Full-fat yogurt compared with non-fat yogurt reduces blood triacylglycerol concentrations and lowers the triacylglycerol content in specific lipoprotein subclasses in adults with prediabetes: an exploratory analysis of a randomized-controlled trial

Abstract

Background

Low- and non-fat dairy foods have long been recommended over full-fat dairy foods due to the negative effect of saturated fatty acids on blood lipids. Recent research, however, suggests saturated fatty acids from dairy foods may not impart these negative health effects. Our objective was to evaluate changes in blood lipids following a diet with full-fat (3.25%) yogurt compared with a diet with non-fat yogurt.

Methods

A randomized, double-masked crossover controlled-feeding trial was performed. Participants with prediabetes (n = 13, 7 female and 6 male participants) consumed three daily servings of full-fat or non-fat yogurt for the three weeks of each experimental diet. A one-week run-in diet preceded each experimental diet period. After each experimental diet period and the first run-in diet period, fasting blood and blood drawn at four post-prandial time points during a mixed meal tolerance test were analyzed for lipoprotein concentrations and contents (i.e., the lipid fractions within the lipoproteins). Statistical analyses were performed using linear mixed models, with values from the first run-in diet as the covariate.

Results

Fasting blood triacylglycerol concentrations were 10% lower in response to the full-fat yogurt diet, compared with the non-fat yogurt diet (P < 0.01). While no diet-induced differences were observed in lipoprotein subclass concentrations, the triacylglycerol contents of smaller very low-density, intermediate-density, and low-density lipoproteins were lower in response to the full-fat yogurt diet (P ≤ 0.01). Trends indicated potentially greater high-density lipoprotein cholesterol concentrations and high-density lipoprotein size following the full-fat yogurt diet (P ≤ 0.05). The ratio of triacylglycerols: high-density lipoprotein cholesterol concentrations was 17% lower following the full-fat yogurt diet (P < 0.01).

Conclusions

This exploratory analysis demonstrates that short-term full-fat yogurt consumption elicits beneficial effects on the blood lipid profile in individuals with prediabetes and highlights the need for further evaluation of the contribution of dairy fat in yogurt and other dairy food matrices in lipid homeostasis and metabolic health.

Trial registration

This trial is registered at clinicaltrials.gov (NCT03577119).

Supplementary Information

The online version contains supplementary material available at 10.1186/s12944-025-02616-4.

Background

Dairy foods, including milk, cheese, and yogurt, are an integral part of dietary guidelines in the United States and internationally [1, 2], as they supply a range of important and shortfall nutrients for human health [3, 4]. Full-fat dairy foods specifically are also dietary sources of saturated fatty acids (SFAs); more than two thirds of the total fatty acids in dairy fat are SFAs [5]. In the dairy cow, this SFA content arises from (i) de novo synthesis using dietary substrates, (ii) biohydrogenation of dietary unsaturated fatty acids by rumen microbes, (iii) endogenous metabolism of dietary fatty acids, and (iv) dietary SFAs [6].

Dietary guidance recommends limiting SFA intake in response to research findings from the mid-20th century that associated greater SFA consumption with greater blood cholesterol (CHOL) concentrations and thereby increased risk of cardiovascular diseases (CVD) [7]. As a result, the advice to opt for low-fat or non-fat dairy food consumption, instead of full-fat dairy foods, appears in early published dietary guidance, i.e., the 1977 Dietary Goals for the United States [8], and has remained steadfast in the Dietary Guidelines for Americans, as well as many recommendations internationally [9]. However, a positive correlation between increased SFA intake from dairy foods and greater CVD risk has not been consistently reported, suggesting that the food source of SFAs influences its metabolic effects [10]. For example, two recent meta-analyses suggested either no or inverse associations between full-fat dairy consumption and risk, incidence, or mortality of hypertension, coronary heart disease, or stroke [11, 12]. For specific blood lipids [i.e., CHOL and triacylglycerol (TAG) concentrations], observational research reports mixed effects of full-fat dairy consumption, but largely points toward a beneficial or no effect of full-fat cheese or yogurt intake on concentrations of total CHOL, low-density lipoprotein (LDL) CHOL, high-density lipoprotein (HDL) CHOL, and TAG [13–15]. Randomized-controlled trials (RCTs) have demonstrated a beneficial effect or no effect on blood lipid concentrations following the consumption of milk and cheese with different fat contents [e.g., intake of whole (2-3.5% fat) milk compared to skim (0.1–0.3% fat) milk or regular-fat (25–32%) compared to reduced-fat (13–16%) cheese] [16–18]. Further, Chen et al. [19] reported a decrease in blood TAG concentrations at 24 weeks following the consumption of full-fat yogurt (FFY; 2.9% fat), compared with the baseline, and a greater decrease in blood TAG and total CHOL concentrations following FFY intake, compared with full-fat (3.7%) milk intake. No differences in HDL or LDL CHOL concentrations were observed at 24 weeks after FFY intake in comparison to baseline concentrations or concentrations following full-fat milk intake [19].

