Replacing dairy fat with rapeseed oil causes rapid improvement of hyperlipidaemia: a randomized controlled study

Abstract

Abstract.  Iggman D, Gustafsson I-B, Berglund L, Vessby B, Marckmann P, Risérus U (Clinical Nutrition and Metabolism, Uppsala University, Uppsala; Center for Clinical Research Dalarna, Falun; Örebro University, Grythyttan, Sweden; Odense University Hospital and Institute of Clinical Research, University of Southern Denmark, Odense, Denmark). Replacing dairy fat with rapeseed oil causes rapid improvement of hyperlipidaemia: a randomized controlled study. J Intern Med 2011; 270: 356–364.

Background.  Rapeseed oil (RO), also known as canola oil, principally contains the unsaturated fatty acids 18:1n-9, 18:2n-6 and 18:3n-3 and may promote cardiometabolic health.

Objective.  To investigate the effects on lipoprotein profile, factors of coagulation and insulin sensitivity of replacing a diet rich in saturated fat from dairy foods (DF diet) with a diet including RO-based fat (RO diet).

Design.  During a 2 × 3-week randomized, controlled, cross-over trial, 20 free-living hyperlipidaemic subjects were provided with isocaloric test diets that differed in fat composition alone. Blood lipoprotein profile, coagulation and fibrinolytic factors and insulin sensitivity (euglycaemic clamp) were determined before and after the dietary intervention.

Results.  All subjects completed the study, and compliance was high according to changes in serum fatty acids. The RO diet, but not the DF diet, reduced the levels of serum cholesterol (−17%), triglycerides (−20%) and low-density lipoprotein cholesterol (−17%), cholesterol/high-density lipoprotein (HDL) cholesterol ratio (−21%), apolipoprotein (apo) B/apo A-I ratio (−4%) and factor VII coagulant activity (FVIIc) (−5%) from baseline. These changes were significantly different between the diets (P = 0.05 to P < 0.0001), except for FVIIc (P = 0.1). The RO diet, but not the DF diet, modestly increased serum lipoprotein(a) (+6%) and tended to increase the glucose disappearance rate (K-value, +33%). HDL cholesterol, insulin sensitivity, fibrinogen and tissue plasminogen activator inhibitor-1 levels did not change from baseline or differ between the two diets.

Conclusions.  In a diet moderately high in total fat, replacing dairy fat with RO causes a rapid and clinically relevant improvement in serum lipoprotein profile including lowering of triglycerides in hyperlipidaemic individuals.

Introduction

Replacing saturated fat (SFA) with monounsaturated fat (MUFA) and polyunsaturated fat (PUFA) from vegetable oils has been associated with reduced cardiovascular disease (CVD) events in clinical trials [1]. Improvement in blood lipid profile has been suggested to be a central mechanism behind the cardioprotective effects of PUFA compared with SFA in these trials [2], supported by results showing more robust risk reductions in those trials in which dietary fat modification caused greatest reductions in serum cholesterol [2, 3]. Epidemiological studies have, however, often failed to show inverse associations between SFA from dairy foods and CVD [4]. Other variables associated with CVD risk are insulin resistance and haemostatic factors. The results of the KANWU study showed impaired insulin sensitivity after 3 months on a diet high in SFA, compared with a MUFA-rich diet [5]. Regarding effects on coagulation, factor VII levels have been influenced by SFA, MUFA and PUFA intake, though not entirely consistently [6], and beneficial effects on factor VII levels have so far primarily been demonstrated for olive oil [7].

