The Soy Treatment Evaluation for Metabolic Health (STEM) Trial

Obesity Clinical Trial

Official title: Role of Soy for Metabolic Health: The Soy Treatment Evaluation for Metabolic Health (STEM) Trial

Status Active, not recruiting

Clinical Trial Summary

Strategies to reduce sugar-sweetened beverages (SSB) have become one of the leading public health targets to address the epidemics of obesity and diabetes. National food, nutrition, and health policies and programs have positioned low-fat milk as the preferred caloric replacement strategy for SSBs. This strategy derives from evidence that replacement of SSBs with low-fat milk is associated with reductions in weight and incident diabetes in prospective cohort studies and reduces liver fat (an important early metabolic lesion linking obesity to diabetes), as well as triglycerides and blood pressure in randomized trials. Whether these benefits hold for soy milk alternatives is unclear. There is an urgent need for studies to clarify the benefits of soy milk as an alternative to cow's milk. Our overarching aim is to produce high-quality clinical evidence that informs the use of soy as a "public health intervention" for addressing the dual epidemics of obesity and diabetes and overall metabolic health. To achieve this aim, we propose to conduct the Soy Treatment Evaluation for Metabolic health (STEM) trial, a large, pragmatic, randomized controlled trial to assess the effect of using 2% soy milk (soy protein vehicle) versus 2% cow's milk (casein and whey vehicle matched for protein and volume) as a "public health intervention" to replace SSBs on liver fat and key cardiometabolic mediators/indicators in an at risk population.

Clinical Trial Description

Rationale:

Soy is at a nutritional "cross-roads". On one hand, it aligns with current dietary advice to consume plant-based dietary patterns [1-6] and has proven advantages [7-9]. On the other hand, it is under threat to have its health claim for CHD risk reduction revoked [10] and is often overshadowed by dairy in public health interventions for metabolic health. Strategies to reduce sugar-sweetened beverages (SSB) have become one of the leading public health targets to address the epidemics of obesity and diabetes [10-22]. National food, nutrition, and health policies and programs have positioned low-fat milk as the preferred replacement strategy for SSBs [23-27]. This strategy derives from evidence that replacement of SSBs with low-fat milk is associated with reductions in weight and incident diabetes in prospective cohort studies [28,29] and reduces liver fat (an important early metabolic lesion linking obesity to diabetes), as well as triglycerides and BP in randomized trials [30]. Whether these benefits hold for soy milk alternatives, which have shown distinct advantages over cow's milk for intermediate metabolic outcomes [7,31-33], is unclear. There is an urgent need for studies to clarify the metabolic benefits of soy milk as an alternative to cow's milk.

Objective:

Our overarching aim is to produce high-quality clinical evidence that informs the use of soy as a "public health intervention" for addressing the dual epidemics of obesity and diabetes and overall metabolic health. To achieve this aim, we propose to conduct the Soy Treatment Evaluation for Metabolic health (STEM) trial, a large, pragmatic, randomized controlled trial to assess the effect of using 2% soy milk (soy protein vehicle) versus 2% cow's milk (casein and whey vehicle matched for protein and volume) as a "public health intervention" to replace SSBs on liver fat and key cardiometabolic mediators/indicators in an at risk population.

Design:

The trial will be a 6-month, 2-phase, 3-arm, non-inferiority, open-label, parallel group, randomized controlled trial to assess the effect of replacing SSBs with 2% soy milk versus 2% cow's milk on the primary outcome liver fat and secondary outcomes muscle fat, insulin sensitivity, beta-cell function, glucose tolerance and established cardiometabolic risk factors in overweight/obese participants with metabolic syndrome (MetS) who are consuming ≥3 SSBs/day. The trial will be conducted according to Good Clinical Practice and reported according to CONSORT and the COSORT extension for non-inferiority trials [34]. The 3-arm design has the advantage of combing a superiority trial with a non-inferiority trial, by allowing us to test the superiority of the active (2% soy milk) and reference (2% cow's milk) treatments over the control (SSBs) and the noninferiority of the active (2% soy milk) compared with the reference (2% cow's milk) treatment simultaneously. We chose a treatment duration of 24-weeks as this was the duration over which milk was shown to reduce liver fat in replacement for SSBs [30] and is a duration that will have a meaningful impact on public health policy [35,36].

Participants:

We will include186 overweight or obese (BMI, 27-40 kg/m2) adult (age, 18-75 years) men and nonpregnant women with MetS based on modified IDF/NHLBI/AHA/WHF/IAAS/IASO [37] criteria (high waist circumference [≥88cm females, ≥102cm males] plus ≥2 of 4 dysglycemia [FPG ≥5.6mmol/L], high triglycerides [≥1.7mmol/L], low HDL-C [<1.0mmol/L males and <1.3 mmol/L females], or high blood pressure [≥130/85mmHg]) who are consuming ≥3 SSBs/day. We have modified the MetS definition to require a high WC to ensure the sample is enriched with participants with NAFLD (defined as hepatic steatosis/intrahepatocellular lipid [IHCL] by 1H-MRS ≥5.0%) [38]. We anticipate > 50% of the sample will have NAFLD based on our entry criteria [38,39]. We will exclude individuals with diabetes or major illness.

