We investigated early changes in T2DM subjects without clinically significant cognitive decline or any significant brain diseases. This study was examined and approved by the Medical Research Ethics Committee of Guangzhou University of Chinese Medicine. Written informed consent was provided by all subjects. T2DM subjects were enrolled from the inpatients of the endocrinology department in The First Affiliated Hospital of Guangzhou University of Chinese Medicine from January 1, 2018, to January 30, 2020 (Table 1). All subjects met the following requirements: age between 40 and 60 years, years of education >6, and right-handed. T2DM subjects met the diagnostic criteria released by the American Diabetes Association (ADA) in 2018 and were treated with insulin therapy via insulin pump or insulin injections. Demographically matched HCs were recruited for comparison; the two groups were matched with respect to age, sex, education, and handedness. After asking the subjects about their medical history and going through a standard set of questionnaires, subjects meeting the following criteria were excluded: a diagnosis of other types of diabetes, history of alcohol or substance abuse, significant head trauma, psychiatric or neurological disease that affects central nervous system function, any systemic disease (e.g., tumor and meningitis) and claustrophobia. Because this study only focused on T2DM subjects without MCI, subjects who scored below 26 on the Montreal Cognitive Assessment (MoCA, Beijing edition) were not enrolled (Li et al., 2018). Ten participants were excluded because their MoCA scores did not meet the requirements. Six more subjects were excluded because of image-related problems, such as substantial motion (>2 mm translation or > 2 rotation in any direction) or image artifacts. Finally, a total of 60 participants (30 T2DM subjects and 30 HCs) were included in this study.

Detailed clinical data were collected, including age, sex, blood pressure (BP), body mass index (BMI), years of education, and medication. Participants’ BP was checked at three different times in 1 day, and the results were averaged. Fasting blood glucose (FBG), hemoglobin A1c (HbA1c), low-density lipoprotein (LDL), total cholesterol (TC), and triglycerides were measured in the morning after overnight fasting.

A range of neuropsychological assessments that covered the overall T2DM-related cognitive field were conducted, including the MoCA (Kawada, 2019), auditory verbal learning test (AVLT) (Zhao et al., 2015), trail making test (TMT; including parts A and B) (Camilleri et al., 2015), grooved pegboard test (GPT) (Tolle et al., 2019), symbol digit test (SDT) (Silva et al., 2019), clock drawing test (CDT) (Zhou et al., 2010), and digital span test (DST, including forward and backward) (Diamond, 2013). These scales are widely used to assess cognitive function in the study of cognitive mechanisms related to T2DM (Yang et al., 2018; Li et al., 2020). This battery of assessments took 50 min to complete.

Resting-state functional magnetic resonance imaging data were acquired at The First Affiliated Hospital of Guangzhou University of Chinese Medicine on a 3-T GE SIGNA MRI scanner with an 8-channel head coil. Subjects lay supine on the scanning bed and had to stay awake and keep their eyes closed during the scanning process. Participants also wore earplugs to block out noise, and a foam pad was used to keep their heads stationary during the scanning process. Conventional sequences were performed first to screen brain lesions, including axial T1-weighted and fluid-attenuated inversion recovery (FLAIR) images. Two radiologists who had more than 5 years of working experience and who were blinded to the subject information inspected the conventional images to rule out intracranial lesions. None of the participants were excluded based on the abovementioned exclusion criteria. Then, the experimental sequences were subjected to data processing. The parameters of the sequences were consistent with those of our previous study (Tan et al., 2019).

Resting-state functional magnetic resonance imaging data were preprocessed using the MATLAB R2014a platform and analysis toolkit RESTplus V1.21¹ (Jia et al., 2020). The preprocessing procedure included removing the first 10 volumes, slice timing, realignment, normalization to Montreal Neurological Institute (MNI) space and re-sampled at a resolution of 3 mm× 3 mm× 3 mm, smoothing (full-width Gaussian kernel = 6 × 6 × 6 mm), detrending and covariance regression. Linear regression was used to remove the covariates of global mean signal, white matter, head-motion, and cerebrospinal fluid signal. All participants’ images were within the defined motion thresholds (i.e., less than 2 mm in translation or more than 2 in rotation in any direction).

The ALFF, PerAF, and seed-based FC methods were conducted successively based on the preprocessed images using RESTplus. ALFF was performed after filtering (0.01–0.08 Hz). The brain regions with altered ALFF and PerAF measurements determined by comparison between the two groups were regarded as seed regions. Peak MNI coordinates of the seed regions were regarded as the center of the region of interest (ROI) with a 6-mm radius. Then, the average values across the time series of seed regions were extracted and correlated with each voxel of the rest of the brain. Fisher’s r-to-z transformation was conducted to increase the normality of the data.

Partial correlation analysis was performed to explore the association between cognitive performance and diabetes-related parameters after controlling for sex, age, and education level. Mean ALFF, PerAF, and FC signal values for each brain region with significant differences were extracted. Pearson’s correlation analysis evaluated associations among the mean z-values, neuropsychological test scores and clinical variables. Significant thresholds were set to p < 0.05.

