies rarely record these pathways simultaneously. H-reflex findings from a remote muscle cannot be assumed to represent quadriceps motoneuron excitability, and a normal resting reflex does not exclude task-dependent inhibition during maximal contraction.

A more defensible formulation is that AMI reflects a failure to translate descending intention into effective motor-unit output in the presence of abnormal sensory weighting. The bottleneck may lie in spinal gain, in the recruitment threshold of the motoneuron pool, or in a protective motor program distributed across several muscles. The relative contribution probably changes with effusion, pain sensitization and task demand.

Cortical Findings Resist a Simple Inhibition Narrative

Transcranial magnetic stimulation studies do not support a uniform cortical lesion. Experimental effusion can increase corticospinal excitability.³¹ In established KOA, corticospinal and intracortical measures vary with pain, disease severity and protocol.⁷⁰ Motor cortical organization differs from that of controls, but topographic reorganization does not identify whether the change is adaptive or pathogenic.⁷¹ Increased intracortical inhibition has also been interpreted as compensation for chronic pain rather than as the primary cause of weakness.⁷² These findings are compatible with a system attempting to stabilize output under altered afferent conditions.

The relevant question is therefore not whether the cortex is “excited” or “inhibited”. It is whether descending drive is appropriately scaled, focused and converted into force. Increased excitability can coexist with poor voluntary activation if downstream gain is reduced; cortical inhibition can limit output or suppress noisy motor commands. Cross-sectional TMS cannot resolve this directionality.

AMI is Heterogeneous and Should Be Phenotyped

Activation deficits modify training response: patients with lower baseline activation may gain less strength from exercise despite similar attendance.⁷³ Antagonist co-contraction can reduce net knee-extension torque and alter joint loading.^74,75 Altered muscle synergies provide a related redistribution mechanism.^76,77 These observations argue against a single AMI phenotype. Candidate joint-related patterns include effusion-dominant, nociception-dominant, proprioceptive/gamma-loop-dominant and central-pain-network-dominant inhibition. Structural neuromuscular impairment is a potentially coexisting limitation or alternative explanation, not an AMI subtype established by weakness alone. The categories are provisional, but they make a testable prediction: interventions that improve activation in one phenotype will fail when the dominant limitation lies elsewhere. Such phenotyping is especially relevant in older adults, in whom joint pathology, pain sensitization, sarcopenia and metabolic comorbidity can coexist.

Domain III: Motor-Unit Remodeling—The Missing Link

Why Reduced Neural Drive Could Matter Beyond the Contraction

A motor unit is repeatedly selected or excluded by the neural strategies used in daily life. Persistent reduction in effective drive can decrease recruitment of high-threshold units, lower discharge rates during maximal effort and shift submaximal work toward a narrower population of active units. The immediate consequence is reduced force and rate of force development. The longer-term consequence is less certain: chronically under-recruited units may receive less mechanical and activity-dependent trophic support, whereas repeatedly recruited units may enlarge their territory or operate closer to fatigue.

Conventional surface electromyography cannot identify this reorganization. High-density recordings and decomposition algorithms now permit non-invasive estimation of multiple concurrent discharge trains, although yield remains force-, anatomy- and algorithm-dependent.⁷⁸ These methods should not be presented as direct motor-neuron counts. They do, however, make it feasible to test whether AMI is associated with altered recruitment thresholds, discharge-rate modulation, common synaptic input and spatial redistribution of motor-unit territories.

The Human Signal is Credible but Not Yet Diagnostic of Reinnervation

Two informative KOA studies compare motor-unit characteristics between participants rather than tracking an AMI-to-denervation sequence. Ling et al reported larger force-normalized motor-unit potentials and severity-dependent recruitment indices.³² These observations do not specifically identify collateral reinnervation because recruitment strategy, muscle geometry, conduction velocity and sampling can alter the signal. Watabe et al prospectively recruited 30 women with early or severe KOA, but the relevant analyses were between-group comparisons and within-group correlations. Discharge differences depended on muscle and contraction phase, while strength related to muscle quantity and firing behavior in early KOA and to computed-tomography attenuation in severe KOA.³³

