Investigation of type 2 diabetes mellitus and mild cognitive impairment: a study based on diffusion tensor imaging analysis along the perivascular space and peak width of skeletonized mean diffusivity

Abstract

Background

Mild cognitive impairment (MCI) is a common complication of type 2 diabetes mellitus (T2DM); however, its underlying pathogenesis remains unclear. This study aimed to employ diffusion tensor imaging analysis along the perivascular space (DTI-ALPS) and peak width of skeletonized mean diffusivity (PSMD) to investigate changes in the perivascular space (PVS) microenvironment in T2DM.

Methods

Patients with T2DM (24 with MCI and 23 without MCI) and healthy controls (HCs) (n=26) were prospectively recruited. All participants underwent the Montreal cognitive assessment (MoCA) and diffusion tensor imaging (DTI). The analysis along the perivascular space (ALPS) index was calculated using the FMRIB software library (FSL), and PSMD was derived based on tract-based spatial statistics (TBSS). One-way analysis of variance (ANOVA) was used for comparisons among three groups. Correlations of the ALPS index and PSMD with MoCA scores, fasting blood glucose (FBG) levels, and disease duration were analyzed.

Results

The ALPS index was significantly lower in the type 2 diabetes mellitus with mild cognitive impairment (T2DM-MCI) (mean ± standard deviation: 1.324±0.170) and type 2 diabetes mellitus without mild cognitive impairment (T2DM-nMCI) (1.362±0.142) groups than in the HC group (1.465±0.104) (P<0.05), while PSMD (×10–4 mm2/s) was significantly higher in the T2DM-MCI group [median (25th and 75th percentile): 2.139 (1.895–2.306)] than in the T2DM-nMCI [1.868 (1.777–2.026)] and HC [1.888 (1.776–2.111)] groups (P<0.05). PSMD was negatively correlated with MoCA scores (r=–0.425, P=0.003) and positively correlated with disease duration (r=0.433, P=0.002). The ALPS index was negatively correlated with FBG levels, disease duration, and PSMD (r=–0.382, P=0.008; r=–0.420, P=0.003; r=–0.333, P=0.022, respectively).

Conclusions

The combined use of DTI-ALPS and PSMD provides a robust, non-invasive dual-biomarker framework that may advance understanding of the mechanisms underlying diabetic MCI and serve as a potential imaging biomarker.

Introduction

Type 2 diabetes mellitus (T2DM) is a chronic metabolic disorder characterized by elevated fasting blood glucose (FBG) levels due to insulin resistance and/or relative insulin deficiency (1). T2DM is associated with accelerated cognitive decline and a significantly increased risk of mild cognitive impairment (MCI) (2). Evidence indicates that T2DM markedly accelerates the progression from MCI to dementia and is an independent risk factor for cognitive dysfunction (3,4), underscoring the critical importance of early detection and intervention. The pathophysiological mechanisms linking T2DM to MCI are multifaceted, involving chronic hyperglycemia, insulin resistance, microvascular dysfunction, and systemic inflammation (5), all of which synergistically contribute to cerebral small vessel disease (CSVD) and neurodegenerative processes (6).

Conventional neuroimaging techniques have revealed macrostructural brain changes associated with T2DM-related cognitive decline (7). However, more sensitive, in vivo biomarkers capable of detecting early and subtle cerebral alterations before irreversible structural damage occurs are urgently needed. Diffusion tensor imaging (DTI) has emerged as a valuable tool in this context, providing quantitative measures of white matter microstructural integrity (8). DTI studies in individuals with T2DM have consistently revealed disruptions in major white matter tracts (9), including the corpus callosum and corona radiata, which are associated with impairments in processing speed, executive function, and memory (10). These findings underscore the utility of DTI in elucidating the neural substrates underlying T2DM-related MCI.

