Brain Activation During Working Memory Task in Amnestic Mild Cognitive Impairment Patients and Its Association with Memory and Attention

Abstract

Background:

Amnestic mild cognitive impairment (aMCI) is regarded as a transitional state of Alzheimer’s disease, with working memory (WM) impairment.

Objective:

To investigate the brain activity in aMCI patients during WM tasks with the functional near-infrared spectroscopy (fNIRS) technique, as well as explore the association between brain activity and cognitive function in multiple domains.

Methods:

This study is a case-control study of 54 aMCI patients and 33 cognitively healthy elderly (NC). All participants underwent neuropsychological assessments. fNIRS was applied to examine the brain activation during the WM task. Multivariable linear regression analysis was applied to evaluate associations between brain activation and cognitive function in multiple domains.

Results:

Compared to NC subjects, aMCI patients had lower activation in the bilateral prefrontal, parietal, and occipital cortex during the WM task. Additionally, activation in the left prefrontal, bilateral parietal, and occipital cortex during the encoding and maintenance phase was positively associated with memory function. During memory retrieval, higher activity in the left prefrontal, parietal, and occipital cortex were correlated with higher memory scores. Besides, a positive association also formed between attention function and the activation in the left prefrontal, parietal, and occipital cortex during the WM task.

Conclusion:

These findings demonstrated that reduced activation in the prefrontal, parietal and occipital cortex during WM might reflect the risk of cognitive impairment, especially memory and attention function in aMCI patients. Given the brain activation visualization, fNIRS may be a convenient and alternative tool for screening the risk of Alzheimer’s disease.

Keywords

Amnestic mild cognitive impairment brain activation cognition functional near-infrared spectroscopy working memory

INTRODUCTION

Alzheimer’s disease (AD) is a prevalent neurodegenerative disorder manifested as cognitive impairment and daily living dysfunction in the elderly. It is becoming an enormous threat to global health owing to the prevalent morbidity and inadequacies of effective treatment and intervention [1]. However, numerous studies have shown that prevention, diagnosis, and intervention treatment in the preclinical stage of AD can reduce the incidence and delay disease development. Mild cognitive impairment (MCI) is considered to be a transitional state between healthy aging and dementia. Amnestic MCI (aMCI), a subtype of MCI, commonly characterized by a memory impairment that exceeds normal aging but does not meet the clinical criteria for AD, is widely regarded as a transitional state of AD [2, 3]. Thus, recognizing aMCI would contribute to the early prevention and intervention of AD.

Working memory (WM), the fundamental function of maintaining information in memory, is a vital component of cognition [4, 5]. The impairment of WM is one of the manifestations of MCI [6]. Increasing functional magnetic resonance imaging (fMRI) studies demonstrated the abnormal brain activity patterns of MCI patients during WM tasks. Compared to healthy elderly, MCI patients showed reduced frontal cortex activation during the WM task’s encoding phase [7]. Similarly, Saykin et al. found that patients with MCI had reduced frontal and parietal cortex activity during the n-back WM task [8]. Lou et al. found that aMCI patients showed significantly decreased activity in the frontal and visual cortices during the encoding and retrieval phase of the WM task [9]. Additionally, recent electroencephalography (EEG) studies demonstrated that MCI patients showed a lower level of theta-gamma coupling during WM task [10] and with lower brain activation, especially the left parietal cortex [11]. These findings suggested that alteration in brain activation during WM may be related to the neurodegenerative processes in MCI patients.

The clinical examination methods applied to detect brain function are mainly fMRI. However, fMRI is costly and technically complex for routine clinical application. Functional near-infrared spectroscopy (fNIRS), an emerging neuroimaging technique, is non-invasive and cost-effective, with few constraints to participants during the assessment. fNIRS can monitor brain activity by measuring the changes in oxyhemoglobin (HbO) and deoxyhemoglobin (HbR) concentrations in the cerebral cortex, based on the neurovascular coupling principle [12].

fNIRS has been applied to explore brain activation during cognitive engagement in recent years [13 –16]. Previous fNIRS studies found that compared to cognitively normal elderly, participants with MCI showed distinct brain activity in the frontal cortex during WM tasks [13, 17]. To date, numerous fNIRS studies focus on the frontal cortex activity in MCI patients during cognitive tasks. However, apart from the prefrontal cortex, the parietal and occipital lobes are also involved in cognitive processes, particularly WM [18 –20]. An fMRI study with 22 healthy participants found that parietal and occipital cortices were associated with memory encoding and retrieval [21]. Even so, few studies have explored prefrontal, parietal, and occipital cortex activity in participants with aMCI during WM tasks. As a subtype of MCI, aMCI may be more predictive of AD development. The lesions of aMCI are not a single brain region problem but involve abnormal brain networks or neural circuits in several key brain regions.

Furthermore, the capabilities of WM are not only concerned with information process and extraction ability but also the attention and executive function [4, 5]. To our knowledge, the association between brain region activation during WM tasks and the impairment of various cognitive domains has not been investigated. Hence, the present study aims to primarily investigate the prefrontal, parietal, and occipital cortex activity in aMCI patients during WM tasks with the fNIRS technique, and secondly explore the association between brain activity and cognitive function in multiple domains.

