https://doi.org/10.

1017/S0007114521004645 Published online by Cambridge University Press

British Journal of Nutrition (2022), 128, 1720–1729 doi:10.1017/S0007114521004645
© The Author(s), 2021. Published by Cambridge University Press on behalf of The Nutrition Society

Effects of diets rich in ghee or olive oil on cardiometabolic risk factors in healthy
adults: a two-period, crossover, randomised trial

Susan Mohammadi Hosseinabadi and Javad Nasrollahzadeh*

Department of Clinical Nutrition and Dietetics, Faculty of Nutrition Sciences and Food Technology, National Nutrition and
Food Technology, Research Institute, Shahid Beheshti University of Medical Sciences, Tehran, Iran
(Submitted 15 May 2021 – Final revision received 22 September 2021 – Accepted 16 November 2021 – First published online 19 November 2021)

Abstract
This study aimed to evaluate the cardiovascular health-related effects of consuming ghee in the usual diet. Thirty healthy men and women were
studied in a free-living outpatient regimen. The participants were instructed for the isoenergetic inclusion of ghee or olive oil in their diets for 4
weeks using a randomised crossover design. At the end of run-in (baseline), 2-week wash-out and interventions, fasting blood samples were
drawn. In addition, 2-h postprandial blood samples were collected after ingestion of a meal containing olive oil or ghee at week 4 of each dietary
intervention. Body weight was not different between the two interventions. Compared with the olive oil, the diet with ghee increased fasting
plasma apo-B (apo B) (0·09, 95 % CI 0·02, 0·17 g/l, P = 0·018), non-HDL-cholesterol (non-HDL-cholesterol) (0·53, 95 % CI 0·01, 1·05 mmol/l,
P = 0·046) and LDL-cholesterol did not differ significantly between diet groups (0·29, 95 % CI –0·05, 0·63 mmol/l, P = 0·092), but had no sig-
nificant effect on total cholesterol:HDL-cholesterol ratio (0·75, 95 % CI − 0·24 to 1·74 mmol/l, P = 0·118). No significant difference was observed
in fasting as well as 2-h postprandial plasma TAG, glucose, insulin and plasminogen activator inhibitor-1 concentrations. This study showed that
ghee that is predominantly saturated fats had an increasing effect on plasma apo B and non-HDL-cholesterol compared with olive oil, adding
further evidence to the existing recommendations to replace dietary fats high in SFA with dietary fats high in unsaturated fats to reduce CVD risk.

Key words: Ghee: Olive oil: SFA: apo B: LDL-cholesterol: Non-HDL-cholesterol: Insulin: PAI-1

Dietary SFA are associated with higher blood concentrations of factors in human studies(11,13–15). Previous studies have shown
LDL-cholesterol(1), a well-established risk factor for CVD(2), and inconsistent results regarding the effect of ghee on plasma lipo-
may also have metabolic effects such as decreased insulin sen- protein cholesterols levels. Results from animal studies indicated
sitivity when compared with unsaturated fatty acids(3,4). that ghee increased serum total cholesterol (TC), LDL-choles-
Although the evidence indicates the substitution of dietary satu- terol and HDL-cholesterol in rabbits(16) or had no significant
rated fats with unsaturated fats is beneficial for CVD risk, and a effect on serum TC, but increased serum TAG concentration
number of health and advisory committees have made recom- in rats when it included at a level of 10 % energy in the diet(17,18).
mendations for reducing the intake of SFA in the diet(1,5), there However, in human studies, the inclusion of ghee in the diet of
are still conflicting reports about the effect of dietary SFA on healthy subjects had no adverse effect on serum LDL-cholesterol
CVD(6,7). This heterogeneity in findings may partly be related levels when compared with mustard oil(19) or hydrogenated veg-
to different dietary sources of SFA(8). etable oil(20).
Ghee is produced from butter by traditional methods and is The effects of ghee consumption on plasma lipids and lipo-
rich in SFA and also contains moderate amounts (about 30 %) of protein might be better interpreted by examining its effects on
MUFA. Palmitic acid, which is associated with an increase in LDL- other risk factors for CVD, including insulin sensitivity, as well
cholesterol(9), accounts for a relatively high percentage of the as hemostatic markers such as plasminogen activator inhibitor
ghee’s SFA (about 62 %), but it also contains stearic acid (about 1 (PAI-1), which is also associated with individual fatty acids
10 %), which has been shown to have a neutral effect on LDL- intake(21,22) and insulin sensitivity(22,23). Few randomised trials
cholesterol(10). In addition to SFA and MUFA, ghee also contains have assessed the effects of consuming ghee on plasma lipids
some conjugated linoleic acid (CLA), a fatty acid that has anti-ath- and lipoproteins. Furthermore, all trials have used a parallel
erosclerotic effects and may improve insulin sensitivity(11,12) in design, making it difficult to distinguish between person from
animal studies, with inconsistent effects on cardiovascular risk variation within person. Moreover, information about the

Abbreviations CLA, conjugated linoleic acid; PAI-1, plasminogen activator inhibitor 1; TC, total cholesterol.

