Enhanced episodic memory and LTP-like plasticity in subjective cognitive decline following 10-Hz repetitive transcranial magnetic stimulation

Abstract

Background

Subjective cognitive decline (SCD) is a self-perceived cognitive complaint in the absence of objective impairment, representing an at-risk state along the continuum of cognitive aging. High-frequency repetitive transcranial magnetic stimulation (rTMS) over the left dorsolateral prefrontal cortex (L-DLPFC) has shown cognitive benefits in Alzheimer’s disease (AD), yet its effects on cognitive functions in SCD remain largely unexplored.

Objective

To evaluate the effects of 10-Hz rTMS over the L-DLPFC in individuals with SCD on cognitive functions and long-term potentiation (LTP)-like plasticity, indexed by changes in motor evoked potential (MEP) following intermittent theta burst stimulation, and examine the relationship between brain plasticity and cognitive outcomes.

Methods

In this randomized, sham-controlled trial, 42 individuals with SCD received 20 sessions of either active or sham 10-Hz rTMS (n = 21 per group) over four weeks. The primary outcome was delayed episodic memory, evaluated using the Auditory Verbal Learning Test-Huashan version (AVLT-H). Secondary outcomes included additional cognitive measures and MEP amplitudes at baseline and at 5 (T₅), 10 (T₁₀), and 30 (T₃₀) minutes post-intervention.

Results

The rTMS group exhibited significant improvements in both delayed episodic memory (AVLT-N5) and MEP amplitudes at T₅ and T₁₀ following the intervention, whereas such changes were not observed in the sham group. Moreover, increased MEP amplitude at T₁₀ was positively correlated with improved AVLT-N5 performance.

Conclusions

These findings provide the first evidence for enhanced delayed episodic memory and LTP-like plasticity in individuals with SCD following 10-Hz rTMS over the L-DLPFC, suggesting a potential role of LTP-like plasticity in elucidating the neurophysiological correlates of cognitive improvements for SCD.

Trial registration

The study design and analysis plan were preregistered on September 7th, 2023 at Chinese Clinical Trial Registry (No. ChiCTR2300075517).

Background

Alzheimer’s disease (AD) is a leading cause of cognitive impairment among older adults, posing a significant global public health challenge [1]. Subjective cognitive decline (SCD) is characterized by persistent, self-reported cognitive complaints in the absence of objective impairment on standardized neuropsychological assessments [2]. SCD has been associated with an elevated risk of progression to mild cognitive impairment (MCI) and AD [3, 4]. Accumulating evidence suggests that SCD may reflect subtle neurobiological changes associated with early AD pathology, including disrupted connectivity and hippocampal dysfunction [5, 6], which positions it a potential window for early intervention. Compared to individuals with MCI or AD, those with SCD typically exhibit preserved cognitive performance and relatively intact brain structure, which may enhance responsiveness to early neuromodulatory or behavioral interventions [2, 7]. Despite promising findings from preliminary trials, there remains no established treatment protocols for SCD [8], highlighting an urgent need to develop novel strategies to address cognitive concerns in this at-risk population.

Episodic memory impairment is among the earliest and most salient cognitive deficits associated with AD [9, 10, 11]. This domain is integral to higher-level processes such as reasoning, judgment, and decision-making [12]. Impairments in episodic memory have been closely linked to disruptions in functional integration among memory-related brain networks in AD [13, 14]. Notably, deficits in episodic memory can be detected as early as the SCD stage [15], and these impairments are not restricted to medial temporal dysfunctions but also involve broader disruptions in cortical and subcortical networks supporting memory consolidation and retrieval processes [16]. Neuropsychological assessments such as the Auditory Verbal Learning Test (AVLT) has shown high sensitivity in detecting early episodic memory deficits, particularly within Chinese populations, where the Huashan version (AVLT-H) is widely used and validated for detecting memory deficits [15].

At the synaptic level, long-term potentiation (LTP)-like plasticity is fundamental to learning and memory processes [17]. Accumulating evidence indicates that disruptions in synaptic plasticity is central to AD pathogenesis and contributes significantly to early memory decline across the cognitive aging spectrum [18, 19]. Therefore, LTP-like plasticity has been emerged as a potential biomarker for cognitive vulnerability, with reduced plasticity associated with greater cognitive impairments in patients with AD [20] and predictive of conversion to dementia in individuals with memory dysfunction [21]. Intermittent theta burst stimulation (iTBS) is a patterned form of repetitive transcranial magnetic stimulation (rTMS) that can induce LTP-like plasticity in the human cortex [21, 22], making it a valuable protocol for investigating early cortical changes in AD [23]. Importantly, iTBS-induced plasticity in the primary motor cortex (M1) shares neruophysiological features with hippocampal plasticity alterations in animal models of AD [24], making it a useful tool for exploring plasticity-related dysfunction in preclinical stages of cognitive decline. However, the extent to which LTP-like plasticity relates to cognitive outcomes in individuals with SCD remains poorly understood.