The same study found no significant between-group differences in vastus medialis or vastus lateralis attenuation values, or in recruitment and derecruitment thresholds; vastus medialis discharge responses also did not show the same pattern as vastus lateralis.³³ These null and heterogeneous results argue against uniform deterioration across every motor-unit and muscle-quality measure. They do not establish equivalence, given the small sample. The female-only, relatively low-pain sample further limits extrapolation. Thus, altered motor-unit behavior is credible, but a monotonic sequence from early to severe disease, or from AMI to reinnervation, is not demonstrated.

The human studies reviewed here do not demonstrate the sequence required to claim denervation–reinnervation: loss of functional units, emergence of denervation activity, collateral expansion of surviving units, and temporal association with persistent AMI. Accordingly, larger motor-unit potentials or lower discharge rates should be described as altered motor-unit behavior rather than evidence of motor-neuron degeneration or established reinnervation.

Competing Explanations and Anatomical Specificity Must Be Separated

The first explanation is functional adaptation: pain, effusion and instability alter recruitment without damaging axons or endplates, so motor-unit behavior should normalize when afferent drive and task strategy normalize. The second is compensatory reinnervation: partial denervation occurs, surviving axons expand and motor-unit potentials enlarge while force is temporarily preserved. The third is generalized aging or comorbidity: motor-unit loss reflects age, diabetes, lumbar disease or systemic sarcopenia, with KOA acting mainly as a marker of reduced activity or vulnerability. A fourth explanation is exposure and sampling: altered activity, adiposity, electrode geometry and task selection can change the motor-unit signal without structural loss. Cross-sectional case-control studies cannot distinguish these models.

A decisive design requires repeated voluntary-activation measures, high-density EMG decomposition, motor-unit-number estimation, muscle composition and activity exposure in the same individuals. Rapid normalization after reduction of joint-related inhibition would favor functional adaptation. Continued unit loss would prompt evaluation of structural injury or systemic disease, rather than identify its cause by itself. A larger effect in the affected quadriceps than in a remote upper-limb muscle would strengthen anatomical specificity, whereas similar remote abnormalities would favor generalized aging, neuropathy or systemic disease.

A Potentially Modifiable Motor-Unit Response

We hypothesize that altered recruitment and rate coding may remain responsive to restored activation and loading while compensatory and regenerative reserve is sufficient. This is a testable opportunity for intervention, not a separate disease stage or a fixed treatment window. Aging, metabolic disease and inactivity could modify recovery, but current data define neither an age boundary nor an obligatory interval preceding structural loss. Reversibility must be measured within individuals rather than inferred from radiographic severity or the order of domains.

Defining Persistent AMI and Cumulative Exposure for Research

No validated duration or severity threshold links persistent AMI to motor-unit loss or NMJ degeneration in human KOA. Acute effusion experiments establish effects over short experimental periods, whereas peri-arthroplasty studies describe marked early inhibition and subsequent recovery over months.^30,79,80 These different contexts cannot identify a biological point at which inhibition becomes structural injury. For a prospective nonoperative KOA study, a pragmatic candidate definition is objectively reduced voluntary activation on at least two standardized visits spanning at least 3 months, with a prespecified 6-month definition as a sensitivity analysis. The 3- and 6-month intervals are proposed sampling rules, not literature-validated critical thresholds or clinical diagnostic criteria.

Activation should be classified against method-specific reference limits and measurement error, using the same evoked-stimulation protocol, joint angle, familiarization and effort criteria at each visit. Pain, swelling, medication, neuropathy and testing tolerance should be recorded; symptom duration alone cannot define persistent AMI. Candidate visits at baseline, 6 weeks and 3, 6 and 12 months would characterize persistence and recovery, with longer follow-up if structural outcomes are sought. Acute postoperative trajectories should be analyzed separately because surgery introduces a new inhibitory exposure.

