You are here

Article Text

Article menu
Cognitive neurology
Original research
Circuit-based neuromodulation enhances delayed recall in amnestic mild cognitive impairment
  1. Jie Ma1,2,
  2. Jia-Jia Wu1,3,
  3. Xiang-Xin Xing1,
  4. Xin Xue1,3,
  5. Yun-Ting Xiang2,
  6. Xiao-Min Zhen4,
  7. Jian-Hua Li5,
  8. Juan-Juan Lu2,
  9. Jun-Peng Zhang2,
  10. Mou-Xiong Zheng3,6,
  11. Xu-Yun Hua3,6,
  12. Jian-Guang Xu2,3
  1. 1 Center of Rehabilitation Medicine, Yueyang Hospital of Integrated Traditional Chinese and Western Medicine, Shanghai University of Traditional Chinese Medicine, Shanghai, China
  2. 2 School of Rehabilitation Science, Shanghai University of Traditional Chinese Medicine, Shanghai, China
  3. 3 Engineering Research Center of Traditional Chinese Medicine Intelligent Rehabilitation, Ministry of Education, Shanghai, China
  4. 4 Department of Neurology, Yueyang Hospital of Integrated Traditional Chinese and Western Medicine, Shanghai University of Traditional Chinese Medicine, Shanghai, China
  5. 5 Department of Heart Disease, Yueyang Hospital of Integrated Traditional Chinese and Western Medicine, Shanghai University of Traditional Chinese Medicine, Shanghai, China
  6. 6 Department of Traumatology and Orthopedics, Yueyang Hospital of Integrated Traditional Chinese and Western Medicine, Shanghai University of Traditional Chinese Medicine, Shanghai, China
  1. Correspondence to Professor Jian-Guang Xu; xjg{at}shutcm.edu.cn; Xu-Yun Hua; huaxuyun{at}shutcm.edu.cn; Mou-Xiong Zheng; zhengmouxiong{at}shutcm.edu.cn

Abstract

Background This study aimed to investigate the efficacy of circuits-based paired associative stimulation (PAS) in adults with amnestic mild cognitive impairment (aMCI).

Methods We conducted a parallel-group, randomised, controlled clinical trial. Initially, a cohort of healthy subjects was recruited to establish the cortical-hippocampal circuits by tracking white matter fibre connections using diffusion tensor imaging. Subsequently, patients diagnosed with aMCI, matched for age and education, were randomly allocated in a 1:1 ratio to undergo a 2-week intervention, either circuit-based PAS or sham PAS. Additionally, we explored the relationship between changes in cognitive performance and the functional connectivity (FC) of cortical-hippocampal circuits.

Results FCs between hippocampus and precuneus and between hippocampus and superior frontal gyrus (orbital part) were most closely associated with the Auditory Verbal Learning Test (AVLT)_N5 score in 42 aMCI patients, thus designated as target circuits. The AVLT_N5 score improved from 2.43 (1.43) to 5.29 (1.98) in the circuit-based PAS group, compared with 2.52 (1.44) to 3.86 (2.39) in the sham PAS group (p=0.003; Cohen’s d=0.97). A significant decrease was noted in FC between the left hippocampus and left precuneus in the circuit-based PAS group from baseline to postintervention (p=0.013). Using a generalised linear model, significant group×FC interaction effects for the improvements in AVLT_N5 scores were found within the circuit-based PAS group (B=3.4, p=0.017).

Conclusions Circuit-based PAS effectively enhances long-term delayed recall in adults diagnosed with aMCI, which includes individuals aged 50–80 years. This enhancement is potentially linked to the decreased functional connectivity between the left hippocampus and left precuneus.

Trial registration number ChiCTR2100053315; Chinese Clinical Trial Registry.

  • CLINICAL NEUROLOGY
  • COGNITION

Data availability statement

Data are available upon reasonable request.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

https://doi.org/10.1136/jnnp-2023-333152

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    WHAT IS ALREADY KNOWN ON THIS TOPIC

    • Repetitive transcranial magnetic stimulation was recognised for its potential to improve cognitive function in amnestic mild cognitive impairment (aMCI). Existing modalities focused on high-frequency or intermittent theta burst stimulation targeting specific brain regions but suggested the need for more effective protocols and targets due to the complex nature of memory processes.

    WHAT THIS STUDY ADDS

    • This study demonstrates that circuit-based paired associative stimulation targeting the cortical-hippocampal circuits can significantly enhance long-term delayed recall in adults with aMCI. The improvements were associated with decreased functional connectivity between the left hippocampus and left precuneus, highlighting the potential of this novel stimulation protocol in enhancing cognitive functions related to aMCI.

    HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

    • The findings suggest a new direction for neuromodulation therapies in aMCI, emphasising the importance of targeting specific brain circuits. This approach could refine therapeutic strategies and stimulate further research into personalised neuromodulation interventions for cognitive impairments, potentially influencing clinical practices and policies regarding the treatment of aMCI and similar conditions.

    Introduction

    Amnestic mild cognitive impairment (aMCI) has garnered significant attention as a pivotal transitional stage in the continuum form normal cognitive function to dementia, making it a focal point of cognitive research in recent years.1–3 Within the realm of neuromodulation, repetitive transcranial magnetic stimulation (rTMS) has gained increasing recognition for its potential to ameliorate cognitive function in aMCI.4–6 Presently, the prevailing stimulation modalities involve high-frequency or intermittent theta burst stimulation applied to brain regions such as the dorsolateral prefrontal cortex, the precuneus cortex and the parietal cortex. Nevertheless, more effective stimulation protocols and targets should be explored.7 One plausible reason is the inadequacy of single-target stimulation in producing sufficient efficacy, as it may not be fully aligned with the complexities of memory processes.8 9 The pursuit of a novel stimulation mode and corresponding targets that better align with the memory processes holds the potential to overcome these therapeutic bottlenecks, bearing substantial clinical and scientific significance.

    Episodic memory reflects the ability to recall the temporal and spatial context of previous experiences. The memory processes encompass three stages: information acquisition, encoding, and storage and retrieval. These stages require collaborative engagement of multiple brain regions.10 The hippocampus stands as the central structure (hub) for episodic memory, yet memory storage engages various brain regions, with diverse types of memory information residing in distinct neural structures. Specifically, the hippocampus plays a pivotal role in integrating memory by fostering information exchange with various brain regions through interconnections among the hippocampus and other neural areas.11–14 Therefore, the neural circuits associated with the hippocampus play a crucial role in the memory process.