* Corresponding author: Javad Nasrollahzadeh, email [email protected]

 https://doi.org/10.1017/S0007114521004645 Published online by Cambridge University Press
Ghee consumption and cardiometabolic risk factors 1721

postprandial metabolic effect after consumption of a test meal equation(27) multiplied by self-reported physical activity). The
containing ghee is sparse. Therefore, we conducted a rando- estimated weight maintenance energy content were distributed
mised crossover trial to determine the effect of the inclusion as ∼55 % from carbohydrates, ∼15 % from protein and ∼30 %
of ghee compared with olive oil in the diet on fasting and post- from fat. Then, the number of servings of each food group
prandial blood cardiometabolic risk markers in healthy men and was determined for each level of energy requirement. The daily
women. Olive oil was considered as the reference oil for com- intake of olive oil or ghee was about 15 % of the total daily energy
parison in this study since it has been promoted as a healthy content, which varied between 20 and 48 g per day (mean and
oil, and previous researches have shown its potential cardiopro- standard deviation of 30·6 ± 6·7), depending on the daily energy
tective benefits(24–26). requirement. Olive oil and ghee were delivered to the subjects
every 2 weeks. Participants received written dietary information
booklets with portion advice. Except for studied oil/fat, all foods
Experimental methods were self-selected by participants. During the study period, par-
ticipants prepared two daily meals (lunch and dinner) using the
Study design and subjects studied oil/fat so that more than 80 % of the daily oil/fat used in
The present study was a randomised, not blinded, crossover, the meal preparation were from study oil/fat. For each person,
clinical trial. The study was conducted between December the amount of oil/fat allowed to prepare food in each of the
2020 and March 2021 at the Shahid Beheshti University of two daily meals was determined, and usually consumed oils
Medical Sciences. Participants were recruited via public adver- were replaced with ghee or olive oil. Participants were instructed
tisements. Adult male and non-menopausal females aged 20– to use each of the studied oil/fat to prepare meals using house-
60 years were included in the study. Participants were excluded hold utensils (such as spoons and cups). During each of the two
if they were on lipid-lowering medication, had a history of type 2 intervention periods, dietary intake was assessed through six
diabetes, history of coronary heart disease, abnormal liver func- telephone-administered 24-h dietary recalls (four weekdays
tion test (alanine aminotransferase and aspartate aminotransfer- and two weekends) obtained from each participant. The mean
ase > 45 U/L), abnormal kidney function test (createnine > 1·4 of the 24-h dietary recalls was calculated for each individual over
mg/dl), were at pregnancy and breastfeeding periods or were a 4-week diet period, and the means were used for data analysis.
on oral contraceptive unless they had no intention to change Diet composition was analysed by use of the Nutritionist soft-
during the study. A simple randomisation method was per- ware (version IV, N-Squared Computing) to which was added
formed using computer-generated random numbers with a 1:1 local food data. In addition, to assess their overall compliance
randomisation between the two treatments. Each number was with consuming the oil/fat, participants were asked to self-rate
placed in a sealed opaque envelope and was consecutively between 0 % and 100 % of their overall experience of consuming
opened by the research assistant. The research assistant involved the assigned oil/fat in the study.
in the procedure of randomisation did not have access to any Participants attended the clinic at baseline and the end of each
information about the laboratory characteristics of the partici- dietary intervention. In the post-intervention period (week 4), par-
pants. In this study, the effect of the inclusion of ghee in the diet ticipants were exposed to a meal challenge that included studied
was compared with a reference oil (olive oil). The study included oil/fat in accordance with the intervention diets. The patients
a 2-week run-in period in which reference oil (olive oil) was attended the clinic in the morning after 12 h fast, and a fasting
used and at the end of which baseline measurements were per- blood sample was collected before ingestion of the meal, in
formed. Subjects were then randomised to one of the two treat- <15 min. The subsequent blood sample was taken at 2 h. For each
ments for 4 weeks in which participants were asked to consume person, the meal was considered to be at 20 % of his/her daily
ghee or olive oil in their diet. The intervention was crossed energy content. This meal consisted of 48 % energy content from
within each group after two weeks wash-out during which refer- carbohydrates, 7 % energy content from protein, 45 % energy con-
ence oil (olive oil) was consumed. Anthropometric and bio- tent from fat so that 40 % of the total energy content, equivalent to
chemical parameters were assessed at baseline and the end of 15–29 g, depending on the daily energy content, were from the
each intervention period. tested fat (olive oil or ghee). The subjects did not consume any
This study was conducted according to the guidelines laid other food for 2 h after the meal challenge but were allowed to
down in the Declaration of Helsinki, and all procedures involv- drink water during the postprandial assessment.
ing human subjects were approved by the Ethical Committee of Ghee was sourced from a commercial manufacturer
National Nutrition and Food Technology Research Institute, (Nikmanesh® Animal Oil) that meets nationally accredited
Tehran, Iran (the Ethical No: IR.SBMU.nnftri. Rec.1399·053), manufacturing standards. In this study, olive oil was used as a
and written informed consent was obtained from all patients. reference oil because it has long been recognised to be favour-
This clinical trial was registered at the Iranian Registry of able for cardiovascular health(28). Refined olive oil was selected
Clinical Trials (IRCT20160702028742N9). since it was preferred by some consumers, probably because of
its milder flavour. Refined olive oil was produced by the com-
pany (Etka) from harvested olive in Rudbar County (Gilan
Diets
Province). The composition of fatty acids in studied oil/fat
Volunteers were instructed to have a diet at their estimated total (Table 1) was determined by GC of the corresponding methyl
energy expenditure. Energy levels were determined using equa- esters using standard methods (ISO 5509:2000 and ISO
tions to determine energy requirements (by Mifflin St Jeor’s 5508:1990).
 https://doi.org/10.1017/S0007114521004645 Published online by Cambridge University Press
1722 S. Mohammadi Hosseinabadi and J. Nasrollahzadeh