Collectively, these results support observational research findings that appear to contradict current dietary recommendations and generate additional research questions based on the heterogeneity of the study designs. Namely, most of the RCTs used a parallel versus a crossover design, provided one to two servings per day, included participants with metabolic risk factors versus healthy participants, and had varying degrees of control regarding other diet components beyond dairy foods (none employed a controlled-feeding design). Of note, there still exists a gap in the literature pertaining to the effects of FFY intake, compared to non-fat yogurt (NFY) intake, on the blood lipid profile. It is particularly important to evaluate yogurt as an individual dairy food given the unique characteristics of its food and fat matrices. Differences in physical structure, composition, and the interaction among nutrients and bioactive compounds across dairy foods may lead to food-specific effects on cardiometabolic health [6, 20].

In this exploratory analysis, our overarching goal was to evaluate the effect of dairy fat per se, within the matrix of yogurt, on blood lipid responses in individuals with prediabetes. Perturbations in blood glucose and lipid control are both defining characteristics of the metabolic syndrome [21] and cardiovascular health problems often cluster with type 2 diabetes (T2D) [22]. Previous RCTs have included participants with risk factors for the metabolic syndrome or CVD, but here we recruited individuals with prediabetes, allowing us to focus specifically on the interplay between blood glucose and blood lipid control. Utilizing our recent randomized-controlled crossover trial comparing the effects of consuming three daily servings of either FFY (3.25% fat) or NFY [23], the objective of this exploratory analysis was to evaluate diet-induced differences in blood lipids (i.e., TAG and CHOL), lipoproteins, and apolipoproteins (Apos).

Methods

Study design

This randomized-controlled crossover trial was conducted at the University of Vermont Medical Center Clinical Research Center in Burlington, Vermont, USA. Briefly, men and post-menopausal women aged 45–75 years with a body mass index of 20–45 kg/m² were recruited. Notably, the inclusion criteria required the condition of prediabetes, defined as a fasting blood glucose concentration of 100–125 mg/dL or a hemoglobin A1C of 5.7–6.4% [24]. Further, the exclusion criteria detailed that participants must not have any other chronic diseases beyond prediabetes. A more detailed description of the study procedures, including participant recruitment, is available elsewhere [23].

Study protocol

All participants completed two, 3-week experimental diet periods, each preceded by a 1-week run-in diet, for a total of eight weeks of intervention. The run-in diet contained no yogurt and was based on the typical U.S. American diet [25–27]. The experimental diet periods were based on the Dietary Approaches to Stop Hypertension (DASH) diet [28] and included three daily servings of either FFY or NFY. All food and beverages, except for water (allowed ad libitum), were provided for the entirety of the study. As previously detailed [23], caloric needs were calculated to match individual energy requirements and participants were instructed to maintain their usual physical activity throughout the eight-week study to ensure weight maintenance. Body weight was monitored at least three times per week. The macronutrient distribution for the diets were as follows: (1) run-in diet comprised 15% protein, 40% fat, and 45% carbohydrate, (2) NFY diet comprised 15% protein, 30% fat, and 55% carbohydrate, and (3) FFY diet comprised 15% protein, 38% fat, and 47% carbohydrate. The 8% difference in fat content between the FFY and NFY diets was solely due to the fat content of the FFY. The fatty acid composition was determined via gas-liquid chromatography, following the methodology previously described in [29], with a modified initial extraction step involving a 2-hour incubation at 70 °C. All yogurt used in the trial was donated by Stonyfield Farm Incorporated (Londonderry, NH, USA). Apart from the FFY, all foods contained only trace quantities of fat, therefore, custom fat blends were created by the Clinical Research Center using plant- and animal-based fats to supplement the fat in the diet, with specific base fat selection to exclude ruminant-specific fatty acids, including odd-chain, branched-chain, and trans-fatty acids [23]. Additional details of the diet have been described previously [23]. Dietary compliance was monitored with the use of daily compliance logs as well as empty food container inspections and confirmed by analysis of FA composition of plasma total lipids via gas-liquid chromatography [23].