    A stimulation pattern that targets these ‘circuits’ seems to be more congruent with the mechanics of memory. Consequently, there is a pressing need to investigate a form of stimulation that operates on these connections, which appears to be in harmony with the memory processes. In recent years, increasing scholarly attention has been directed towards the pivotal role of neural circuits in both brain function and structure. Fox has contended that disparate sites of brain damage can lead to similar clinical symptoms, and complex symptoms can be mapped to larger distributed brain networks rather than being limited to isolated brain regions.15

    Damage to specific areas or connections between regions can lead to intricate ‘disconnection’ syndromes. He has introduced the innovative concept of ‘identifying treatment targets based on the connectome’. In 2022, a series of articles have laid the theoretical and methodological foundation for neuromodulation strategies rooted in connections.16–19 Building on this foundation, we introduce an innovative approach for defining stimulation targets: employing the hippocampus as the seed region and tracing white matter fibre connections with cortical areas in healthy individuals to identify pre-existing structural connections within the hippocampus-cortex circuit. Among these connections, we select those with significant functional associations with long-delayed recall function, a paramount function in aMCI patients,20 21 were selected as the target circuits.22

    Subsequently, we delve into the exploration of stimulation patterns and optimal parameters, following the identification of subject-specific target circuits. The theoretical underpinning for the circuit-based stimulation pattern lies in synaptic plasticity mechanisms, predominantly Hebbian plasticity.23 Paired associative stimulation (PAS) represents a neuromodulation approach potentially rooted in Hebbian theory, as it induces repeated coupling activity between interconnected neuron populations.24 By adjusting the interstimulus interval between two stimuli, we can invoke spike-timing-dependent plasticity and selectively modulate physiological connections among brain regions.24 Our proposal involves the application of PAS patterns in aMCI patients, targeting cortical-hippocampal circuits that exhibit the strongest associations with cognitive symptoms. Addressing the technical challenge of stimulating these subject-specific circuits is accomplished through connectivity-based segmentation and a pilot study (online supplemental material 1).

    Supplemental material

    In summary, we introduce a novel PAS protocol designed to modulate symptom-related cortical-hippocampal circuits with the aim of enhancing long-delayed recall function in aMCI patients. To substantiate our hypothesis, we conducted a parallel-group, randomised, controlled clinical trial among aMCI patients, placing particular emphasis on evaluating the strength of cortical-hippocampal connectivity and its correlation with cognitive functions.

    Methods

    Study design and participants

    This study was designed as a parallel-group, randomised, controlled clinical trial. Participants were recruited from the rehabilitation centre at Yueyang Hospital of Integrated Traditional Chinese and Western Medicine, Shanghai University of Traditional Chinese Medicine (online supplemental file 1).

    Supplemental material

    We initially recruited healthy adults to track cortical-hippocampal circuits. These right-handed individuals aged 50–80 years underwent neuropsychological assessments to confirm normal cognitive function. Inclusion criteria required participants to have at least 6 years of education, no reported cognitive decline verified by neuropsychological assessments, and informed consent.

    Eligible right-handed participants diagnosed with aMCI were aged 50–80 years. The diagnosis followed Jak/Bondi’s diagnostic criteria25 (online supplemental material 2), did not have taken any cognitive medication, had an education of 6 years or more, complained of memory loss with a decreased long-term delayed recall score in Auditory Verbal Learning Test (AVLT) (an impaired score defined as >1 SD below the age-corrected normative mean). Exclusion criteria for all participants included comorbid tumours, severe heart, liver, kidney, haematologic disorders or infectious diseases as well as a history of neurological disorders (such as cerebrovascular disease, Parkinson’s syndrome, epilepsy, dementia from various causes) or psychiatric disorders (such as anxiety, depression, schizophrenia, etc), severe visual or hearing impairment, drug or alcohol abuse and contraindications to MRI and TMS treatment (eg, pacemaker, cardiac stent, artificial heart valve, fixed plate after fracture surgery, etc) (figure 1).

    Supplemental material

    Figure 1
    Figure 1

    Trial profile. aMCI, amnestic mild cognitive impairment; TMS, transcranial magnetic stimulation.

    Procedures

    The study comprised three main phases: identification of subject-specific cortical-hippocampal circuits related to cognitive function in aMCI, modulation of target circuits, and exploration of their relationship with clinical efficacy. These steps included:

    1. Obtaining white matter fibre maps of cortical-hippocampal circuits using probabilistic fibre tracking and diffusion tensor imaging (DTI) in healthy subjects matched in age to aMCI patients.

    2. Selecting cortical-hippocampal circuits whose functional connectivity (FC) was significantly related to long-term delayed recall scores in aMCI patients for circuit-based PAS.

    3. Randomly assigning aMCI patients to either the circuit-based PAS group or the sham PAS group and locating the target circuits for each patient.

    4. Administering 2 weeks of circuits-based PAS or sham PAS.

    5. Comparing changes in cognitive functions before and after intervention between the two groups.

    6. Comparing changes in functional magnetic resonance imaging (fMRI) data before and after intervention and examining the correlation between these changes and cognitive function (figure 2).

    Figure 2
    Figure 2

    The research procedure. (A) Identification of subject-specific cortical-hippocampal circuits related to cognitive function in aMCI: obtaining white matter fibre maps of cortical-hippocampal circuits using probabilistic fibre tracking and DTI in healthy subjects matched in age to aMCI patients, And selecting cortical-hippocampal circuits whose functional connectivity was significantly related to long-term delayed recall scores in aMCI patients for circuit-based PAS; (B) modulation of target circuits: randomly assigning aMCI patients to either the circuit-based PAS group or the sham PAS group and locating the target circuits for each patient; (C) processing the data and constructing individualised target circuits and 3D precise localisation, and administering 2 weeks of circuit-based PAS or sham PAS; (D) stimulus pattern: all patients received 10 trials with an inter-stimulus interval of 2 ms, 900 pairs pulses in total (five trials a week for 2 weeks). aMCI, amnestic mild cognitive impairment; DTI, diffusion tensor imaging; PAS, paired associative stimulation.

    Randomisation and blinding

    Patients with aMCI were randomly assigned to receive either circuits-based PAS or sham PAS using a computer-based algorithm. Researchers assigned random identification numbers to participants after they met inclusion and exclusion criteria and signed informed consent forms. Treating therapists were aware of treatment allocation, while outcome assessors were blinded to treatment assignment. Participants were also blinded to treatment allocation.