The diffusion tensor imaging analysis along the perivascular space (DTI-ALPS) is a novel, non-invasive neuroimaging biomarker that serves as a proxy for evaluating the perivascular space (PVS) microenvironment—the brain’s primary waste clearance pathway (11). It measures the anisotropy of water diffusion along the PVS, with a reduced analysis along the perivascular space (ALPS) index reflecting impaired metabolic waste clearance (12). Such dysfunction has been associated with the accumulation of neurotoxic metabolites, including amyloid β-protein and hyperphosphorylated microtubule-associated protein tau (13), and is increasingly recognized as a contributing factor to cognitive decline in various neurodegenerative and neurovascular conditions, such as CSVD (14,15) and Parkinson’s disease (PD) (16). Given that T2DM is characterized by microvascular damage and chronic low-grade inflammation (17)—both of which are known to compromise the integrity of the perivascular unit and disrupt glymphatic flow (18)—investigating the ALPS index in individuals with T2DM is highly warranted.

Peak width of skeletonized mean diffusivity (PSMD) is a fully automated and robust imaging metric specifically developed to reflect diffuse cerebrovascular injury, particularly within the white matter skeleton (19). By quantifying the heterogeneity of water diffusion, PSMD demonstrates higher sensitivity to widespread, subtle microstructural white matter damage that may escape detection by conventional lesion-based volumetric methods (20). PSMD has been shown to correlate more strongly with deficits in processing speed than traditional magnetic resonance imaging (MRI) markers (21) and is closely associated with white matter hyperintensity (WMH) burden (22). Given that T2DM is a major risk factor for CSVD, PSMD represents a promising biomarker for detecting underlying microstructural white matter alterations associated with MCI in this population.

Both the ALPS index and PSMD are derived from DTI and are associated with cognitive impairment; however, they are thought to reflect distinct—yet potentially interrelated—pathophysiological pathways. The ALPS index primarily serves as an indirect marker of diffusion properties along the PVS, while PSMD captures the impact of diffuse cerebral microvascular injury on white matter microstructure. To date, previous studies have largely investigated these metrics in isolation across different diseases, leaving it unclear whether the ALPS index and PSMD are independently or synergistically associated with MCI in T2DM, and which metric offers superior explanatory power or clinical utility.

Therefore, this study aimed to investigate the association between T2DM and MCI using DTI-ALPS and PSMD, to assess differences in the ALPS index and PSMD between T2DM patients with and without MCI, and to evaluate their correlations with key clinical characteristics. The combination of these two indicators forms a dual-biomarker framework that enables comprehensive evaluation of neurovascular pathological changes in T2DM from the perspectives of perivascular fluid dynamics and white matter microstructure. This integrated approach is expected to provide a more comprehensive imaging basis and address the limitations of single biomarkers in exploring the pathological mechanisms underlying T2DM-related cognitive impairment. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2814/rc).

Methods

Participants

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Dongfang Hospital, Beijing University of Chinese Medicine (Nos. JDF-IRB-2023051302 and JDF-IRB-2024057001), and informed consent was obtained from all individual participants. T2DM patients were recruited from the Endocrinology Department, Dongfang Hospital, Beijing University of Chinese Medicine between May 2023 and March 2025. The inclusion criteria were as follows: (I) FBG ≥6.1 mmol/L or hemoglobin A1c (HbA1c) ≥6.5% (23); (II) age between 18 and 70 years, regardless of sex. The upper age limit of 70 years was set to minimize potential confounding effects of age-related irreversible brain structural and functional changes; (III) right-handedness; and (IV) education duration ≥6 years. The exclusion criteria were as follows: (I) failure to meet the diagnostic criteria for T2DM; (II) presence of other psychiatric disorders such as Alzheimer’s disease or PD; and/or (III) MRI contraindications or an inability to cooperate. Age-, sex-, education-, and handedness-matched healthy controls (HCs) were also recruited.

All participants completed the domain-specific neuropsychological assessment and Beijing version of the Montreal cognitive assessment (MoCA) (24) under the supervision of two experienced neurologists. One point was added for participants with ≤12 years of education; otherwise no additional point was added. Based on the clinical assessment, education-corrected MoCA scores, neuropsychological test results, and the 2018 Chinese Guidelines for Dementia and Cognitive Impairment (25), participants were classified into three groups: (I) experimental group [type 2 diabetes mellitus with mild cognitive impairment (T2DM-MCI) group]: MoCA score <26, with FBG ≥6.1 mmol/L or HbA1c ≥6.5%; (II) positive control group [type 2 diabetes mellitus without mild cognitive impairment (T2DM-nMCI) group]: MoCA score ≥26, with FBG ≥6.1 mmol/L or HbA1c ≥6.5%; and (III) HC group: MoCA score ≥26, with 3.9< FBG ≤6.1 mmol/L and HbA1c <6.5% (Figure 1).

