This study is a randomized, assessor-blind, three-arm, superiority, parallel controlled trial. Participants will be randomly assigned to the HIIT group, MICT group, or guideline group. Testing will be performed at baseline and after the 12-week intervention. The minimum attendance rate will be established in light of the least beneficial dose of exercise. According to previous research (Abreu et al., 2021), participants who complete fewer than 24 training sessions will be considered to drop out of the study. All health and fitness tests will be conducted using the same equipment at the different testing sites. All blood samples will be measured by the same type of facility, and all blood samples will be stored in the Experimental Center of China Medical University for subsequent analysis. An overview of the trial procedures is displayed in Figure 1. This study was approved by the Ethics Review Committee of the Reproductive Hospital Affiliated to China Medical University (approval no. 202203, China clinical trial registration number: ChiCTR2200061573¹). All patients will provide written informed consent prior to enrollment, and the results of the study will be reported following the Consolidated Standards of Reporting Trials (CONSORT) guidelines.

We will recruit 330 eligible T2DM participants from 3 communities in Shenyang by posters, leaflets and brochures and by conducting health seminars; community physicians will be involved in participant recruitment. All diagnostic participants will be evaluated by clinicians. The exercise interventions and measurements of anthropometric parameters and physiological indicators will be performed at community exercise clinics. The research staff who blinded to group allocation will collect baseline data, supervise the training, and finish outcomes related test in clinical exercise physiology. Participants was initial enrolled in June 2022 and the recruitment is still undergoing.

Participants will be included for both of the following criteria: (1) self-reported physical activity of < 150 min per week of moderate-intensity for the past 6 months, (2) accelerometer measured energy consumption of ≤ 1.5 metabolic equivalents (METs) according to previous study of Ku et al. (2017). Briefly, participants will wear the accelerometers continuously for 7 days (5 weekdays and 2 weekend days, excepting bedtime). The average energy consumption will be calculated with a requirement of for a minimum of 10 h per day of 7 days. (3) Have T2DM and regularly take oral hypoglycemic drugs, (4) age ≥ 65 years, (5) consciousness and ability to volunteer for exercise intervention, (6) lack a self-reported hypertension diagnosed by physician or hypertension with well-controlled blood pressure (< 140/90 mmHg).

Patients who meet any of the following criteria will be excluded: patients with (1) inability to exercise or other diseases that may affect their physical activity; (2) cognitive disorder; (3) serious comorbidities or complications, such as heart failure, autonomic neuropathy, and recent stroke; (4) a high risk of falling; (5) a history of cancer; or (6) end-stage renal disease dialysis.

After the baseline assessment, we will randomly assign the participants into three equal groups (1:1:1), namely, the HIIT group, MICT group, and guideline group. This study is an assessor-blind randomized controlled trial, and indicators will be measured at baseline and after 12 weeks by assessors blinded to group allocation. Study interventions will be delivered by study staff. A randomization table will be provided by the Department of Statistics, School of Public Health, Fudan University (Excel software; Microsoft). Study staff who blinded to the group allocation will collect the measurements and maintain possession of the randomization table. The group allocation of participants will be recorded on a paper in an opaque envelope that will only be opened after the baseline assessment of participants has been completed.

We assumed MICT is superior to guideline-based group in glucose control and cognitive function. Moreover, we also test whether HIIT was superior to MICT group. Limited studies have compared the HIIT or MICT and guideline groups on HbA1c or BDNF. We assumed a glycosylated hemoglobin (HbA1c) mean (SD) difference in change of 0.2 (0.3)% between MICT groups and guideline groups. Based on previous study (Terada et al., 2013), an additional mean (SD) difference in change of HbA1c of 0.2 (0.3)% between HIIT and MICT was assumed. The mean difference (SD) in BDNF levels was assumed to be 1.2 (1.4) ng/ml between MICT and guideline group. Additional 1.2 (1.4) ng/ml for BDNF between HIIT and MICT was assigned (Kovacevic et al., 2020). We will adopt the larger sample size to ensure adequate power. Finally, we intend to enroll 330 participants (110 per group) with a power of 90%, a one-sided α of 0.025 and an expected dropout rate of 20%. The α value of 0.025 will be equal assigned to HbA1c and BDNF. PASS software (version 15.0, NCSS Statistical Software) was used to calculate sample size. If the upper confidence interval of the MD in HbA1c levels between the HIIT or MICT interventions and the reference group falls below zero (<0.0%), HIIT or MICT is considered superior to the reference group in terms of improving glucose control. When the lower confidence interval > 0.0% of the MD in BDNF, it would be declared superior to the reference group for improving cognitive function.

