2.1 Subjects

A total of 85 subjects with AD were consecutively enrolled in the cross-sectional study between September 2021 and November 2022 from the Center for Cognitive Neurology, Department of Neurology, Beijing Tiantan Hospital, Capital Medical University. All subjects were diagnosed with MCI due to AD (AD-MCI)¹¹ or dementia due to AD (AD-D)¹² according to the National Institute of Aging and Alzheimer's Association (NIA-AA) criteria and divided into the AD-MCI group and the AD-D group, respectively. AD patients meeting the following criteria were excluded from this study: (1) patients with other neurological diseases that affected cognition, including Parkinson's disease, dementia with Lewy bodies, frontotemporal dementia, corticobasal degeneration, acute cerebrovascular disease involving the cerebral cortex, and acute stroke that was temporally associated with cognitive impairment; (2) patients with diseases affecting gait, including lower extremity fracture, femoral head necrosis, and lower extremity arteriosclerosis obliterans; (3) patients with other conditions leading to malnutrition, including hematological tumors, liver cirrhosis, severe systemic diseases, or subtotal gastrectomy; and (4) patients who were unable to cooperate with all the examinations for various reasons.

2.2 Demographic data

The general demographic data, including sex, age, age of onset, education level, marital status, apolipoprotein E (APOE) genotype, and several comorbidities, including hypertension, hyperlipidemia, diabetes mellitus, hyperhomocysteinemia, myocardial infarction, atrial fibrillation, cerebral infarction, cerebral hemorrhage, thyroid disease, smoking, and drinking, were collected.

The following five aspects of nutrition-related demographic data were collected according to the Global Leadership Initiative on Malnutrition (GLIM) criterion¹³:

2.3 Clinical symptoms

The global cognition of patients was assessed by the scales of Mini-Mental State Examination (MMSE)¹⁹ and the Montreal Cognitive Assessment (MoCA).²⁰ Neuropsychiatric symptoms were assessed by the Neuropsychiatric Inventory (NPI) scale.²¹ ADL were assessed by the ADL scale, which includes basic ADL (BADL) and instrumental ADL (IADL).²²

The nutritional status of patients was assessed by the Mini-Nutritional Assessment (MNA) scale.²³ The scoring categorizes subjects: well-nourished (≥24 points), at risk of malnutrition (17–23.5 points), and malnourished (<17 points). To avoid interference of cognitive performance and depression with nutritional assessment, we also analyzed a modified MNA (MNA^m) scale, leaving out the question of “neuropsychological problem”.²⁴

The detailed description of these scales was provided in the Appendix S1.

2.4 Nutrition-related laboratory variables

The venous blood samples of AD patients were collected from the median elbow under fasting condition the next morning after admission, and then sent to the clinical laboratory of Beijing Tiantan Hospital.

We collected a variety of nutrition-related laboratory variables in blood to assess the nutritional status of patients, including hemoglobin A1c, fasting blood glucose, hemoglobin, urea nitrogen, creatinine, albumin, prealbumin, homocysteine, folic acid, and vitamin B₁₂, in serum.

2.5 Gait parameters

The Codamotion 3D motion capture system (Charnwood Dynamics, Ltd., United Kingdom) was used to collect the bilateral gait parameters of patients by active infrared capture.

To record steady-state walking patterns, subjects were instructed to start walking approximately 1 m before entering the infrared capture range. A successful trial is defined as the acquisition of more than three consecutive complete cycles of stereo-motion data, and each task requires the acquisition of at least six successful trials.

The gait parameters, including speed (m/s), step length (m), step length time (s), stride length (m), stride length time (s), cadence (steps/min), cadence (strides/min), and percentage of support (%), were collected. Cadence (steps/min) is defined as the number of steps per minute, and cadence (strides/min) is defined as the number of strides per minute. The variability of each gait parameter is represented by the coefficient of variation (CV): standard deviation (SD)/mean × 100.²⁵

2.6 Statistical analysis

Statistical analyses were performed by SPSS Statistics 25.0 (IBM Corporation, New York, USA). Statistical significance was defined as a two-sided p < 0.05.