Figure 1.

Patient selection flowchart. FBG, fasting blood glucose; HbA1c, hemoglobin A1c; HC, healthy control; MoCA, Montreal cognitive assessment; MRI, magnetic resonance imaging; PD, Parkinson’s disease; T2DM-MCI, type 2 diabetes mellitus with mild cognitive impairment; T2DM-nMCI, type 2 diabetes mellitus without mild cognitive impairment.

Image acquisition

MRI data for all participants were collected using a 3.0 Tesla MRI scanner (Discovery MRI 750, GE Healthcare, Milwaukee, USA) with an eight-channel phased-array head coil. The DTI parameters were as follows: repetition time =6,000 ms; echo time = minimum; slice thickness =3 mm; inter-slice gap =0 mm; matrix size =128×128; field of view =256 mm × 256 mm; number of diffusion directions =64; and b values =0 and 1,000 s/mm².

MRI processing

All DTI data underwent eddy current and motion correction during preprocessing using MRtrix3. For each participant, total motion in each diffusion volume was quantified as the root mean square (RMS) displacement across all intracerebral voxels, calculated as the square root of the mean squared voxel-wise displacements. Two participants with an RMS displacement exceeding 2 mm were excluded.

DTI data were analyzed using tools from the FMRIB Software Library (FSL; https://fsl.fmrib.ox.ac.uk/fsl/) (26). At the level of the bilateral centrum semiovale, three circular regions of interest (ROIs) with a diameter of approximately 5 mm were manually placed using FSLeyes software within the projection fibers (corticospinal tract) and association fibers (superior longitudinal fasciculus). Care was taken to ensure that each ROI was positioned within the identified fiber tract, avoiding the ventricles, gray matter, and visible WMHs. All ROI placements were performed by a neuroimaging researcher blinded to the participants’ clinical information. To assess the reproducibility, the ROI placements were reviewed by another experienced rater to confirm that the placements met the required standards.

Within FSL, diffusion metrics were automatically extracted, including Dx values from the projection (Dxxproj) and association (Dxxassoc) fiber regions on the Dx map, Dy values (Dyyproj and Dyyassoc) on the Dy map, and Dz values (Dzzproj and Dzzassoc) on the Dz map. The ALPS index was then calculated using the following formula:

$$\text{ALPS\hspace{0.17em}index}=\frac{\text{mean}\left(\text{Dxxproj},\text{Dxxassoc}\right)}{\text{mean}\left(\text{Dyyproj},\text{Dzzassoc}\right)}$$ [1]

A color tensor map was also generated as part of the ALPS index analysis.

PSMD was automatically calculated using the PSMD tool (http://www.psmd-marker.com) (19). DTI data were skeletonized using tract-based spatial statistics (TBSS) (27). The mean diffusivity (MD) images were projected onto the white matter skeleton using projection parameters derived from the fractional anisotropy images. A standard skeleton threshold and a custom mask (provided by the PSMD tool) were applied to exclude regions near the ventricles, and the resulting skeletonized maps were aligned to the Montreal Neurological Institute standard space. Histogram analysis was then performed by calculating the difference between the 5th and 95th percentiles, yielding PSMD values for the left and right hemispheres of each participant. These values were averaged to obtain the final PSMD measurement.

Statistical analysis

All clinical data were analyzed using SPSS version 25.0. A P value <0.05 was considered statistically significant. Continuous variables were expressed as mean ± standard deviation (x ± s) if normally distributed, or as median (interquartile range) if not normally distributed. Categorical variables were presented as frequencies (percentages).

For comparisons between the T2DM-MCI and T2DM-nMCI groups, the independent samples t-test was used for normally distributed data with equal variances; otherwise, the Mann-Whitney U test was used. For comparisons among the T2DM-MCI, T2DM-nMCI, and HC groups, one-way analysis of variance (ANOVA) was used when assumptions of normality and homogeneity of variance were met; otherwise, the Kruskal-Wallis test was applied. Categorical variables were compared using the Chi-squared (χ²) test.

Correlation between continuous variables was assessed using Pearson’s correlation coefficient for normally distributed data or Spearman’s rank correlation coefficient otherwise.