Table 1. Fatty acid composition of ghee and olive oil and apolipoprotein B (apo B) were measured in batches using
commercial kits (Pars-Azmoon by an automated analyser
Fatty acids Ghee Refined olive oil
(Selectra ProXL, Vital Scientific). The intra-assay coefficient of
% Total fatty acids % Total fatty acids variations were 1·49 %, 1·82 %, 1·62 %, 0·82 %, 0·67 % and
C4:0 1·07 ND 2·18 % for glucose, TAG, TC, HDL-cholesterol, LDL-cholesterol
C6:0 1·07 ND
C8:0 0·82 ND and apo B, respectively. Plasma insulin (Monobind, Inc., Lake
C10:0 2·20 ND Forest) and PAI-1 (R and D system) were measured in batches
C10:1 0·20 ND in frozen plasma samples via enzyme-linked immunosorbent
C11:0 0·05 ND
C12:0 3·12 ND
assays kit according to the manufacturer’s protocol. The intra-
C14:0 10·46 0·14 assay coefficient of variations were 5·1 % and 6·8 % for insulin
C14:1 1·50 ND and PAI-1, respectively. All measurements of each of the studied
C15:0 1·03 ND parameters were analysed in batch and together (baseline sam-
C15:1 0·37 ND
C16:0 31·59 12·33 ples along with end-of-study samples) to minimise inter-assay
C16:1 1·99 0·47 variation. Homoeostasis model assessment-insulin resistance
C17:0 0·57 0·17 was calculated according to the formula: fasting insulin (mIU/
C17:1 0·24 0·13
ml) × fasting glucose (mmol/l)/22·5.
C18:0 10·15 5·31
C18:1t 2·64 0·27
C18:1c 23·71 58·88 Statistical analysis
C18:2t 0·33 0·04
C18:2c 3·38 19·96 The sample size was calculated considering a change in LDL-
C20:0 0·25 0·85 cholesterol as the primary outcome. A previous study that com-
C18:3n3 0·33 0·52
C20:1 0·16 0·42
pared the effect of diets enriched with olive oil or butter for 4
C21:0 0·04 ND weeks generated an approximately 0·38 mmol/l between-treat-
CLAc9t11 0·63 ND ment difference in plasma LDL-cholesterol concentrations and
CLAt10c12 0·03 ND standard deviations of 0·39 and 0·48 mmol/l(8). Accordingly,
C20:2 0·04 ND
C20:3n3 0·12 ND using the G × Power 3.1 software, it was determined that a sam-
C22:0 0·08 0·21 ple size of n 28 would allow the power of 80 % at the level of
C20:3n6 0·19 ND significance of 5 % and a correlation between treatment values
C20:4n6 0·03 ND
C24:0 0·05 0·18
of 0·4. By considering a dropout, the required sample size
Others 1·60 0·12 was n 30.
∑SFA 62·51 19·19 The normality of continuous variables was assessed by using
∑MUFA 30·81 60·17 the Kolmogorov–Smirnov test. All studied variables had normal
∑PUFA 5·08 20·52
distribution (P > 0·05). A paired t test was performed to assess
ND, not detected. Expressed as % of total fatty acids. the effect of each intervention compared with its baseline.
Differences in study outcomes between treatments were ana-
lysed using a two-factor repeated-measures ANOVA, with treat-
The level of physical activity over each treatment period was
ment and time (baseline and week 4 within each treatment
determined two times by using the short form of a valid
period) as within-subject factors. The main effects of treatment
International physical activity questionnaire(29), and the mean
and time, as well as the time × treatment interaction, were inves-
of them was regarded in data analysis.
tigated. The possible carryover effect was determined by includ-
ing allocation order as a fixed factor in the model and
Anthropometric and blood pressure measurement
investigating its interaction with treatment × time. No statistically
Height and weight were measured during the baseline and post- significant carryover was evident; therefore, data from both treat-
intervention visits. Weight was measured in light street clothes. ment sequences were pooled. For postprandial data, the base-
Height was measured without shoes. Waist circumference was line to postprandial response was analysed by a two-factor
measured midway between the iliac crest and the lower rib mar- repeated-measures ANOVA, with treatment and time (baseline
gin by an anthropometric flexible tape. During each visit, while and 2-h postprandial in each treatment period) as within-subject
the participant was seated after 5 min of rest, two blood pressure factors. To analyse the dietary intake and physical activity level
measurements were obtained using a digital arm sphygmoma- values, a paired t test was used. Data are presented as means ±
nometer (Omron digital automatic blood pressure monitor standard deviations (SD) unless otherwise mentioned. SPSS
HEM-907). Statistics software version 21.0 (IBM Corporation) was used
for the statistical analyses and significance was defined
Biochemical measurement as P ≤ 0·05.
Blood samples were collected in the heparin tubes in the fasting
(at least 12 h) and 2-h postprandial state and plasma samples
Results
were separated using centrifugation at 2400 rpm for 10 min at
20°C and were stored at −80°C until they were analysed. All thirty individuals randomised in the two interventions com-
Plasma glucose, TAG, TC, HDL-cholesterol, LDL-cholesterol pleted both 4 week intervention periods and were included in
 https://doi.org/10.1017/S0007114521004645 Published online by Cambridge University Press
Ghee consumption and cardiometabolic risk factors 1723