It is noteworthy that rTMS has emerged as a promising non-invasive neuromodulatory intervention for cognitive impairments in AD [25]. At the neurophysiological level, high-frequency rTMS (HF-rTMS) is thought to facilitate synaptic plasticity by reducing magnesium blockade of N-methyl-D-aspartate (NMDA) receptors during neuronal depolarization, thereby enhancing postsynaptic excitability and promoting LTP-like mechanisms [26]. Although rTMS-induced plasticity changes typically last 30 to 60 min, their longer-term consolidation may be influenced by individual differences in genetic predispositions and brain network integrity, potentially leading to sustained cognitive benefits [27]. Although limited, several studies have investigated the effects of HF-rTMS on domain-specific cognitive functions in SCD. For example, Sole-Padulles et al. [28] found that 5-Hz rTMS over the left prefrontal cortex enhanced face-name associative memory, while Liu et al. [29] reported that 10-Hz rTMS over the left dorsolateral prefrontal cortex (L-DLPFC) enhanced visual working memory, attention, and executive function. In addition, Liang et al. [30] found that rTMS over the precuneus in individuals with SCD modulated their abnormal effective connectivity in the default mode network (DMN) and the executive control network (CEN), resulting in improved episodic memory performance. Despite these encouraging findings, whether delayed episodic memory, one of the most sensitive cognitive domains in individuals with SCD, can benefit from HF-rTMS remain uncertain.

The L-DLPFC is often chosen as a target in rTMS interventions for cognitive impairments, as it plays an important role in encoding and retrieving episodic memory [31, 32, 33, 34]. In particular, HF-rTMS over the DLPFC is generally associated with excitatory effects, increasing local cortical activity, enhancing functional connectivity, and improving the allocation of cognitive resource within the PFC [35, 36]. There is evidence showing a positive correlation between increased cortical plasticity in the M1 and improved cognitive function in AD patients following HF-rTMS over the DLPFC [37]. Complementing this finding, our previous study also found that 4 weeks of HF-rTMS over the prefrontal cortex in healthy adult rats enhanced spatial episodic learning and memory and modulated receptor-dependent brain plasticity in the PFC, hippocampus, and M1 [38]. Collectively, these findings raise the possibility that HF-rTMS may enhance cognitive function by promoting LTP-like plasticity.

Therefore, the present study investigated whether HF-rTMS (10 Hz) over the L-DLPFC could improve cognitive functions, particularly delayed episodic memory, in individuals with SCD. To explore potential neurophysiological correlates of treatment effects, we also measured changes in LTP-like plasticity, typically indexed by iTBS-induced motor evoked potential (MEP) amplitudes in M1, before and after the intervention. We hypothesized that HF-rTMS would enhance episodic memory performance and promote LTP-like plasticity in individuals with SCD and that these improvements would be associated with changes in LTP-like plasticity. These findings would provide insights into the neurophysiological mechanisms by which rTMS may exert beneficial effects on cognitive declines in SCD.

Methods

Participants

This study was designed as a randomized, sham-controlled clinical trial. Participants were recruited from local communities and memory clinics at the Department of Rehabilitation Medicine of The Affiliated Jiangsu Shengze Hospital of Nanjing Medical University in China between September 2023 and March 2024. Ethical approval was granted by the Human Research Ethics Committee of The Affiliated Jiangsu Shengze Hospital (No. 2023-002-01), and all participants provided written informed consent prior to enrollment. The study protocol was preregistered with the Chinese Clinical Trial Registry (No. ChiCTR2300075517) and published [39] in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines for randomized clinical trials. A detailed flow diagram of participant enrollment, allocation, and analysis is presented in Fig. 1.

Fig. 1.

The CONSORT diagram of the clinical trial

Inclusion/exclusion criteria

Participants were included in this study if they met the following inclusion criteria [40, 41]: (1) not meeting the diagnostic criteria for MCI as determined by standardized neuropsychological assessments of memory, executive function, and language; (2) self-reported memory decline persisting for at least six months; (3) personal concerns about memory performance being worse than that of age-matched peers; (4) age between 55 and 80 years old; and (5) ability and willingness to provide written informed consent.

Patients were excluded if they if they met any of the following criteria: (1) a Modified Hachinski Ischemic Score > 4, or a diagnosis of vascular dementia or other forms of dementia based on the NINDS-AIREN criteria as assessed by two psychiatrists; (2) medical or neurological conditions that could interfere with cognitive function or study participation; (3) symptoms of depression, defined by a score of ≥ 6 on the Geriatric Depression Scale; (4) any contraindications to TMS (e.g., history of seizures, metallic implants in the head); (5) a history of substance or alcohol abuse within the past six months; (6) recent treatment with central nervous system medications within the past three months; or (7) a history of severe cardiovascular or cerebrovascular diseases.