A candidate cumulative exposure is the area under the activation-deficit–time curve, calculated from repeated percentage-point deficits relative to a prespecified reference and expressed, for example, in percentage-point months. An alternative is time spent below a method-specific activation limit. Both are unvalidated summaries, vulnerable to fluctuating inhibition and sparse sampling. Dose–response analyses should test continuous and nonlinear associations with subsequent motor-unit and muscle-quality change, account for baseline outcomes and measured activity, and test whether any association differs by age, sex or metabolic status. The absence of longitudinal studies coupling these measures currently prevents estimation of a critical exposure dose, latency or irreversible threshold.

Domain IV: Muscle Quality Records the History of Neural and Mechanical Loading

Weakness without Marked Fiber Atrophy is Biologically Informative

Human biopsy data challenge the assumption that KOA weakness is primarily a problem of muscle quantity. Adults with moderate KOA showed increased extracellular matrix and collagen, fewer satellite cells, profibrotic signaling and fiber-type remodeling despite limited differences in mean fiber cross-sectional area.³⁴ Histological abnormalities have also been described in vastus medialis tissue from patients undergoing arthroplasty.⁸¹ Imaging studies link weakness to non-contractile tissue within the muscle.³⁵ These findings identify a tissue phenotype capable of lowering force per unit muscle size.

Fibrosis is not a passive scar. It can impair lateral force transmission, capillary exchange, fascicle shortening and satellite-cell migration. Inflammatory and NF-kB-related pathways promote proteolysis and interact with TGF-beta-dependent extracellular-matrix deposition.^82,83 Whether the signal originates in the joint, systemic metabolic inflammation or reduced loading is unresolved. The most parsimonious interpretation is convergence: local pain and inhibition reduce mechanical input while obesity, age and systemic inflammation alter the tissue response to that reduction.

Adipose Infiltration is Both a Marker and a Candidate Mediator

Thigh muscle fat is increased in KOA across MRI-based studies, and individual quadriceps muscles show differential atrophy.^84,85 Ultrasound meta-analysis similarly supports altered quadriceps thickness and echo characteristics, although acquisition and normalization remain inconsistent.⁸⁶ Single-nucleus profiling of intermuscular adipose tissue in older women with KOA has identified an inflammation-associated cellular profile.⁵⁶ This is one of the few tissue-level observations directly anchored in an older-adult KOA cohort. The comparison included six women with KOA and knee pain and five women without either condition; inferred leptin signaling was enriched in the KOA group. This small, cross-sectional, female-only study supports a local inflammatory hypothesis but cannot establish causation or a sex difference.

Several pathways could connect metabolic comorbidity to muscle quality. Conditioned media from human intermuscular adipose tissue reduced insulin sensitivity in cultured muscle cells, providing evidence for a local secretory effect outside a KOA-specific causal model.⁸⁷ In aged female rats and muscle-derived cells, leptin was linked to IL-6-related inflammation; this supports a candidate adipose–muscle pathway rather than direct human KOA evidence.⁸⁸ Adiponectin can stimulate AMPK-dependent glucose utilization and fatty-acid oxidation in experimental systems, illustrating that adipokines should not be treated as uniformly proinflammatory.⁸⁹ Local lipid exposure, impaired insulin signaling and inflammatory or extracellular-matrix signals may therefore converge with underloading, but their contributions and directionality in KOA remain unresolved.

Fiber Phenotype and Mitochondrial Quality are Downstream Integrators

Motor-unit activity helps determine myosin expression and fatigue resistance.⁹⁰ The reduction in type I fibers and increase in hybrid fibers reported in KOA muscle are therefore compatible with altered neural input, deconditioning or denervation–reinnervation; histology alone cannot choose among them.³⁴ Mitochondrial dynamics and autophagic quality control regulate muscle mass and stress tolerance.⁹¹ KOA-specific human evidence is limited, but mitochondrial dysfunction is a plausible amplifier of low activity, oxidative stress and impaired recovery rather than a proven primary driver.