final analyses (Fig. 1 and Table 2). Table 3 shows energy and There were no significant differences in blood pressure, body
macronutrient intake during each of the two dietary interven- weight, central adiposity as measured by waist circumference
tions. The energy content, carbohydrates, protein and fibre and physical activity between the two treatments; therefore,
intake did not differ between the two diet periods. The total the observed changes in plasma lipid and lipoprotein concentra-
amount of fat intake did not differ between the two diet periods, tions can be attributed to changes in the dietary intake of partic-
but the dietary intervention changed the composition of the fat ipants. The subjects reported increased SFA and cholesterol
intake. During the diet with olive oil, MUFA, as well as PUFA intake by the ghee intervention, whereas MUFA and PUFA
intake, were higher, whereas SFA and dietary cholesterol were intakes were increased by olive oil intervention. This could indi-
higher during the diet with ghee intervention. Regarding self- cate good compliance with consuming the studied dietary fats.
reported compliance, participants reported 80 % or more com- The clinical relevance of the increase in apo B after consum-
pliance with the studied oil/fat over the 4 weeks. Evaluation ing the diet containing ghee compared with the diet with olive oil
of daily physical activity by physical activity questionnaire indi- remains unclear given that LDL-cholesterol increased margin-
cated no difference between the two diet periods. ally. However, the plasma apo B concentration has been consid-
In Table 4, values of anthropometric measures are shown. ered to be a more accurate representation of atherogenic
There was no statistical difference between the two diet periods particles(2,30). A meta-analysis reported that each 0·1 g/L
in body weight, waist circumference and blood pressure. decrease in apo B was associated with a 9 % decrease in coro-
A decrease in the consumption of olive oil and an increase in nary heart disease and a 6 % decrease in major CVD risk(31). In
the intake of ghee in the diet increased plasma apo B concentra- addition to apo B, ghee consumption increased the non-HDL-
tion (treatment × time, P = 0·018) and in non-HDL-cholesterol cholesterol level. Non-HDL-cholesterol represents the choles-
(treatment × time, P = 0·046) (Table 5). However, despite the dif- terol in all particles causing CVD(32). Non-HDL-cholesterol is
ference in apo B concentration between the two diets, the an important target of therapy for the prevention of coronary
plasma LDL-cholesterol concentrations were not different after heart disease and there is a consistent direct relationship
consumption of the two diets (treatment × time, P = 0·092). between cardiovascular risk reduction and the magnitude of
Further, there were no significant differences in plasma HDL- non-HDL-cholesterol lowering(2,33,34). Other studies examining
cholesterol and TAG concentrations, or the ratio of TC: HDL-cho- the effect of ghee on plasma lipid and lipoproteins concentra-
lesterol between the two diets. tions have yielded conflicting results. In an experimental study
Fasting plasma glucose had no significant changes, but fast- in which male rabbits were fed a normal chow diet or diets con-
ing insulin tended to increase after both olive oil and ghee con- taining ghee or olive oil, ghee significantly increased TC, LDL-
sumption (time effect, P = 0·059). However, neither of the cholesterol and HDL-cholesterol as compared with chow or
dietary interventions had a significant time nor time × treatment olive diet(16). Replacing ghee in the diet of healthy young people
effect on insulin resistance as assessed by homoeostasis model on a vegetarian diet did not increase LDL-cholesterol concentra-
assessment-insulin resistance (Table 5). tion compared with their baseline, but in the control group who
Although fasting TAG concentration before ingestion of the consumed mustard oil, atrend towards a decline in LDL-choles-
olive oil was slightly and non-significantly (P = 0·18) lower than terol was observed(19). In another parallel study in healthy sub-
and ghee meals, the 2-h postprandial TAG response of meals jects who used mostly hydrogenated vegetable oil in their usual
was not different. TAG concentrations increased (P < 0·01) after diets before the study, replacing hydrogenated oil with ghee did
both test meals, but no significant difference was observed in the not change serum LDL-cholesterol and apo B levels, but did
2-h postprandial TAG response of meals (Fig. 2(a)). Meals had no increase HDL-cholesterol and lowered TAG levels. On the other
significant effect on postprandial plasma HDL-cholesterol con- hand, in the group that replaced hydrogenated oils with non-
centration (Fig. 2(b)). Glucose concentration return to lower hydrogenated ones, the levels of total cholesterol, apo B and
than basal values after 2 h and the response was similar after TG were decreased(20). The discrepancy between the results
ingestion of both test meals (Fig. 2(b)). Both meals markedly may be caused by differences in study design and subjects.
increased (P < 0·001) plasma insulin concentrations at 2 h with These studies were not a crossover, thus between-person
no significant difference between meals (Fig. 2(c)). Likewise, differences may have influenced the observed results.
PAI-1 concentrations markedly increased after both meals, but Moreover, in the first study(19), all analyses compared final values
there was no difference between the two meals (Fig. 2(e)). to their baselines, and between groups, analysis was not per-
formed, and the second study(20) did not have a lead-in period
to control participants’ baseline diets.
There was no difference between the effects of ghee or olive
Discussion
oil intake on fasting plasma concentrations of glucose and insu-
This study aimed to evaluate the cardiovascular health-related lin or insulin resistance index. Fasting plasma insulin tended to
effects of consuming ghee in the usual diet. Compared with increase after both dietary interventions. This tendency to
the diet containing olive oil, consumption of ghee instead of increase was observed despite no change in body weight or
olive oil in the diet, significantly increased non-HDL and apo waist circumference and also without significant change in
B concentrations, but it had a neutral effect on insulin resistance energy, carbohydrate or fibre intake. Regarding postprandial
and other cardiovascular risk markers. In addition, similar post- glycaemic response, the two test meals showed similar postpran-
prandial TAG and glycaemic responses were observed after dial glucose and insulin responses. These findings are somewhat
ingestion of meals containing ghee or olive oil. consistent with the results of other studies in which the effects of
 https://doi.org/10.1017/S0007114521004645 Published online by Cambridge University Press
1724 S. Mohammadi Hosseinabadi and J. Nasrollahzadeh