A possible advantage of rapeseed (or canola) oil (RO) is the alpha-linolenic acid (ALA) content (∼11%), in addition to linoleic acid (LA, ∼19%) and oleic acid (∼56%). Whereas LA has been inversely related to both CVD and diabetes [8–11], sufficient amounts of ALA may also be important for reducing CVD risk [12, 13]. The results of the Lyon Diet Heart Study demonstrated substantial risk reduction in cardiovascular mortality after adherence to a Mediterranean diet rich in MUFA and ALA, as compared with a Western diet rich in SFA and low in vegetable fats [13]. In this trial, margarine enriched with RO was provided to participants as a part of the intervention diet [13]. RO may exert even more favourable effects than olive oil with regard to blood lipid profile [14]. Whereas attention has focused on the potentially beneficial effects of olive oil, data from controlled studies investigating the metabolic and clinical effects of RO in comparison with SFA are limited. Also, the metabolic effects of an SFA diet based on dairy fat (DF) are of great interest because high-fat dairy foods have been proposed to be less atherogenic than other SFA-rich foods [4]. The aim of this strictly controlled cross-over study was to investigate the effect of replacing DF with RO for 3 weeks on blood lipids, glucose metabolism and coagulation factors in hyperlipidaemic weight-stable subjects.

Materials and methods

Study design

This was a 3-week, randomized, controlled, two-period, cross-over intervention with an intermediate wash-out period of 3 weeks (Fig. 1). All subjects were free-living during the study. Individuals were assessed at baseline and thereafter received a DF diet and an RO diet in a randomized order. Thus, all subjects took part in the study for 9 weeks in total, i.e. three consecutive 3-week periods. The primary outcome measures were serum blood lipid levels, and secondary outcome measures were insulin sensitivity and coagulation factors. The sample size was calculated based on lipid-lowering effects, rather than insulin sensitivity or the glucose disappearance coefficient (K-value).

Figure 1.  Flow chart showing the enrolment of participants (14 men and six women) and completion of the cross-over study.

Subjects

The subjects, 14 men and six women, were recruited from an ongoing health survey at a Swedish telephone company [15]. Men and women, 25–68 years of age, without a diagnosis of serious disease and not using any prescription drugs were eligible for study inclusion. All participants gave written informed consent and the study was approved by the Ethics Committee at Uppsala University, Sweden.

Dietary intervention

All foods were provided to participants. The diets were planned and calculated by the research dietician and cooked at a metabolic ward kitchen from where participants collected all the foods that were consumed during the test periods. The menu was planned with ordinary food, i.e. meals commonly consumed in the population and without functional foods or supplements, as a 7-day menu including breakfast, lunch and dinner with two snacks in between. The energy intake requirements were estimated to be 30 kcal or 126 kJ kg⁻¹ body weight for women and 35 kcal or 147 kJ kg⁻¹ body weight for men. If participants lost weight during the intervention, they were instructed to consume more bread with the diet-specific spread (i.e. butter or margarine). All additional foods were reported and included in the calculated nutrient intake (Table 1).

Table 1.   The calculated daily intake of energy and nutrients at baseline and during the two test periods^a

Nutrients	Baseline		DF diet		RO diet
	LSM	SD	LSM	SD	LSM	SD
^aLSM, least square mean; E% percentage energy; DF, dairy fat; RO, rapeseed oil. ^b P < 0.001 for difference between diets (anova). ^c P < 0.05 for difference between diets (anova). Baseline dietary and energy data were calculated from 7-day weighed food records at baseline and from food and nutrient tables from the Swedish Food Adminstration. Data from the test diets were calculated in the same way but the fat content was also determined by NMR. All meals were provided for participants during the test periods. These data have been previously reported [28].
Energy (MJ)	7.7	1.7	11.4	1.6	11.3	1.5
Energy (kcal)	1860	405	2720	395	2700	365
Protein (E%)	15.8	1.6	13.8	0.4	13.5^b	0.3
Total fat (E%)	34.5	3.4	35.7	1.0	36.1^c	1.0
Saturated fat (E%)	14.2	1.5	18.8	0.4	7.8^b	0.2
Monounsaturated fat (E%)	12.5	1.3	10.7	0.4	16.2^b	0.5
Trans fatty acids (E%)			0.9	0.2	0.8	0.2
Polyunsaturated fat (E%)	5.3	1.0	3.6	0.2	8.7^b	0.3
18:3n-3 (E%)	0.6	0.1	0.4	0.1	2.2	0.3
20:5–22:6n-3 (E%)	0.2	0.2	0.2	0.0	0.2	0.0
Carbohydrates (E%)	46.9	5.3	49.2	0.9	49.2	0.9