    Clinical and neuropsychological assessments

    Demographic data were collected at baseline, and clinical and neuropsychological assessments were conducted at baseline (2 weeks before treatment for designation of optimised target) and at the end of the 2-week treatment. Two senior neuropsychologists with >10 years of work experience performed the neuropsychological evaluation without knowledge of the clinical diagnosis; another senior neuropsychologist then reviewed the assessment results. Demographic data include gender, age, handedness, height, weight, average daily sleep time, average daily exercise time, years of education, marital status, times of general anaesthesia and previous history. Clinical data include Hamilton Anxiety Scale,26 Hamilton Depression Scale27 and Functional Activities Questionnaire.28 Neuropsychological evaluations include Mini-Mental State Examination (MMSE),29 AVLT,30 shape trails test (STT),31 symbol digit modalities test (SDMT),32 Boston naming test (BNT)33 and complex figure test (CFT).34 Detailed assessment standards of these scales have been included in online supplemental material 3.

    Supplemental material

    Imaging data

    Imaging data were collected at two time points: baseline (2 weeks before intervention) and at the end of the 2-week intervention. MRI scans were scheduled between 16:00 and 18:00 to maintain consistent data quality. Half an hour before the MRI scan, the subject will enter a quiet preparation area without bright light. During the scan, the subject was placed in a lying position with the head fixed in the coil. Before scanning, subjects will wear noise-proof earplugs and will be asked to follow the instructions during the MRI scan: breath calmly, close your eyes, stay awake, stay relaxed, keep your body free of movement, especially your head and raise your hand to indicate any discomfort during the scan.

    Image acquisition

    MRI data were acquired using a MAGNETOM Verio 3.0-Tesla scanner (Siemens, Erlangen, Germany). Resting-state fMRI data were obtained with a gradient-recalled echo-planar imaging (EPI) sequence with the following parameters: transverse plane; repetition time (TR), 3000 ms; echo time (TE), 30 ms; flip angle, 90°; slice thickness, 3.0 mm; slice number, 43; matrix size, 64×64; field of view (FOV), 230 mm×230 mm; voxel size=3.6 mm×3.6 mm×3.0 mm; and number of acquisitions, 200.

    A T1-weighted magnetisation-prepared rapid acquisition gradient echo scan was then performed, with the following parameters: TR, 1900 ms; inversion time, 900 ms; TE, 2.93 ms; flip angle, 9°; FOV, 256 mm×256 mm; slice thickness, 1 mm.

    DTI was performed using a single-shot spin EPI in the axial plane: TR, 10 000 ms; TE, 89 ms; flip angle, 90°; slice thickness, 2.0 mm; in-plane resolution, 1.875 mm; 60 non-colinear directions (b, 1000 s/mm2), and two b0 images.

    Imaging data preprocessing and processing

    DTI data were processed in the Camino (http://www.cs.ucl.ac.uk/research/medic/camino/) and FMRIB Software Library V.5.0 (University of Oxford Center for Functional MRI of the Brain, http://www.fmrib.ox.ac.uk/)35 software. Functional image preprocessing and FC calculation were performed using MATLAB 2013b platform (The Mathworks, Natick, USA), Statistical Parametric Mapping V.12 (SPM12) (http://www.fil.ion.ucl.ac.uk/spm/), and Data Processing Assistant for Resting-State fMRI36 (V.5.0) (http://www.rfmri.org/dpabi). The data processing and construction of individualised target circuits are described in detail in the online supplemental material 4.

    Supplemental material

    Interventions

    Stimulus was delivered with a MagPro X100 stimulator equipped with the B70 fluid-cooled coil (MagVenture). The maximum surface magnetic field intensity of the coil was 4.2T. All patients received 10 trials (five trials a week for 2 weeks). Each patient’s resting motor threshold (RMT) was tested in accordance with the standard practice37 (online supplemental material 5).

    Supplemental material

    Circuit-based PAS

    Two B70 fluid-cooled eight-figure coils were applied to left prefrontal lobe and left precuneus (online supplemental material 4). Paired stimulation included a 80% RMT stimulus on the left prefrontal lobe, followed by a 120% RMT stimulus on the left precuneus. A trial of circuits-based PAS consisted of trains of 5 Hz paired stimulations with an interstimulus interval of 2 ms, 900 pairs pulses in total.

    Sham PAS

    Two sham coils were placed vertically on the scalp of the patients. The coils generated stimulating sounds but with no virtual effect on the brain. Other parts of protocol were the same as the circuits-based PAS group (figure 2B).

    Outcome measures

    The primary outcome measure was the change in long-term delayed recall performance following the 2-week intervention, measured by the age- and education-normalised AVLT_N5 score (Delta AVLT_N5). Secondary outcome measures included MMSE, AVLT items (excluding AVLT_N5), STT, SDMT, BNT, CFT and fMRI data. Correlations between changes in cognitive performance and functional connections (FC) were explored.

    Safety measures

    Adverse effects and accidents were monitored and recorded during the course of the intervention. Serious adverse events (SAEs) and treatment discontinuation were documented and evaluated for their relevance to clinical interventions.

    Sample size calculation

    Sample size calculation was calculated based on long delay recall score as the main observation index. As this study adopts a new treatment method, sample size was acquired from our preliminary study. As a result, the mean (5.1) and SD (2.23) after treatment, as well as the mean (2.6) after false stimulation, a minimum total sample size of 33 treatment completers was required to achieve 90% power at α=0·05. To account for attrition and ensure adequate power at 2 weeks after treatment, 42 participants need to be included, with 21 subjects in each group.

    Statistical analysis

    SPSS V.21.0 (SPSS, Chicago, USA) software was used for statistical analysis for clinical data. Continuous variables were expressed as mean±SD. Independent sample t-test was used in the comparison between two groups when the data conform to the normal distribution and homogeneity test of variance.

    A generalised linear model was constructed with AVLT_N5 score as dependent variable and every FC value of cortical-hippocampal circuits as independent variable.

    The repeated measurement analysis of variance (ANOVA) was performed in comparison between two groups before and after treatment, the test of within-subject effect was corrected by Greenhouse-Geisser, and the post test was performed by least significant difference (LSD) method. Effect sizes (Cohen’s d)38 were interpreted as small=0.2, medium=0.5, large=0.8. Non-parametric test was adopted when the data did not conform to normal distribution or homogeneity of variance. The discrete data were expressed by frequency/rate, and the comparison between the two groups was conducted by χ2 test or Fisher exact probability method. The p<0.05 (two-sided) indicated significant statistical difference.