Randomization:

The Applied Health Research Centre (AHRC) will perform block randomization with allocation concealment through the Research Electronic Data Capture (REDCap) program. Following successful completion of the run-in phase, participants will be randomized into 3 groups from each strata, using random permuted blocks with unequal sizes.

Protocol:

The trial will be conducted at the CFI-funded Toronto 3D Clinical Research Centre and MRI Research Unit and at the Clinical Nutrition and Risk Factor Modification Centre at St. Michael's Hospital. Eligible participants will undergo a 4-week run-in phase prior to randomization. Participants who wish to continue will then undergo a 24-week intervention phase in which they are randomized (see above) to 1 of 3 interventions: ≥3 servings of (1) 2% soy milk (test treatment), (2) 2% cow's milk (reference treatment/positive control) or (3) their usual SSBs (negative control). All study beverages will be provided with the instruction to replace usual SSBs intake while maintaining background diets. Participants will attend clinic visits every 4-weeks for review of beverage logs, collection of study foods, and clinical assessments of weight, waist circumference [40], and blood pressure. 1H-MRS measurements and 75g oral glucose tolerance tests (75g-OGTT; World Health Organization protocol [41]) with blood samples at -30, 0, 30, 60, 90, and 120 minutes will be performed at 0 and 24 weeks as will collection of the Harvard Willett FFQ), and blood and fecal samples. Adherence biomarker samples will be collected at 0, 12, and 24 weeks.

Interventions:

The 3 interventions will consist of ≥3 servings of (1) 2% soy milk (Alpro Soya, 250mL single-serve shelf-stable packs, 100Kcal with 8g soy protein per 250mL), (2) 2% cow's milk (Organic Meadows®, 250mL, single-serve shelf-stable packs, 130Kcal with 8g casein/whey protein per 250mL), or (3) usual SSBs (355mL single-serve cans, 130-140kcal per 355mL). The participant's usual level of intake of SSBs will determine the dose (number of servings), providing a range of doses (that is, ≥ 3-servings or 24g protein for the soy and cow's milk) for dose response analyses. Although the soy and milk interventions will be matched for protein and serving size, we intentionally did not make the 3 interventions isocaloric, as the goal was to use "real-world" product substitutions.

Outcomes:

All outcome assessments will be blinded and assessed as the end value at week 24. The primary outcome will be liver fat (intrahepatocellular lipid [IHCL]) by 1H-MRS). Secondary outcomes will include change from baseline at week 24 of 75g-OGTT derived changes in insulin sensitivity (Matsuda insulin sensitivity index [ISI] [42]), beta-cell function (insulin secretion-sensitivity index-2 [ISSI-2]) [43,44], and glucose tolerance (2hPG, iAUC). Exploratory outcomes will include ectopic muscle fat (intramyocellular lipid [IMCL]), continuous MetS criteria (waist circumference [WC], FPG, triglycerides [TG], HDL-C, and blood pressure [BP]), reversion of MetS (defined as a reversion to <3 of 5 IDF/NHLBI/AHA/WHF/IAAS/IASO criteria [37]; diabetes incidence (Diabetes Canada 2018 FPG, 2h-PG (75g-OGTT), or HbA1c criteria), and body weight, gut microbiome diversity (16S rRNA sequencing), markers of NAFLD (liver enzymes [ALT, AST, GGT, ALP], fatty liver index (FLI) [45]), hepatic insulin resistance (HOMA-IR), inflammation (hs-CRP), uric acid, established lipid targets (LDL-C, non-HDL-C), and diet quality (Alternative Healthy Eating Index [AHEI] using the Harvard Willett FFQ [46]). Adherence outcomes will include study beverage logs and objective biomarkers of intake of soy milk (urinary isoflavonoid excretion [UIE][47]), cow's milk (dairy-derived serum fatty acids [15:0, 17:0, CLA, trans-palmitoleic acid][48]), and SSBs (isotopic ratios of 13C/12C in serum fatty acids [49]).Other post hoc exploratory outcomes will be decided based on the availability of funding.

Data Management:

The Applied Health Research Centre (AHRC), St. Michael's Hospital, will manage the digital database (REDCap) securely and ensure data integrity. This independent third party data management will be an important strength, bringing a high degree of credibility to the research.

Power:

A total of 186 participants will be randomized. N=62 per group will allow us to test the non-inferiority of the active (2% soy milk) compared with the reference (2% cow's milk) treatment with a non-inferiority margin (δ) of 75% of the effect of the reference treatment on the primary outcome ntra-hepatocellular lipid (IHCL) by 1H-MRS (1.5% absolute difference in IHCL based on a 75% preserved fraction of an expected 2% absolute difference in IHCL) by difference of means [57], assuming an SD of 3%, α=0.05, 80% power (β=0.20), and 20% attrition. A 2% absolute difference in IHCL was chosen as it is the minimum mean difference associated with clinically meaningful changes in downstream glycemic control, insulin sensitivity, blood pressure, and/or blood lipids [51-56]. A non-inferiority margin (δ) of 75% was selected based on guidance from the Food and Drug Administration (FDA) and European Medicines Agency [58-60]. This sample size will provide us ample power to test the superiority of the active (2% soy milk) and the reference (2% cow's milk) treatments compared with the control (SSBs) treatment assuming the same 2% absolute difference in IHCL, SD of 3%, α=0.05, 80% power (β=0.20), and 20% attrition. It will also provide us with sufficient power for superiority testing of the secondary outcomes.