Endpoint testing

A schematic of endpoint testing is depicted in Fig. 1. The primary endpoints were the diet-induced effects on fasting blood lipids (total CHOL, HDL CHOL, LDL CHOL, non-HDL CHOL, and TAG concentrations as well as the ratio of total CHOL: HDL CHOL and TAG: HDL CHOL). On the last day of the first run-in period and each experimental diet (i.e., study days 8, 29, and 57), blood samples were collected after a 12-hour fast. For the analysis of fasting concentrations of blood lipids, whole blood was collected in SST BD Vacutainer^® tubes (Becton Dickinson, Franklin Lakes, NJ, USA) and the resulting serum was analyzed via colorimetric assay on the Ortho Vitros 5600 system by the University of Vermont Medical Center Laboratory (Burlington, VT, USA). For analysis of lipoprotein A concentrations, whole blood was collected in K₃ EDTA tubes (Medtronic, Minneapolis, MN, USA), centrifuged (1200 x g, 10 min), and plasma was frozen at -80 °C. The Human Lipoprotein A ELISA kit (Abcam, Cambridge, UK) was then used according to manufacturer instructions to quantify plasma lipoprotein A concentrations.

Fig. 1.

Schematic of study procedures. Blood draws measured in minutes, relative to the start of the breakfast meal. Created with BioRender.com

The secondary endpoints were diet-induced changes in concentrations of Apos as well as the concentration and TAG content of lipoprotein subclasses. On the penultimate day of the first run-in period and each experimental diet (i.e., study days 7, 28, and 56), a mixed meal tolerance test was performed. Participants were fasted for 12 h prior to receiving a standardized breakfast (see Table 1 for meal composition), specific to each study diet, to be consumed within 30 min. Blood was collected into 2 mL K₂ EDTA BD Vacutainer^® tubes at the − 10-, 30-, 60-, 120-, and 180-minute time points, in relation to the start of the breakfast meal. Subsequently, blood was centrifuged (1694 x g, 10 min, 4 °C) and plasma was aliquoted and frozen at -80 °C until Apo and lipoprotein particle subclass analysis via nuclear-magnetic-resonance-spectrometry-based biomarker analysis by Nightingale Health (Helsinki, FI). The total area under the curves (AUCs) for the Apos as well as lipoprotein subclass concentrations and contents were calculated using the trapezoid method (GraphPad Prism version 10.2.3, Boston, MA, USA).

Table 1.

Composition of mixed meal tolerance test meal per 2000 Kcal [23]

Component	Run-in diet	Non-fat yogurt diet	Full-fat yogurt diet
Energy (kcal)1	569.8	611.4	647.9
Protein (% energy)1	13.6	13.9	13.9
Fat (% energy)1	45.3	31.4	36.8
Saturated fat (% total fat)2	44.3	32.3	42.7
Monounsaturated fat (% total fat)2	40.1	48.8	42.2
Polyunsaturated fat (% total fat)2	15.6	18.9	15.1
Carbohydrate (% energy)1	41.1	54.7	49.4
Fiber, total dietary (g)1	2.0	1.9	2.1
Calcium (mg)1	139.9	367.2	334.0
Magnesium (mg)1	30.0	59.8	53.8
Potassium (mg)1	195.3	1001.8	819.8
Sodium (mg)1	828.5	478.6	512.2
Vitamin D (D2 + D3; µg)1	10.7	1.9	1.9

¹Calculated by ProNutra. ²Calculated from gas-liquid chromatography analysis

Statistical analysis

A power calculation was performed based on the primary outcome measure of the clinical trial at large [23]. Data are expressed as non-transformed means ± SEM. The linear mixed model procedure in SAS (Version 9.4, SAS Institute Inc., Cary, NC, USA) was used with values from the run-in period as a covariate to compare outcome measures following each of the experimental diets in the crossover design. In this model, day, diet, and sex were used as fixed effects, and the repeated measures were day and diet. The diet by sex interaction effect was assessed using a separate model in which diet by sex was the only fixed effect. Normality of variables was assessed using the univariate procedure in SAS and data were transformed to reach normality or improve model fit statistics as determined by a smaller absolute value of the − 2 log likelihood, Akaike Information Criterion (original and corrected), and Schwarz’s Bayesian Information Criterion statistics. Variables were transformed using the natural log transformation except those that were transformed using the square root transformation (body weight, fasting total CHOL concentration, fasting chylomicron TAG content, average VLDL particle size AUC, average LDL particle size AUC, and average HDL particle size AUC) and those that were not transformed (fasting ratio of total CHOL: HDL CHOL concentrations, fasting ApoA1 concentration, 120-minute ApoA1 concentration, fasting average VLDL particle size, 120-minute average VLDL particle size, fasting extra-large VLDL particle concentration, fasting average LDL particle size, 120-minute average LDL particle size, fasting large LDL TAG content, medium LDL TAG content AUC, small LDL TAG content AUC, fasting average HDL particle size, 120-minute average HDL particle size, and small HDL concentration AUC). Significance was determined at P ≤ 0.01 and trends were determined at a P-value less than or equal to 0.05, but greater than 0.01. The use of a more stringent P-value was chosen to minimize the risk of false positives given the smaller sample size.