METHODS

Participants

This study recruited 54 aMCI patients (mean age 66.78±4.51) from the Cognitive Disorders Clinics in the First People’s Hospital of Foshan between September 2020 and December 2021. Thirty-three cognitively normal control (NC) subjects (mean age 68.03±3.96) were recruited from the community. The First People’s Hospital of Foshan Research ethics committee has approved the research proposal. The written informed consent to the present study of each participant has been obtained.

All the participants with normal visual acuity or corrected acuity. The inclusion criteria for NC participants: 1) cognitively normal; 2) The Montreal Cognitive Assessment (MoCA) score > 14 for illiterate,>20 for individuals with 1–6 years of education, and > 25 for individuals with seven or more years of education [22]; 3) Clinical Dementia Rating (CDR) score = 0.

The diagnostic criteria for aMCI [2]: 1) have a subjective cognitive complaint; 2) clinically confirmed daily living ability preserved; 3) memory test confirmed objective memory impairment, with a cutoff of 1.5 standard deviations below age, gender, and education matched-specific norms; 4) one or more domain cognitive decline; 5) global CDR score = 0.5; 6) absence of dementia according to DSM-V.

The exclusion criteria: 1) neurologic disorder (cerebral hemorrhage or infarction, epilepsy, brain surgery, Parkinson’s disease); 2) severe cardiopulmonary disease, liver or kidney failure; 3) autoimmune or peripheral vascular disease; 4) cancer histories; 5) severe emotional disorder; 6) Severe visual impairment or ocular disease.

Medical and cognitive assessment

Data on participants’ medical history, physical examination, and demographic characteristics were collected. All participants completed a battery neuropsychological test. Global cognition was assessed by Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA). Memory was evaluated by the Auditory Verbal Learning Test-Huashan version (AVLT-H). Attention was measured by Symbol Digit Modalities Test (SDMT). Besides, Stroop’s Color Word Test was used to evaluate the executive function, Boston Naming Test (BNT) was used to determine the language function, and the Clock Drawing Test was used to assess visuospatial function. Moreover, the Hamilton rating scale for depression (HAMD) score was collected to adjust the effects of subsyndromal depression.

Experiment procedures

fNIRS data were acquired while participants performed a delayed matching to sample (DMTS) task [13] in a quiet room. In the beginning, each participant was required to rest for about 300 s and then complete the DMTS task. The task includes ten trials, and each trial consisted of four phases, including rest, encoding, maintenance, and retrieval (Fig. 1A). Firstly, they were asked to stare at the “+” on the screen for a 15-s rest. In the encoding phase, participants were required to remember the set of graphics on the screen for 12 s. Then keep the memory and stare at the “+” on the screen for 12 s during the maintenance phase. Finally, participants need to identify the image presented in the encoding phase among four groups of images in 10 s in the retrieval phase.

Fig. 1

Experimental paradigm and setup. A) DMTS task-based experiment protocol. B) Channels distribution on brain cortex.

fNIRS data acquisition and processing

The fNIRS system (NirSmart-6000A; Danyang Huichuang Medical Equipment Co., Ltd., China) with wavelengths of 730 nm and 850 nm, was used to measure the concentration changes in HbO and HbR at a sampling rate of 11 Hz. Thirty-nine channels are composed of 23 sources and 13 detectors, with a fixed source-detector distance of 3 cm. The spatial locations of sources, detectors, and anchor points (located at Nz, Cz, Al, Ar, Iz referring to the standard international 10–20 system of electrode placement) were measured by an electromagnetic 3D digitizer device (Patriot, Polhemus, USA) on subject. The acquired coordinates were then transformed into MNI coordinates and further projected to the MNI standard brain template using spatial registration approach in NirSpace (Danyang Huichuang Medical Equipment Co., Ltd., China) (Fig. 1B).

The NirSpark software (HuiChuang, China) package was used to preprocess fNIRS signals [23]. Motion artifacts were manifested as the impulse or cliff-type jumps caused by the relative sliding of the scalp and probes [24]. Any signal change beyond six standard deviations of the entire time series was considered a motion artifact. Spline interpolation is a commonly used correction method. The advantage of spline interpolation is that it only corrects the pre-localized artifacts. Therefore, a spline interpolation algorithm was used to the resulting signals to amend motion artifacts by channels. And then, during the preprocessing, the raw data were band-pass filtered between 0.01 and 0.1 Hz to remove physiological noise (e.g., respiration, cardiac activity, and low-frequency signal drift). Then the modified Beer–Lambert law was used to calculate the relative hemoglobin concentration changes in HbO and HbR. The parameter differential path factor was set at 6 for wavelength one (730 nm) and 6 for wavelength two (850 nm).

Compared to HbR signals, HbO signals can provide a better signal-to-noise ratio and reflect regional cerebral blood oxygenation changes [25, 26]. Thus, we only explored the HbO variations during the WM task. The individual concentration of HbO in each channel was obtained by averaging the ten trials separately. The mean of trials was performed with baseline correction, the baseline set to two seconds before the task. The initial time of the hemodynamic response function was set to –2 s (i.e., baseline state), the HbO concentration of encoding is obtained from the average concentration of 0 to 12 s in the encoding phase, while the HbO concentration of retrieval is the average of 0 to 10 s in the retrieval phase.