Regenerative Reserve Links Muscle Tissue to Innervation

Satellite cells respond to loading, inflammation, extracellular matrix and innervation.⁹² Their reduction in KOA muscle may limit repair after repeated microinjury and constrain adaptation to renewed loading. Denervation can in turn alter the satellite-cell and fibro-adipogenic niche. This reciprocity matters: a muscle with severe fibrosis and reduced regenerative reserve may not recover simply because voluntary activation improves, while persistent neural underuse may prevent a viable muscle from receiving an adequate anabolic stimulus.

Human type 2 diabetes provides an additional, non-KOA tissue mechanism: expansion of CD90-positive fibro-adipogenic progenitors is associated with extracellular-matrix remodeling, and PDGF promotes their proliferation and collagen production.⁹³ This offers a plausible route by which diabetes could constrain adaptation independently of AMI. It does not demonstrate that the same pathway mediates KOA-related fibrosis or that improving activation reverses established matrix change. Metabolic disease may therefore be a parallel cause, an effect modifier or a mediator in different causal models; its role must be specified before statistical adjustment.

Domain V: NMJ Remodeling and the Threshold to Degeneration

Direct Evidence is Preclinical and Should Remain Labeled as Such

The strongest evidence connecting KOA to the NMJ comes from a rat anterior-cruciate-ligament-transection model. Quadriceps and tibialis anterior atrophy were accompanied by altered endplate dimensions and muscle-specific changes in MuSK and acetylcholine-receptor subunits.³⁶ Reduced MuSK and alpha-subunit expression, together with increased gamma-subunit expression in quadriceps, is compatible with impaired postsynaptic organization and a denervation-like transcriptional response. It is not proof of synaptic transmission failure. Whole-muscle expression does not localize the signal to endplates, and gamma-subunit re-expression can accompany regeneration as well as denervation.

Other KOA models demonstrate muscle wasting, and experimentally induced atrophy or weakness can worsen joint degeneration.^94–96 These experiments support a bidirectional joint–muscle relationship but do not establish that AMI caused NMJ remodeling. Surgical models produce abrupt instability and inflammation; the trajectory differs from slowly progressive human KOA. Model selection, histological grading and timing should follow standardized recommendations.^97,98 NMJ outcomes must be related to spontaneous activity, joint severity and muscle-specific loading.

What Would Constitute Convincing NMJ Pathology?

The agrin–LRP4–MuSK pathway organizes the postsynaptic apparatus.^50–52 LRP4 provides the receptor link between agrin and MuSK.⁵³ A change in one transcript is therefore insufficient to diagnose an NMJ lesion. Convincing evidence requires convergent measures: preterminal axon integrity, synaptic vesicle markers, terminal Schwann cells, postsynaptic acetylcholine-receptor topology, pre/postsynaptic apposition and, ideally, neuromuscular transmission. Fiber type and regenerative status must be mapped because endplate size and receptor expression vary with both.

Rodent morphology also requires cautious translation. Exercise and caloric restriction attenuate age-related synaptic changes in mice, showing that endplate architecture is plastic.³⁷ Reviews often describe age-related fragmentation as degeneration, yet fragmentation may coexist with preserved transmission.³⁸ Human NMJs are substantially smaller and less elaborate than mouse NMJs.³⁹ A rodent endplate that appears “simplified” after KOA cannot be assumed to represent the same biological state in humans.

Metabolic Injury as a Parallel Route to NMJ Dysfunction

Diabetes can affect axons and endplates independently of joint disease. In a long-duration insulin-deficient mouse model, distal motor-terminal and endplate innervation loss occurred despite preserved motor-neuron numbers; insulin delivery was associated with protection of innervation and PI3K–Akt signaling.⁹⁹ In 32-week-old db/db mice, partial NMJ denervation, acetylcholine-receptor fragmentation and attenuated transmission accompanied sensorimotor dysfunction.¹⁰⁰ These studies support a metabolic route to NMJ injury, but neither isolates insulin resistance as the cause of accelerated denervation in human KOA. The db/db model also has a leptin-receptor defect, so it cannot establish that elevated leptin directly damages an otherwise intact human NMJ.