Data were tested for normal distribution using the Kolmogorov–Smirnov test. Clinical characteristics, nutritional status, and gait performance were compared between the AD-MCI and the AD-D groups. Continuous variables conforming to normal distribution were presented as means ± SD and compared by two-tailed t test, non-normal distributed variables were presented as median (quartile) and compared by non-parametric test, and categorical variables were presented as number (percentage) and compared by chi-squared test. Binary logistic regression analyses were performed to assess the association of nutritional status and gait performance with dementia, with Model 1 was adjusted for age and gender, and Model 2 was further adjusted for age of onset, educational level, and the scores of NPI and ADL scales based on Model 1. Pearson correlation analysis and Spearman correlation analysis were performed to evaluate the correlations between nutritional status and gait performance in the total patients, the AD-MCI group, and the AD-D group, respectively, and the results were presented by heat maps.

3.1 Clinical characteristics of AD patients

The clinical characteristics of AD patients were shown in Table 1. It was showed that 58.82% of patients were females, the average age was 64.13 ± 10.26 years old, the average age of onset was 59.54 ± 11.92 years old, and 32.90% of cases carried APOE ε4 allele. There were no statistically significant differences in all demographic variables between the AD-MCI and the AD-D groups.

3.1.3 Gait assessment of AD patients

Figure 2 showed the differences in the gait performance between the AD-MCI and AD-D groups. Patients in the AD-D group had slower step speed (p = 0.004), shorter step length (p = 0.012), and shorter stride length (p = 0.014) than those in the AD-MCI group. The variability of these gait parameters did not differ between the two groups.

Binary logistic regression analysis showed that patients with slower step speed [adjusted OR 0.02 (0.00, 0.62), p = 0.017], shorter step length [adjusted OR 0.00 (0.00, 0.48), p = 0.035], and shorter stride length [adjusted OR 0.01 (0.00, 0.69), p = 0.035] were associated with dementia after adjusting for sex, age, age of onset, educational level, and the scores of NPI and ADL scales (Table 2).

In addition, step speed, step length, stride length, and cadence (steps/min) were positively associated with the scores of MMSE and MoCA scales, and negatively associated with the score of ADL scale (all p < 0.05) (Table S1).

3.2 Association between nutritional status and gait performance of AD patients

The correlations between the nutritional status and gait performance in AD patients were shown in Figure 3.

The score of MNA scale was negatively correlated with step length time (r = −0.28, p = 0.011), stride length time (r = −0.23, p = 0.038), and stride length CV (r = −0.22, p = 0.043), and positively correlated with cadence (steps/min) (r = 0.26, p = 0.020) and cadence (strides/min) (r = 0.23, p = 0.036). The score of MNA^m scale was similarly correlated with step length time (r = −0.30, p = 0.006), stride length time (r = −0.25, p = 0.023), stride length CV (r = −0.23, p = 0.037), cadence (steps/min) (r = 0.26, p = 0.018), and cadence (strides/min) (r = 0.24, p = 0.027).

In the nutrition-related laboratory variables, homocysteine level was negatively correlated with step speed (r = −0.31, p = 0.005), step length (r = −0.24, p = 0.030), stride length (r = −0.23, p = 0.037), cadence (steps/min) (r = −0.24, p = 0.034), and cadence (strides/min) (r = −0.28, p = 0.011), and was positively correlated with stride length time (r = 0.29, p = 0.009). Blood Vitamin B₁₂ level was positively correlated with step speed (r = 0.24, p = 0.036). Urea nitrogen level was negatively correlated with stride length CV (r = −0.24, p = 0.029).