Statistical analysis

The characteristics were compared between groups using the Student’s t-test for continuous variables and χ²-test or Fisher exact test for the categorical variables. The False Discovery Rate had been used to perform the correction of the comparisons. The association between brain activity and cognitive function was evaluated by multivariable linear regression analysis adjusted for age, gender, education levels, HAMD score, and histories of hypertension and diabetes. EmpowerStats (https://www.empowerstats.com, X&Y solutions, Inc., Boston, MA) and statistical software packages R (https://www.R-project.org, The R Foundation) were used for data analyses. All the p-value<0.05 were considered statistically significant.

RESULTS

The characteristics of the study population

Compared to NC elderly, participants with aMCI had lower scores of MMSE (24.54±2.58 versus 26.73±1.38, p < 0.001), MoCA (19.43±3.32 versus24.45±2.08, p < 0.001), AVLT immediate recall (3.75±1.75 versus 5.76±1.62, p < 0.001), AVLT 5-min delayed recall (4.28±2.39 versus 7.33±1.57, p < 0.001), AVLT 20-min delayed recall (2.77±2.43 versus 6.58±1.94, p < 0.001), AVLT recognition (3.06±2.58 versus 6.58±1.85, p < 0.001), SDMT (31.82±11.84 versus 39.82±6.60, p < 0.001), and BNT (19.89±3.93 versus 23.58±3.00, p < 0.001). Besides, the time of Stroop test in aMCI patients was longer than that in NC elderly (20.08±7.13 versus 16.03±6.44, p = 0.010). The clock drawing test (CDT) score, age, gender, education levels, and histories of hypertension and diabetes did not differ between NC and aMCI groups (p > 0.05) (Table 1).

Table 1

Characteristics of the study population

	NC (n = 33)	aMCI (n = 54)	p
Age (y)	68.03±3.96	66.78±4.51	0.192
Male, n (%)	14 (42.42)	27 (50.00)	0.492
HBP, n (%)	15 (45.45)	17 (31.48)	0.190
DM, n (%)	5 (15.15)	3 (5.56)	0.133
Education (y)	12.94±2.71	12.57±2.84	0.555
HAMD score	5.91±3.12	6.54±3.54	0.404
Global cognition
MMSE	26.73±1.38	24.54±2.58	<0.001
MOCA	24.45±2.08	19.43±3.32	<0.001
Memory
AVLT_immediate	5.76±1.62	3.75±1.75	<0.001
AVLT_5 min	7.33±1.57	4.28±2.39	<0.001
AVLT_20 min	6.58±1.94	2.77±2.43	<0.001
AVLT_recognition	6.58±1.85	3.06±2.58	<0.001
Attention
SDMT	39.82±6.60	31.82±11.84	<0.001
Executive function
Stroop test time	16.03±6.44	20.08±7.13	0.010
Language
BNT	23.58±3.00	19.89±3.93	<0.001
Visuospatial function
CDT	9.00±1.48	8.17±2.15	0.055

Data expressed as mean±standard deviation or percentage (%). HBP, hypertension; DM, diabetes; HAMD, Hamilton rating scale for depression; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; AVLT, auditory verbal learning test; SDMT, symbol digit modalities test; BNT, Boston naming test; CDT, clock drawing test.

The characteristics of brain HbO concentration changes in NC and aMCI subjects during WM task

In the case of all channels average, HbO concentration increased in the encoding phase, decreased slightly during the maintenance stage, and subsequently rose during the retrieval phase in both NC and aMCI elderly (Fig. 2G). The magnitude of HbO concentration changes in aMCI patients during the WM task was less compared to those in the NC subjects. Additionally, we further investigated the HbO concentration changes in brain regions (Fig. 2A–F). The difference lies in the magnitude of HbO concentration changes in each brain region at different stages of the WM task. The concentrations of HbO in the prefrontal cortex peaked during the retrieval phase (Fig. 2A, D). However, the peak of HbO concentration in the parietal and occipital cortex was in the encoding phase (Fig. 2B, C, E, F). Moreover, aMCI patients displayed lower HbO concentration relative to the NC elderly during the WM task. The brain activation reduction was particularly pronounced during the encoding stage. (Fig. 2A–G, Supplementary Table 1).

Fig. 2

The changes of HbO concentration during DMTS task. HbO changes averaged over left prefrontal (A), right prefrontal (B), left parietal (C), right parietal (D), left occipital (E), and right occipital cortex (F) during DMTS task. G) HbO changes averaged over all channels during DMTS task.

The HbO concentration in NC and aMCI groups during the encoding and retrieval phases also overlayed on brain surface for a better visual presentation (Fig. 3). Compared to NC elderly, participants with aMCI had a lower concentration of HbO in bilateral prefrontal (channel 9/16/19), parietal (channel 1/12/30/33/34/35/36/37/38), occipital (channel 24/26/27/29) cortex during encoding graphics (Fig. 3A, Supplementary Table 2). During the maintenance stage, aMCI patients with reduced HbO concentration in bilateral prefrontal (channel 16/17), right parietal (channel 34/37/38), and left occipital (channel 29) cortex (Fig 3B, Supplementary Table 3). During the retrieval phase, the HbO concentration in the bilateral prefrontal (channel 4/8/9/15/16/17/21), parietal (channel 12/23/33/34/35/36/37/38), occipital (channel 24/32) cortex showed a significant decrease in aMCI patients compared to those in the NC group (Fig. 3C, Supplementary Table 4).