Sample size calculation

The sample size was determined based on an anticipated effect size of Cohen’s d = 0.80 [42]. With an alpha level of 0.05 and a power (1 − β) of 80%, the minimum sample size per group was calculated using G*Power software (version 3.1).To account for an estimated 5% attrition rate, a total of 24 participants were recruited in the present study.

TMS intervention

TMS was administered using a D-MT500 magnetic biphasic stimulator (Neurosoft Ltd., Russia) equipped with a figure-of-eight coil (AFEC-02-100-C; Neurosoft Ltd., Russia). The stimulation target was the L-DLPFC, localized using the F3 scalp position according to the international 10–20 EEG system [43]. Participants in the active rTMS group received 10 Hz stimulation over the L-DLPFC at 90% of their resting motor threshold (RMT). Each session consisted of 50 pulses that were delivered in 5-second trains with a 10-second intertrain interval, resulting in a total of 40 trains (2,000 pulses). Sessions lasted approximately 10 min and were administered 5 times per week over a 4-week period (total of 20 sessions). Participants in the sham group underwent an identical procedure using a self-contained sham stimulation protocol to maintain blinding and ensure comparability.

Primary outcomes

The primary outcome was episodic memory performance, assessed using the AVLT-H. This assessment has been validated as an effective and robust measure of verbal memory and semantic categorization in mainland Chinese populations [44]. The AVLT-H is widely used in clinical and research settings to detect cognitive impairment in individuals with SCD and MCI [45]. Previous studies have shown that measures such as delayed recall and memory retention are effective in identifying memory impairments associated with AD [44].

Secondary outcomes

Secondary outcomes included assessments of additional cognitive domains and LTP-like cortical plasticity. To minimize individual variability, all assessments were conducted on the same day for each participant, with cognitive tests performed in the morning and neurophysiological evaluations scheduled in the afternoon. Global cognitive function was assessed using the Mini-Mental State Examination (MMSE). Language function was evaluated using the Animal Fluency Test (AFT) and the Boston Naming Test China version (BNT-C). Executive function was assessed by measuring the completion time on the Trail Making Test (TMT) parts A and B [40]. In addition, working memory and inhibitory control were assessed using a 1-back task and a Go/No-Go task, respectively. These tasks were administered via E-Prime 2.0 software (Psychology Software Tools Inc., Pittsburgh, PA), which recorded both task accuracy and reaction times for further analysis.

To evaluate LTP-like plasticity, MEPs were recorded from the abductor pollicis brevis (APB) muscle of the right hand using SKINTACT RT-34 electrodes (Fannin Ltd., Dublin, Ireland; 10.5 mm × 25 mm) connected to an electromyography (EMG) system (Neuro-MEP-Micro, Russia). Participants were seated in a relaxed, semi-reclined position during assessment. The TMS coil was positioned over the left M1, with the handle angled 45° backward from the midline of the sagittal plane of the brain. The individual RMT, defined as the minimum stimulation intensity required to elicit MEPs ≥ 50 µV in at least 5 out of 10 consecutive trials, was used to determine subsequent stimulation intensities.

LTP-like plasticity was indexed by changes in MEP amplitude in response to iTBS. Single-pulse TMS was applied at the left motor hotspot at 120% of the RMT, and 20 consecutive MEPs were recorded per time point. MEP measurements were obtained at five time points: two baseline recordings (separated by 5 min) and three recordings at 5 (T₅), 10 (T₁₀), and 30 (T₃₀) minutes post-iTBS (see Fig. 2). To ensure baseline reliability, iTBS was administered only if the difference between the two baseline MEP averages was less than 20% [46]. The iTBS protocol consisted of 2-second trains of three stimuli delivered at 50 Hz, repeated every 10 s over a total duration of 192 s, resulting in 600 pulses at 80% of the RMT [46, 47]. All MEPs were peak-to-peak amplitude averaged and normalized relative to baseline to quantify the degree of induced plasticity.

Fig. 2.