The demographic, clinical and neuropsychological information of all subjects was presented in Table 1. There was no significant difference in age, sex, or education levels between the two groups. The T2DM subjects showed significantly worse performance on the MoCA and SDT tests than HCs (p < 0.05).

Intergroup comparison showed that compared with HCs, T2DM subjects had significantly increased ALFF in the right orbital inferior frontal gyrus, decreased ALFF in the right calcarine (Table 2 and Figure 1A), and decreased PerAF in the bilateral middle occipital gyrus and the right calcarine (Table 2 and Figure 1B).

Seed-based FC analysis showed that compared with HCs, T2DM subjects exhibited significantly decreased FC between the right middle occipital gyrus and the right basal ganglia, which contains the right caudate nucleus and right putamen (Table 3 and Figure 2). However, there was no correlation between the right orbital inferior frontal gyrus/the right calcarine and the rest of the brain regions in T2DM subjects.

According to the partial correlation analyses, HbA1c and AVLT immediate memory scores were negatively correlated in the T2DM group (P = 0.045, r = −0.316, Figure 3A). However, we found that TC was positively correlated with SDT in the T2DM group (P = 0.047, r = 0.366, Figure 3B).

In the T2DM group, correlations were not found between the values of these image indicators and cognitive assessment scores.

The decreased PerAF values may indicate that BOLD signal fluctuations were reduced. In the current study, PerAF values of the bilateral middle occipital gyrus and right calcarine were significantly decreased, which implies decreased neural activity.

Previous studies revealed that decreased spontaneous neural activity in the occipital lobe was relevant to impaired visual performance (Cui et al., 2014). Also, T2DM subjects are prone to retinal damage (Yao et al., 2021). The occipital cortex is generally deemed to dominate visual function, especially the primary visual cortex (Wandell, 1999). The altered spontaneous activity in the occipital cortex may suggest visual information processing dysfunction in T2DM subjects. It has been reported that there is aberrant spontaneous neural activity in the occipital cortex in T2DM subjects, which is consistent with our research (Xia et al., 2013; Cui et al., 2014).

In the current study, T2DM subjects did not have alterations in visual function, implying that decreased spontaneous neural activity may be an early sign that occurs before the appearance of clinically measurable symptoms. It would be interesting to follow these subjects over time to explore the clinical significance of these findings.

Strong activation of the visual cortex is associated with motor execution (Lebon et al., 2018). This relationship between spontaneous nerve activity and cognitive function supports our hypothesis that spontaneous neural activity has a significant impact on cognitive function in T2DM subjects.

It is necessary to explore how the various brain areas work together to better investigate encephalic function (Liu et al., 2018). Decreased FC suggests that temporal consistency between the two brain regions has been reduced. In the present study, compared with the HC group, the T2DM group exhibited significantly decreased FC between the right middle occipital gyrus, which was measured by PerAF, and the right basal ganglia (caudate nucleus and putamen). However, there was no correlation between right orbital inferior frontal gyrus/right calcarine which was measured by ALFF and the rest of the brain regions.

The caudate nucleus is involved not only in complicated emotional regulation but also in executive function (Yang et al., 2015; Huang et al., 2016). Recent evidence has suggested that altered FC of the middle occipital gyrus and caudate nucleus may represent deficiencies in visual information processing and executive function in T2DM subjects (Liu et al., 2020). The putamen is involved in learning and motor control, including language and cognitive functions (Ghandili and Munakomi, 2021). Also, the caudate nucleus and putamen are important parts of the basal ganglia and are responsible for motor function regulation (Andres and Darbin, 2018; Hu et al., 2019). More Pertinently, the basal ganglia structure has been reported to dominate motor function. However, research now shows that the basal ganglia also deal with more complex goal-directed behavior, especially cognitive function (Haber, 2016). We infer that changes in these brain regions contribute to the early decline in cognitive function observed.

In the neuropsychological results analysis, the T2DM subjects showed significantly worse performance on the MoCA and SDT. The MoCA mainly evaluates memory state, language ability, visual condition, and abstract thinking ability. The SDT measures learning ability, recognition speed and flexibility. The reduced MoCA and SDT scores reflect that decreased cognitive function manifested in the incipient stage of T2DM.

Many studies have found a negative correlation between HbA1c levels and cognitive functions (Abbatecola et al., 2010; Reynolds et al., 2010), including memory (Ravona-Springer et al., 2014), executive function (Munshi et al., 2006), attention and information processing speed (Cukierman-Yaffe et al., 2009). In particular, a systematic review provided evidence that high concentrations of HbA1c were negatively associated with cognitive function in T2DM subjects without dementia (Geijselaers et al., 2015). In this study, HbA1c levels were negatively correlated with AVLT immediate recall scores, indicating that increased HbA1c levels may be a risk factor for early cognitive decline in T2DM subjects.

However, TC was positively correlated with SDT scores. Most of the T2DM subjects’ TC levels were below 7 mmol/L, which may suggest that cognitive function improve as cholesterol levels rise within the normal range. There was no correlation between the remaining clinical variables and cognitive scores. The reason for this lack of correlation may be that since all the subjects scored higher than 26 on the MoCA test, their overall cognitive level was considered in the normal range, and the correlation was not strong enough to be detected. In addition, these factors may not major causes that cause early cognitive decline, at least in the present research.