The evidence reviewed here does not establish direct modulation of agrin–LRP4–MuSK signaling by leptin or adiponectin in human KOA. Proposed connections through inflammation, lipid exposure, impaired insulin signaling or altered repair remain mechanistic hypotheses and should not be represented as verified receptor-level links. Testing them would require pathway-specific assays alongside axonal occupancy and transmission measures. Clinically and experimentally, diabetic neuropathy is therefore a competing source of denervation as well as a possible modifier of recovery; it cannot be used as evidence that AMI caused NMJ failure.

Human Biomarkers are Indirect and Vulnerable to Confounding

Circulating C-terminal agrin fragments have been proposed as markers of NMJ instability and identify subsets of older adults with sarcopenia.^101–103 Their interpretation in KOA is problematic. Agrin fragments are systemic, influenced by renal handling, and are not specific to the quadriceps. In older adults with previous total knee arthroplasty, elevated agrin-related measures coexisted with features of musculoskeletal aging, but the study could not localize the source to the operated limb or establish KOA causality.¹⁰⁴ CAF should therefore be considered an exploratory systemic covariate, not a validated biomarker of local NMJ degeneration.

Remodeling Becomes Degeneration Only When Compensation Fails

We propose three operational evidence states for interpreting future NMJ findings. Adaptive remodeling changes recruitment, terminal architecture, or receptor organization while preserving transmission and reversibility. Maladaptive remodeling would comprise persistent partial denervation, enlarged or lost motor units, unstable endplates, or incomplete reinnervation while residual compensation remains. Degeneration should be reserved for sustained axonal or terminal loss, impaired transmission, and exhaustion of reinnervation reserve. These definitions are proposed operational criteria rather than validated clinical stages and are intended to prevent structural change from being treated as proof of degeneration without functional and temporal evidence.

Closing the Loop: Neuromuscular Failure Can Alter the Joint Environment

Joint–muscle interactions may be bidirectional. Quadriceps weakness reduces control of knee flexion and the capacity to respond to perturbation.^14,15 Patients may compensate by increasing co-contraction or altering coordination.^74,75 Shortened stance time and related strategies can further change loading.⁷⁶ These strategies can improve perceived stability while increasing compressive load and metabolic cost. Knee buckling, avoidance of activity and loss of strength then reinforce one another.^10,105

Human longitudinal studies do not support a universal protective effect of greater quadriceps strength. Associations vary by sex, alignment, compartment and symptom phenotype.^11–13 The implication is not that strength is harmful, but that force capacity and force use are different variables. A strong quadriceps acting through maladaptive alignment or coordination may not reduce compartmental loading. Conversely, animal experiments show that imposed weakness can accelerate degeneration.^95,96 The feedback loop should therefore be quantified with gait mechanics and activity exposure rather than inferred from peak torque.

Mechanism-Based Therapeutic Targets for Older Adults

In older adults with KOA, identical reductions in knee-extension strength can arise from different mixtures of abnormal joint afference, activation failure, motor-unit adaptation, fibrofatty muscle change, and generalized neuromuscular aging. Figure 2 presents coexisting mechanism profiles without a fixed order or ranking of reversibility. The therapeutic goal is to identify the dominant modifiable constraint, measure its response and match treatment to the capacity to express an adequate training stimulus. More than one target may require attention at the same time.

Figure 2.

Coexisting mechanism profiles in knee osteoarthritis, presented without a prescribed order or fixed ranking of recovery. The four panels describe activation failure, muscle-quality impairment, altered motor-unit behavior and suspected structural neuromuscular impairment. Profiles may overlap, and reversibility must be tested rather than inferred from panel position. A KOA-specific structural NMJ-failure phenotype has not been established in humans and would require convergent physiological and structural evidence. Age, sex and metabolic status may modify reserve and recovery across all profiles. Drawings and the waveform are schematic, not patient data.

Abbreviations: AMI, arthrogenic muscle inhibition; KOA, knee osteoarthritis; NMJ, neuromuscular junction.

