Study design and participants
This was a single-centre, randomized controlled, open-label, parallel-arm trial. Participants without a diagnosis of dementia who were older than 60 years of age; had been attending Soushinkai, a daycare rehabilitation facility in Okayama, for at least three months; and were in stable condition were eligible for inclusion. The procedures were in accordance with the guidelines of the Declaration of Helsinki (2000) for human experimentation and informed consent was obtained from all subjects.
At the time of entry, interviews, evaluations, blood tests, and medical examinations were conducted. Patients with cognitive dysfunction, severe anaemia, terminal cancer, or brain degenerative diseases such as Parkinson's disease were excluded. Soushinkai rehabilitation facilities provide not only passive rehabilitation but also voluntary rehabilitation, visitation, and assistance so that rehabilitation can be performed at home. The facility also has a daycare centre and exercise machines.
People who consented to participate the study were randomly divided into three groups. Group C, a normal rehabilitation-only group as control group; Group D, a dietary guidance group, who received dietary guidance and rehabilitation by the same physician; and Group M, who received one capsule of MAF triple in the morning upon waking and one capsule before sleep and the same dietary guidance as Group D. Groups C and D received placebo in the same manner, with one capsule upon waking and one capsule before sleep. MAF triple capsules are dietary supplements produced by Saisei Pharma, Japan, that are designed to modulate the mucosal immunity of the intestine.
Before assessing the Mild Cognitive Impairment Screening (MCIS) score, AGE levels were measured with an AGE reader. The plasma amyloid-β 40/42 ratio, general blood parameters (peripheral blood, GOT, GPT, gamma GPT, ALP, HbA1C, BUN, Cr, uric acid, total cholesterol, HDL cholesterol, triglycerides, ferritin, 25OH vitamin D, CRP, free T4, TSH, albumin, zinc, copper, sodium, potassium, chloride, folic acid, vitamin B12, vitamin B1), and dietary AGE intake were measured. The same procedure was used after 6 and 12 months. The study was approved by SaiseiMirai Ethics Comiittee (No. 2022-001).
Randomization
After screening, the data of each participant were linked to the participant identification code provided and anonymized. Next, the participants were randomized to Groups M, D and C at a ratio of 1:1:1. Randomized allocation was performed via the simple randomization method. The allocation table was generated and maintained by a third party at the facility not involved in the study.
Dietary guidance
Dietary guidance was provided to reduce the amount of AGEs in the diet as much as possible. It is generally known that AGEs are produced in large amounts when meat products and fat-rich foods are fried or baked at high temperatures. On the other hand, cooking methods such as slow steaming or boiling of foods with a high water content are less likely to produce AGEs, which have long been known to be produced in Maillard reactions or browning reactions. The browning of foods is a rough indication of their AGE content. Fructose is approximately 10 times more likely to form AGEs than glucose. It is thought that avoiding drinking carbonated beverages that contain a large amount of fructose corn syrup, avoiding excessive consumption of fructose, and avoiding consuming foods with significant browning can reduce the amount of exogenous AGE intake30. Lemon and vinegar can also reduce the formation of AGEs during the cooking process. A well-balanced diet rich in vegetables, seaweed, and mushrooms is also recommended31.
First, all groups were asked to describe their meals for one week, or if they were unable to do so, they were asked to do so with the help of their family members, who were interviewed in more detail by a nutritionist. Calorie calculations were then performed. AGE calculations were performed with the exAGE formula devised by the AGE Research Association. These calculations were performed before the study and at 6 months. For Group D and Group M, at the start of the study, materials were first distributed to the group, and dietary guidance was given by the same physician in the form of a lecture. The same doctor showed them pictures and instructed them to eat less processed meat (sausage, bacon, etc.), packaged foods, deep fried foods, sugary sweets, and snacks, and more steamed foods, fresh vegetables, and fruits. Afterwards, during the individual interviews, we provided advice on what to maintain and what to add to the diet of each person based on the dietary chart. Individual interviews were held at the start of the study, at 6 months, and at 12 months.