    The statistical analysis of magnetic resonance data was carried out using SPM12, GRETNA and the Resting State fMRI Data Analysis Toolkit39 67 (RESTV.1.8) based on Matlab 2013b (The MathWorks, USA). The flexible factorial of the second order analysis was used to design the statistical matrix, and the three factors included subjects, group and time, in which group and time were set as fixed factors, while the number of subjects was random factors, and then the main effect and Group×Time interaction matrix were set for statistical analysis, respectively. Brain regions with p<0.05 and voxel >50 were extracted for post hoc analysis. The results were reported using bspmview software (https://www.bobspunt.com/software/bspmview/), and the brain regions were referred to anatomical automatic labeling (AAL) template. For correlation analysis, a generalised linear model was fitted with the delta AVLT_ N5 as the dependent variable, group and the delta FC value between left hippocampus and left precuneus as independent/ interactive variables, using age and years of education as covariates.

    Results

    Characteristics of participants

    Participants were recruited between 1 November 2021 and 30 February 2022. A total of 26 healthy subjects underwent initial screening, with 21 (80.77%) subsequently enrolled for tracking cortical-hippocampal circuits. Eligibility assessments were conducted on 70 aMCI patients, resulting in the enrolment of 42 (60%) who were randomly assigned to either the circuit-based PAS or sham PAS groups. Notably, significant disparities in cognitive performance scores existed between the healthy subjects and aMCI patients (p<0.05), while no significant difference was found between the circuit-based PAS and sham PAS groups at baseline (all p>0.05) (table 1).

    View this table:
    Table 1

    Characteristics of the healthy subjects and patients with aMCI

    Target circuits related to long-term delayed recall function

    The FC value between the left hippocampus and the left superior frontal gyrus (orbital part) exhibited a positive correlation with the AVLT_N5 scores in aMCI patients (B=3.44; p=0.045). Conversely, the FC values between the left hippocampus and the left precuneus, as well as between the hippocampus and the left temporal pole (superior temporal gyrus), demonstrated negative correlations with the AVLT_N5 scores in aMCI patients, respectively (B=−4.71, p=0.032; B=−3.09; p=0.025) (figure 2A). Notably, the interaction effect of age and years of education on AVLT_N5 scores had not been found (p=0.759). FC between hippocampus and precuneus, and between hippocampus and superior frontal gyrus (orbital part), were most closely associated with the AVLT_N5 score in 42 aMCI patients, thus designated as target circuits.

    Efficacy outcomes

    Primary outcome

    At baseline, the mean AVLT_N5 score was 2.43 (1.43) in the circuit-based PAS group and 2.52 (1.44) in the sham PAS group, consistent with the diagnostic criteria for aMCI. The mean change from baseline to the end of the 2-week intervention was 2.86 (1.80) in the circuit-based PAS group and 1.33 (1.32) in the sham PAS group. An estimated adjusted difference of 1.52 (95% CI 0.54 to 2.51; p=0.003) was observed between the two groups, favouring circuit-based PAS (figure 3A, table 2).

    Figure 3
    Figure 3

    Primary and key secondary endpoints. (A) showed the results for the primary endpoint (Delta AVLT_N5), the score on the AVLT_N5. Scores range from 0 to 12, with lower scores indicating greater impairment. (B–F) showed the results for the key secondary endpoints; values were calculated in the same manner as those for the primary endpoint. (G) showed estimated adjusted difference shown with 2-sided lower and upper 90% CIs. (H) showed the mean change from baseline in functional connectivity between left hippocampus and left precuneus. Green represents circuit-based PAS group, red represents Sham PAS group, and asterisk represents p<0.05. AVLT_IR, Auditory Verbal Learning Test, total score of immediate recall; AVLT_N1, Auditory Verbal Learning Test, first immediate recall; AVLT_N5, Auditory Verbal Learning Test, long-term delay recall; AVLT_N6, Auditory Verbal Learning Test, long delay cued recall; AVLT, Auditory Verbal Learning Test; BNT, Boston Naming Test; HIP.L, left hippocampus; PAS, paired associative stimulation; PCUN.L, left precuneus.

    View this table:
    Table 2

    Primary and secondary outcomes

    Secondary outcomes

    At baseline, the mean AVLT score was 17.81 (6.07) in the circuit-based PAS group and 18.62 (6.02) in the sham PAS group. The mean change from baseline to the end of the 2-week intervention was 11.43 (6.22) in the circuit-based PAS group and 6.38 (5.27) in the sham PAS group (estimated adjusted difference, 5.06; 95% CI 1.46 to 8.65; p=0.007). The mean AVLT_IR score at baseline was 12.43 (3.30) in the circuit-based PAS group and 13.19 (3.80) in the sham PAS group. The mean change of AVLT_IR was 6.00 (3.54) in the circuit-based PAS group and 3.38 (3.81) in the sham PAS group (estimated adjusted difference, 2.46; 95% CI 0.20 to 4.73; p=0.034). The mean AVLT_N1 score at baseline was 2.71 (1.15) in the circuit-based PAS group and 2.91 (1.09) in the sham PAS group. The mean change of AVLT_N1 was 2.29 (1.42) in the circuit-based PAS group and 0.95 (1.83) in the sham PAS group (estimated adjusted difference, 1.27; 95% CI 0.27 to 2.28; p=0.014). The mean AVLT_N6 score at baseline was 2.29 (1.85) in the circuit-based PAS group and 2.86 (1.56) in the sham PAS group. The mean change of AVLT_N6 was 2.86 (1.68) in the circuit-based PAS group and 1.05 (1.86) in the sham PAS group (estimated adjusted difference, 1.57; 95% CI 0.52 to 2.62; p=0.004). The mean BNT score at baseline was 19.81 (3.78) in the circuit-based PAS group and 21.33 (3.43) in the sham PAS group. The mean change in BNT was 2.67 (2.63) in the circuit-based PAS group and 0.62 (2.36) in the sham PAS group (estimated adjusted difference, 1.37; 95% CI 0.99 to 2.63; p=0.035) (figure 3E and tables 1 and 2). No between-group differences were found in AVLT_N2, AVLT_N3, AVLT_N4, AVLT_N7, MMSE, STT-B, SDMT_correct or CFT_copy scores (online supplemental material 6).