Fig. 3

Group mean HbO hemodynamic responses overlayed on the brain surface for graphics encoding (A), maintenance (B), and retrieval (C).

The association between brain HbO concentration and cognitive domains

A multivariable linear regression analysis was conducted for all subjects to evaluate the association between HbO concentration and cognitive domain (Fig. 4). During the encoding process, the concentration of HbO in the left prefrontal cortex (LPFC, channel 9), right occipital cortex (ROC, channel 25) were positively associated with AVLT_immediate score after adjusting for age, gender, education level, histories of hypertension and diabetes, and the score of HAMD (LPFC: β= 11.48, 95% CI 1.20–21.76; ROC: β= 12.54, 95% CI 1.00–24.07). Higher concentrations of HbO in the LPFC (channel 9), parietal cortex (PC, channel 12/37) and ROC (channel 25/26) were associated with higher AVLT-20 min score (LPFC: β= 22.66, 95% CI 7.38–37.93; LPC: β= 20.79, 95% CI 3.39–38.20; RPC: β= 18.27, 95% CI 0.62–35.93; ROC: β= 22.49, 95% CI 4.67–40.30). The higher HbO concentration in ROC (channel 25) also associated with a higher AVLT_recognition score (β= 19.16, 95% CI 1.99–36.33). Besides, the concentration of HbO in LPFC (channel 19) and LPC (channel 12) were positively correlated to the SDMT score (LPFC: β= 69.37, 95% CI 8.15–130.59; LPC: β= 83.52, 95% CI 19.98–147.07) (Fig. 4A).

Fig. 4

A) The association between brain HbO concentration during encoding and retrieval phase and cognitive domains. B) The association between brain HbO concentration during maintenance phase and cognitive domains. CI, confidence interval; LPFC, left prefrontal cortex; LPC, left parietal cortex; RPC, right parietal cortex; LOC, left occipital cortex; ROC, right occipital cortex; AVLT, auditory verbal learning test; SDMT, symbol digit modalities test.

During the maintenance phase, higher HbO concentration in LPFC (channel 9), parietal cortex (channel 12/34/35), and occipital cortex (channel 25/29) were correlated with higher AVLT_immediate score (LPFC: β= 14.49, 95% CI 3.02–25.96; LPC: β= 12.47, 95% CI 0.39–24.56; RPC: β= 17.35, 95% CI 3.45–31.26; LOC: β= 10.36, 95% CI 0.79–19.94; ROC: β= 13.83, 95% CI 1.85–25.81). The increased HbO concentration in LPFC (channel 9), LOC (channel 28), and ROC (channel 25/26) were associated with higher AVLT-20 min score (LPFC: β= 22.82, 95% CI 5.46–40.19; LOC: β= 19.74, 95% CI 0.38–39.10; ROC: β= 22.16, 95% CI 4.78–39.54). Higher HbO concentration in prefrontal (channel 9/15) and occipital cortex (channel 25/26/32) were correlated with higher AVLT recognition score (LPFC: β= 21.84, 95% CI 4.86–38.82; RPFC: β= 16.49, 95% CI 0.54–32.43; LOC: β= 14.05, 95% CI 0.69–27.40; ROC: β= 23.43, 95% CI 6.53–40.32). In addition, the HbO concentration in LPC (channel 35), RPC (channel 12) and ROC (channel 28) were positively associated with SDMT score (LPC: β= 72.46, 95% CI 1.70–143.21; RPC: β= 92.99, 95% CI 27.36–158.61; ROC: β= 85.73, 95% CI 14.75–156.72) (Fig. 4B).

During the retrieval phase, the increase of HbO concentration in LPFC (channel 8/9/21/22), LPC (channel 12/13/36), and LOC (channel 32) were correlated with higher AVLT_immediate score (LPFC: β= 29.67, 95% CI 11.01–48.33; LPC: β= 22.16, 95% CI 4.25–40.07; LOC: β= 14.15, 95% CI 0.78–27.53). The HbO concentration in LPFC (channel 9) and LOC (channel 32) also positively associated with AVLT-20 min score (LPFC: β= 29.67, 95% CI 11.01–48.33; LOC: β= 14.15, 95% CI 0.78–27.53). Higher HbO concentration in LPFC (channel 8/9/21), LPC (channel 1), LOC (channel 32) and ROC (channel 24) were correlated with higher AVLT_recognition score (LPFC: β= 33.22, 95% CI 7.68–58.77; LPC: β= 27.61, 95% CI 2.22–52.99; LOC: β= 23.74, 95% CI 3.95–43.54; ROC: β= 20.42, 95% CI 0.95–39.89). Additionally, positive correlation also could be found between HbO concentration in LOC (channel 32) and SDMT score (β= 101.40, 95% CI 18.07–184.72) (Fig. 4A). However, brain HbO concentration in the WM task was not significantly associated with Stroop test time, BNT, or CDT scores (data not shown).