All P values were adjusted for multiple comparisons using the false discovery rate (FDR) with the Benjamini-Hochberg procedure, and statistical significance was set at P<0.05.

Results

Participant characteristics

A total of 73 subjects were enrolled in this study, including 47 patients with T2DM and 26 HCs. Among the T2DM patients, 24 were diagnosed with MCI, and 23 were without MCI. Both the T2DM-MCI and T2DM-nMCI groups had significantly higher FBG and HbA1c levels than the HC group (both P<0.001). The T2DM-MCI group exhibited lower MoCA scores than both the T2DM-nMCI and HC groups (P<0.001). However, no significant differences were observed in FBG, HbA1c, or disease duration between the T2DM-MCI and T2DM-nMCI groups (all P>0.05). There were no significant differences in sex, age, or years of education among the three groups (Table 1).

Table 1. Participant characteristics, laboratory tests, and cognitive assessment.

Variables	T2DM-MCI (n=24)	T2DM-nMCI (n=23)	HCs (n=26)	PFDR
Male	16 (66.7)	11 (47.8)	10 (38.5)	0.190a
Age (years)	60.0 [54.5, 62.0]	58.0 [46.0, 62.0]	53.5 [51.0, 61.3]	0.264c
Years of education	12 [12, 16]	16 [12, 16]	12 [12, 13]	0.190c
MoCA	22 [20, 25]†,‡	26 [26, 27]	27 [26, 27]	<0.001c
FBG (mmol/L)	7.75 [6.02, 11.17]‡	7.45 [6.71, 8.86]‡	5.04 [4.78, 5.45]	<0.001c
HbA1c (mg/dL)	7.80±1.28‡	7.35±1.30‡	5.66±0.44	<0.001d
Disease duration (years)	13.0 [5.5, 20.0]	11.0 [7.0, 15.0]	Null	0.840b

Data are presented as n (%), mean ± standard deviation or median [25th and 75th percentile]. ^†, compared to the T2DM-nMCI group, P<0.05; ^‡, Compared to the HC group, P<0.05; ^a, Chi-squared test; ^b, Mann-Whitney U test; ^c, Kruskal-Wallis test; ^d, one-way ANOVA. ANOVA, analysis of variance; FBG, fasting blood glucose; FDR, false discovery rate; HbA1c, glycosylated hemoglobin; HCs, healthy controls; MoCA, Montreal cognitive assessment; T2DM-MCI, type 2 diabetes mellitus with mild cognitive impairment; T2DM-nMCI, type 2 diabetes mellitus without mild cognitive impairment.

Comparison of the ALPS index and PSMD among the three groups

ANOVA revealed significant differences in the ALPS index among the three groups (P=0.005). Pairwise comparison showed that both the T2DM-MCI and T2DM-nMCI groups had lower ALPS index values than the HC group (P=0.002 and P=0.020, respectively) (Figure 2A). Similarly, significant differences in the PSMD values were observed among the three groups (P=0.025). Pairwise comparisons indicated that the T2DM-MCI group had higher PSMD values than both the T2DM-nMCI and HC groups (P=0.022 and P=0.027, respectively) (Figure 2B). No significant difference was found in the ALPS index between the T2DM-MCI and T2DM-nMCI groups, or in PSMD between the T2DM-nMCI and HC groups (Table 2).

Figure 2.

Group comparisons of the ALPS index and PSMD. (A) Comparisons of ALPS index values among the T2DM-MCI, T2DM-nMCI, and HC groups. (B) Comparisons of PSMD values among the T2DM-MCI, T2DM-nMCI, and HC groups. The results were corrected using the false discovery rate method. *, P<0.05; **, P<0.01. ALPS, analysis along the perivascular space; HC, healthy control; PSMD, peak width of skeletonized mean diffusivity; T2DM-MCI, type 2 diabetes mellitus with mild cognitive impairment; T2DM-nMCI, type 2 diabetes mellitus without mild cognitive impairment.

Table 2. Group comparisons of imaging parameters.

