Neurology · Medical Literature
Changes in cortical plasticity induced by paired associative stimulation in people with mild cognitive impairment and Alzheimer's disease: a systematic review and meta-analysis.
Abstract
Paired associative stimulation (PAS) is a non-invasive neuromodulation paradigm capable of inducing long-term potentiation (LTP)- or long-term depression (LTD)-like plasticity. It has been used to probe neuroplasticity and corticospinal excitability alterations in Alzheimer's disease (AD) and mild cognitive impairment (MCI). However, existing studies report inconsistent directions and magnitudes of PAS-induced plasticity changes across the AD continuum. We searched PubMed, Web of Science, Embase, and the Cochrane Library from database inception to December 2025. Eligible studies were case-control studies and randomized controlled trials (RCTs) that assessed PAS-induced cortical plasticity in individuals with AD or MCI, with healthy older adults as controls. Study quality was evaluated using the Newcastle-Ottawa Scale (NOS) for non-randomized studies and the Cochrane Risk of Bias tool (RoB 2) for RCTs. This review was registered in PROSPERO (CRD420251178441). Five studies met the inclusion criteria. Four studies quantified motor cortical plasticity using changes in motor-evoked potential (MEP) amplitude (total N = 135; 64 MCI/AD and 71 controls). The pooled analysis showed no significant difference in PAS-induced MEP changes between participants with MCI/AD and healthy controls [mean difference (MD) = 0.06, 95% CI (-0.05, 0.17), p = 0.32]. This pooled estimate was restricted to MEP-based outcomes and should be interpreted as an exploratory synthesis of motor-system readouts rather than definitive evidence of preserved cortical plasticity in MCI or AD. The remaining RCT, which assessed dorsolateral prefrontal cortex (DLPFC) plasticity using a repetitive PAS (rPAS) intervention without MEP outcomes, also found no significant improvement in prefrontal plasticity relative to control stimulation. Based on the currently available MEP-based evidence, PAS-induced motor-system plasticity findings in individuals with MCI or AD remain inconclusive. Given the limited number of studies and heterogeneity in experimental designs, the pooled negative results should not be interpreted as evidence that cortical plasticity is preserved. Corticospinal hyperexcitability and network-level PAS findings should be regarded as preliminary, hypothesis-generating observations requiring validation in larger, longitudinal, and biomarker-characterized cohorts.
Clinical relevance
Does paired associative stimulation (PAS) induce measurable and consistent differences in cortical plasticity—as indexed by motor-evoked potential (MEP) amplitude changes—between individuals with MCI or Alzheimer's disease and healthy older adult controls, and does repetitive PAS targeting the DLPFC improve prefrontal plasticity or cognition in this population? __SNAPSHOT__ {"studyDesign":"Systematic review and meta-analysis of case-control studies and randomized controlled trials (RCTs); PROSPERO-registered (CRD420251178441)","setting":"International; multi-database search (PubMed, Web of Science, Embase, Cochrane Library) from inception to December 2025","sampleSize":"5 studies met inclusion criteria; MEP-based meta-analysis: n=135 (64 MCI/AD, 71 healthy controls); 1 additional RCT assessed DLPFC plasticity via rPAS without MEP outcomes (sample size not separately reported in available abstract/full text excerpt)","population":"Adults with clinically defined MCI or Alzheimer's disease compared with healthy older adult controls; biomarker confirmation of AD pathology not consistently available across included studies","intervention":"Paired associative stimulation (PAS) paradigms, including classical peripheral nerve–M1 PAS (LTP- or LTD-inducing protocols), cortico-cortical PAS (cc-PAS targeting PPC–M1 pathways), and repetitive PAS (rPAS) targeting the dorsolateral prefrontal cortex (DLPFC)","comparator":"Healthy older adults (for case-control studies); control/sham stimulation (for the included RCT)","primaryOutcome":"PAS-induced change in MEP amplitude (as a neurophysiological index of motor cortical plasticity); prefrontal plasticity and cognitive outcomes for the rPAS RCT","followUp":"Not reported uniformly across included studies; acute post-PAS neurophysiological readouts were primary; durability of rPAS effects assessed within the RCT but timeframe not specified in available text","fundingSource":"Not reported in the available full text"}
Key findings
Four of five included studies quantified motor cortical plasticity using PAS-induced MEP amplitude changes (total n=135; 64 MCI/AD, 71 controls). The pooled meta-analytic result showed no statistically significant difference in PAS-induced MEP changes between participants with MCI/AD and healthy controls [mean difference (MD) = 0.06, 95% CI (−0.05, 0.17), p = 0.32]. NNT/NNH are not applicable to this neurophysiological synthesis. Heterogeneity statistics (I², τ²) were not reported in the extractable text, but the authors attributed divergent individual-study findings to methodological and biological heterogeneity across PAS paradigm type, cortical target, and disease stage. The fifth study—an RCT—assessed DLPFC plasticity using a repetitive PAS (rPAS) protocol; it found no significant improvement in prefrontal plasticity relative to control stimulation. Exploratory within-group analyses in the rPAS RCT suggested transient modulation of DLPFC plasticity and working memory after active rPAS, but these effects were not sustained. No adverse events were specifically reported in the available text.