Flow of LTP-like plasticity assessment. MEPs were elicited by delivering 20 consecutive single-pulse TMS stimuli on the motor hotspot in the left M1. The peak-to-peak amplitudes of the 20 MEPs were measured and averaged for further analysis. MEPs were assessed at five time points: two baseline assessments with 5 min apart (the second used as the reference baseline) and three measurements obtained at 5, 10, and 30 min post-iTBS

Statistical analysis

Statistical analyses were performed using SPSS (v. 23.0; IBM Corp., Armonk, NY). Statistical significance was defined as p < 0.05 (two-tailed), with a 95% confidence interval. Demographic and baseline characteristics were compared between groups using independent samples t-tests for continuous variables and Fisher’s exact tests for categorical variables. Cognitive function performance was analyzed using two-way repeated-measures analysis of variance (RM-ANOVA), with session (pre- and post-intervention) as a within-subject factor, group (rTMS and sham) as a between-subjects factor. To evaluate LTP-like plasticity, independent-samples t-tests were first performed on raw MEP amplitudes before and after the intervention to determine whether both groups were comparable at baseline. Next, two-way RM-ANOVAs with factors group (rTMS vs. sham) and timepoint (T₅, T₁₀, T₃₀) were performed on normalized post-iTBS MEP amplitudes before and after the intervention to examine between-group differences in post-iTBS changes. Finally, separate one-way RM-ANOVAs were conducted within each group to compare normalized MEP amplitudes across four timepoints (T₀, T₅, T₁₀, T₃₀) before and after the intervention to determine rTMS-induced changes in LTP-like plasticity. Bonferroni correction was applied for adjustment of multiple comparisons. In addition, to assess the association between changes in cognitive performance and LTP-like plasticity, we calculated post–pre differences for both memory performance scores and MEP amplitudes, standardized these values into z-scores, and examined their relationship using Spearman’s rank-order correlation. Effect sizes were quantified using partial eta-squared (η²ₚ) values, with η²ₚ values of 0.01–0.05, 0.06–0.13, and > 0.14 indicating small, medium, and large effects, respectively [48].

Results

Participants

Table 1 presents the demographic and baseline characteristics of both the rTMS and sham groups. No significant differences were observed between the two groups in terms of age, sex, education background, MMSE scores, and PSQI scores, indicating that the two groups were well-matched across these variables.

Table 1.

Demographic and clinical characteristics of participants in the rTMS and Sham groups

	rTMS group	Sham group	t/χ2	P value
Age (years)	69.48 ± 6.86	65.38 ± 6.52	1.982	0.054
Sex (F, %)	13 (60%)	16 (75%)	1.000	0.317
Education (years)	10.10 ± 2.61	8.76 ± 2.95	-1.553	0.128
MMSE (scores)	28.76 ± 1.30	29.33 ± 1.15	1.506	0.140
PSQI (scores)	6.84 ± 4.37	8.38 ± 4.99	-1.032	0.309

Note: Between-group comparisons of categorical variables (sex distribution) were analyzed using chi-square tests, while continuous variables were assessed using independent samples t-tests. The results were reported as mean ± standard deviation (SD). MMSE: Mini Mental State Examination; PSQI: Pittsburgh sleep quality index

Episodic memory

Table 2 shows the results of statistical analyses examining performance on all AVLT-H subsets as a function of group and time. Two-way RM-ANOVAs revealed significant main effects of time for all AVLT-H subtests (p < 0.003) except AVLT-N7 (p = 0.443), reflecting overall improvements from pre- to post-intervention across the two groups. In contrast, no significant main effects of group were observed for any subtest (all p > 0.05). However, a significant time × group interaction was found specifically for AVLT-N5 (F(1, 40) = 9.774, p = 0.003, η²ₚ=0.196). Follow-up analyses revealed a significant improvement in AVLT-N5 in the rTMS group (t=-1.952, p < 0.001), whereas no significant difference was observed in the sham group (t=-0.048, p = 0.913).

Table 2.

Changes of AVLT-H subscales before and after the intervention in the rTMS and Sham groups

	rTMS group		Sham group		P value			η2p
	Baseline	Post-treatment	Baseline	Post-treatment	Group	Time	Group x Time	Group	Time	Group x Time
AVLT-N1 + N2 + N3	15.05 ± 3.88	18.33 ± 3.81	15.33 ± 3.45	17.14 ± 3.44	0.617	0.001*	0.286	0.006	0.258	0.028
AVLT-N4	4.71 ± 2.33	6.29 ± 1.52	5.52 ± 1.81	6.86 ± 1.82	0.175	< 0.001*	0.694	0.045	0.369	0.004
AVLT-N5	3.90 ± 1.79	5.86 ± 1.98	4.86 ± 1.42	4.90 ± 1.79	0.842	0.002*	0.003*	0.000	0.212	0.196
AVLT-N1 + N2 + N3 + N4 + N5	23.67 ± 6.18	30.48 ± 6.64	25.72 ± 4.98	28.90 ± 5.01	0.873	< 0.001*	0.063	0.001	0.412	0.084
AVLT-N7	21.42 ± 1.78	21.19 ± 3.03	20.43 ± 2.25	21.48 ± 2.42	0.504	0.443	0.226	0.011	0.015	0.036