Variables	T2DM-MCI (n=24)	T2DM-nMCI (n=23)	HCs (n=26)	PFDR
DTI-ALPS	1.324±0.170‡	1.362±0.142‡	1.465±0.104	0.005a
PSMD (×10−4 mm2/s)	2.139 (1.895, 2.306)†,‡	1.868 (1.777, 2.026)	1.888 (1.776, 2.111)	0.025b

Data are presented as mean ± standard deviation or median (25^th and 75^th percentile). ^†, compared to the T2DM-nMCI group, P<0.05; ^‡, compared to the HC group, P<0.05; ^a, one-way ANOVA; ^b, Kruskal-Wallis test. ANOVA, analysis of variance; DTI-ALPS, diffusion tensor imaging along the perivascular space; FDR, false discovery rate; HCs, healthy controls; PSMD, peak width of skeletonized mean diffusivity; T2DM-MCI, type 2 diabetes mellitus with mild cognitive impairment; T2DM-nMCI, type 2 diabetes mellitus without mild cognitive impairment.

For the ALPS index, Cohen’s d was 0.43, with an achieved power of 0.82. For PSMD, Cohen’s d was 0.34, with an achieved power of 0.81. These results indicate that, despite the relatively small sample sizes, the study was adequately powered (achieved power >0.8).

Correlation among the ALPS index and cognitive function, FBG, HbA1c, and disease duration

The correlation analysis revealed that the ALPS index was negatively correlated with FBG [r=–0.382, 95% confidence interval (CI): –0.609 to –0.098, P=0.008]. This association remained significant after adjusting for age, sex, and years of education in the partial correlation analysis (r=–0.458, P=0.002). Additionally, the ALPS index was significantly negatively correlated with disease duration (r=–0.420, 95% CI: –0.637 to –0.143, P=0.003). After adjusting for the same covariates (sex, age, and years of education), the association between the ALPS index and disease duration remained significant in the partial correlation analysis (r=–0.340, P=0.024) (Figure 3A,3B; Table 3).

Figure 3.

Correlations of the ALPS index and PSMD with cognitive function, FBG levels, and disease duration. (A) A negative correlation was observed between the ALPS index and FBG levels. (B) A negative correlation was observed between the ALPS index and disease duration. (C) A negative correlation was observed between PSMD and MoCA. (D) A positive correlation was observed between PSMD and disease duration. ALPS, analysis along the perivascular space; FBG, fasting blood glucose; MoCA, Montreal cognitive assessment; PSMD, peak width of skeletonized mean diffusivity.

Table 3. Correlations between FBG, HbA1c, MoCA, disease duration, and imaging parameters.

Parameters	MoCA					FBG					HbA1c					Disease duration
	r	95% CI	P	PFDR		r	95% CI	P	PFDR		r	95% CI	P	PFDR		r	95% CI	P	PFDR
ALPS	0.080	–0.220, 0.367	0.592†	0.592		–0.382	–0.609, –0.098	0.008†	0.016		–0.088	–0.366, 0.205	0.557‡	0.592		–0.420	–0.637, –0.143	0.003†	0.008
PSMD	–0.425	–0.640, –0.149	0.003†	0.008		0.176	–0.126, 0.448	0.236†	0.378		0.105	–0.196, 0.388	0.483†	0.592		0.433	0.158, 0.646	0.002†	0.008

^†, Spearman; ^‡, Pearson. ALPS, analysis along the perivascular space; CI, confidence interval; FBG, fasting blood glucose; FDR, false discovery rate; HbA1c, glycosylated hemoglobin; MoCA, Montreal cognitive assessment; PSMD, peak width of skeletonized mean diffusivity.

Correlation among PSMD and cognitive function, FBG, HbA1c, and disease duration

The Spearman correlation analysis revealed that PSMD was negatively correlated with MoCA scores (r=–0.425, 95% CI: –0.640 to –0.149, P=0.003). After adjusting for sex, age, and years of education, the partial correlation analysis indicated that the correlation remained significant (r=–0.355, P=0.018). Further, PSMD showed a significant positive correlation with disease duration (r=0.433, 95% CI: 0.158 to 0.646, P=0.002). After adjusting for the same covariates (sex, age, and years of education), the correlation between PSMD and disease duration remained significant in the partial correlation analysis (r=0.328, P=0.030) (Figure 3C,3D, and Table 3).