Abnormal Afference and Impaired Activation

Effusion and active synovitis are rational proximal targets when activation loss varies with swelling. Aspiration or anti-inflammatory treatment may reduce abnormal afferent input, although evidence that either strategy prevents downstream motor-unit remodeling is absent. Cryotherapy can transiently reduce experimentally induced AMI.³⁰ Transcutaneous electrical nerve stimulation may improve selected functional outcomes, but its mechanism and durability remain uncertain.¹⁰⁶ In an older patient, the clinically relevant endpoint is not pain relief alone; it is recovery of voluntary activation and sufficient neural access to load the intended motor units. If pain improves but activation remains depressed, additional neural, task-related or tissue constraints should be assessed; persistence alone does not localize a lesion.

Inadequate Motor-Unit Recruitment and Loading

Neuromuscular electrical stimulation can recruit fibers when voluntary drive is inadequate and has evidence in the early period after total knee arthroplasty.¹⁰⁷ It should be viewed as a bridge to effective loading, not as proof that the primary lesion is electrical. Stimulation intensity, tolerance and superimposition on voluntary contractions determine whether high-threshold units are meaningfully engaged. Biofeedback and task-specific practice may improve force control, but trials should measure activation or motor-unit behavior rather than infer a neural mechanism from questionnaire improvement alone. The practical objective for this profile is to widen the effectively recruited motor-unit pool before increasing external resistance.

Muscle-Quality and Metabolic Constraints

Exercise improves pain and function and is recommended across international guidance.^108–110 External load, however, is not equivalent to biological dose. Marked AMI can prevent the target muscle from experiencing the intended stimulus, while fibrofatty replacement, frailty and comorbidity may further reduce the tissue response to a given workload. Progressive resistance, aerobic conditioning and neuromuscular training should therefore be adapted to activation status, pain phenotype, tissue quality and recovery capacity. MRI data show that neuromuscular control exercise, with or without added strengthening, can reduce intermuscular adipose tissue and improve specific strength.¹¹¹ For this profile, improvement in specific strength, force steadiness, mobility and participation is more informative than muscle size alone.

Peri-Arthroplasty Studies as a Distinct Clinical Context

Total knee arthroplasty removes severely diseased surfaces and often relieves pain, but it also creates acute inflammation, swelling and tissue injury. Voluntary activation and strength fall sharply after surgery, and pre-existing deficits may persist.^79,80 A randomized comparison of surgical and non-surgical care confirms substantial clinical benefit for selected patients but does not resolve the mechanism of residual weakness.¹¹² Peri-arthroplasty studies are valuable because they separate preoperative chronic afferent disturbance, early postoperative inhibition and later recovery after the joint environment stabilizes. Persistent abnormalities after pain and swelling improve would prompt assessment of residual activation failure, tissue quality and systemic disease; they would not by themselves prove irreversible local injury.

Suspected Structural Neuromuscular Impairment

Exercise is currently the most defensible NMJ-supportive strategy because it promotes motor-unit recruitment and attenuates age-related NMJ remodeling in experimental systems.^27,37 Pharmacological manipulation of agrin–LRP4–MuSK signaling, terminal Schwann cells or reinnervation pathways is biologically attractive but unsupported in KOA. An NMJ-directed trial should not enroll patients on the basis of radiographic severity or chronological age. It should require convergent evidence of motor-unit loss, transmission instability or endplate pathology and should exclude common mimics such as diabetic neuropathy and lumbar radiculopathy. Until such a subgroup is demonstrated in humans, the clinically defensible target is preservation of activation, loading and reinnervation reserve rather than direct NMJ repair.

Table 4 summarizes parallel treatment options and their measurement targets; its rows do not prescribe a treatment sequence. The practical implication is mechanism-specific combination therapy. An inhibition-dominant phenotype requires control of abnormal afference and restoration of activation. A motor-unit adaptation phenotype requires sufficiently intense, task-specific recruitment. A muscle-quality phenotype requires prolonged loading, metabolic risk management and nutritional adequacy. A structural neuromuscular-failure phenotype may have a lower ceiling on recovery even after joint replacement, but that ceiling should be demonstrated rather than assumed. In older adults, generalized sarcopenia can coexist with any of these phenotypes; consensus sarcopenia criteria can identify low strength and muscle quantity or quality, but sarcopenia should not be used as a synonym for KOA-related weakness.¹¹³ The treatment question is therefore which constraint is limiting effective force production, mobility and participation, and whether that constraint remains reversible.