AVLT-H: auditory verbal learning Test-Huashan; *P < 0.05

Secondary cognitive functions

Table 3 shows the statistical results regarding changes in performance on secondary cognitive functions as a function of group and time. There were significant main effects of time on AFT (F(1, 40) = 6.949, p = 0.012, η²ₚ=0.148), BNT-C (F(1, 40) = 4.915, p = 0.032, η²ₚ=0.109), and TMT-A-C assessments (F(1, 40) = 4.401, p = 0.042, η²ₚ=0.099), while the main effects of time failed to reach significance for the other subscales (p > 0.2). However, no significant main effects of group (all p > 0.5) or time × group interactions were found across any secondary cognitive functions (p > 0.05).

Table 3.

Changes of secondary cognitive functions before and after the intervention in the rTMS and Sham groups

	rTMS group		Sham group		P value			η2p
	Baseline	Post-treatment	Baseline	Post-treatment	Group	Time	Group x Time	Group	Time	Group x Time
AFT	16.00 ± 3.48	17.52 ± 3.98	17.19 ± 3.37	18.42 ± 3.19	0.277	0.012*	0.786	0.029	0.148	0.002
BNT-C	23.57 ± 2.62	24.81 ± 3.78	21.90 ± 3.43	23.05 ± 3.73	0.067	0.032*	0.930	0.082	0.109	0.000
TMT-A-C (s)	69.24 ± 25.42	62.57 ± 28.20	69.14 ± 24.53	61.28 ± 17.14	0.917	0.042*	0.864	0.000	0.099	0.001
TMT-B-C (s)	153.81 ± 38.47	145.57 ± 37.77	158.62 ± 65.68	158.38 ± 34.85	0.714	0.294	0.632	0.003	0.028	0.006
Go RT (ms)	336.17 ± 17.78	329.07 ± 14.85	336.44 ± 17.18	328.45 ± 19.08	0.164	0.930	0.818	0.053	0.000	0.001
No-Go ERR (%)	0.06 ± 0.11	0.06 ± 0.11	0.07 ± 0.13	0.07 ± 0.13	0.838	0.249	0.616	0.001	0.034	0.007
1-back ACC (%)	0.83 ± 0.22	0.87 ± 0.11	0.88 ± 0.14	0.89 ± 0.07	0.452	0.308	0.521	0.015	0.027	0.011

AFT: animal fluency test; BNT-C: Boston naming test China version; TMT-A/B-C: trail making test parts A and B China version; go RT: the average response time of the go task; No-Go ERR: No-Go task error rate; 1-back ACC: 1-back task accuracy; *P < 0.05

Long-term potentiation-like plasticity

Table 4 shows changes in LTP-like plasticity, indexed by normalized MEP amplitudes in M1, across the rTMS and sham groups before and after the intervention. At baseline, independent sample t-tests on raw MEPs confirmed that the rTMS and sham groups did not differ significantly before (t=-0.359, p = 0.722) or after the intervention (t=-0.873, p = 0.388). To assess group differences in iTBS-induced plasticity before and after the intervention, two-way RM-ANOVAs were conducted on normalized post-iTBS MEPs at T₅, T₁₀, and T₃₀. Before the intervention, no significant main effects of timepoint (F(2, 80) = 2.729, p = 0.071) or group (F(1, 40) = 0.571, p = 0.454) as well as their interaction (F(2, 80) = 1.609, p = 0.206) were observed. After the intervention, there was a significant main effect of timepoint (F(2, 40) = 6.596, p = 0.004, η²ₚ=0.142), with MEP amplitudes at T₃₀ significantly smaller that at T₅ (p = 0.032) and T₁₀ (p = 0.015). However, neither the main effect of group (F(1, 40) = 2.775, p = 0.104) nor group × timepoint interaction (F(2, 80) = 1.491, p = 0.231) reached significance.

Table 4.

LTP-like responses indexed by normalized MEP amplitudes at T₅, T₁₀, and T₃₀ time points before and after the intervention in the rTMS and Sham groups

	Baseline (mV)	T5 (%)	T10 (%)	T30 (%)	Group (P)	Time (P)	Group x Time (P)	Group (η2p)	Time (η2p)	Group x Time (η2p)
Pre-intervention					0.454	0.071	0.206	0.014	0.064	0.039
rTMS group	0.81 ± 0.50	101 ± 21.20	114 ± 28.74	106 ± 19.60
Sham group	0.86 ± 0.48	115 ± 30.18	118 ± 31.02	103 ± 28.25
Post-intervention					0.104	0.004*	0.231	0.065	0.142	0.036
rTMS group	0.66 ± 0.29	130 ± 36.45	143 ± 38.01	109 ± 24.92
Sham group	0.76 ± 0.09	118 ± 32.07	119 ± 32.39	108 ± 36.01