Fig. 1. Flow diagram of participants.

Table 2. Baseline characteristics of participants by treatment sequence*

(Mean values and standard deviations)

Characteristic Total (n 30) Olive oil then ghee (n 15) Ghee then olive oil (n 15)
Mean SD Mean SD Mean SD

Age, years 37·50 9·52 38·75 10·90 36·07 7·80

Height, cm 164·92 9·56 163·81 11·71 166·18 6·50
Weight, kg 70·95 16·73 70·69 19·59 71·25 13·47
BMI, kg/m2 25·90 4·41 26·02 4·98 25·70 3·89
waist circumference, cm 89·75 11·40 90·53 13·69 88·86 8·49
Systolic BP, mmHg 113·07 11·70 115·69 10·67 110·07 12·48
Diastolic BP, mmHg 73·53 8·44 75·69 8·45 71·07 8·02
Plasma glucose, mmol/l 4·38 0·71 4·46 0·73 4·30 0·71
Plasma total cholesterol, mmol/l 4·62 0·84 4·33 0·86 4·93 0·70
Plasma LDL-cholesterol, mmol/l 2·80 0·70 2·60 0·73 3·02 0·62
Plasma HDL-cholesterol, mmol/l 1·11 0·24 1·11 0·23 1·12 0·25
Plasma TAG, mmol/l 1·57 1·12 1·44 0·97 1·72 1·29
Plasma creatinine mg/dl 0·98 0·18 0·99 0·20 0·97 0·16

* Data are means ± SD. BP, blood pressure.

 https://doi.org/10.1017/S0007114521004645 Published online by Cambridge University Press
Ghee consumption and cardiometabolic risk factors 1725

Table 3. Dietary intake and physical activity level over the 4-week consumption of diets rich in olive oil or ghee*
(Mean values and standard deviations; mean values and 95 % confidence intervals)

Diet with olive oil (n 30) Diet with ghee (n 30) Difference
Mean SD Mean SD Mean 95 %CI P†

Energy, kcal/d 1791·42 528·97 1778·50 484·76 –12·91 –88·09, 62·26 0·728
Carbohydrate, % of energy/d 52·67 4·35 53·63 3·78 0·97 –1·01, 2·94 0·326
Protein, % of energy/d 14·96 1·54 15·17 1·27 0·21 –0·41, 0·83 0·492
Total fat, % of energy/d 32·37 3·76 31·19 3·43 –1·18 –2·98, 0·32 0·156
SFA, % of energy/d 8·21 0·98 14·91 2·29 6·70 5·77, 7·63 < 0·001
MUFA, % of energy/d 16·32 2·23 9·63 1·20 –6·69 –7·36, −6·01 < 0·001
PUFA, % of energy/d 5·22 1·99 3·60 0·77 –1·62 –2·39, −0·85 < 0·001
Cholesterol, mg/d 272·02 144·08 314·68 107·42 42·66 3·12, 82·19 0·035
Total fibre, g/d 14·83 5·66 15·08 4·85 0·25 –1·68, 2·17 0·795
Physical activity (Met-h/d) 25·03 3·15 24·69 3·61 0·34 –2·19, 2·86 0·483

* For dietary intake assessment, six 24-h dietary recalls were taken from each participant over each 4-week diet period, the mean of which was calculated for each individual, and the
means were used for data analysis.
† Data were analysed using a paired t test.