Results

Participant cohort

In total, 13 participants (n = 7 female and n = 6 male), completed the study. Characteristics at screening are detailed in Table 2 and have previously been reported [23]. Body weights at the start of each experimental period did not differ by diet (88.1 ± 5.6 kg at the start of the FFY diet versus 88.2 ± 5.6 kg at the start of the NFY diet; P = 0.32).

Table 2.

Baseline blood lipid characteristics at screening (n = 13)

						Sex1
		All				Female				Male
Parameter	Unit	Mean		SEM		Mean		SEM		Mean		SEM
Total CHOL2	mg/dL	174	±	11		165	±	15		184	±	17
Triacyglycerols2	mg/dL	89	±	11		95	±	18		82	±	14
HDL CHOL2	mg/dL	58	±	5		52	±	5		64	±	8
Low-density lipoprotein CHOL3	mg/dL	98	±	9		94	±	12		103	±	14
Non- HDL CHOL2	mg/dL	116	±	10		113	±	14		120	±	14
Total CHOL: HDL CHOL Ratio3		3.2	±	0.2		3.3	±	0.4		3.0	±	0.3
Triacylglycerol: HDL CHOL Ratio3		1.8	±	0.4		2.1	±	0.6		1.5	±	0.5

¹Values expressed as mean ± SEM. ²Fasting blood measurement. ³Calculated using fasting blood measurements. Abbreviations: CHOL cholesterol, HDL high-density lipoprotein

Fasting lipid panel and lipoprotein A

Following the FFY diet, fasting TAG concentrations were 10% lower (P < 0.01; Table 3; Supplemental Table 1), and the ratio of TAG: HDL CHOL concentrations was 17% lower (P < 0.01). Fasting total CHOL, HDL CHOL, LDL CHOL, non-HDL CHOL, and lipoprotein A concentrations as well as the ratio of total CHOL: HDL CHOL concentrations did not differ by diet. However, HDL CHOL concentrations and the ratio of total CHOL: HDL CHOL concentrations trended greater (P = 0.04 and 0.02, respectively) as a result of the FFY diet.

Table 3.

Selected blood lipoprotein and apolipoprotein concentrations and characteristics following three weeks of experimental diet consumption (n = 13)

		Diet1
		NFY				FFY				P
Parameter	Unit	Mean		SEM		Mean		SEM		Diet	Sex	Day	Diet*Sex
Fasting TAG	mg/dL	100	±	12		90	±	11		< 0.01	0.22	0.72	0.42
Fasting HDL CHOL	mg/dL	50	±	4		53	±	4		0.04	0.27	0.29	0.82
Fasting Total CHOL: HDL CHOL Ratio		3.4	±	0.2		3.3	±	0.2		0.02	0.03	0.27	0.18
TAG: HDL CHOL Ratio		2.3	±	0.4		1.9	±	0.3		< 0.01	0.05	0.68	0.52
120-minute ApoA1	mg/dL	116	±	5		121	±	5		0.05	0.80	0.50	0.88
ApoA1 AUC	mg/dL x minutes	22,300	±	920		22,900	±	853		0.04	0.89	0.33	0.85
120-minute ApoB: ApoA1		0.561	±	0.035		0.537	±	0.032		0.01	0.60	0.09	0.73
ApoB: ApoA1 AUC		106	±	7		102	±	6		0.02	0.53	0.05	0.75
Chylomicron2 AUC	nmol/L x minutes	213	±	46		197	±	53		0.05	0.28	0.30	0.09
Fasting XL VLDL3	nmol/L	2.20	±	0.39		1.91	±	0.38		0.03	0.18	0.59	0.90
Fasting M VLDL4	nmol/L	23.2	±	2.0		22.0	±	1.9		0.05	0.51	0.13	0.43
Fasting M VLDL TAG	mmol/L	0.211	±	0.021		0.191	±	0.021		0.04	0.37	0.64	0.78
Fasting S VLDL5	nmol/L	26.7	±	2.0		24.9	±	1.9		0.02	0.34	0.33	0.78
120-minute S VLDL	nmol/L	27.2	±	1.8		26.0	±	1.7		0.03	0.49	0.61	0.60
S VLDL AUC	nmol/L x minutes	5120	±	349		4860	±	342		0.02	0.31	0.31	0.87
Fasting S VLDL TAG	mmol/L	0.113	±	0.009		0.101	±	0.010		0.01	0.26	0.55	1.00
S VLDL TAG AUC	mmol/L x minutes	22.4	±	1.87		21.1	±	1.97		0.05	0.15	0.51	0.67
L LDL6 TAG AUC	mmol/L x minutes	13.0	±	0.6		12.6	±	0.6		0.04	0.66	0.37	0.60
120-minute Average HDL Size	nm	9.68	±	0.07		9.72	±	0.07		0.04	0.03	0.58	0.66
Average HDL Size AUC	nm x minutes	1840	±	14		1850	±	14		0.04	0.04	0.31	0.87
120-minute XL HDL7	µmol/L	0.236	±	0.028		0.251	±	0.025		0.04	0.07	0.27	0.73
120-minute M HDL8	µmol/L	2.79	±	0.19		2.96	±	0.16		0.02	0.76	0.59	0.89
M HDL AUC	µmol/L x minutes	536	±	34		561	±	31		0.03	0.95	0.44	0.77
Fasting M HDL TAG	mmol/L	0.029	±	0.002		0.027	±	0.002		0.05	0.23	0.82	0.89
Fasting S HDL9 TAG	mmol/L	0.034	±	0.003		0.031	±	0.003		0.04	0.09	0.57	0.66