The diets were planned to reflect the habitual Swedish diet, containing about 50% of energy from carbohydrates, 35% from fat and 14% from protein. The two diets contained equal amounts of macronutrients and energy, dietary fibre and cholesterol; they differed only in fatty acid composition. The DF diet included butter, cream and high-fat cheese. The RO diet was planned with the same food as the DF diet but was prepared with a special spread high in RO (a liquid margarine made of RO and pure canola oil), resulting in a diet high in MUFA with relatively high amounts of ALA and moderate amounts of LA (Tables 1 and 2).

Table 2.   The analysed fatty acid composition (% of total) of the dairy fat (DF)-rich diet and the rapeseed oil (RO)-rich diet^a

Fatty acid (% of total)	DF diet		RO diet
	LSM	SD	LSM	SD
^aLeast square mean (LSM) of 7-day menu as analysed in 14 duplicate portions. This data has been previously reported [28].
8:0 (Caprylic acid)	0.6	0.2	0.1	0.0
10:0 (Capric acid)	2.4	0.3	0.4	0.0
12:0 (Lauric acid)	3.5	0.5	0.9	0.2
14:0 (Myristic acid)	10.1	0.8	1.9	0.3
16:0 (Palmitic acid)	29.4	1.1	14.2	2.1
16:1n-7 (Palmitoleic acid)	1.7	0.2	0.8	0.3
18:0 (Stearic acid)	11.3	0.7	5.3	0.5
18:1n-9 (Oleic Acid)	30.6	1.7	49.3	1.1
18:2n-6 (Linoleic acid)	7.6	0.6	18.7	0.8
18:3n-3 (α-Linolenic acid)	2.0	0.3	7.6	0.4
20:3n-6 (Dihomo-γ-linolenic acid)	0.1	0.0	–	–
20:4n-6 (Arachidonic acid)	0.2	0.0	0.2	0.0
20:5n-3 (Eicosapentaenoic acid)	0.5	0.2	0.5	0.3
22:5n-3 (Docosapentaenoic acid)	0.2	0.0	0.2	0.0
22:6n-3 (Docosahexaenoc acid)	0.7	0.4	0.6	0.5

Dietary assessment and food analyses

Before random assignment to diet, subjects were required to weigh and record all food and drink consumed for 7 consecutive days to assess their habitual dietary intake. All subjects received instructions on how to record their dietary intake including correct use of scales. Nutrient intakes were calculated using a computer database (MATs, Rudans Lättdata, Västerås, Sweden) with nutrient tables from the Swedish Food Adminstration. To further assess compliance with the dietary fat interventions, changes in the fatty acid composition of plasma cholesterol esters and phospholipids were determined. Subjects were encouraged to maintain their usual lifestyle habits and not undertake any changes in diet or physical activity during and between the test periods. To achieve accurate quantitative data of the fat content of the test diets, all foods were chemically analysed using nuclear magnetic resonance (NMR) imaging (Table 2).

Laboratory methods

The fatty acid composition of the diets (Table 2) was analysed using the NMR technique, and the fatty acid composition of plasma cholesterol esters and of phospholipids were determined by gas–liquid chromatography as previously described [16]. Anthropometric measurements and blood sampling were carried out after an overnight fast. Triglyceride and cholesterol concentrations were measured in serum and in high-density lipoproteins (HDLs) by enzymatic methods, using the IL Test Cholesterol Triander’s method 181618-80 and IL Test Enzymatic-Colorimetric-Method 181709-00 for use in a Monarch 2000 centrifugal analyser (Instrumentation Laboratory, Lexington, MA, USA). Very low-density lipoproteins, low-density lipoproteins (LDLs) and HDLs were isolated using a combination of preparative ultracentrifugation [17] and precipitation with sodium phosphotungstate and magnesium chloride solution [18]. The concentrations of serum apolipoprotein (apo) B and A-I were determined by immunoturbidimetry using the Monarch apparatus with monospecific polyclonal antibodies against apo B and A-I (Orion, Espoo, Finland). The samples were preincubated prior to the assay, as suggested by DaCol and Kostner [19]. Lipoprotein(a) (Lp(a)) was measured using the Pharmacia apo(a) radioimmunoassay method (Pharmacia Diagnostics AB, Uppsala, Sweden).