Furthermore, analyses were conducted to explore the nutritional status and gait performance in the AD-MCI and AD-D groups (Figure 4). In the AD-MCI group, the nutritional status was related to gait variability. The scores of MNA (r = −0.36, p = 0.040) and MNA^m scores (r = −0.39, p = 0.025) were all negatively correlated with stride length CV. Folic acid level was negatively correlated with stride length time CV (r = −0.39, p = 0.037), cadence (steps/min) (r = −0.48, p = 0.009), and cadence (strides/min) (r = −0.38, p = 0.045). In the AD-D group, nutrition-related variables were significantly correlated with gait parameters. BMI was negatively correlated with step length time (r = −0.30, p = 0.031). Loss of weight was negatively correlated with step speed CV (r = −0.30, p = 0.033). The score of MNA scale was negatively correlated with step length time (r = −0.35, p = 0.013) and stride length time (r = −0.29, p = 0.044), and positively correlated with cadence (steps/min) (r = 0.33, p = 0.020) and cadence (strides/min) (r = 0.29, p = 0.042). Furthermore, vitamin B₁₂ level was positively correlated with step speed (r = 0.32, p = 0.032), cadence (steps/min) (r = 0.35, p = 0.017), and cadence (strides/min) (r = 0.31, p = 0.035), and negatively correlated with step length time (r = −0.32, p = 0.028) and stride length time (r = 0.31, p = 0.033).

We further explored the differences in gait performance between the stages of MCI and dementia by dividing the patients into the well-nourished and undernourished risk/malnourished groups according to the score of the MNA scale. In the well-nourished subgroup, there were no statistical differences in gait parameters and their variability between the stages of MCI and dementia (Figure S1). Nevertheless, in the undernourished risk/malnourished group, step speed (p = 0.007), step length (p = 0.027), and stride length (p = 0.034) in the AD-MCI group were all significantly lower than that in the AD-D group (Figure S2).

4.1 The nutritional status of AD patients was related to disease severity

As weight loss and malnutrition are common problems in AD patients as well as predictors of disease progression and mortality, the nutritional status of individual patient needs to be seriously taken into consideration.^2-4 We found that lower score of MNA scale predicted the occurrence of dementia in AD patients. To improve the reliability of the results, we adopted the revised MNA (MNA^m) scale and adjusted for sex, age, age of onset, educational level, the score of NPI scale, and the score of ADL scale, and still found a significant and negative association between score of MNA scale and dementia.

The lower scores of MNA and MNA^m scales were associated with the lower scores of MMSE and MoCA scales. So far, the factors of malnutrition in AD are not fully clear. It was suggested that the factors that affected food intake, such as olfactory dysfunction, taste disorders, and compromised appetite were the primary reasons for malnutrition in the early stages of AD.^26-28 As the disease progressed, the chronic inflammatory response was gradually increased, leading to excessive protein and energy consumption as the primary cause of malnutrition in the middle and late stages of AD.^{29, 30} In addition, eating disorder, dysphagia, and neuropsychiatric symptoms at the later stage of AD exacerbated malnutrition.³¹

In addition, we analyzed the relationship between nutrition-related laboratory variables and dementia, and found that lower levels of blood urea nitrogen, folic acid, and vitamin B₁₂, and higher homocysteine level had good predictive value for dementia. Blood urea nitrogen is the main end product of protein metabolism and one of the indicators reflecting the nutritional status of the body, which was influenced by multiple factors, including protein intake, dehydration, and gastrointestinal bleeding.³² When insufficient protein intake or excessive consumption, blood urea nitrogen level will decline. Up to now, almost no literature has reported the relationship between blood urea nitrogen and AD. In this study, the results showed that blood urea nitrogen level in AD-D patients was significantly lower than that in AD-MCI patients, and it is reasonable to speculate that this may be related to malnutrition caused by excessive protein and energy consumption at the dementia stage.²⁹

B vitamins, including vitamin B₁₂ and folic acid, are mainly derived from food intake and are indispensable substances for the metabolism of sugars, fats, and proteins in the human body. These are the components of mono-carbon metabolism and essential for maintaining cellular methylation capacity.³³ Multiple studies suggested that low levels of plasma B vitamins in AD patients led to not only the disturbed metabolism of homocysteine but also the impaired release of docosahexaenoic acid and choline into blood, which resulted in increased homocysteine and decreased docosahexaenoic acid and choline, and thus impaired synaptic synthesis and cognition.^34-36