¹Data expressed as means ± standard error of the mean. ²Chylomicron particle diameter > 75 nm. ³XL VLDL average particle diameter 64 nm. ⁴M VLDL average particle diameter 44.5 nm. ⁵S VLDL average particle diameter 36.8 nm. ⁶L LDL average particle diameter 25.5 nm. ⁷XL HDL average particle diameter 14.3 nm. ⁸M HDL average particle diameter 10.9 nm. ⁹S HDL average particle diameter 8.7 nm. Abbreviations: Apo Apolipoprotein, AUC Area under the curve, CHOL Cholesterol, FFY Full-fat yogurt, HDL High-density lipoprotein, L Large, LDL Low-density lipoprotein, M Medium, NFY Non-fat yogurt, S Small, TAG Triacylglycerol, VLDL Very low-density lipoprotein, XL Extra-large, XS Extra-small

Apolipoproteins

ApoB concentrations did not differ by diet (Supplemental Table 2). ApoA1 concentrations trended lower at the 120-minute time point (P = 0.05) and for the AUC (P = 0.04) in response to the NFY diet, but there was no difference observed at fasting. Similarly, there was no difference in the ratio of ApoB: ApoA1 concentrations at fasting or for the AUC, but this ratio was lower in response to the FFY diet at the 120-minute time point (0.537 compared with 0.561; P = 0.01; Table 3).

Very low-density lipoprotein subclasses

The average VLDL particle size did not differ by diet (Supplemental Table 3). Small VLDL particle (average diameter: 36.8 nm) concentrations were not different by diet, but the TAG content in small VLDL particles was 11% lower as a result of the FFY diet at fasting (P = 0.01) but was not different at the 120-minute time point or for the AUC between diets. There was no difference by diet for extra-small VLDL particle (average diameter: 31.3 nm) concentrations. However, the TAG content in extra-small VLDL particles was lower following the FFY diet at fasting (8% lower; P < 0.01; Fig. 2A), at the 120-minute time point (4% lower; P < 0.01; Fig. 2B), and for the AUC (6% lower; P = 0.01; Fig. 2C). No differences by diet were observed for chylomicrons and extremely-large VLDL, extra-large VLDL, large VLDL, and medium VLDL particle concentrations or their respective TAG contents. Trends observed support these results and indicated potentially lower concentrations and TAG contents of VLDL subclasses in response to the FFY diet (Table 3).

Fig. 2.

Triacyclglycerol (TAG) content in extra-small (XS) very low-density lipoprotein (VLDL) at fasting (panel A), the 120-minute time point (panel B), and for the area under the curve (AUC; panel C) as well as in intermediate-density lipoprotein (IDL) at fasting (panel D), the 120-minute time point (panel E), and for the AUC (panel F) as a result of consuming a diet with non-fat yogurt (NFY) or full-fat yogurt (FFY). Values are expressed individually and as mean ± SEM from data collected after three weeks of the experimental diet. Data were analyzed using a linear mixed model with the fixed effects of day, diet, and sex, and the repeated measures of day and diet. Created using GraphPad Prism (version 10.2.3, Boston, MA, USA)

Intermediate-density lipoproteins

IDL particle (average diameter: 28.6 nm) concentrations did not differ by diet (Supplemental Table 4). The TAG content in IDL particles, however, was lower as a result of the FFY diet at fasting (5% lower; Fig. 2D), the 120-minute time point (4% lower; Fig. 2E), and for the AUC (4% lower; Fig. 2F; P = 0.01 for all).