Insulin sensitivity was determined by hyperinsulinaemic euglycaemic clamp according to the method of DeFronzo et al. [20], with some modifications as described previously [21]; insulin was infused at a constant rate of 56 mU/(min per m²). The insulin sensitivity index (M I⁻¹) was calculated as the glucose infusion rate (M) during the second h (mg kg⁻¹ body weight per min) divided by the mean insulin concentrations (I) during the same period (100*mg kg⁻¹ body weight min⁻¹/[mU l⁻¹]). Glucose tolerance was assessed by calculating the glucose disappearance rate (K-value) after an intravenous glucose tolerance test. A 25% glucose solution (25 mg) was rapidly injected. Capillary blood glucose was measured every 5 min over a duration of 70 min, and values were plotted semi-logarithmically against time. The near-straight line obtained after a 25-min equilibration period was used to calculate the K-value (% per min), i.e. the disappearance rate of glucose [22]. Blood glucose concentration was measured using the glucose oxidase assay [23].

The activity of tissue plasminogen activator inhibitor-1 (PAI-1) was measured using the Spectolyse/pL kit (Biopool AB, Umeå, Sweden) with stimulation by polylysine [24]. Plasma fibrinogen concentrations were assessed by a modified Clauss assay [25]. Plasma factor VII coagulant activity (FVIIc, expressed in %) was measured in a one-stage clotting assay using factor VII-deficient plasma (Biopool AB, Umeå, Sweden) [26].

Statistical analysis

Data are presented as means ± SD. The normality of the variables was tested using the Shapiro–Wilk’s test. Skewed variables (W < 0.95) were log-transformed before analysis. If variables were still skewed, nonparametric testing (Wilcoxon test) was performed. For continuous variables with normal distribution (after log-transformation if necessary), an anova model with factors for patient, time-point, sequence, sequence by time-point interaction (= dietary intervention), was used. A test for carry-over effects was carried out according to Jones and Kenward [27]. Least square means, which take into account imbalances between groups, formed the basis for all tests and estimates in the analyses. The statistical analyses were carried out by a statistician using SAS software, version 9.1 (SAS Institute, Cary, NC, USA). All tests were two-tailed, and probability values <0.05 were considered statistically significant.

Results

The mean age (SD) amongst participants was 50.9 ± 10.0 years. None of the patients dropped out during the trial (Fig. 1), and there were no significant carry-over effects.

Dietary compliance

The serum fatty acid composition changed according to the composition of the diets, indicating good compliance. Following 3 weeks of the RO diet, serum levels of 15:0, 16:0, 17:0, 18:0, 16:1n-7 and 18:1n-9 were lower, whereas LA and ALA levels were significantly higher compared with the DF diet. We have previously reported these results in further detail [28].

Effects on body weight

At baseline, body weight (mean ± SD) was 77.5 ± 9.9 kg and body mass index (BMI) was 26.3 ± 2.7 kg m⁻². After 3 weeks of the RO diet and the DF diet, body weight was 76.9 ± 9.3 and 77.0 ± 9.7 kg and BMI was 26.2 ± 2.4 and 26.2 ± 2.5 kg m⁻², respectively. Thus, there was a slight, but similar decrease in mean body weight and BMI as a result of both diets (p = 0.96 for difference between groups). All data have been adjusted for changes in body weight; however, this only marginally influenced the effects on outcome measures.