*P < 0.05

To further determine changes in LTP-like plasticity within each group, separate one-way RM-ANOVAs were performed on normalized MEPs across T₀, T₅, T₁₀, and T₃₀. In the rTMS group, no significant main effect of timepoint was found before the intervention (F(3, 60) = 2.655, p = 0.057), but there was a significant main effect of time after the intervention (F(3, 60) = 11.067, p < 0.001, η²ₚ=0.356). Post hoc comparisons revealed significantly higher MEP amplitudes at T₅ (p = 0.007) and T₁₀ (p < 0.001) relative to baseline (see Fig. 3). In contrast, the sham group exhibited no significant main effects of timepoint either before (F(3, 60) = 2.548, p = 0.088) or after the intervention (F(3, 60) = 2.988, p = 0.099).

Fig. 3.

Changes in LTP-like plasticity levels at the T₅, T₁₀, and T₃₀ for the rTMS and sham groups before and after the intervention. a: comparison of normalized MEP amplitudes between T₀ and T₅ in the rTMS group after the intervention; b: comparison of normalized MEP amplitudes between T₀ and T₁₀ in the rTMS group after the intervention; *P < 0.05

Association between cognitive function with LTP-like plasticity

To explore whether changes in LTP-like plasticity were associated with improvements in memory performance, we examined the relationship between z-score–transformed post–pre changes in AVLT-N5 and normalized MEP amplitudes at T₁₀ within the rTMS group. The results revealed a significant positive correlation (ρ = 0.453, p = 0.045; Fig. 4), indicating that individuals with greater increases in MEP amplitudes tended to show larger gains in delayed recall following rTMS over the left DLPFC. However, when accounting for individual variability in baseline memory performance using a linear regression model with post-intervention AVLT-N5 as the dependent variable, baseline AVLT-N5 as a covariate, and T₁₀ MEP change as the predictor, the results did not to show a significant correlation (F(2, 20) = 2.294, p = 0.131).

Fig. 4.

Scatter plots showing significant correlations between post-pre changes in AVLT-N5 and iTBS-induced MEP at T₁₀ in the rTMS group. AVLT-H: Auditory Verbal Learning Test-Huashan; *P < 0.05

Discussion

This study investigated the cognitive and neurophysiological effects of HF-rTMS in individuals with SCD. Following a four-week session of 10-Hz rTMS over the L-DLPFC, participants in the rTMS group exhibited improved delayed episodic memory and LTP-like plasticity, as reflected by enhanced AVLT-N5 scores and MEP amplitudes at T₅ and T₁₀. Importantly, improvement of AVLT-N5 scores was significantly correlated with enhancement of MEP amplitudes at T₁₀. In contrast, the sham group exhibited no significant changes in cognitive outcome or MEP amplitudes. These findings provide the first evidence that 10-Hz rTMS over the L-DLPFC can lead to domain-specific cognitive benefits in individuals with SCD and highlight the potential role of LTP-like plasticity in elucidating the neurophysiological correlates of treatment-induced cognitive improvement in this at-risk population.

Previous research has demonstrated the beneficial effects of rTMS on episodic memory in healthy individuals [49, 50]. Our study extends these findings to a cognitively at-risk population, showing that individuals with SCD following HF-rTMS over the L-DLPFC exhibited significant improvements in episodic memory, particularly tasks requiring delayed recall. This finding is consistent with prior studies in AD, where rTMS was found to improve both immediate and delayed recall [51, 52]. Our findings suggest that delayed episodic memory may be particularly sensitive to HF-rTMS, potentially because it is one of the earliest cognitive functions to deteriorate along the AD continuum. The observed improvements in episodic memory may reflect enhanced functional integration or compensatory recruitment within fronto-hippocampal networks, a hypothesis supported by as previous studies linking episodic memory performance (e.g., Rey AVLT) and hippocampal activity in patients with AD [53]. Notably, no significant improvements were observed in other cognitive domains such as language or executive function, consistent with prior studies reporting domain-specific benefits of rTMS in episodic memory without generalized cognitive enhancement [52, 54]. This selective effect may be attributed to the vulnerability of episodic memory in SCD, which is closely associated with the function integrity of the L-DLPFC.