    Supplemental material

    Correlation between changes of FC and cognitive performance

    On reanalysis, we observed a significant decrease in the FC between the left hippocampus and left precuneus across both groups from baseline to the conclusion of the 2-week intervention period (estimated adjusted difference, −0.02; 95% CI −0.12 to 0.07, p=0.67). Specifically, the circuit-based PAS group exhibited a change of −0.13±0.23 (p=0.013), while the sham PAS group showed a change of −0.05±0.10 (p=0.318) (figure 3H). A generalised linear model was fitted, with the delta AVLT_N5 as the dependent variable and the group×delta FC between the left hippocampus and the left precuneus as group×FCs as interactive variable, using age and years of education as covariates. A significant group×FC interaction effects for the improvements in AVLT_N5 scores were found within the circuit-based PAS group (B=3.4, p=0.017), which was absent in the sham PAS group (p=0.533). Another generalised linear model was fitted with the delta AVLT, delta AVLT_IR, delta AVLT_N1, delta AVLT_N6, delta BNT as the dependent variable and the delta FC between the left hippocampus and the left precuneus as independent variables, using age and years of education as covariates. No significant differences were found in the five secondary endpoints (p>0.05).

    Safety

    Adverse events related to the intervention included headache (n=3), fatigue (n=2), nausea (n=1) and dizziness (n=1) in the circuit-based PAS group, compared with two cases of headache and one case of fatigue in the sham PAS group, as detailed in online supplemental material 7. None of these participants reported any SAE.

    Supplemental material

    Discussion

    The concept of PAS was initially reported two decades ago, introducing a non-invasive brain stimulation protocol involving paired stimuli with a fixed repetition interval.40 Initially, PAS employed a pair of stimuli, one in the periphery and the other in the cortex, capable of inducing Hebbian plasticity changes.40 Subsequently, this protocol has been extensively replicated and led to the development of cortico-cortical PAS (ccPAS), which has found application in various functional systems. It is clear that ccPAS exhibits feasibility and potential for research and application in neural plasticity. In this study, we chose the left frontal lobe and left precuneus for paired stimulation, based on prior research, a preliminary research foundation and clinical experience. Throughout the study, participants did not report any discomfort, and the results demonstrated a significant improvement in long-term delayed recall scores in the circuit-based PAS group compared with the baseline and the sham PAS group. This provides initial confirmation of the effectiveness of the paired stimulation protocol using the ‘Hebbian pattern’ for enhancing cognitive function in aMCI patients. Changes in FC between the left hippocampus and left precuneus appear to be a potential central mechanism for improving long-term delayed recall function.

    Presently, in most studies, the construction of functional networks is grounded in brain regions with statistically significant differences in functional indicators, such as amplitude of low frequency fluctuation (ALFF), regional homogeneity (ReHo) and FC, or through the utilisation of existing functional network templates.41 In our review of literature on aMCI-related functional networks, we observed widespread abnormalities in brain regions and network function, predominantly involving the default network, salience network and visual network.42–46 The construction of symptom-related functional networks based on existing research results displays significant heterogeneity. In our study, we amalgamated brain structure and function to construct the ‘hippocampus-cortex’ network, revealing circuits closely linked to the AVLT long-term delayed recall score. The prefrontal lobe and precuneus were chosen as stimulation targets. The results exhibited significant improvement in the AVLT long-term delayed recall scores in the circuit-based PAS group compared with baseline and the sham PAS group. This demonstrates the feasibility of utilising structural and FC to determine brain network targets and identify clinically relevant circuits for neuroregulatory treatment. Notably, executive function, attention and visuospatial ability exhibited no significant changes post-intervention, indirectly signifying the precision of this method. Zhao et al also found that structural damage and functional changes in aMCI are interconnected.47 Grey matter volume reductions were observed in several regions in aMCI patients, and ALFF values in these regions also exhibited variable changes. In the network model fitted by Zhu et al for predicting AVLT delayed recall, the prefrontal lobe and parietal lobe were key nodes.48 Cui et al found that spontaneous neural activity in the left prefrontal lobe was positively correlated with AVLT long-term delayed recall scores.49 The severity of cognitive impairment in aMCI is related to spontaneous activity in the cuneus gyrus/precuneus cortex.42 These studies support our research results. In our study, we noted a significant increase in BNT scores in the circuits-based PAS group after paired stimulation using the “Hebbian pattern”. Early symptoms of AD encompass progressive episodic memory impairment, followed by other cognitive deficits, including language.50 Aphasia may be related to AD progression, and early intervention is effective.51 Therefore, evaluating naming function is crucial in clinical practice.52 BNT is among the most commonly used naming function assessment scales. MCI patients with reduced BNT scores are at a higher risk of converting to AD, closely linked to episodic memory.53 The BNT scores in AD patients are significantly lower than those in aMCI patients and the normal population. While BNT scores in aMCI patients are not significantly lower than in the normal population, semantic errors significantly increase.54 In AD patients, the decrease in BNT scores positively correlates with hippocampus volume reduction and is closely related to the temporal lobe, thalamus and prefrontal cortex.55 56 These may underlie the improvement in BNT scores with stimulation of the ‘hippocampus-cortex’ circuit. We also extended our analysis to include a comprehensive evaluation of global network metrics that there were no statistically significant changes across these metrics (online supplemental material 8).

    Supplemental material

    In our study, we observed that stimulating the ‘cortical-hippocampal’ circuits led to decreased abnormal FC between the left hippocampus and the left precuneus cortex in aMCI patients compared with baseline. We also found that FC changes between the left hippocampus and the left precuneus cortex likely play a central role in improving long-term delayed recall function. Qin 46 et al demonstrated through linear regression analysis that atrophy of the left precuneus is a risk factor for memory impairment in aMCI patients.57 Chen et al applied rTMS to the anterior precuneus in patients with subjective cognitive decline and observed functional changes in the anterior precuneus-hippocampal subregion and improved episodic memory.58 Although their subjects and treatment methods differed from ours, the results were consistent.

    Limitations

    While this study has yielded promising results, there are limitations, including a small sample size, single-centre design and a single control group. In future studies, we plan to expand the sample size and extend the observation period to verify the reproducibility and sustainability of therapeutic efficacy. Additionally, we will consider increasing the number of pulses per session and treatment duration to evaluate the potential for enhanced efficacy. Furthermore, a comparison of efficacy between the two stimulation modes, using the currently clinically used single target point stimulation pattern as a control group, should be explored.