Table 4.

Parallel Mechanism-Based Intervention Options for Older Adults with Knee Osteoarthritis

Predominant Mechanism/Constraint	Mechanistic Objective	Candidate Intervention	Proximal Endpoint/Functional Readout	Reason for Failure if Unselected
Abnormal afference/AMI	Reduce inhibitory input and restore effective drive	Effusion/synovitis control, cryotherapy, TENS, NMES-assisted activation	Voluntary activation, force development, mobility confidence	Pain relief may occur without activation recovery
Altered motor-unit behavior/inadequate recruitment	Recruit a broader unit pool and restore rate coding	Progressive resistance, task-specific training, NMES-assisted loading	Discharge behavior, force steadiness, rate of force development	External load may not reach under-recruited units; low reserve may limit adaptation
Muscle-quality impairment	Increase contractile fraction and regenerative response	Combined resistance/aerobic exercise, nutrition, metabolic risk control	Specific strength, MRI fat fraction, ultrasound quality, mobility and participation	Fibrosis, fat and low regenerative reserve may limit short-term reversibility
Suspected structural neuromuscular impairment	Preserve transmission and remaining reinnervation reserve	Exercise; experimental NMJ-directed strategies only in future defined subgroups	Convergent EMG, transmission, tissue and functional outcomes	No validated human KOA NMJ-defined subgroup; age alone is insufficient

Abbreviations: AMI, arthrogenic muscle inhibition; EMG, electromyography; MRI, magnetic resonance imaging; NMES, neuromuscular electrical stimulation; NMJ, neuromuscular junction; TENS, transcutaneous electrical nerve stimulation. These are parallel, potentially coexisting targets; row order does not indicate progression, mandatory treatment order or a fixed ranking of reversibility.

Future Directions: Experiments That Can Refute the Framework

Replace Cross-Sectional Association with Temporal Ordering

The central prediction is temporal: if persistent AMI contributes to downstream neuromuscular remodeling, activation deficits should precede measurable changes in motor-unit behavior and muscle quality. A prospective cohort should recruit early, moderate, and end-stage KOA, include age- and activity-matched controls, prespecify older-adult age strata, and repeat synchronized measurements using the proposed initial visit schedule in Figure 3, followed by longer observation for structural outcomes. Table 3 specifies the measurement domains and modifiers. Joint imaging should quantify effusion and synovitis; voluntary activation should be measured with a standardized evoked method; reflex and TMS measures should define neural level; high-density EMG and motor-unit-number estimation should quantify unit behavior; and MRI should separate contractile tissue from fat and fibrosis. Wearable activity and gait loading are required because biological exposure is the product of impairment and behavior.

Figure 3.

Multimodal assessment, parallel therapeutic targets and longitudinal validation. Joint, activation, motor-unit and muscle/NMJ measurements are synchronized, with age, sex and metabolic status recorded and remote-muscle controls used to assess anatomical specificity. Treatment targets are selected from the measured constraints and may be combined; the horizontal arrangement does not prescribe a sequence. NMJ-directed therapy remains investigational. Baseline, 6-week and 3-, 6- and 12-month visits are a proposed research schedule for quantifying repeated activation deficit and cumulative exposure, not validated diagnostic thresholds. Peri-arthroplasty trajectories require separate analysis. Drawings and waveforms are schematic.

Abbreviations: EMG, electromyography; HD-sEMG, high-density surface electromyography; KOA, knee osteoarthritis; MUNE, motor-unit-number estimation; NMES, neuromuscular electrical stimulation; NMJ, neuromuscular junction; TMS, transcranial magnetic stimulation.