Effects on blood lipids and lipoproteins

The concentrations of serum total cholesterol (P < 0.0001), LDL cholesterol (P < 0.0001), the ratio of total/HDL cholesterol (P < 0.0001), apo B/A-I ratio (P = 0.002) and triglycerides (P = 0.02) were lower at the end of the period of the RO diet, compared with the DF diet, whereas Lp(a) was lower after the DF diet (P = 0.02) (Table 3). HDL cholesterol levels were not affected by either diet (P = 0.75 for difference between diets).

Table 3.   Serum lipid concentrations before and after 3 weeks on the dairy fat (DF)-rich and the rapeseed oil (RO)-rich diets

	Baseline Mean	SD	DF diet Mean	SD	Percentage change %	RO diet Mean	SD	Percentage change %	P-value for changes between the diets
Data are least squares mean ± SD. *P < 0.05 compared to baseline values; **P < 0.01 compared to baseline values; ***P < 0.001 compared to baseline values. HDL, high-density lipoprotein; VLDL, very low-density lipoproteins; LDL, low-density lipoproteins.
Total cholesterol (mmol L^−1)	6.70	1.20	6.66	1.04	−1	5.59***	0.91	−17***	<0.0001
Triglycerides (mmol L^−1)	2.21	1.19	2.03	0.73	−8	1.77*	0.82	−20*	<0.05
VLDL cholesterol (mmol L^−1)	0.70	0.57	0.67	0.35	−4	0.53	0.35	−24	<0.05
VLDL triglycerides (mmol L^−1)	1.54	1.06	1.43	0.70	−7	1.25	0.79	−19	0.09
LDL cholesterol (mmol L^−1)	4.76	1.13	4.91	1.01	+3	3.95***	0.91	−17***	<0.0001
LDL triglycerides (mmol L^−1)	0.52	0.15	0.50	0.12	−4	0.42***	0.12	−19***	<0.001
HDL cholesterol (mmol L^−1)	0.98	0.31	1.01	0.34	+3	1.02	0.29	+4	0.75
HDL triglycerides (mmol L^−1)	0.13	0.08	0.11	0.05	−15	0.11	0.05	−15	1.00
LDL/HDL cholesterol	5.16	1.63	5.17	1.63	0	4.11***	1.38	−20***	<0.0001
Total/HDL cholesterol	7.33	2.41	7.03	2.00	−4	5.80***	1.49	−21***	<0.001
Apo B (g L^−1)	1.21	0.20	1.19	0.11	−2	1.05***	0.10	−13***	<0.0001
Apo A-I (g L^−1)	1.31	0.21	1.24**	0.18	−5**	1.20***	0.20	−8***	0.05
Apo B/Apo A-I	0.94	0.18	0.98	0.17	+4	0.90	0.15	−4	<0.01
Lp(a) (U L^−1)	172	167	154	126	−10	182*	146	+6*	<0.05

Effects on glucose metabolism and coagulation

There were no differences between the diets with regard to insulin sensitivity or fasting glucose concentrations. The K-value for glucose tolerance did however increase (+33%) significantly compared with baseline values after 3 weeks of the RO diet (Table 4). FVIIc was decreased after the RO diet (−5%) compared with baseline, but no significant differences between diets were observed for FVIIc (P = 0.1). Neither fibrinogen nor PAI-1 concentrations changed significantly or differed between diets (Table 4).

Table 4.   Glucose metabolism and coagulation factors in plasma before and after 3 weeks on the dairy fat (DF)-rich and the rapeseed oil (RO)-rich diets

	Baseline Mean	SD	DF diet Mean	SD	Percentage change	RO diet Mean	SD	Percentage change
Least squares mean ± SD n = 20 (six women, 14 men). *P < 0.05 compared to baseline. ^aGlucose infusion rate between 60 and 120 min during the clamp divided by insulin concentrations.
Insulin sensitivity (M/I^a)	5.6	2.3	5.6	2.2	0	5.7	2.1	2
K-value (% per min)	1.2	0.4	1.4	0.7	17	1.6*	1.0	33
Fasting blood glucose (mmol L^−1)	5.8	1.1	5.5*	1.0	−5	5.5*	0.8	−5
Fibrinogen (μmol L^−1)	7.30	1.29	7.55	1.22	+3	7.55	0.95	+3
Factor VIIc (% act.)	109.6	21.8	108.2	21.0	−1	103.7*	21.1	−5
PAI-1 activity (kIU t-PA L^−1)	31.0	28.0	26.2	14.0	−15	26.3	15.4	−15