Practice implications
This meta-analysis established that current MEP-based evidence is insufficient to demonstrate a consistent difference in PAS-induced motor-system plasticity between individuals with MCI or AD and healthy older adults. Critically, the null pooled result must not be interpreted as evidence of preserved cortical plasticity; rather, it reflects an underpowered, methodologically heterogeneous evidence base. The included studies differed substantially in PAS paradigm (peripheral nerve–M1 vs. cc-PAS vs. rPAS), cortical target, disease stage, and theoretical assumptions, making pooling of MEP outcomes an exploratory exercise rather than a definitive synthesis. MEP amplitude changes primarily reflect plasticity within the primary motor cortex and corticospinal pathways—a region less prominently affected in early AD compared with medial temporal and associative cortices—limiting the ecological validity of this biomarker for the clinical question of interest. Because biomarker confirmation of AD pathology (amyloid PET, CSF tau/Aβ) was not consistently available across included studies, findings apply to clinically defined MCI/AD cohorts only, not to biologically confirmed AD. The rPAS findings at the DLPFC are preliminary and hypothesis-generating only. Practice should not change based on these data. Only the abstract was available at the time of initial review; the full-text article was subsequently accessed, but quantitative heterogeneity statistics (I², τ²) and individual study-level effect sizes were not fully extractable from web-accessible content. __LIMITATIONS__ Extremely small evidence base: only 5 studies met inclusion criteria, with MEP-based pooled analysis restricted to n=135 across 4 studies—insufficient for reliable meta-analytic inference.,Clinically defined populations: absence of consistent biomarker confirmation (amyloid PET, CSF biomarkers) means included cohorts may represent heterogeneous or etiologically mixed groups, undermining specificity of findings to AD biology.,Substantial methodological heterogeneity: included studies used distinct PAS paradigms (peripheral nerve–M1, cc-PAS, rPAS), different cortical targets, variable inter-stimulus intervals, and differing disease stages, making pooling of MEP outcomes scientifically questionable.,Surrogate/proxy endpoint: MEP amplitude reflects corticospinal and primary motor cortex plasticity, which is neither the primary site of early AD pathology nor a validated surrogate for synaptic or cognitive plasticity in clinically relevant networks.,Lack of stage stratification: MCI and AD were pooled without separation by disease stage, amyloid status, or network phenotype, obscuring any potential stage-dependent plasticity signal.,No longitudinal design: all studies provided cross-sectional or acute post-PAS assessments; no data on whether PAS-indexed plasticity predicts disease progression or treatment response.,rPAS RCT limitations: single RCT for prefrontal plasticity with non-sustained within-group effects; adequacy of blinding, sample size, and protocol optimization not verifiable from available text.,Funding source not reported: independence of the synthesis from commercial interests cannot be confirmed.,Publication bias assessment: with only 5 eligible studies, formal funnel plot asymmetry testing is unreliable and was not interpretable.,Statistical vs. clinical significance: even if a future powered study detected a statistically significant MEP difference, the clinical meaning of MEP changes as a biomarker of AD-relevant plasticity remains unestablished. __BOTTOM_LINE_CATEGORY__ low_certainty __BOTTOM_LINE__ This systematic review and meta-analysis found no statistically significant difference in PAS-induced MEP amplitude changes between clinically defined MCI/AD patients and healthy older adults (MD = 0.06, 95% CI −0.05 to 0.17, p = 0.32), but this null result must not be interpreted as evidence of intact cortical plasticity given the critically small evidence base (5 studies, n=135 for MEP analysis), substantial paradigm heterogeneity, and use of a motor cortex proxy that may not capture plasticity in AD-affected networks. Repetitive PAS targeting the DLPFC showed no durable superiority over control stimulation for prefrontal plasticity or cognition in the single included RCT. PAS as a clinical or diagnostic neuromodulation tool in MCI/AD cannot be recommended based on current evidence; adequately powered, longitudinal, biomarker-stratified studies using multimodal neurophysiological readouts (e.g., TMS-EEG) are needed before any clinical application can be considered.