    Summary

    This study proposes the use of paired TMS to modulate the ‘cortical-hippocampus’ circuits for the treatment of aMCI. The method combines individualised FC based on white matter fibre tracking related to symptoms to determine stimulation targets. The originality of employing the ‘Hebbian pattern’ paired stimulation in precise neuromodulation treatment has been demonstrated and its efficacy has been confirmed.

    Data availability statement

    Data are available upon reasonable request.

    Ethics statements

    Patient consent for publication

    Ethics approval

    The study protocol was approved by the institutional ethics committee of Yueyang Hospital of Integrated Traditional Chinese and Western Medicine Hospital (#2021-103). Participants gave informed consent to participate in the study before taking part.

    Acknowledgments

    We thank all contributors and participants for their contribution to this study.

    References

    1. ↵
      2. Derby CA ,
      3. Katz MJ ,
      4. Rozner S , et al
      . A birth cohort analysis of amnestic mild cognitive impairment incidence in the einstein aging study (EAS). J Alzheimers Dis 2019;70:S271–81. doi:10.3233/JAD-181141
      OpenUrl
    2. ↵
      2. Harwood RH ,
      3. Goldberg SE ,
      4. Brand A , et al
      . Promoting activity, independence, and stability in early dementia and mild cognitive impairment (praised): randomised controlled trial. BMJ 2023;382:e074787. doi:10.1136/bmj-2023-074787
    3. ↵
      2. Mayor S
      . One in three with mild cognitive impairment has depression, review finds. BMJ 2016:i6387. doi:10.1136/bmj.i6387
    4. ↵
      2. Chen J ,
      3. Chen R ,
      4. Xue C , et al
      . Hippocampal-subregion mechanisms of repetitive transcranial magnetic stimulation causally associated with amelioration of episodic memory in amnestic mild cognitive impairment. JAD 2022;85:1329–42. doi:10.3233/JAD-210661
      OpenUrl
    5. ↵
      2. Yuan L-Q ,
      3. Zeng Q ,
      4. Wang D , et al
      . Neuroimaging mechanisms of high-frequency repetitive transcranial magnetic stimulation for treatment of amnestic mild cognitive impairment: a double-blind randomized sham-controlled trial. Neural Regen Res 2021;16:707–13. doi:10.4103/1673-5374.295345
      OpenUrl
    6. ↵
      2. Cui H ,
      3. Ren R ,
      4. Lin G , et al
      . Repetitive transcranial magnetic stimulation induced hypoconnectivity within the default mode network yields cognitive improvements in amnestic mild cognitive impairment: a randomized controlled study. J Alzheimers Dis 2019;69:1137–51. doi:10.3233/JAD-181296
      OpenUrl
    7. ↵
      2. Licht C ,
      3. Herbrandt S ,
      4. van Meegen C , et al
      . A systematic review and meta-analysis of cognitive effects of rTMS in caucasian patients with mild cognitive impairment. Brain Sci 2023;13:1335. doi:10.3390/brainsci13091335
    8. ↵
      2. Miyashita Y
      . Cognitive memory: cellular and network machineries and their top-down control. Science 2004;306:435–40. doi:10.1126/science.1101864
      OpenUrlAbstract/FREE Full Text
    9. ↵
      2. Eichenbaum H
      . Prefrontal-hippocampal interactions in episodic memory. Nat Rev Neurosci 2017;18:547–58. doi:10.1038/nrn.2017.74
      OpenUrlCrossRefPubMed
    10. ↵
      2. Yonelinas AP ,
      3. Ranganath C ,
      4. Ekstrom AD , et al
      . A contextual binding theory of episodic memory: systems consolidation reconsidered. Nat Rev Neurosci 2019;20:364–75. doi:10.1038/s41583-019-0150-4
      OpenUrlCrossRefPubMed
    11. ↵
      2. Yang Z ,
      3. Zhuang X ,
      4. Sreenivasan K , et al
      . A robust deep neural network for denoising task-based fMRI data: an application to working memory and episodic memory. Med Image Anal 2020;60:101622. doi:10.1016/j.media.2019.101622
      OpenUrl
    12. ↵
      2. He H ,
      3. Shang Y ,
      4. Yang X , et al
      . Constructing an associative memory system using spiking neural network. Front Neurosci 2019;13:650. doi:10.3389/fnins.2019.00650
    13. ↵
      2. Khadka S ,
      3. Chung JJ ,
      4. Tumer K
      . Neuroevolution of a modular memory-augmented neural network for deep memory problems. Evol Comput 2019;27:639–64. doi:10.1162/evco_a_00239
      OpenUrl
    14. ↵
      2. Li F ,
      3. Xiang W ,
      4. Wang J , et al
      . Quantum weighted long short-term memory neural network and its application in state degradation trend prediction of rotating machinery. Neural Netw 2018;106:237–48. doi:10.1016/j.neunet.2018.07.004
      OpenUrl
    15. ↵
      2. Fox MD
      . Mapping symptoms to brain networks with the human connectome. N Engl J Med 2018;379:2237–45. doi:10.1056/NEJMra1706158
      OpenUrlCrossRefPubMed
    16. ↵
      2. Leergaard TB ,
      3. Bjaalie JG
      . Atlas-based data integration for mapping the connections and architecture of the brain. Science 2022;378:488–92. doi:10.1126/science.abq2594
      OpenUrl
    17. ↵
      2. Axer M ,
      3. Amunts K
      . Scale matters: the nested human connectome. Science 2022;378:500–4. doi:10.1126/science.abq2599
      OpenUrlCrossRef
    18. ↵
      2. Lee JH ,
      3. Liu Q ,
      4. Dadgar-Kiani E
      . Solving brain circuit function and dysfunction with computational modeling and optogenetic fMRI. Science 2022;378:493–9. doi:10.1126/science.abq3868
      OpenUrlCrossRef
    19. ↵
      2. Thiebaut de Schotten M ,
      3. Forkel SJ
      . The emergent properties of the connected brain. Science 2022;378:505–10. doi:10.1126/science.abq2591
      OpenUrlCrossRef
    20. ↵