All patients will be advised (50 min seminar) on the health benefit of exercise according to current guidelines before intervention. We report our intervention according to the TIDieR checklist. After the informed consent and screening steps are completed, patients will be randomly assigned to three groups (in a 1:1:1 ratio): the MICT group, involving moderate-intensity continuous training (MICT); the HIIT group, involving high-intensity interval training (HIIT); and the guideline group, involving guideline-based physical activity (Figure 1). Participant compliance is of paramount importance for this study, and every possible measure will be taken to maintain adherence to the exercise intervention and follow-up plans. If a participant fails to attend three consecutive sessions, the researcher will actively inquire about the reasons for their absence and provide encouragement for continued participation. To account for the minimal effective dose of exercise, a minimum attendance rate has been established based on previous research findings (Abreu et al., 2021). Specifically, participants who complete fewer than 24 training sessions will be classified as dropouts. The adherence to the training program will be meticulously monitored and regularly documented throughout the trial, allowing for comprehensive analysis and evaluation. Reasons for terminating the test: if serious adverse events occur during exercise, the test should be terminated in time. Adverse events: (1) severe hypoglycemia and complications: shaking, dizziness, abnormal sweating, headache, blurred vision, (2) chest pain, arrhythmia, (3) breathing difficulties and asthma.

Participants in the MICT and HIIT groups will perform exercise of different intensity on a treadmill. Exercise intensity, duration, the heart rate (HR), and blood pressure will be recorded before and after exercise, along with any adverse events during exercise. To minimize the risk of bias in the interpretation of the results, we matched the amount of exercise between the HIIT and MICT groups based on workload. Exercise training will be terminated at signs of any serious adverse event (SAE), such as, severe hypoglycemia and complications: shaking, dizziness, abnormal sweating, headache, blurred vision; chest pain, arrhythmia; breathing difficulties and asthma; pain and fatigue.

The research assistants who blinded to group allocation will collect baseline data from the patients, including age, course of the disease, height, weight, blood pressure, smoking history, socioeconomic status, time since retirement, and history of antidiabetic medication. These assessments will provide a comprehensive overview of the participants’ characteristics. The primary outcome measures of the study will be the HbA1c and BDNF levels. Additionally, the secondary outcome measures will include physical activity levels, anthropometric parameters (weight, waist circumference, hip circumference, and BMI), physical parameters (including fat percentage, muscle percentage, and the fitness rate), cardiorespiratory fitness indicators (blood pressure, the heart rate, vital capacity, and maximum oxygen consumption), biochemical indicators (high-density lipoprotein, low-density lipoprotein, triglyceride, total cholesterol, and HbA1c), inflammation level (C-reactive protein), cognitive function (assessed through reaction time and dual-task gait test performance), and motor function (evaluated through static balance, dynamic balance, single-task gait test performance, and grip strength). To collect these data, participants will be assessed before the start of the intervention and again after the 12-week intervention. Notably, all exercise tests and data collection will be conducted by research assistants blinded to patient group allocation.

Fasting participants will provide blood samples before and after the 12-week exercise intervention to measure serum BDNF levels. The analysis of serum BDNF levels will be performed using an enzyme-linked immunosorbent assay (ELISA) kit in accordance with the instructions provided by the manufacturer. BDNF is considered an indicator of neuroplasticity and cognitive function, and decreased BDNF levels have been associated with cognitive (Zhen et al., 2013).