Although our study revealed a significant improvement in AVLT-N5 for the rTMS group relative to the sham group, no between-group differences were found for other AVLT subtests or broader cognitive functions. The general improvement in episodic memory performance across both the rTMS and sham groups may reflect non-specific influences such as practice effects or increased familiarity with the testing procedures, which are likely to enhance performance on tasks requiring immediate recall (e.g., AVLT-H) [55]. However, the significant time-by-group interaction exclusively for AVLT-N5 (delayed recall following interference) suggests the involvement of a more specific neurobiological mechanism beyond generalized learning or expectancy effects [56]. Unlike immediate recall, delayed recall after interference places greater demands on executive control and resistance to proactive interference, processes that are critically supported by the integrity of the prefrontal-hippocampal network [57, 58]. HF-rTMS over the left DLPFC may enhance synaptic efficacy and strengthen functional connectivity within these circuits, thereby improving memory consolidation and retrieval under high interference conditions [59, 60]. In contrast, sham stimulation did not provide sufficient neuromodulatory influence to support delayed recall after interference, resulting in the absence of corresponding improvement. This task-specific dissociation underscores the role of L-DLPFC in mediating rTMS-induced cognitive benefits and highlights AVLT-N5 as a sensitive behavioral marker of plasticity following this neuromodulation.

The L-DLPFC has been widely recognized as an optimal neuromodulatory target for interventions for improving cognitive functions in patients with MCI and AD [32], which is primarily due to its critical role in processing and recalling verbal information, working memory, and executive control [34]. Applying rTMS to the L-DLPFC has been proposed to enhance episodic memory by facilitating semantic processing and promoting memory reconsolidation [61]. In supporting of this hypothesis, Manenti et al. [62] found that anodal transcranial direct current stimulation (tDCS) applied to the left parietal cortex and L-DLPFC significantly improved verbal memory retrieval in older adults. Similarly, anodal tDCS over the left PFC in individuals with SCD enhanced delayed memory retrieval compared to sham stimulation, with effects persisting for up to 30 days [63]. Consistent with these findings, our results provide additional evidence that HF-rTMS over the L-DLPFC may enhance and consolidate verbal episodic memory in individuals with SCD.

Hyperactivation of the DLPFC has been consistently associated with compensatory responses to cognitive decline across the AD spectrum [64]. In MCI, increased DLPFC activation is thought to compensate for structural and functional deficits, enhancing performance across various cognitive domains like semantic retrieval [65]. A similar compensatory process may occur in SCD, where increased DLPFC activation may counterbalance reduced activation of the right hippocampus during the retrieval of episodic memory, enabling performance comparable to healthy individuals [66]. Notably, this compensatory hyperactivation is not confined to the DLPFC itself but extends to other cortical regions through distal clusters of neurons. For example, Solé-Padullé et al. [28] found that 5-Hz rTMS over the L-DLPFC in individuals with SCD enhanced the recruitment of the right PFC and bilateral posterior cortical regions during associative memory tasks. Similarly, Plas et al. [50] found that 1-Hz rTMS over the L-DLPFC in healthy individuals modulated beta power in the posterior brain regions to facilitate memory encoding. These findings suggest that rTMS over the L-DLPFC may not only enhance local cortical excitability but also modulate neural activity across connected networks, particularly involving parietal and posterior regions. Such network-level modulation highlights the potential of L-DLPFC neuromodulation as a circuit-based intervention strategy for mitigating early cognitive decline.

LTP-like plasticity provides a fundamental neurophysiological basis for learning and memory by maintaining the integrity of distributed neural networks [67, 68]. Disruptions in LTP-like plasticity reflect synaptic dysfunction and reduced network stability [17], which has been observed in early AD [24]. In line with these findings, our results found that individuals with SCD exhibited no significant LTP-like plasticity before the intervention, suggesting the presence of early, subclinical deficits in synaptic responses. Importantly, following a four-week intervention of 10-Hz rTMS over the L-DLPFC, the rTMS group exhibited significantly increased MEP amplitudes at T₅ and T₁₀, whereas no such modulation occurred in the sham group. These findings suggest that HF-rTMS may reinstate LTP-like plasticity, possibly via NMDA receptor-dependent mechanisms that recalibrate excitatory-inhibitory balance within compromised memory networks [69]. This interpretation is supported by Freedberg et al. [49], showing that rTMS over the inferior parietal cortex enhanced episodic memory performance while strengthening hippocampal-cortical connectivity. Similarly, rTMS over the precuneus in AD patients improved episodic memory and enhanced functional connectivity within the DMN, particularly between the precuneus and the medial frontal lobe [54]. These converging lines of evidence highlight the potential of rTMS to restore synaptic plasticity and functional integration within compromised memory networks.