DISCUSSION

The present study investigates the characteristic of brain activity in aMCI patients during WM tasks with the fNIRS technique and explores the association between brain activation during the WM tasks and multiple cognitive domains. Compared to the NC elderly, participants with aMCI have lower activation in the bilateral prefrontal, parietal, and occipital cortex during the WM task. Additionally, activation in the LPFC, LPC, RPC, LOC, and ROC during the encoding and maintenance phase are positively associated with memory function. Higher activity in LPFC, LPC, and LOC during memory retrieval are correlated with higher memory scores. Besides, a positive association also formed between attention function and the activation in LPFC, LPC, and LOC during the WM task.

WM is suggested to be a cognitive system to manipulate information storage and process, which is processed through the encoding, maintenance, and retrieval phase [27]. In recent studies, neurons in the prefrontal, parietal, and occipital have shown sustained activity during WM tasks [28 –31]. The feature-specific information of visual working memory can be represented in the prefrontal, parietal, and occipital cortex [32, 33]. When performing the WM task, robust spatial tuning topographic maps were shown in the prefrontal, parietal and occipital lobes [34]. In the present study, brain activity increased in the encoding phase, decreased slightly during maintenance, and rose during the retrieval phase in NC and aMCI subjects. Prefrontal neuron activity peaked during the retrieval phase, while the parietal and occipital cortex activation peaked in the encoding phase. In agreement with the present study, a previous fMRI study of MCI patients demonstrated that activation of the prefrontal cortex during the retrieval process was more significant than those in the encoding phase of the WM task [35]. Notably, significantly increasing LPFC activity was observed under the memory retrieval condition [36]. Since retrieval requires the activation of more reserve resources than encoding, both the degree and extent of prefrontal activation are increased compared with activation during encoding. Distinct from prefrontal neurons, the parietal and occipital neurons focus more on information encoding during the visual WM task. During encoding the information of targets, the parietal and occipital cortex showed significant sustained multiunit activity [18 , 37]. Stimulation of the parietal lobe is associated with spatial attention. Parietal neuron is strongly activated during the encoding phase of spatial or verbal WM tasks [38]. An fMRI study found that instead of memory retrieval, the parietal lobe focuses more on the encoding of graphic memory [39]. Berryhill et al. found that damage to the parietal cortex did not eliminate the ability to recognize WM judgments [40]. Although there is no evidence for a separate function of the occipital lobe in working memory encoding and retrieval, there is considerable evidence that the occipital lobe is involved in the encoding of picture information in working memory tasks [19 , 42]. These results confirmed the role of the prefrontal, parietal and occipital cortex in WM encoding and retrieval. Further illustrated that the parietal and occipital cortex might be associated preferentially involved in working memory encoding, while the prefrontal cortex, especially LPFC activation, was biased towards memory retrieval. Nevertheless, the specific mechanism needs to be further explored.

Previous fMRI studies indicated that compared to the NC elderly, participants with aMCI showed decreased activation in bilateral prefrontal and temporoparietal cortex during WM task [9, 43]. Consistent with these findings, the present study showed that compared to NC elderly, aMCI patients had lower activation in bilateral prefrontal, parietal and occipital cortex during WM task. Notably, the brain activation reduction was particularly pronounced during the encoding stage. The ability to encode information is an important component of WM. Previous studies have confirmed that impaired information encoding is a characteristic manifestation of MCI and early AD [44, 45]. These results demonstrated that aMCI patients might be fail to recruit sufficient brain resources for the WM task. The inefficient recruitment of the prefrontal, parietal and occipital cortex hinders the encoding of visual material and results in memory maintenance and retrieval impairment in the aMCI group. Nevertheless, cognitive impairment brain activation changes varied across studies. Some fNIRS and fMRI studies found that brain activation in the prefrontal and parietal cortex of the MCI patients was higher than that of the NC subjects during the WM task [17, 35]. The conflict might be due to the compensatory mechanism. As a compensatory mechanism, increased activation in the prefrontal cortex during memory encoding was found at the early stage of MCI; however, this compensatory mechanism was not found in MCI patients with relatively severe symptoms [46].

Furthermore, the present study analyzed the association between brain activation and neuropsychological scores. AVLT is an effective test for evaluating memory impairment and can reflect immediate, delay (short-term and long-term), and recognition memory function [47]. The spike rate, associated with short-term and long-term memory, shows a varying and persistent change in prefrontal neurons [48 –50]. The present study found that the left hemisphere, especially LPFC activity, is positively associated with AVLT immediate, 20-min delay recall, and recognition scores. In agreement with our findings, a resting fMRI study showed that the functional activity of LPFC in aMCI patients was positively correlated with AVLT long delay recall score [51]. Likewise, fNIRS research demonstrated a significant association between higher AVLT recognition scores and increased left dorsolateral prefrontal cortex activity during n-back tasks in subjective cognitive decline people [52]. These results confirmed the contribution of the LPFC in immediate, delay, and recognition memory.

Attention is considered a cornerstone of WM processes, creating, and maintaining the representation accessible for instantaneous cognition. The information encoding in WM is the outcome of attention processes interaction [53]. During WM and visual attention tasks, manipulation of attention to visual characteristics can modulate characteristic-selective signals in the frontal and parietal cortex and the early visual cortex [54 –56]. Evidence revealed that the prefrontal and parietal cortex controlled spatial attention [57, 58]. In the present study, improved prefrontal, parietal, and occipital cortex activation during memory encoding is significantly associated with a higher SDMT score. Given that SDMT is a reliable and sensitive neuropsychological test in assessing attention and information processing speed [59], the activation of prefrontal, parietal, and occipital cortices may be a vital part of executing representation attention in WM task.