Homocysteine, a kind of sulfur-containing amino acid in the human body, is either metabolized to cysteine by the transsulfuration pathway or converted back to methionine through the remethylation pathway by using cofactors like folic acid and vitamin B₁₂. It was recognized as proatherogenic and prothrombotic, and accepted in many studies as an independent risk for cardiovascular and cerebrovascular diseases.³⁷ In addition, homocysteine was reported to be associated with dementia, such as AD-D, due to vascular damage and oxidative stress.^{38, 39}

Abnormal nutrition-related variables are modifiable risk factors for AD; hence, it is important to reveal whether there is a relationship between the levels of above variables and dementia. In this study, AD patients in the dementia stage exhibited lower levels of folic acid and vitamin B₁₂. Furthermore, the increased homocysteine and decreased folic acid and vitamin B₁₂ were associated with worse global cognition and ADL. These results could potentially be attributed to decreased food intake and increased protein and energy consumptions, which were common in the stage of dementia. The reduced folic acid and vitamin B₁₂ in turn impaired homocysteine synthesis, resulting in higher homocysteine level during the dementia stage.

4.2 The gait performance of AD patients was related to disease severity

Walking is a complex process related to cognitive ability as well as motor function, and it is sensitive to the loss of a higher level of cognitive control.^{7, 40} Previous studies found that patients with cognitive impairment had worse gait performance than those with normal cognition, and poor gait performance acted as a factor predicting progression from MCI to dementia.^{9, 41, 42} In this study, we comprehensively analyzed the relationship between multiple gait parameters and different stages of AD, and found that slower step speed, shorter step length, and stride length significantly predicted dementia and were associated with worse global cognition and ADL.

Altered gait variability is a common feature of several neurodegenerative diseases, including AD, and its increase reflected worse gait rhythm and was associated with the increased risk of falling, and a harbinger of cognitive decline and disease severity.^{42, 43} A previous study found increased gait variability in AD patients compared to cognitively normal individuals, and might distinguish MCI from cognitively normal individuals.⁴⁴ However, the underlying mechanisms of altered gait variability in AD patients are poorly understood. In this study, gait variability between AD-MCI and AD-D patients was not different and the association of gait variability with cognition, neuropsychiatric symptoms, and ADL was not found. These results suggested that gait variability might be more useful in distinguishing dementia from non-dementia than AD at different stages.

4.3 The associations between nutritional status and gait performance in AD patients

This study for the first time found a significant correlation between nutritional status and gait performance in AD patients. Patients with lower scores of MNA and MNA^m scales, and lower levels of urea nitrogen and vitamin B₁₂ had slower gait speed and gait cadence, and longer step length time and stride length time. Patients with higher homocysteine level had slower step speed and cadence, shorter step length and stride length, and longer stride length time. Poor nutritional status was associated with poor muscle strength, leading to poorer gait performance.^{10, 45}

It is worth noting that the correlation between nutrition and gait is more significant at the dementia phase than at the MCI phase. Gait performance was worse, and the risk of falling was higher in the advanced-stage patients with AD.^{9, 41} The above results indicate that improving nutritional status of patients at the last stage has a greater benefit in improving gait performance, and consequently may be more effective in reducing the risk of falling and mortality compared to the patients at the early stage of AD. More interestingly, we found that nutritional status was markedly related to gait variability in patients at the MCI stage, which was revealed by the results that the score of MNA scale was significantly and negatively correlated with stride length CV, and the folic acid level was significantly and negatively correlated with stride length CV and cadence CV. However, the association was not significant in AD patients at the dementia stage, except that weight loss was negatively correlated with step speed CV. Higher gait variability was associated with worse gait rhythm, walking stability, and a higher risk of falling.^{42, 43} The above results were critical for early identification of AD patients at high risk of falling and cognitive decline.

In this study, in AD patients with impaired nutrition, patients at the dementia stage had slower step speed, shorter step length, and stride length than those at the MCI stage; however, in the well-nourished AD patients, this differences in gait performance were not evident between the cases at different stages. We speculate that AD patients with malnutrition may have lower muscle mass and worse muscle strength, which may lead to impaired gait performance. Thus, nutritional status may accelerate the deterioration of gait performance as the disease progresses.^{10, 45}