BP will be measured using the Accoson sphygmomanometer (Accoson, Ayrshire, UK), and HR will be measured using the Omron HEM 3 automated blood pressure machine (Omron, Osaka, Japan). Doctors will conduct the tests while participants maintain a seated position.

Vital capacity will be tested using an electronic spirometer. Cardiopulmonary exercise testing (CPET) will be performed with a cycle ergometer (Ergoline, Germany), and a ramped test protocol of 15 W/min at 60 ± 5 rpm, continued until volitional exhaustion. A 12-lead electrocardiogram (ECG) device will provide continuous monitoring during the test, and respiratory gas analysis will be assessed in a breath-by-breath manner with a computerized metabolic cart (Schiller CS-200 Ergospirometry, Switzerland). The ventilatory anaerobic threshold (VO₂AT) will be assessed using the V-slope method by two independent clinicians. In cases where the VO₂AT deviation is over 10% between the two clinicians, a third clinician will be asked to provide an additional assessment of the VO₂AT. An average of the two nearest values will be selected. VO₂peak will be recorded as the average value of VO₂ during the final 30 s of the test and expressed in ml/kg/min. Cardiopulmonary exercise testing (CPX) will be performed at baseline and at the end of 12 weeks. CPET will be repeated using the same protocol as the baseline test and performed at the same time of day. Cardiorespiratory fitness is an objective measure of physical fitness and has been recognized as protective indictor against cardiovascular events, T2DM and all-cause death.

A Seca 213 stadiometer (Seca; Hamburg, Germany) will be used to measure height and weight, and tape will be used to measure the waist and hip circumference. Finally, the BMI (kg/m²) of participants will be calculated using the following formula: body weight (kg)/[(body height*body height)m²]. BMI is a widely used indictor for obesity and risk factor for cardiovascular events in patients with T2DM.

An Inbody 770 (Inbody Co., LTD., Seoul, Republic of Korea) scan will be used to conduct body fat tests. The Inbody 770 device has reliability and validity for body composition assessment (Poblete Aro et al., 2015). Participants will stand barefoot on a platform with electrodes and undergo measurement through hand-held handles. Finally, the body fat percentage of participants will be calculated using the following formula: body fat weight/total weight. There is a significant association between body fat percentage and the risk of cardiovascular events in patients with T2DM.

During the study, blood samples will be collected from participants before and the 12-week exercise intervention. Blood collection will be performed in the next morning after overnight fast. Four milliliters of venous blood will be drawn, thoroughly mixed by shaking, and allowed to stand at room temperature for 2 h. Subsequently, the blood will be centrifuged at 4°C and 3,000 rpm for 5 min. The resulting supernatant will be stored in a refrigerator at −70°C. To assess various parameters, including glucose, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), triglyceride (TG), total cholesterol (TC), and C-reactive protein (CRP), a fully automated analyzer (7,071°) will be utilized. Each index will be measured twice, and the average value will be used for statistical analysis. Fasting blood glucose is an indicator for blood control during the night. Similarly, blood lipid levels are strong risk factors for incidence of cardiovascular events, regardless presence of T2DM.

Before the test, the entire testing process will be explained by researchers. Response times reflect cognitive function, impaired response times is an indicate of impaired daily function in patients with T2DM (Wong et al., 2014). During the test, the participants will press the start button and wait for the signal to be sent. When any signal is presented, the participants should press the button as fast as possible. After the signal disappears, the participants should press the start button again and wait for the next signal to be presented, for a total of 5 signals. After the participants complete the fifth signal response, the signal key will emit a prompt sound, which indicates the end of the test. There will be three tests in total, and the best score will be recorded.