The present study provided evidence of the impairment of cognitive abilities and functional deficits in the prefrontal, parietal, and occipital cortex in aMCI patients, and suggested fNIRS may be a reliable and valuable clinical diagnostic tool for aMCI. Compared to PIB-PET brain amyloid imaging or fMRI, fNIRS is cost-effective, portable, has no ionizing radiation, and lowers movement artifacts susceptibility, allowing for evaluating brain activity in large cohorts. Additionally, although some previous studies had tested the correlation between brain activity and clinical cognitive scores, most of them used Pearson’s correlation analysis [16 , 52], which could not exclude potential confounders. The present study characterized the influence of sub-regions cortical activity on multiple cognitive domain functions after adjusting potential confounders by using multiple linear regression analysis. Unlike previous fNIRS studies on MCI subjects, our study focused on aMCI patients. Given that aMCI is a transitional state of AD, our findings might contribute to further explore the clinical value of fNIRS-derived neural biomarkers in the early diagnosis of AD.

Several limitations of the present study need to be mentioned. Firstly, we did not measure the response time, from the start time of the retrieval phase to the answer time, which could help to demonstrate the difference in WM between aMCI and NC elderly. Hence, the memory retrieval response time must be recorded in successive research. Secondly, the area measured by the fNIRS was restricted to the prefrontal, parietal and occipital cortex, due to the limited number of channels. Whereas WM might also depend on the activation of target characteristics in the temporal cortex [60, 61], the activity of the temporal cortex must be further explored in future studies.

Conclusion

This fNIRS study indicated that compared to NC elderly, aMCI patients have a lower activation in bilateral prefrontal, parietal, and occipital cortex during WM task. Moreover, the sub-regions cortical activation during memory encoding and retrieval are positively associated with memory and attention function. These findings demonstrated that reduced prefrontal, parietal, and occipital cortex activation during WM might reflect the risk of cognitive impairment in aMCI patients. Besides, fNIRS might be an alternative and reliable tool for screening for early AD diagnosis.

Footnotes

ACKNOWLEDGMENTS

This work was supported by the National Key R&D Program of China (Grant No. 2018YFC2001700), the Medical scientific research project of Foshan Municipal Health Bureau (Grant No. 20220319), Luoyang Program for Development of Science and Technology (Grant No. 2001028A), and Joint programs for science and technology development of Henan province (Grant No. LHGJ20200575).

We would like to thank Ms. Biqing Lin, Ms. Weiping Zhang and Ms. Qiulan Huang for the screening tests in our clinic population and data collection. We thank Mr. Hang Liang for the fNIRS technical assistance.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Jia

, Quan

, Fu

, Zhao

, Li

, Wei

, Tang

, Qin

, Wang

, Qiao

, Shi

, Wang

, Du

, Zhang

, Luo

, Qu

, Zhou

, Gauthier

, Jia

(2020) Dementia in China: Epidemiology, clinical management, and research advances. Lancet Neurol 19, 81–92.

Petersen

(2004) Mild cognitive impairment as a diagnostic entity. J Intern Med 256, 183–194.

, Tan

, Wang

, Tan

, Xu

, Zhao

, Wang

, Jiang

, Yu

(2016) Risk factors for predicting progression from mild cognitive impairment to Alzheimer’s disease: A systematic review and meta-analysis of cohort studies. J Neurol Neurosurg Psychiatry 87, 476–484.

Miller

, Lundqvist

, Bastos

(2018) Working Memory 2.0. Neuron 100, 463–475.

D’Esposito

, Postle

(2015) The cognitive neuroscience of working memory. Annu Rev Psychol 66, 115–142.

Germano

, Kinsella

(2005) Working memory and learning in early Alzheimer’s disease. Neuropsychol Rev 15, 1–10.

Mandzia

, McAndrews

, Grady

, Graham

, Black

(2009) Neural correlates of incidental memory in mild cognitive impairment: An fMRI study. Neurobiol Aging 30, 717–730.

Saykin

, Wishart

, Rabin

, Flashman

, McHugh

, Mamourian

, Santulli

(2004) Cholinergic enhancement of frontal lobe activity in mild cognitive impairment. Brain 127, 1574–1583.

Lou

, Shi

, Wang

, Tam

, Chu

, Mok

, Cheng

, Lam

(2015) Decreased activity with increased background network efficiency in amnestic MCI during a visuospatial working memory task. Hum Brain Mapp 36, 3387–3403.

10.

Goodman

, Kumar

, Zomorrodi

, Ghazala

, Cheam

, Barr

, Daskalakis

, Blumberger

, Fischer

, Flint

, Mah

, Herrmann

, Bowie

, Mulsant

, Rajji

(2018) Theta-gamma coupling and working memory in Alzheimer’s dementia and mild cognitive impairment. Front Aging Neurosci 10, 101.

11.

Cicalese

, Li

, Ahmadi

, Wang

, Francis

, Selvaraj

, Schulz

, Zhang

(2020) An EEG-fNIRS hybridization technique in the four-class classification of Alzheimer’s disease. J Neurosci Methods 336, 108618.