diets or meals rich in SFA or MUFA have been compared. Chang In the present study, the fasting and postprandial TAG con-
et al. investigated the effect of MUFA or SFA-enriched diets in centrations were not statistically significant between the two
centrally obese subjects and found no significant difference interventions. The relative effects of SFA and MUFA rich meals
between SFA or MUFA diets on fasting and postprandial insulin on postprandial TAG concentrations have been investigated in
and glucose secretion(35). Roche et al. evaluated the postprandial only a small number of studies, and none of these studies have
response during 9 h after the ingestion of meals containing differ- evaluated the effect of ghee. However, the postprandial TAG
ent proportions of MUFA and SFA in healthy males and reported response may be best measured at times longer than 2 h since
no significant difference in postprandial plasma glucose and similar previous studies have shown that peak TAG concentra-
insulin response(36). Itoh et al. evaluated the glucose and insulin tions occur between 3 and 5 h after high-fat meals(3,43).
responses after the ingestion of an SFA-enriched high-fat meal or With respect to PAI-1 concentration, the main regulator of
a reduced SFA high-fat meal in healthy women and found no dif- fibrinolysis, the effect of ghee consumption was not significantly
ference in insulin response between SFA enriched and reduced different from olive oil. This suggests that a diet rich in ghee does
SFAs high-fat meals(37). However, our finding may not be consis- not increase PAI-1. No previous study has reported the effect of
tent with some previous studies in which the consumption of ghee consumption on haemostatic factors. It has been shown
SFA instead of MUFA changed the insulin sensitivity index. that enrichment of test meals with refined olive oil, butter and
Lopez et al. evaluated acute postprandial insulin sensitivity effect high-palmitic sunflower oil induces a peak in PAI-1 2-h post-
over the 8 h after ingestion of high-fat meals enriched in SFA prandially(21). Tholstrup et al. compared the effects of test meals
(high-palmitic sunflower oil and butter) or MUFA (refined olive rich in stearic acid, palmitic acid or oleic acid on haemostatic pro-
oil) in normotriglyceridaemic subjects and found that the early file, including PAI-1, in 16 young men and consistent with our
postprandial insulin response (0–2 h) increased as the ratio of finding, they reported no significant difference in postprandial
dietary MUFA:SFA decreased(3). Christiansen et al. studied the PAI-1 between test fats(44). Oakley et al. compared the postpran-
effects of diets (for 6 weeks) enriched in SFA or MUFA in obese dial effects of three high-fat (95 g) meals (butter, high oleate and
patients with non-insulin-dependent diabetes mellitus and oleate þ medium-chain triacylglycerols) with an isoenergetic
found that in the presence of unchanged fasting glycaemia, low-fat meal (18 g medium-chain triacylglycerols) in twelve