Low-density lipoprotein subclasses

Average LDL particle sizes did not differ by diet (Supplemental Table 5). Large LDL particle (average diameter: 25.5 nm) concentrations were also not different by diet. The TAG content in large LDL particles trended lower in response to the FFY diet at fasting (P = 0.02; Fig. 3) and for the AUC (P = 0.04; Table 3), but not at the 120-minute time point. No diet-induced changes were observed for the concentrations of medium LDL particles (average diameter 23.0 nm), but the TAG content in medium LDL particles was approximately 9% lower as a result of the FFY diet at fasting only (P < 0.01; Fig. 3), not at the 120-minute time point or for the AUC. Concentrations of small LDL particles (average diameter: 18.7 nm) and their TAG contents followed the same pattern: no difference by diet for small LDL particle concentrations, and approximately 9% lower TAG content in small LDL particles following the FFY diet at fasting (P < 0.01; Fig. 3), but not the 120-minute time point or for the AUC.

Fig. 3.

Triacylglycerol (TAG) content in large (L; panel A), medium (M; panel B), and small (S; panel C) low-density lipoprotein (LDL) as a result of consuming a diet with non-fat yogurt (NFY) or full-fat yogurt (FFY). Values are expressed individually and as mean ± SEM from data collected after three weeks of the experimental diet. Data were analyzed using a linear mixed model with the fixed effects of day, diet, and sex, and the repeated measures of day and diet. Created using GraphPad Prism (version 10.2.3, Boston, MA, USA)

High-density lipoproteins subclasses

The average HDL particle size did not differ by diet (Supplemental Table 6), nor did the concentration of any HDL subclasses or the TAG contents of the subclasses. Observed trends indicated potentially greater HDL particle sizes and subclass concentrations as well as lower TAG contents in HDL subclasses as a result of the FFY diet (Table 3).

Discussion

Our objective was to evaluate differences in the blood lipid profile following the consumption of a diet with FFY, compared with NFY, in individuals with prediabetes. The FFY diet resulted in beneficial effects on the blood TAG concentration, the TAG content of specific VLDL, IDL, and LDL subclasses, and the ratio of TAG: HDL CHOL concentrations. These results align with the findings of observational studies that largely demonstrate beneficial or no effects of FFY consumption on CVD outcomes [11, 13–15].

Our primary finding was that fasting blood TAG concentrations were 10% lower after the consumption of the FFY diet, compared with the NFY diet. Hypertriglyceridemia is a risk factor for T2D development [30, 31]; thus, a lower fasting blood TAG concentration indicates a particularly substantial improvement in metabolic health for these individuals. This differs from the results of other RCTs that did not show effects of dairy fat consumption on TAG concentration, both in studies focused on specific dairy foods [16, 32] and total intake of low- or high-fat dairy foods [33, 34]. However, Chiu et al. [35] also reported approximately 13% lower fasting plasma TAG concentrations following consumption of a diet with full-fat dairy foods (primarily milk, cheese, and yogurt), compared with the same diet, instead comprised of low-fat dairy foods. Notably, the dairy foods in the aforementioned study were consumed within the DASH diet, just as the experimental diets in our current study were based on the DASH diet. Traditionally, the DASH diet promotes the consumption of low- or non-fat dairy products and specifically discourages full-fat dairy consumption [28]. Therefore, we highlight the findings from Chiu et al. [35] and our study, as both studies demonstrate improvements in blood lipid profiles when the DASH diet was modified to substitute full-fat dairy foods in place of low- or non-fat dairy alternatives. Chiu et al. [35] also hypothesized that their findings may be in part due to the lower carbohydrate content of the high-fat dairy diet, based on the 12% difference between their experimental diets and the relationship between high carbohydrate intake and greater blood TAG concentrations [36]. In our study, the difference in carbohydrate contents between the diets was 8%, thus, this difference may also influence the difference in blood TAG concentrations. Previous research has demonstrated a decrease in TAG as a result of replacing carbohydrates with SFAs from butter [37], but not cheese [18, 37, 38], and research is limited for the yogurt matrix. Together, the substitution of full-fat dairy foods in place of low- or non-fat dairy foods can yield beneficial effects on the blood lipid profile, a result that may be aided, at least in part, by a simultaneous decrease in dietary carbohydrate content.