Discussion

The results of this randomized controlled study demonstrate that replacing SFA from high-fat dairy foods with unsaturated fats from RO for 3 weeks causes clinically relevant beneficial effects on the serum lipid profile in weight-stable hyperlipidaemic subjects. However, there were no significant differences in insulin sensitivity or coagulation factors between groups.

Few strictly controlled studies have compared RO with DF with respect to lipid and glucose metabolism in hyperlipidaemic subjects. In previous controlled trials, it was shown that replacing SFA with MUFA or PUFA lowers LDL cholesterol and the total/HDL cholesterol ratio [2], although the degree of change may depend on the type of food investigated and the specific fatty acid profile of the test diets. It has been suggested that RO (or canola oil), based on its fatty acid profile, may be optimal with regard to metabolic health effects [29], and RO has shown more favourable effects on blood lipid levels, compared with olive oil [14, 30], sunflower oil [31], and palm and soybean oil [32]. In the latter study, when 20% of daily energy intake as vegetable SFA (palm oil) was replaced with RO for 5 weeks, LDL cholesterol levels decreased by 18%, which is in line with results from the present study, despite the fact that slightly less fat was substituted in that study [32]. The reduction from baseline values in total (17%) and LDL (17%) cholesterol levels was greater in the present study than in a similar Swedish 4-week controlled study in healthy subjects [33] (11% and 11%, respectively), although both the amounts of fat and the fatty acid compositions of the diets were comparable in the two studies. The greater lipid-lowering effect seen in the present study might be because of the somewhat higher blood lipid levels of the current hyperlipidaemic patients or may in part be explained by different baseline or control diets and the amounts of fat in the intervention diets. For instance, in a study in which larger amounts of butter were replaced with MUFA from an RO-based margarine, LDL cholesterol decreased by 29.5% [34]. However, the reduction in total/HDL cholesterol ratio by 21% in the current study is greater than that observed following most previous interventions using RO (or canola oil) [2, 3, 8, 33, 35–37] and even exceeds mathematical predictions [29]. We did not observe any effects on HDL cholesterol, which is in line with previous studies [2, 3, 33–37]. The current marked effect on the total/HDL cholesterol ratio after RO is relevant, as this ratio may be a particularly important predictor of CVD mortality [38, 39]. Differences in Lp(a) concentrations between diets, and a modest (but statistically significant) increase after the RO diet have been described previously [40], but the clinical relevance is unclear and requires further study. The combined reduction in LDL cholesterol, total/HDL cholesterol ratio, apo B/A-I ratio and triglycerides instead clearly suggests a reduced overall CVD risk profile [31, 35]. By adapting the Framingham risk score for 10-year coronary heart disease risk, the current lowering of LDL cholesterol translates into approximately a 22% relative risk reduction in men and 13% in women [41].

The reduction in plasma triglycerides after 3 weeks of RO diet is especially relevant for insulin-resistant dyslipidaemic individuals. To our knowledge, previous studies have shown triglyceride-lowering effects of ALA only at high intakes of flaxseed oil (>38 g day⁻¹) [42], whereas oils rich in LA reduce triglycerides to a greater extent than oils rich in ALA such as RO [43]. In the study by Södergren et al. [33], the reduction in triglyceride levels (−11%) was not significant (P = 0.18). No significant effects on triglyceride levels were observed after increased intake of RO in other similar studies [34–37]. The present participants were hypertriglyceridaemic which may explain this novel finding; however, this warrants further study.