Before conducting the test, the researcher will explain the specific requirements of the single- and dual-task gait tests and allow the participants to practice on the Footscan gait system (RsScan International, Olen, Belgium). Our previous study demonstrated the reliability and validity of the Footscan gait system in testing gait function in the elderly population (Yan et al., 2022). Gait can reflect motor and cognitive function in patients with T2DM, and the decline in dual-task gait test performance was significantly associated with motor and cognitive dysfunction in patients with T2DM (Herings et al., 2022). During the formal test, participants will be instructed to walk at their normal pace for a total of five trials. Following a brief break, a dual-task gait test will be conducted, in which participants will be instructed to walk at their normal pace while simultaneously performing addition and subtraction tasks aloud, with numbers ranging up to 100. The results obtained will be based on the average of the five trials conducted. The main parameters measured during the single- and dual-task gait tests will include gait velocity, stride length, stride width, and gait cycle. These parameters provide valuable insights into the participants’ walking abilities and help to identify any potential changes or differences between single- and dual-task walking conditions.

Grip strength will be measured using a grip strength dynamometer. As a convenient test, grip strength has been shown to reflect muscle quality (Boonpor et al., 2021). Participants will be instructed to assume a standing position with their arms fully extended. They will be asked to maintain a firm grip on a dynamometer, with the size of the grip adjusted to accommodate their palm size. Three measurements will be taken for each hand to ensure accuracy and reliability. To prevent any potential errors, the dynamometer will be cleared before each test. This will ensure that any residual tension or previous measurements do not influence the results. The maximum grip strength achieved during the three measurements for each hand will be recorded as the final maximum grip strength value for that hand.

Static balance function will be assessed by instructing the participant to stand on one foot with their eyes closed. Impaired static balance is a potential risk factor for fall (Dixon et al., 2017). During the test, participants will be instructed to lift one foot, and the tester will start the timer. The test will continue until the supporting foot moves or the raised foot lands on the ground, indicating the end of the trial. Both legs will be tested three times each to determine the maximum value achieved. To ensure accurate results, a 30-s break will be provided between each test. This break serves to minimize the effects of fatigue and other factors that could influence performance.

The Y balance test (YBT) will be used to test the dynamic balance function of the participant. Our previous study demonstrated the reliability and validity of the YBT in testing dynamic balance function in the elderly population (Yan et al., 2022). Decreased dynamic balance function static balance is a potential risk factor for fall (Dixon et al., 2017). Prior to the commencement of the YBT, participants will receive clear instructions and engage in practice sessions to ensure their understanding and proficiency in performing the test. During the test, participants will stand barefoot and begin by placing their dominant leg on a support board, with their hands at their waist. They will aim to maintain their balance while exerting force with the non-supporting leg to push the indicator as far as possible. Subsequently, they will carefully return their body back to the initial test position. The same procedure will be repeated with the non-dominant leg. To obtain accurate and reliable results, each participant will perform three trials with each leg. The average of the three trials for each leg will be used for subsequent analysis. To account for the influence of participants’ height on the test results, the YBT data will be normalized. Normalization will be achieved by calculating the ratio of the extension distance to the leg length distance and multiplying it by 100%. By normalizing the YBT data, we can effectively account for the individual’s height and obtain a standardized measure of their performance. This allows for meaningful comparisons and analysis across different participants.

The intention-to-treat principle will be employed for all analyses, ensuring that participants are analyzed according to their assigned treatment groups. Linear mixed models will be utilized, incorporating participant and baseline values as random effects, to estimate changes within groups (with a difference between groups of P < 0.1) for all outcome measures. To address missing data, the multiple-imputation method will be employed to test its influence on the analysis. All statistical analyses will be conducted using R software (version 4.0.2, the R Project for Statistical Computing²). The hypothesis tests will be two-sided, and statistical significance will be defined as p < 0.05. For the primary outcomes, multiple will be not necessary as the α value have been allocation in the sample calculation. Importantly, no adjustments for multiple comparisons will be made for secondary outcomes, and therefore, their interpretation should be considered exploratory in nature. Subgroup analyses based on sex, duration of T2DM, and BMI will also be performed to examine potential variations within these subgroups. A per-protocol sensitivity analysis will be applied.