In the specific lipoprotein subclasses, there were lower TAG contents in smaller VLDL and LDL particles as well as IDL particles in response to the FFY diet. Chylomicrons and larger VLDL species are the most TAG-rich lipoproteins, and increases in these particles are associated with increased plasma TAG concentrations [39]. However, no diet-induced differences were observed in the TAG content of chylomicrons or larger VLDL species. Recently, the TAG content of LDL particles has been gaining recognition as it may have important implications for cardiovascular health [40, 41]. The TAG content in LDL particles has been positively associated with many, but not all, CVD outcomes, including atherosclerosis [40, 41], coronary heart disease [40, 42, 43], and peripheral artery disease [44]. Specifically, the TAG content of smaller LDL particles has also been positively associated with certain aspects of CVD [45–49], such as coronary heart disease risk. Moreover, the TAG contents of LDL particles [50], and specifically smaller LDL particles as well [30, 51, 52], have been associated with increased T2D risk and incidence. Thus, the lower TAG contents of smaller LDL particles observed following the consumption of the FFY diet indicates a beneficial impact on the blood lipid profile. Similarly, the TAG contents in smaller VLDL particles and IDL particles have been positively associated with cardiometabolic disease risk [30, 45–49, 51, 52]. Therefore, the lower TAG contents in these lipoprotein particles observed in our study also suggest benefits of the FFY diet. Hidaka et al. [32] measured the TAG contents in LDL and VLDL particles and reported no differences following the consumption of whole (3.6% fat) or non-fat (0.1%) milk. There are many differences between this study and our study, including, but not limited to, the dairy food matrix (milk versus yogurt), food fermentation status, study design (parallel versus crossover), length of treatment (two versus three weeks), and diet considerations (controlled dairy food versus controlled diet), thus future research evaluating the interaction of these methodological considerations is needed. Collectively, the lower TAG contents of specific lipoprotein particles observed in response to FFY intake indicate beneficial effects on metabolic health.

Suggestions for potential mechanisms for the decrease in TAG concentration and contents have predominantly focused on the actions of dairy polar lipids (i.e., phospholipids and sphingolipids) [53–55], and, to a lesser extent, on odd- and branched-chain fatty acids [55–57]. Reis et al. [58] demonstrated in a mouse model that dairy polar lipids may reduce blood TAG concentrations by reducing de novo hepatic lipid synthesis. However, a recent meta-analysis of RCTs did not report differences in blood TAG concentration as a result of milk fat globule membrane phospholipid consumption, despite the range of dietary phospholipids provided [59]. Observational studies have reported inverse associations between proportions of odd-chain fatty acids in blood (i.e., plasma or serum phospholipids and erythrocyte membranes) and blood TAG concentrations [60–63]. Moreover, Venn-Watson et al. [57] demonstrated that supplementing diets of rabbits with diet-induced hypercholesterolemia, anemia, and nonalcoholic steatohepatitis with 15:0 resulted in lower blood TAG concentrations, but the mechanisms are yet to be clarified. Blood iso-branched-chain fatty acid proportions were also inversely correlated with blood TAG concentrations in a cross-sectional study [64]. Follow-up in vitro research in the hepatocarcinoma cell line HepG2 demonstrated that incubation with the branched-chain fatty acid 16:0-iso reduced transcription of genes relating to fatty acid synthesis, suggesting iso-branched-chain fatty acids may decrease hepatic TAG production [65]. Taken together, accumulating evidence indicates that dairy fat components may lower blood TAG concentration via reduced hepatic TAG production. Additionally, there is the potential for their combination within the dairy fat matrix to produce a synergistic effect to lower blood TAG concentrations, thus future mechanistic research evaluating the dairy-fat matrix is warranted.

We did not observe diet-induced effects on the fasting concentrations of total, HDL, or LDL CHOL, which aligns with previous RCTs [16, 18, 33–35, 66]. However, there was a trend toward 6% greater HDL CHOL concentrations, along with greater HDL particle sizes following the consumption of the FFY diet. Blood HDL CHOL concentrations have been inversely related to T2D risk [67, 68], making these results particularly relevant to our trial of individuals with prediabetes and a focus for further investigation. Engel et al. [16] reported an increase in HDL CHOL concentration in adults following the consumption of whole (3.5%) milk, compared with skim (0.1%) milk, while Raziani et al. [18] did not report differences in HDL CHOL concentrations between regular-fat (25–32%) and reduced-fat (13–16%) cheese intake. Only the trial conducted by Engel et al. [16] used a crossover design and neither used yogurt nor controlled other aspects of the diet (i.e., beyond the dairy foods provided), thus there is still much to learn about the relationship between FFY intake and HDL CHOL concentrations. The functionality and size of HDL particles are also gaining attention as HDL is a heterogenous lipoprotein class [69–71] and HDL size has been shown to be inversely related with T2D development [72, 73]. Brassard et al. [74] reported an increase in HDL CHOL efflux capacity after butter intake, compared with cheese intake (both diets had comparable total fat and SFA contents), but did not report differences in HDL particle size or HDL subclass concentrations. Together, the trends we have observed suggest that FFY intake may exert beneficial effects on HDL CHOL concentration as well as HDL particle size and functionality, but further research is needed to support these findings.