Neither fasting plasma glucose nor insulin sensitivity assessed by the gold standard hyperinsulinaemic euglycaemic clamp method was significantly affected by diet. As has been previously suggested, 3 weeks may be insufficient time to alter the fatty acid composition of the skeletal muscle cell membranes and thereby affect insulin signalling and cellular glucose transport [5, 44]. By contrast, fasting plasma glucose was decreased after a 4-week RO diet [33]. A larger sample size is probably needed to detect potential differences in insulin sensitivity [5] although a 3-week study in 10 healthy women [34] indicated slightly improved glucose tolerance after replacing SFA (butter) with RO. In that study, the intake of RO was higher than in the present study, which may be relevant. The glucose disappearance rate was not significantly different between the two diet groups, but significantly increased by 33% from baseline within the RO diet group in the present study. The K-value is mainly determined by first-phase insulin secretion and a lower value predicted impaired glucose tolerance over time [45]. This finding merits further investigation considering the potential to prevent diabetes by replacing SFA with MUFA and PUFA [11].

In agreement with most previous studies, amongst coagulation factors, only FVIIc tended to decrease after 3 weeks of the RO diet, which may suggest that unsaturated fatty acids could be preferable to SFA (especially stearic acid) in subjects with elevated coagulation factors [6, 46]. In contrast to our results, those of a Finnish study showed decreased fibrinogen levels after replacing SFA with RO in individuals with elevated baseline fibrinogen levels [47].

The strengths of the study include the randomized design and strictly controlled intervention, with all foods provided to subjects and only dietary fat quality differing between the diets. It is striking that another important strength was the completion of this study by all subjects, thus indicating exceptional high acceptance of the diets which is of practical and clinical importance. Indeed, dietary compliance was excellent as shown by corresponding changes in plasma lipids. Diets were planned to resemble the background diet of the population, i.e. moderately high in fat (35% of energy) which increases the generalizability of the results. Changes in lipoprotein levels are close to those achieved by lipid-lowering drugs.

Limitations of the study include its short duration (3 weeks) and the fact that, for practical reasons, it was not possible to blind the intervention. Also, the results achieved from this strictly controlled design may not translate into more uncontrolled conditions. However, this study provides proof-of-principle evidence that a simple dietary fat modification (i.e. replacing butter with RO-based margarine or RO) can produce clinically relevant improvement of blood lipid profile. The rapid effects could increase the motivation and awareness of the importance of diet in hyperlipidaemic patients without necessarily achieving weight loss. This is relevant because weight loss is difficult to achieve and also because hyperlipidaemia is common in the absence of obesity. The relevance of the results also increases because the improvement of serum lipids was observed for a diet containing a moderately elevated total fat intake as seen in most Western countries as well as in this population [48]. Thus, our results suggest that a low-fat diet may not be necessary to improve serum lipid profile as long as dietary fat quality is improved. These favourable lipid-lowering effects by replacing SFA with plant-based oils rich in MUFA and PUFA are also relevant with regard to the questioned role of SFA for CVD risk [49]. Our results support the current dietary guidelines [50] and encourage the aim of improvement of dietary fat quality.

In conclusion, replacing DF with RO causes marked improvement of serum lipoprotein profile including lowering of LDL cholesterol and triglycerides after only 3 weeks. These results are observed without significant weight loss and are especially clinically relevant for hyperlipidaemic patients at high risk of CVD.

Conflict of interest statement

No conflicts of interest to declare.

Acknowledgements

The nuclear magnetic resonance work by Ia Torelm, of the Swedish Food Administration, is gratefully acknowledged. David Iggman was funded by the Center for Clinical Research Dalarna, Sweden. Ulf Risérus was supported by grants from NordForsk (Nordic Centre of Excellence in Food, Nutrition and Health [SYSDIET]), the Swedish Council for Working Life and Social Research (FAS), the Family Ernfors Fund for Diabetes, the Erik, Karin and Gösta Selanders Foundation, Diabetesfonden (The Research Foundation of the Swedish Diabetes Federation) and the Swedish Heart-Lung Foundation.