      2. Wolk DA ,
      3. Dickerson BC
      . Fractionating verbal episodic memory in alzheimer’s disease. NeuroImage 2011;54:1530–9. doi:10.1016/j.neuroimage.2010.09.005
      OpenUrlCrossRefPubMedWeb of Science
    21. ↵
      2. Shankle WR ,
      3. Romney AK ,
      4. Hara J , et al
      . Methods to improve the detection of mild cognitive impairment. Proc Natl Acad Sci U S A 2005;102:4919–24. doi:10.1073/pnas.0501157102
      OpenUrlAbstract/FREE Full Text
    22. ↵
      2. Ma J ,
      3. Zheng M-X ,
      4. Wu J-J , et al
      . Mapping the long-term delayed recall-based cortex-hippocampus network constrained by the structural and functional connectome: a case-control multimodal MRI study. Alzheimers Res Ther 2023;15:61. doi:10.1186/s13195-023-01197-7
    23. ↵
      2. Rowe JB ,
      3. Chan V ,
      4. Ingemanson ML , et al
      . Robotic assistance for training finger movement using a hebbian model: a randomized controlled trial. Neurorehabil Neural Repair 2017;31:769–80. doi:10.1177/1545968317721975
      OpenUrlCrossRef
    24. ↵
      2. Koch G ,
      3. Ponzo V ,
      4. Di Lorenzo F , et al
      . Hebbian and anti-hebbian spike-timing-dependent plasticity of human cortico-cortical connections. J Neurosci 2013;33:9725–33. doi:10.1523/JNEUROSCI.4988-12.2013
      OpenUrlAbstract/FREE Full Text
    25. ↵
      2. Bondi MW ,
      3. Edmonds EC ,
      4. Jak AJ , et al
      . Neuropsychological criteria for mild cognitive impairment improves diagnostic precision, biomarker associations, and progression rates. J Alzheimers Dis 2014;42:275–89. doi:10.3233/JAD-140276
      OpenUrl
    26. ↵
      2. Maier W ,
      3. Buller R ,
      4. Philipp M , et al
      . The hamilton anxiety scale: reliability, validity and sensitivity to change in anxiety and depressive disorders. J Affect Disord 1988;14:61–8. doi:10.1016/0165-0327(88)90072-9
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      2. Bech P ,
      3. Paykel E ,
      4. Sireling L , et al
      . Rating scales in general practice depression: psychometric analyses of the clinical interview for depression and the hamilton rating scale. J Affect Disord 2015;171:68–73. doi:10.1016/j.jad.2014.09.013
      OpenUrl
    28. ↵
      2. Marshall GA ,
      3. Zoller AS ,
      4. Lorius N , et al
      . Functional activities questionnaire items that best discriminate and predict progression from clinically normal to mild cognitive impairment. Curr Alzheimer Res 2015;12:493–502. doi:10.2174/156720501205150526115003
      OpenUrlCrossRefPubMed
    29. ↵
      2. Folstein MF ,
      3. Folstein SE ,
      4. McHugh PR
      . Mini-mental state". A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189–98. doi:10.1016/0022-3956(75)90026-6
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      2. Zhao Q ,
      3. Guo Q ,
      4. Liang X , et al
      . Auditory verbal learning test is superior to rey-osterrieth complex figure memory for predicting mild cognitive impairment to alzheimer’s disease. Curr Alzheimer Res 2015;12:520–6. doi:10.2174/1567205012666150530202729
      OpenUrlPubMed
    31. ↵
      2. Zhao Q ,
      3. Guo Q ,
      4. Li F , et al
      . The shape trail test: application of a new variant of the trail making test. PLoS ONE 2013;8:e57333. doi:10.1371/journal.pone.0057333
    32. ↵
      2. Wang B ,
      3. Guo Q ,
      4. Zhao Q , et al
      . Memory deficits for non-amnestic mild cognitive impairment. J Neuropsychol 2012;6:232–41. doi:10.1111/j.1748-6653.2011.02024.x
      OpenUrlPubMed
    33. ↵
      2. Mack WJ ,
      3. Freed DM ,
      4. Williams BW , et al
      . Boston naming test: shortened versions for use in alzheimer’s disease. J Gerontol 1992;47:P154–8. doi:10.1093/geronj/47.3.p154
      OpenUrlCrossRefPubMed
    34. ↵
      2. Youn YC ,
      3. Pyun J-M ,
      4. Ryu N , et al
      . Use of the clock drawing test and the rey-osterrieth complex figure test-copy with convolutional neural networks to predict cognitive impairment. Alzheimers Res Ther 2021;13:85. doi:10.1186/s13195-021-00821-8
    35. ↵
      2. Smith SM ,
      3. Jenkinson M ,
      4. Woolrich MW , et al
      . Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 2004;23 Suppl 1:S208–19. doi:10.1016/j.neuroimage.2004.07.051
      OpenUrlCrossRefPubMed
    36. ↵
      2. Chao-Gan Y ,
      3. Yu-Feng Z
      . “DPARSF: a MATLAB toolbox for "pipeline" data analysis of resting-state fMRI”. Front Syst Neurosci 2010;4:13. doi:10.3389/fnsys.2010.00013
    37. ↵
      2. Pitcher JB ,
      3. Doeltgen SH ,
      4. Goldsworthy MR , et al
      . A comparison of two methods for estimating 50% of the maximal motor evoked potential. Clin Neurophysiol 2015;126:2337–41. doi:10.1016/j.clinph.2015.02.011
      OpenUrlCrossRefPubMed
    38. ↵
      2. Tipton E ,
      3. Hallberg K ,
      4. Hedges LV , et al
      . Implications of small samples for generalization: adjustments and rules of thumb. Eval Rev 2017;41:472–505. doi:10.1177/0193841X16655665
      OpenUrlCrossRef
    39. ↵
      SONG X W, DONG Z Y, LONG X Y, et al. REST: a Toolkit for resting-state functional magnetic resonance imaging data Processing[J]. PLoS One 2011;6:e25031.
    40. ↵
      2. Stefan K ,
      3. Kunesch E ,
      4. Cohen LG , et al
      . Induction of plasticity in the human motor cortex by paired associative stimulation. Brain 2000;123 Pt 3:572–84. doi:10.1093/brain/123.3.572
      OpenUrl
    41. ↵
      2. Yeo BTT ,
      3. Krienen FM ,
      4. Sepulcre J , et al