Though the difference in the HDL CHOL concentrations did not reach statistical significance, we did observe a lower ratio of TAG: HDL CHOL concentrations following the FFY diet. This ratio is a more recently proposed indirect marker of insulin resistance, drawing from the risk factors of high TAG and low HDL CHOL concentrations that accompany the metabolic syndrome and T2D [21, 75]. There appear to be strong race, ethnicity, and sex differences, thus cut-off points for a desirable ratio are yet to be solidified, but a lower ratio is proposed to indicate greater insulin sensitivity [76, 77] and a lower T2D risk [78]. Herein, we demonstrate that the consumption of FFY, compared with the consumption of NFY, resulted in a lower ratio of TAG: HDL CHOL concentrations, indicating potential benefits to not only blood lipid metabolism, but also blood glucose metabolism. This aligns with our previous findings that FFY intake beneficially affected aspects of whole-body glucose handling [23]. As dyslipidemia and insulin resistance form an interconnected feedback loop of metabolic dysfunction [79], the observed benefits of FFY intake on T2D risk reduction may be multifaceted, supporting both lipid regulation and glycemic control. To our knowledge, we are among the first to evaluate changes in this ratio in response to dairy fat consumption in an RCT, but our results support the findings of Yuan et al. [80] that demonstrate that a lower ratio of TAG: HDL CHOL concentrations was associated with greater intake of SFAs from dairy foods. We encourage additional investigation using the ratio of TAG: HDL CHOL concentrations to further our understanding of dairy fat on glucose and lipid homeostasis, as well as their intersectionality, particularly in individuals with prediabetes.

Beyond the methodological strengths of the trial design [23], the level of detail on the content and composition of the 14 lipoprotein subclasses provided by the biomarker analysis is a key strength of the study. Additionally, the study sample featured participants with prediabetes, which provides insight into a particularly vulnerable population due to the increased risk of CVD experienced by individuals with prediabetes and T2D [81, 82]. The primary limitation for this trial was the small sample size because of the direct and indirect impacts of the COVID-19 pandemic [23]. The beneficial effects of FFY intake on the blood lipid profile demonstrated here justify the need for additional research with a larger sample size. Finally, the generalizability of our results is limited due to our specific recruitment of individuals with prediabetes, thus, RCTs with diverse participant cohorts are also warranted.

Conclusions

Our preliminary results demonstrate lower blood TAG concentrations, TAG contents of specific VLDL, IDL, and LDL subclasses, and ratio of TAG: HDL CHOL concentrations in response to short-term intake of FFY, compared with NFY, in individuals with prediabetes. Moreover, trends toward greater HDL CHOL concentration and particle size were observed following the FFY diet, thus this relationship is worth exploring in a larger sample of participants. Analysis of lipoprotein subclass concentration and content is not yet common practice but provided our trial with additional insight and avenues to evaluate the effect of FFY intake on the blood lipid profile. Further, the lower ratio of TAG: HDL CHOL concentrations as a result of the FFY diet reinforces the relationship between glucose and lipid metabolism and highlights the specific benefits for individuals with prediabetes. Collectively, our work supports epidemiological findings that uncouple the intake of full-fat dairy foods from detrimental metabolic health effects. Additional clinical research on FFY as well as other full-fat dairy foods can bolster the evidence base available to elucidate the effect of dairy fat on metabolic health.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

AUC

Area under the curve

Apo

Apolipoprotein

CHOL

Cholesterol

CVD

Cardiovascular diseases

DASH

Dietary Approaches to Stop Hypertension

FFY

Full-fat yogurt

HDL

High-density lipoprotein

IDL

Intermediate-density lipoprotein

NFY

Non-fat yogurt

LDL

Low-density lipoprotein

RCT

Randomized-controlled trial

SFA

Saturated fatty acid

TAG

Triacylglycerol

T2D

Type 2 diabetes

VLDL

Very low-density lipoprotein

Author contributions

Research design: JK, MEP, CLK; Research conduct: VMT, JK, SE; Data analysis: VMT; Physician of record: MPG; Manuscript preparation: VMT, SE, MPG, MEP, CLK, JK; Primary responsibility for final content: JK. All authors read and approved the final manuscript.

Funding

This project was funded by the National Dairy Council, Vermont Dairy Promotion Board, US Department of Agriculture National Institute of Food and Agriculture (2022-67011-36572), and the University of Vermont Food Systems Research Center.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This study was approved by the University of Vermont Institutional Review Board (18–0295) and informed consent was obtained from all participants prior to study procedures.

Competing interests

The authors declare no competing interests.

Consent for publication

Not applicable.