A causal hypothesis would be strengthened if repeated activation deficits precede motor-unit change and predict later tissue-quality loss beyond baseline abnormalities and relevant competing causes. Analyses should prespecify age-, sex- and metabolic-status interactions, with sufficient sampling to estimate them. A causal diagram should distinguish confounders from potential mediators, including activity and muscle fat, rather than adjusting for every recorded variable indiscriminately. Associations remain observational even after adjustment. Equivalent abnormalities in remote muscles, abnormalities predating joint symptoms, or loss of association after appropriate control for systemic disease would weaken a joint-specific explanation.

Use Perturbation and Reversibility to Distinguish Function from Structure

Joint aspiration, targeted anti-inflammatory treatment and arthroplasty create perturbations of afferent input. Rapid improvement in activation with stable motor-unit number would indicate functional inhibition. Improvement in pain without recovery of discharge behavior or muscle quality would warrant investigation of additional constraints but would not prove structural injury. Continued loss of functional units after the joint signal is reduced would challenge the assumption that AMI is the principal driver and direct attention to age, neuropathy or irreversible local pathology.

Direct Human NMJ Analysis is the Decisive Gap

Tissue obtained during arthroplasty offers an opportunity for fibrosis quantification, single-nucleus or spatial transcriptomics and, where sampling permits, three-dimensional NMJ imaging. Older adults undergoing arthroplasty are a clinically relevant access point, but age alone cannot establish that any NMJ lesion is KOA-specific. The minimum NMJ dataset should include axonal occupancy, pre/post-synaptic apposition, terminal Schwann cells, acetylcholine-receptor topology, fiber type and denervation markers. Whole-muscle MuSK or receptor transcripts are insufficient. Affected quadriceps should be compared with an anatomically remote muscle or with carefully matched non-KOA tissue to determine whether the lesion is local or systemic.

Trials Should Enroll Mechanisms, Not Diagnoses

KOA trials commonly enroll by pain and radiographic criteria, then interpret heterogeneous responses as treatment failure. The framework predicts that enrichment by mechanism will increase signal. An AMI trial should require objectively low voluntary activation. A motor-unit trial should require abnormal discharge or number estimates. A muscle-quality trial should require low specific strength or imaging-defined fibrofatty replacement. An NMJ trial should require convergent structural and physiological evidence. Age, sex and metabolic status should be prespecified as potential modifiers rather than used as substitutes for mechanism. Core outcomes should extend beyond WOMAC and peak torque to activation, rate of force development, force steadiness, motor-unit discharge, muscle fat fraction, gait loading and participation.

Priority Falsification Criteria

Five results would materially weaken the model: no longitudinal association between AMI and motor-unit change; equivalent motor-unit abnormalities in unaffected and remote muscles; complete normalization of motor-unit behavior after analgesia despite persistent muscle pathology; absence of NMJ abnormalities in well-phenotyped human KOA tissue; and no relationship between preoperative neuromuscular phenotype and recovery after arthroplasty. Stating these criteria is essential. Without them, the framework would be an explanatory narrative capable of absorbing any result.

Conclusion

Human KOA evidence supports the coexistence of activation failure, altered motor-unit behavior and impaired muscle quality. It does not establish that persistent AMI causes denervation–reinnervation or progressive NMJ degeneration; the KOA-specific NMJ extension remains predominantly preclinical. Aging, sex and metabolic disease may modify reserve and recovery while also providing competing explanations. The framework organizes these uneven findings into testable relationships, not obligatory disease stages.

The priority is to determine whether repeated activation deficits precede motor-unit or tissue change and whether those changes respond to restored activation and loading. Proposed persistence definitions and cumulative-exposure measures require validation in longitudinal cohorts with anatomical controls and sex- and age-aware analyses. Until then, mechanism-directed care should target measured constraints and functional participation; NMJ-directed treatment remains investigational.

Funding Statement

This work was supported by the Lanzhou Science and Technology Development Guiding Plan (No. 2025‑5‑134), and the Gansu Provincial Clinical Key Specialty Capacity Improvement Project (Document No. Gan Zhong Yi Yao Yi Zheng [2026] No.1).

Disclosure

The authors report no conflicts of interest in this work.