Statistical analysis:

All statistical analyses will be conducted using STATA 14 (StataCorp, Texas, USA) or a suitable statistical software package on end values at week-24 . The 3-arm design of the trial is intended to test the superiority of the active (2% soy milk) and reference (2% cow's milk) treatments compared with the control treatment (SSBs) and the noninferiority of the active treatment (2% soy milk) compared with the reference treatment (2% cow's milk).

The testing of the primary outcome will be done in a stepwise manner to decrease the familywise error rate. We will first undertake superiority testing of the active treatment (2% soy milk) and the reference treatment (2% cow's milk) compared with the control treatment (SSB). The primary analysis will be conducted according to the intention to treat (ITT) principle with inverse probability weighting (IPW) [https://www.bmj.com/content/370/bmj.m2215] to account for missing values. Data will be analyzed using ANCOVA models with significance set at a p < 0.05. Adjustments will made for for age, sex, NAFLD status, medication use, intervention dose (servings/day), and baseline level. Sensitivity analyses will include completers, per protocol, and ITT with multiple imputations (MI). If superiority is established in the primary analysis (confirming the "assay sensitivity"), then we will undertake non-inferiority testing of the active (2% soy milk) compared with the reference (2% cow's milk) treatment using per protocol analysis. If the upper bound of the 90% CI is less than the 1.5% non-inferiority margin [δ] by difference of means, then the active treatment (2% soy milk) will be considered non-inferior to the reference treatment (2% cow's milk). If the upper bound of the 90% CI of the difference of means is less than both the 1.5% non-inferiority margin [δ] and the 0 value (unity), then the active treatment (2% soy milk) will be considered superior to the reference treatment (2% cow's milk). The primary analysis for non-inferiority will be a per protocol analysis. Adjustments will made for for age, sex, NAFLD status, medication use, intervention dose (servings/day), and baseline level. Sensitivity analyses will include completers, ITT with IPW and ITT with MI.

Secondary, exploratory, and adherence outcomes will be assessed by superiority testing among the active treatment (2% soy milk), reference treatment (2% cow's milk), and control treatment (SSB).By reporting all estimated effects and confidence intervals with p-values, the necessary context is provided for a proper interpretation of the statistical evidence [https://www.tandfonline.com/doi/full/10.1080/00031305.2016.1154108]. If it is required for a future regulatory submission, then the secondary outcomes will be adjusted for false discovery. The approach will depend on the results of the primary outcome. If the superiority testing of the primary outcome is significant, then we will analyze the secondary outcomes using the Benjamini-Hochberg false discovery rate controlling method with a starting alpha of 0.05 to correct for false discovery [80]. The sample size was selected to allow sufficient power for this analytical approach. The primary analysis will be conducted according to the intention to treat (ITT) principle with IPW to account for any missing values. Data will be analyzed using ANCOVA models for continuous data (mean differences with 95% CIs) and logistic regression models for categorical data (odds ratio with 95% CI for MetS reversion and relative risk with 95% CI for diabetes incidence). Adjustments will made for for age, sex, NAFLD status, medication use, intervention dose (servings/day), and baseline level of the end value. Sensitivity analyses will include completers, per protocol analyses, and ITT with MI.

Prespecified subgroup analyses will be conducted using Chi-square tests by age, sex, NAFLD status, caffeinated beverage use (cola vs non-cola drinkers), intervention dose (servings/day), and baseline waist circumference, FPG, 2hPG, Triglycerides, HDL-C, BP. Continuous linear (dose response gradient ) and non-linear (dose response threshold) dose-response analyses will be undertaken over the natural dose response range (3, 4, 5, or 6 servings per day) by multiple linear regression and piecewise regression analyses, respectively.

Knowledge translation:

We will follow the Ottawa model (http://www.nccmt.ca/registry/resource/pdf/65.pdf) of knowledge translation with guidance from The Keenan Research Centre - Joint Program in Knowledge Translation, a collaboration between St. Michael's Hospital and University of Toronto. ;

Related Conditions & MeSH terms

• Cardiovascular Diseases
• Diabetes
• Dysglycemia
• Dyslipidemias
• Healthy Obesity, Metabolically
• Heart Diseases
• Hypertension
• Metabolic Syndrome
• Obesity
• Obesity, Metabolically Benign
• Overweight
• Overweight and Obesity
• PreDiabetes

• NCT number: NCT05191160
• Study type: Interventional
• Source: University of Toronto
• Status: Active, not recruiting
• Phase: N/A
• Start date: November 2, 2021
• Completion date: March 30, 2025