12.

Boas

, Elwell

, Ferrari

, Taga

(2014) Twenty years of functional near-infrared spectroscopy: Introduction for the special issue. Neuroimage 85 Pt 1, 1–5.

13.

Niu

, Li

, Chen

, Ma

, Zhang

(2013) Reduced frontal activation during a working memory task in mild cognitive impairment: A non-invasive near-infrared spectroscopy study. CNS Neurosci Ther 19, 125–131.

14.

Arai

, Takano

, Miyakawa

, Ota

, Takahashi

, Asaka

, Kawaguchi

(2006) A quantitative near-infrared spectroscopy study: A decrease in cerebral hemoglobin oxygenation in Alzheimer’s disease and mild cognitive impairment. Brain Cogn 61, 189–194.

15.

Yang

, Hong

, Yoo

, Kim

(2019) Evaluation of neural degeneration biomarkers in the prefrontal cortex for early identification of patients with mild cognitive impairment: An fNIRS study. Front Hum Neurosci 13, 317.

16.

Uemura

, Shimada

, Doi

, Makizako

, Tsutsumimoto

, Park

, Suzuki

, (2016) Reduced prefrontal oxygenation in mild cognitive impairment during memory retrieval. Int J Geriatr Psychiatry 31, 583–591.

17.

, Lim

, Kim

, Kyu

, Seok

, Sakong

, Choi

(2020) Prefrontal functional connectivity analysis of cognitive decline for early diagnosis of mild cognitive impairment: A functional near-infrared spectroscopy study. Biomed Opt Express 11, 1725–1741.

18.

Hahn

, Robinson

, Leonard

, Luck

, Gold

(2018) Posterior parietal cortex dysfunction is central to working memory storage and broad cognitive deficits in schizophrenia. J Neurosci 38, 8378–8387.

19.

Makovski

, Lavidor

(2014) Stimulating occipital cortex enhances visual working memory consolidation. Behav Brain Res 275, 84–87.

20.

, Shim

(2017) Occipital, parietal, and frontal cortices selectively maintain task-relevant features of multi-feature objects in visual working memory. Neuroimage 157, 97–107.

21.

Bomyea

, Taylor

, Spadoni

, Simmons

(2018) Neural mechanisms of interference control in working memory capacity. Hum Brain Mapp 39, 772–782.

22.

, Li

, Zhou

, Wang

, Zuo

, Jia

, Song

, Jia

(2011) Montreal cognitive assessment in detecting cognitive impairment in Chinese elderly individuals: A population-based study. J Geriatr Psychiatry Neurol 24, 184–190.

23.

, Feng

, Guo

, Wang

, Li

, Liu

, Fan

(2020) Effects of the multisensory rehabilitation product for home-based hand training after stroke on cortical activation by using NIRS methods. Neurosci Lett 717, 134682.

24.

Sutoko

, Monden

, Tokuda

, Ikeda

, Nagashima

, Funane

, Atsumori

, Kiguchi

, Maki

, Yamagata

, Dan

(2020) Atypical dynamic-connectivity recruitment in attention-deficit/hyperactivity disorder children: An insight into task-based dynamic connectivity through an fNIRS study. Front Hum Neurosci 14, 3.

25.

Homae

, Watanabe

, Otobe

, Nakano

, Go

, Konishi

, Taga

(2010) Development of global cortical networks in early infancy. J Neurosci 30, 4877–4882.

26.

Hoshi

(2003) Functional near-infrared optical imaging: Utility and limitations in human brain mapping. Psychophysiology 40, 511–520.

27.

Kim

(2019) Neural activity during working memory encoding, maintenance, and retrieval: A network-based model and meta-analysis. Hum Brain Mapp 40, 4912–4933.

28.

Leavitt

, Pieper

, Sachs

, Martinez-Trujillo

(2017) Correlated variability modifies working memory fidelity in primate prefrontal neuronal ensembles. Proc Natl Acad Sci U S A 114, E2494–E2503.

29.

Curtis

, Sprague

(2021) Persistent activity during working memory from front to back. Front Neural Circuits 15, 696060.

30.

Rawley

, Constantinidis

(2009) Neural correlates of learning and working memory in the primate posterior parietal cortex. Neurobiol Learn Mem 91, 129–138.

31.

van Kerkoerle

, Self

, Roelfsema

(2017) Layer-specificity in the effects of attention and working memory on activity in primary visual cortex. Nat Commun 8, 13804.

32.

Bettencourt

, Xu

(2016) Decoding the content of visual short-term memory under distraction in occipital and parietal areas. Nat Neurosci 19, 150–157.

33.

Lee

, Kravitz

, Baker

(2013) Goal-dependent dissociation of visual and prefrontal cortices during working memory. Nat Neurosci 16, 997–999.

34.

Sprague

, Serences

(2013) Attention modulates spatial priority maps in the human occipital, parietal and frontal cortices. Nat Neurosci 16, 1879–1887.

35.

Bokde

, Karmann

, Born

, Teipel

, Omerovic

, Ewers

, Frodl

, Meisenzahl

, Reiser

, Moller

, Hampel

(2010) Altered brain activation during a verbal working memory task in subjects with amnestic mild cognitive impairment. J Alzheimers Dis 21, 103–118.