Our study revealed a significant positive correlation between improvements in AVLT-N5 scores and increases in LTP-like plasticity at T₁₀ in the rTMS group, suggesting a potential functional link between rTMS-induced synaptic plasticity and delayed episodic memory improvement. However, it is noteworthy that this association did not remain significant when baseline memory performance was controlled for using regression analysis, indicating that the observed relationship should be viewed as preliminary. Nevertheless, previous research has shown that disrupted LTP-like plasticity is associated with memory deficits and elevated cerebrospinal fluid (CSF) tau levels in AD [70]. However, other reports have reported an inverse relationship between LTP-like plasticity and memory function in amyloid-positive patients with MCI, where greater plastic responses may reflect maladaptive hyperexcitability or compensatory overactivation during prodromal stages [71]. These discrepancies highlight the complex and context-dependent nature of neuroplasticity in early cognitive decline. While cortical hyperexcitability and disrupted homeostatic regulation have been implicated in progressive memory dysfunction [72, 73], reduced LTP-like responses in MCI may alternatively represent an adaptive mechanism to limit excitotoxicity and preserve neural integrity [73]. Taken together, these findings suggest that LTP-like plasticity may serve as a neurophysiological correlate of treatment-induced cognitive improvements in SCD, warranting further investigation in larger and longitudinal cohorts to clarify the robustness and significance of this relationship.

This study has several limitations that should be addressed. First, the relatively small sample size with a predominantly female cohort may restrict the generalizability of our findings. Although the trial was preregistered with a larger target sample, unforeseen logistical challenges including COVID-19–related disruptions to recruitment and temporary closures of clinical research sites impeded our ability to achieve the planned recruitment. Second, we did not evaluate psychological or emotional states such as anxiety, depression, or stress, which are known to influence cognitive performance and could confound treatment effects. Third, the lack of neuroimaging assessments limits our ability to characterize prefrontal plasticity and its network-level contributions underlying the observed behavioral improvement. Future research should address these limitations by recruiting larger, demographically balanced cohorts, incorporating assessments of psychological well-being, and integrating neuroimaging approaches to elucidate the neural mechanisms underlying rTMS-induced cognitive benefits in individuals with SCD.

Conclusion

In summary, our study provides evidence that 10-Hz rTMS over the L-DLPFC can selectively enhance delayed episodic memory and promote LTP-like plasticity in individuals with SCD, as reflected by improved AVLT-N5 scores and increased MEP amplitudes at T₅ and T₁₀. A significant correlation was also found between improved AVLT-N5 and increased MEP amplitudes at T₁₀, although this relationship should be considered preliminary given its failure to reach significance after adjusting for baseline memory performance. These findings highlight the potential of HF-rTMS over the L-DLPFC for restoring synaptic plasticity and enhancing cognitive function in SCD, suggesting that LTP-like plasticity may provide important insights into the neurophysiological basis underlying treatment-induced cognitive benefits.

Abbreviations

AD

Alzheimer’s disease

ADAS-cog

Alzheimer’s Disease Assessment Scale-Cognitive Subscale

AFT

Animal Fluency Test

APB

Abductor pollicis brevis

BNT-C

Boston Naming Test China version

CEN

Executive control network

CSF

Cerebrospinal fluid

DMN

Default mode network

EMG

Electromyography

HF-rTMS

High-frequency repetitive transcranial magnetic stimulation

iTBS

Intermittent theta burst stimulation

L-DLPFC

Left dorsolateral prefrontal cortex

LTP

Long-term potentiation

M1

Primary motor cortex

MCI

Mild cognitive impairment

MEP

Motor evoked potential

MMSE

Mini-Mental State Examination

NMDA

N-methyl-D-aspartate

PFC

Prefrontal cortex

PSQI

Pittsburgh Sleep Quality Index

RMT

Resting motor threshold

RM-ANOVA

Repeated-measures analysis of variance

SCD

Subjective cognitive decline

SPSS

Statistical Package for the Social Sciences

TMT

Trail Making Test

tDCS

Transcranial direct current stimulation

Author contributions

Authors’ contributionsTianjiao Zhang contributed to the writing of the original draft, conducted formal analysis, and curated data. Qian Lu developed the methodology and contributed to the software implementation. Jie Song, Manyu Dong, Han Yang and Yilun Qian performed the assessments and intervention for the patients. Chuan He and Zude Zhu provided resources and supervised the study. Hanjun Liu participated in writing, reviewing, and editing the manuscript. Tong Wang managed project administration. Ying Shen secured funding and contributed to the conceptualization of the study. All authors reviewed and approved the final manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (No.82372582), the Competitive Project of Jiangsu Province’s Key Research and Development Program (No. BE2023034), and the Jiangsu Province Hospital clinical diagnosis and treatment of technological innovation “Open bidding for selecting the best candidates” project (No. JBGS202414).

Data availability

The datasets used and/or analyzed in the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Human Research Ethics Committee of The Affiliated Jiangsu Shengze Hospital (No. 2023-002-01). All the patients signed the written informed consent before enrollment.

Consent for publication

Consent for publication were given by all participants.

Competing interests

The authors declare no competing interests.