dietary SFA induced an increase in postprandial insulinaemia men and found neither the amount nor type of fat meals influ-
compared with a diet with MUFA(38). However, the health status enced PAI-1 concentration(45). Similarly, Stonehouse et al.
of participants (healthy v. diabetes mellitus), as well as the differ- showed that consumption of meals containing palmolein or
ence in the composition of fatty acids oil/fat, may have influ- olive oil did not affect postprandial plasma PAI-1 in overweight
enced the results. Ghee’s fatty acid composition is somewhat and obese men(46). However, Pacheco et al. reported that when
different from the SFA-rich fats used in the above studies. In addi- the ratio of MUFA:SFA in dietary fats (from butter, refined olive
tion to being rich in SFA, ghee is also a relatively good source of oil and high-palmitic sunflower oil) was lower, postprandial con-
MUFA and CLA. Although the effects of CLA on glucose metabo- centrations of PAI-1 increased in healthy subjects(21).
lism are controversial, the administration of CLA has improved The main strengths of this study are the randomised crossover
insulin sensitivity in non-obese, regularly exercising women(39) design, pragmatic trial in free-living individuals with a high com-
or in young sedentary humans(14). However, CLA supplementa- pletion rate over 4 weeks. The trial has some limitations. The
tion did not affect plasma glucose and insulin concentrations or sample size was rather small, and the power to detect subtle
insulin resistance index in healthy subjects(40,41) or even changes is therefore restricted. Furthermore, the generalisability
impaired insulin sensitivity in non-diabetic abdominally obese of the data is limited to healthy men and women. In addition, it
men(15,42). was a short-term trial of 4 weeks intervention, and the study
 1726
Table 4. Anthropometry and blood pressure in healthy adults at baseline and following 4-week consumption of diets rich in olive oil or ghee
(Mean values and standard deviations)

Diet with olive oil (n 30) Diet with ghee (n 30)

Basal 4 weeks Δ Basal 4 weeks Δ Δ olive oil v. Δ ghee
Mean SD Mean SD Mean 95 % CI Mean SD Mean SD Mean 95 %CI Mean 95 % CI P*

Body mass (kg) 70·93 16·64 71·02 17·18 0·09 –0·34, 0·51 70·93 17·36 71·05 17·67 0·12 –0·29, 0·53 –0·03 –0·52, 0·46 0·891
BMI (kg/m2) 25·87 4·43 25·88 4·49 0 –0·14, 0·15 25·84 4·51 25·88 4·67 0·04 –0·10, 0·20 –0·04 –0·21, 0·14 0·676
WC (cm) 89·85 11·50 89·88 11·40 0·03 –0·42, 0·48 89·58 11·42 89·80 12·15 0·22 –0·24, 0·68 –0·18 –0·75, 0·39 0·516
SBP (mmHg) 113·63 11·32 111·9 12·77 –1·73 –5·27, 1·81 110·03 12·34 110·27 14·12 0·23 –4·00, 4·47 –1·97 –7·91, 3·98 0·504

S. Mohammadi Hosseinabadi and J. Nasrollahzadeh

DBP (mmHg) 71·70 8·52 67·47 10·01 –4·23 –6·61, −1·85 69·47 9·29 65·33 7·42 –4·13 –7·42, −0·84 –0·10 –4·21, 4·01 0·961

* P for time × treatment, analysed using a two-factor repeated-measures ANOVA. Δ, change from baseline; DBP, diastolic blood pressure; SBP, systolic blood pressure; WC, waist circumference.