36.

Badgaiyan

, Schacter

, Alpert

(2002) Retrieval of relational information: A role for the left inferior prefrontal cortex. Neuroimage 17, 393–400.

37.

Supeèr

, Spekreijse

, Lamme

VAF

(2001) A neural correlate of working memory in the monkey primary visual cortex. Science 293, 120–124.

38.

Chein

, Moore

, Conway

(2011) Domain-general mechanisms of complex working memory span. Neuroimage 54, 550–559.

39.

Xie

, Xiao

, Bai

, Jiang

(2004) Memory tasks in normal aged volunteers: An fMRI study. Chin J Med Imaging Technol 20, 571–574.

40.

Berryhill

, Chein

, Olson

(2011) At the intersection of attention and memory: The mechanistic role of the posterior parietal lobe in working memory. Neuropsychologia 49, 1306–1315.

41.

Takase

, Boasen

, Kuriki

, Toyomura

, Yokosawa

(2022) Processing time affects sequential memory performance beginning at the level of visual encoding. PLoS One 17, e0265719.

42.

Xie

, Li

, Duan

, Xu

, Zhang

, Fang

(2021) Theta oscillations and source connectivity during complex audiovisual object encoding in working memory. Front Hum Neurosci 15, 614950.

43.

Machulda

, Senjem

, Weigand

, Smith

, Ivnik

, Boeve

, Knopman

, Petersen

, Jack

(2009) Functional magnetic resonance imaging changes in amnestic and nonamnestic mild cognitive impairment during encoding and recognition tasks. J Int Neuropsychol Soc 15, 372–382.

44.

Pasquier

, Grymonprez

, Lebert

, Van der Linden

(2001) Memory impairment differs in frontotemporal dementia and Alzheimer’s disease. Neurocase 7, 161–171.

45.

Karrasch

, Laine

, Rinne

, Rapinoja

, Sinerva

, Krause

(2006) Brain oscillatory responses to an auditory-verbal working memory task in mild cognitive impairment and Alzheimer’s disease. Int J Psychophysiol 59, 168–178.

46.

Kochan

, Breakspear

, Slavin

, Valenzuela

, McCraw

, Brodaty

, Sachdev

(2010) Functional alterations in brain activation and deactivation in mild cognitive impairment in response to a graded working memory challenge. Dement Geriatr Cogn Disord 30, 553–568.

47.

Rosen

(1983) Clinical and neuropsychological assessment of Alzheimer disease. Adv Neurol 38, 51–64.

48.

Romo

, Brody

, Hernandez

, Lemus

(1999) Neuronal correlates of parametric working memory in the prefrontal cortex. Nature 399, 470–473.

49.

Funahashi

, Bruce

, Goldman-Rakic

(1989) Mnemonic coding of visual space in the monkey’s dorsolateral prefrontal cortex. J Neurophysiol 61, 331–349.

50.

Inagaki

, Fontolan

, Romani

, Svoboda

(2019) Discrete attractor dynamics underlies persistent activity in the frontal cortex. Nature 566, 212–217.

51.

Cui

, Zhang

, Zac

, Guo

(2021) Local functional MR change pattern and its association with cognitive function in objectively-defined subtle cognitive decline. Front Aging Neurosci 13, 684918.

52.

, Zhao

, Hu

, Yang

, Zhao

, Niu

, Lin

(2021) Neural mechanism of Shentai tea polyphenols on cognitive improvements for individuals with subjective cognitive decline: A functional near-infrared spectroscopy study. J Alzheimers Dis 82, 1137–1145.

53.

Ricker

, Nieuwenstein

, Bayliss

, Barrouillet

(2018) Working memory consolidation: Insights from studies on attention and working memory. Ann N Y Acad Sci 1424, 8–18.

54.

Ester

, Sutterer

, Serences

, Awh

(2016) Feature-selective attentional modulations in human frontoparietal cortex. J Neurosci 36, 8188–8199.

55.

Liu

, Hospadaruk

, Zhu

, Gardner

(2011) Feature-specific attentional priority signals in human cortex. J Neurosci 31, 4484–4495.

56.

Jehee

, Brady

, Tong

(2011) Attention improves encoding of task-relevant features in the human visual cortex. J Neurosci 31, 8210–8219.

57.

Kastner

, DeSimone

, Konen

, Szczepanski

, Weiner

, Schneider

(2007) Topographic maps in human frontal cortex revealed in memory-guided saccade and spatial working-memory tasks. J Neurophysiol 97, 3494–3507.

58.

Silver

, Ress

, Heeger

(2005) Topographic maps of visual spatial attention in human parietal cortex. J Neurophysiol 94, 1358–1371.

59.

Silva

, Spedo

, Barreira

, Leoni

(2018) Symbol digit modalities test adaptation for magnetic resonance imaging environment: A systematic review and meta-analysis. Mult Scler Relat Disord 20, 136–143.

60.

Jeneson

, Squire

(2012) Working memory, long-term memory, and medial temporal lobe function. Learn Mem 19, 15–25.

61.

Ranganath

(2006) Working memory for visual objects: Complementary roles of inferior temporal, medial temporal, and prefrontal cortex. {Neuroscience 139, 277–289.