Correlation between ALPS index and PSMD

The Spearman correlation analysis revealed a significant negative correlation between the ALPS index and PSMD (r=–0.333, P=0.022). However, after adjusting for the covariates of sex, age, and years of education in the partial correlation analysis, the correlation between the two metrics was no longer significant (r=–0.270, P=0.077) (Figure 4).

Figure 4.

Correlations between ALPS index and PSMD. Partial correlation analysis revealed no significant correlation between ALPS index and PSMD. ALPS, analysis along the perivascular space; PSMD, peak width of skeletonized mean diffusivity.

Discussion

This study employed DTI to investigate white matter microstructural alterations and the PVS microenvironment in patients with T2DM, with a particular focus on those with MCI. Our findings revealed that the ALPS index was significantly lower in both the T2DM-MCI and T2DM-nMCI groups than in the HC group. Conversely, PSMD was significantly higher in the T2DM-MCI group than in both the T2DM-nMCI and HC groups. These results suggest impairment of the PVS microenvironment and white matter microstructural damage in individuals with T2DM, with more pronounced abnormalities observed in the T2DM-MCI group.

The observed reduction in the ALPS index in both the T2DM-MCI and T2DM-nMCI groups, irrespective of cognitive status, suggests that dysfunction of the brain’s perivascular fluid transport system represents an important pathological event in the diabetic brain. The DTI-ALPS method, first proposed in 2017 (11), quantifies the ratio of water diffusivity along the PVS to that in orthogonal directions at the level of the lateral ventricle body. A higher ALPS index is presumed to reflect more efficient water diffusion along the PVS, which is conceptually associated with the PVS microenvironment (28). Our finding of a negative correlation between the ALPS index and FBG levels establishes a direct association between hyperglycemia and impaired PVS dynamics. This observation is consistent with substantial evidence demonstrating that hyperglycemia compromises blood-brain barrier (BBB) integrity, a structure closely integrated with PVS function (29). Chronic hyperglycemia may induce endothelial cell dysfunction (30), downregulate tight junction proteins expression (31), and promote oxidative stress (32), all of which collectively impair the vascular interface critical for effective interstitial fluid (ISF) and solute exchange.

PSMD is derived from the histogram width of MD values along the skeleton of major white matter tracts, a method specifically designed to detect widespread and subtle microstructural damage while minimizing contamination from cerebrospinal fluid (CSF) (33). It has been robustly validated as a marker of overall CSVD burden, demonstrating significant associations not only with WMH but also with other neuroimaging features of CSVD (34). Our finding that PSMD values were elevated in the T2DM-MCI group compared with both the T2DM-nMCI and HC groups suggests that cognitive impairment in T2DM is likely associated with the accumulation of diffuse white matter injury. This damage likely reflects a combination of axonal loss, demyelination, and gliosis resulting from chronic hypoperfusion, BBB disruption, and inflammation driven by diabetic microangiopathy (29,35). The selective elevation of PSMD in the T2DM-MCI group, along with its significant negative correlation with MoCA scores, underscores the role of PSMD as a key neuroimaging biomarker linking cerebrovascular pathology to cognitive impairment in diabetes. Additionally, the PSMD values in this study ranged from 1.868 to 2.152 (×10^–4 mm²/s), consistent with previous PSMD studies in T2DM patients with CSVD, supporting the reliability of the results.

Further, the negative correlation between the ALPS index and disease duration suggests a progressive decline in the PVS microenvironment over time. This deterioration may be attributed to cumulative metabolic injury, which contributes to PVS enlargement and may impede CSF movement (36). Notably, a recent study in patients with T2DM reported a reduced ALPS index, particularly in those with MCI, and further reported associations between ALPS index reduction and both cognitive deficits and white matter damage (37), thereby supporting our findings. However, the ALPS index should be interpreted with caution. It provides only an indirect assessment of the PVS microenvironment and does not directly quantify CSF-ISF exchange. As a measure of macroscopic water diffusion along the PVS, it cannot be directly equated with glymphatic function without additional supporting evidence (28). Indeed, a previous study found no association between the ALPS index and measures of CSF tracer dynamics at 48 hours (38). Moreover, a reduction in the ALPS index in T2DM may have multiple potential interpretations beyond PVS microenvironment dysfunction, including microvascular remodeling in CSVD (39), PVS enlargement resulting from chronic hyperglycemia (40), and interstitial edema in the brain parenchyma (41). These pathological changes can all affect water diffusion along the PVS and lead to a decreased ALPS index. Future studies integrating additional techniques are needed to more directly assess the PVS microenvironment.