Calculation of AGEs in the diet
The CML (carboxymethyllysine; a type of AGE) coefficient is determined by dividing foodstuffs into 18 categories, including grains, meats, fats, oils, and vegetables, and multiplying the coefficient of cooking (1 for raw foodstuffs and 1 for boiled or baked foodstuffs) by the coefficient of each category. The following original formula is used to estimate the CML (AGE) content of food products.
CML content (in kilograms) = total protein and fat content (g) per 100 g of food material × processing factor × cooking factor × CML factor. This formula was derived by determining the protein and fat contents of ingredients from the Japanese Food Composition Table. There was a strong positive correlation (p<0.0001) between the measured CML content in foods reported by Uribarri et al. and the estimates obtained in the four factors described above, with a correlation coefficient of 0.52. The average CML intake for the American population is reported to be approximately 15,000 KU30.
Cognitive function assessment
The MCIS was used because it is applicable for those with dementia or mild dementia. The MCIS, derived from the Word List Memory test, differentiates cognitive changes associated with normal ageing from mild cognitive impairment as well as dementia32. The MCIS is a brief, electronically scored, verbally administered test that uses correspondence analysis to calculate an individual’s memory capabilities, which is then reported as the Memory Performance Index33-35. The MCIS has been validated in both academic and community clinical settings and in multiple languages36,37. Testing was performed at before the start of the study and at 6 and 12 months. Since large social fluctuations in participants have a direct impact on the score, this index is best for use in studies of populations without large social fluctuations.
AGE measurement
Glycation levels were measured on the dominant volar forearm using an AGE Reader mu (Diagnoptics Technologies B. V., Groningen, Netherlands)38, 39. Pentosidine and crossline have structural properties that cause them to emit fluorescent light across a specific range of wavelengths upon excitation by ultraviolet light. This unique characteristic has been used to develop technology that quantifies the accumulated AGEs within human skin. Although several confounding factors exist, such as other fluorophores, skincare cream use, and skin pigmentation, skin autofluorescence (SAF) is a prominent biomarker that may reflect tissue accumulation of AGEs29,40,41. Certain types of AGEs autofluoresce when exposed to ultraviolet light. The AGE Reader illuminates skin with ultraviolet light (excitation range= 300-420 nm) and then detects the resulting fluorescent light (emission range = 420-600 nm) while simultaneously detecting light reflected from the skin in the 300- to 420-nm range. SAF, reported in arbitrary units, is defined as the ratio of the intensity of the emitted fluorescent light to that of the reflected light42-44.
Plasma ratio of amyloid-β40 to amyloid-β42
Recent technological advancements in mass spectrometry have led to improvements in instrument sensitivity and precision, resulting in the development of improved plasma Aβ assays. Many studies have reported encouraging results for plasma Aβ as a biomarker for Alzheimer’s disease. Tests for the risk of developing dementia were conducted at the starting, 6-month, and 12-month time points45.
Sample size
The study protocol was designed to detect a large effect size (effect size = 0.8) between groups for the change in endpoint over the 12-month study period. Based on that assumption and for a power of 80% and a 5% significance level, the sample size was calculated to be 13 participants per group.
Statistical methods
All data were analysed based on the intention-to-treat principle. The primary analysis was performed on the full analysis set (FAS), while the MCIS score, which is heavily influenced by social variables, was evaluated in the pre-protocol set (PPS). The absolute standard difference (ASD) was used to assess the balance of patient backgrounds by random assignment, and the ASD was considered in balance if the 95% confidence interval in this population was less than 0.755 as a threshold value that did not cross zero.
The repeated-measures endpoints were analysed with linear mixed models that included intervention, dummy variables for time, intervention-by-time interactions and baseline parameters of endpoints as covariates and participants as a random effect. The covariance structure was a completely general covariance matrix. The results are reported as the least squares means with 95% confidence intervals (CIs) at each time point. A P value < 0.05 was considered to indicate statistical significance, and all P values were two-sided with multiplicity adjustment by the Bonferroni method. All the statistical analyses were performed using SPSS version 24.0 (IBM Japan, Ltd.).