      . The organization of the human cerebral cortex estimated by intrinsic functional connectivity. J Neurophysiol 2011;106:1125–65. doi:10.1152/jn.00338.2011
      OpenUrlCrossRefPubMedWeb of Science
    42. ↵
      2. Pan P ,
      3. Zhu L ,
      4. Yu T , et al
      . Aberrant spontaneous low-frequency brain activity in amnestic mild cognitive impairment: a meta-analysis of resting-state fMRI studies. Ageing Res Rev 2017;35:12–21. doi:10.1016/j.arr.2016.12.001
      OpenUrlPubMed
    43. ↵
      2. Zhao Z ,
      3. Lu J ,
      4. Jia X , et al
      . Selective changes of resting-state brain oscillations in aMCI: an fMRI study using ALFF. Biomed Res Int 2014;2014:920902. doi:10.1155/2014/920902
    44. ↵
      2. Lin H ,
      3. Li M ,
      4. Zhan Y , et al
      . Disrupted white matter functional connectivity in aMCI APOEε4 carriers: a resting-state study. Brain Imaging Behav 2021;15:1739–47. doi:10.1007/s11682-020-00367-7
      OpenUrl
    45. ↵
      2. Zhu L ,
      3. Shu H ,
      4. Liu D , et al
      . Apolipoprotein E Ε4 specifically modulates the hippocampus functional connectivity network in patients with amnestic mild cognitive impairment. Front Aging Neurosci 2018;10:289. doi:10.3389/fnagi.2018.00289
    46. ↵
      2. Zhou Q-H ,
      3. Wang K ,
      4. Zhang X-M , et al
      . Differential regional brain spontaneous activity in subgroups of mild cognitive impairment. Front Hum Neurosci 2020;14:2. doi:10.3389/fnhum.2020.00002
    47. ↵
      2. Zhao Z-L ,
      3. Fan F-M ,
      4. Lu J , et al
      . Changes of gray matter volume and amplitude of low-frequency oscillations in amnestic MCI: an integrative multi-modal MRI study. Acta Radiol 2015;56:614–21. doi:10.1177/0284185114533329
      OpenUrlCrossRefPubMed
    48. ↵
      2. Zhu Y ,
      3. Zang F ,
      4. Wang Q , et al
      . Connectome-based model predicts episodic memory performance in individuals with subjective cognitive decline and amnestic mild cognitive impairment. Behav Brain Res 2021;411:113387. doi:10.1016/j.bbr.2021.113387
      OpenUrl
    49. ↵
      2. Cui L ,
      3. Zhang Z ,
      4. Zac Lo C-Y , et al
      . Local functional MR change pattern and its association with cognitive function in objectively-defined subtle cognitive decline. Front Aging Neurosci 2021;13:684918. doi:10.3389/fnagi.2021.684918
    50. ↵
      2. Silagi ML ,
      3. Bertolucci PHF ,
      4. Ortiz KZ
      . Naming ability in patients with mild to moderate alzheimer’s disease: what changes occur with the evolution of the disease Clinics (Sao Paulo) 2015;70:423–8. doi:10.6061/clinics/2015(06)07
      OpenUrlCrossRefPubMed
    51. ↵
      2. Noonan KA ,
      3. Pryer LR ,
      4. Jones RW , et al
      . A direct comparison of errorless and errorful therapy for object name relearning in alzheimer’s disease. Neuropsychol Rehabil 2012;22:215–34. doi:10.1080/09602011.2012.655002
      OpenUrlPubMed
    52. ↵
      2. Aniwattanapong D ,
      3. Tangwongchai S ,
      4. Supasitthumrong T , et al
      . Validation of the Thai version of the short Boston naming test (T-BNT) in patients with alzheimer’s dementia and mild cognitive impairment: clinical and biomarker correlates. Aging Ment Health 2019;23:840–50. doi:10.1080/13607863.2018.1501668
      OpenUrlCrossRefPubMed
    53. ↵
      2. Bozoki A ,
      3. Giordani B ,
      4. Heidebrink JL , et al
      . Mild cognitive impairments predict dementia in nondemented elderly patients with memory loss. Arch Neurol 2001;58:411–6. doi:10.1001/archneur.58.3.411
      OpenUrlCrossRefPubMedWeb of Science
    54. ↵
      2. Willers IF ,
      3. Feldman ML ,
      4. Allegri RF
      . Subclinical naming errors in mild cognitive impairment: a semantic deficit Dement Neuropsychol 2008;2:217–22. doi:10.1590/S1980-57642009DN20300010
      OpenUrl
    55. ↵
      2. Peng G-P ,
      3. Feng Z ,
      4. He F-P , et al
      . Correlation of hippocampal volume and cognitive performances in patients with either mild cognitive impairment or alzheimer’s disease. CNS Neurosci Ther 2015;21:15–22. doi:10.1111/cns.12317
      OpenUrl
    56. ↵
      2. Balthazar MLF ,
      3. Yasuda CL ,
      4. Lopes TM , et al
      . Neural correlates of lexical-semantic memory: a voxel-based morphometry study in mild AD, aMCI and normal aging. Dement Neuropsychol 2011;5:69–77. doi:10.1590/S1980-57642011DN05020003
      OpenUrl
    57. ↵
      2. Qin R ,
      3. Li M ,
      4. Luo R , et al
      . The efficacy of gray matter atrophy and cognitive assessment in differentiation of aMCI and naMCI. Appl Neuropsychol Adult 2022;29:83–9. doi:10.1080/23279095.2019.1710509
      OpenUrl
    58. ↵
      2. Chen J ,
      3. Ma N ,
      4. Hu G , et al
      . rTMS modulates precuneus-hippocampal subregion circuit in patients with subjective cognitive decline. Aging (Albany NY) 2020;13:1314–31. doi:10.18632/aging.202313
      OpenUrl

    Supplementary materials

    Footnotes

    • JM and J-JW contributed equally.

    • Contributors Design and conceptualisation of the study: J-GX, X-YH and M-XZ. Acquisition and analysis of the data: X-XX, XX and Y-TX. Recruitment of the sample: X-MZ and J-HL. TMS Plan execution: J-JL and J-PZ. Interpretation of the data: X-YH and M-XZ. Drafting and revising the manuscript: JM and J-JW. Supervision: J-GX. J-GX is the guarantor for the overall content.

    • Funding This work was supported by the National Key R&D Program of China [grant numbers 2018YFC2001600, 2018YFC2001604].

    • Competing interests None declared.

    • Provenance and peer review Not commissioned; externally peer-reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.