The positive correlation between PSMD and disease duration further supports the notion of progressive and cumulative vascular injury. Notably, evidence suggests that PSMD is specifically linked to vascular pathology rather than generalized neurodegeneration, positioning it as a relatively selective biomarker for vascular cognitive impairment (22). This specificity is particularly valuable in T2DM, where cognitive decline frequently arises from mixed underlying etiologies. Further, the observed significant negative correlation, coupled with the non-significant partial correlation between the ALPS index and PSMD, supports the view that these two markers reflect interrelated yet distinct pathological processes in T2DM-related brain injury. Alterations in the PVS microenvironment may exacerbate white matter microstructural damage by impairing ISF drainage and metabolite clearance, although this hypothesis warrants further investigation.

The combined analysis of DTI-ALPS and PSMD offers distinct advantages over the use of either marker in isolation. First, DTI-ALPS and PSMD assess different but interrelated aspects of neurovascular pathology: the ALPS index provides a functional measure of the PVS microenvironment, while PSMD reflects structural alterations associated with diffuse white matter integrity loss. Together, they enable a more comprehensive assessment of both the PVS microenvironment and white matter tract integrity in the diabetic brain. Second, our findings indicate that while a reduced ALPS index is present in patients with T2DM, a significant increase in PSMD is specifically evident in those with MCI. Based on these cross-sectional results, we hypothesize that a reduced ALPS index may represent an early potential vulnerability factor, while the accumulation of diffuse white matter microstructural damage may serve as a key factor associated with the emergence of MCI in T2DM. However, this hypothesis requires validation through prospective longitudinal studies, mediation analysis, and interaction modeling in future research. Third, both DTI-ALPS and TBSS-based PSMD are advanced DTI-based techniques. TBSS mitigates registration inaccuracies by projecting data onto a standardized white matter skeleton (42), enabling robust, whole-brain, voxel-wise comparisons of microstructural integrity without bias from predefined ROI (43). The integration of DTI-ALPS with PSMD thus constitutes an efficient, multi-parametric approach to leveraging DTI data, capturing both localized fluid dynamics and global tissue pathology.

This study had several limitations. First, the cross-sectional design and relatively small sample size limit causal inference and generalizability; future prospective longitudinal studies with large, multi-center cohorts are required to validate the findings. Second, the ALPS index provides only an indirect assessment of the PVS microenvironment and cannot directly measure CSF‑ISF exchange; subsequent studies combining DTI-ALPS with CSF tracer imaging are needed to confirm its relationship with glymphatic clearance. Third, the use of single-b-value DTI (b=1,000 s/mm²) restricts microstructural accuracy, and the absence of WMH quantification prevents differentiation between microstructural damage and visible WMH lesions; future work should employ multi-shell DTI and incorporate WMH burden as a key covariate. Fourth, patients over 70 years of age were excluded, limiting applicability to older T2DM populations; further studies should include this age group to assess the dual-biomarker framework. Fifth, although age, sex, and education were adjusted for, other factors influencing microvascular pathology and cognitive function (e.g., hypertension, dyslipidemia, sleep disorders, smoking, and medication use) were not accounted for; future investigations should systematically collect these data to enable more comprehensive covariate adjustment.

Conclusions

This study proposed a preliminary hypothesis regarding the association of the ALPS index and PSMD with T2DM-related cognitive impairment. However their temporal and causal relationships require further validation in future studies. The combined use of DTI-ALPS and PSMD provides a robust, non-invasive dual-biomarker framework that may advance understanding of the mechanisms underlying diabetic cognitive impairment and serve as a potential imaging biomarker. Nonetheless, prospective studies are needed to confirm these findings.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Dongfang Hospital, Beijing University of Chinese Medicine (Nos. JDF-IRB-2023051302 and JDF-IRB-2024057001) and informed consent was taken from all individual participants.

Data Sharing Statement

Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1-2814/dss

qims-16-08-654-dss.pdf ^{(76.4KB, pdf)}

DOI: 10.21037/qims-2025-1-2814