Table 5. Fasting plasma concentrations of glucose, insulin, lipids and plasminogen activator inhibitor-1 (PAI-1) in healthy adults at baseline and following 4-week consumption of diets rich in olive oil or ghee
(Mean values and standard deviations)

Diet with olive oil (n 30) Diet with ghee (n 30)

Basal 4 weeks Δ Basal 4 weeks Δ Δ olive oil v. Δ ghee
Mean SD Mean SD Mean 95 % CI Mean SD Mean SD Mean 95 % CI Mean 95 % CI P*

TC, mmol/l 4·63 0·86 4·38 0·98 –0·24 –0·52, 0·02 4·33 1·17 4·59 1·07 0·25 –0·06, 0·57 –0·50 –1·00, 0 0·050
TAG, mmol/l 1·54 1·02 1·47 1·24 –0·07 –0·35, 0·21 1·38 0·99 1·82 1·73 0·44 –0·10, 0·97 –0·50 –1·20, 0·19 0·150
LDL-cholesterol, mmol/l 2·83 0·77 2·65 0·77 –0·18 –0·40, 0·03 2·63 0·97 2·74 0·75 0·10 –0·09, 0·30 –0·29 –0·63, 0·05 0·092
HDL-cholesterol, mmol/l 1·11 0·25 1·07 0·27 –0·04 –0·09, 0·01 1·09 0·24 1·04 0·27 –0·05 –0·12, 0·03 0·01 –0·10, 0·11 0·924
Apo B, g/l 0·99 0·24 0·94 0·24 –0·05 –0·10, −0·01 0·94 0·26 0·98 0·24 0·04 –0·01, 0·09 –0·09 –0·17, −0·02 0·018
TC/HDL-cholesterol 4·44 1·64 4·42 1·88 –0·02 –0·40, 0·36 4·01 1·41 4·83 2·25 0·73 –0·6, 1·51 –0·75 –1·74, 0·24 0·118
Non-HDL-cholesterol mmol/l 3·55 0·92 3·33 0·98 –0·21 –0·49, 0·06 3·24 1·10 3·54 1·08 0·30 –0·02, 0·63 –0·53 –1·05, −0·01 0·046
Glucose, mmol/l 4·35 0·70 4·30 .064 –0·06 –0·28,0·17 4·18 0·61 4·24 0·69 0·06 –0·13, 0·25 –0·11 –0·37, 0·15 0·392
Insulin, mIU/ml 6·39 2·46 7·29 3·78 0·90 –0·17, 1·98 6·25 3·48 6·86 2·99 0·61 –0·81, 2·03 0·29 –1·68, 2·26 0·764
HOMA-IR 1·25 0·51 1·38 0·73 0·13 –0·07, 0·34 1·20 0·74 1·30 0·54 0·08 –0·19, 0·36 0·05 –0·32, 0·42 0·544
PAI-1 ng/ml 9·36 8·94 10·22 10·80 0·86 –2·86, 4·58 11·62 11·87 12·44 17·91 0·82 –5·16, 6·80 –0·001 –0·02, 0·02 0·989

* P for time × treatment, analysed using a two-factor repeated-measures ANOVA. Δ, change from baseline; TC, total cholesterol.

https://doi.org/10.1017/S0007114521004645 Published online by Cambridge University Press

 https://doi.org/10.1017/S0007114521004645 Published online by Cambridge University Press
Ghee consumption and cardiometabolic risk factors 1727

(a) Olive oil (b)

Ghee
Treatment: p 0·147 Treatment: p 0·841
2·5 Time: p ·005 Time: p ·123
Treatment*Time: 0·511 Treatment*Time: 0·342

Plasma HDL-C (mmol/L)

Plasma TAG (mmol/L)

1·5
2·0

1·5 1·0

1·0
0·5
Fasting 2-h postprandial Fasting 2-h postprandial

(c) (d)
Treatment: p 0·832
Treatment: p 0·625
4·6 15 Time: p ·000
Time: p ·005
Treatment*Time: 0·825
Treatment*Time: 0·989

Plasma insulin (mIU/mL)

Plasma glucose (mmol/L)

4·4

4·2
10
4·0

3·8

3·6 5
Fasting 2-h postprandial Fasting 2-h postprandial

(e) Treatment: p 0·455

50 Time: p ·000
Treatment*Time: 0·522

40
Plasma PAI-1(ng/mL)

30

20

10

0
Fasting 2-h postprandial

Fig. 2. Comparison of plasma 2-h postprandial responses of study participants to meals containing olive oil or ghee. Values are the means ± standard error of the mean.
2-h postprandial values are after ingestion of a meal containing olive oil or ghee in accordance with the background diets at week-4 of each dietary intervention. Data were
analysed using a two-factor repeated-measures ANOVA. Δ, change from baselinePAI-1, plasminogen activator inhibitor-1.

would have benefited if continued for a longer duration to assess such as the blood fatty acid profile were measured in
long-term effects. Moreover, the postprandial period was too participants.
short and limited by only two measurements (fasting and 2 h).
Ghee as one of the treatment oils in this study was included at
more than 80 % of the total cooking fat which may be higher than Conclusion
what is expected for the usual intake of this oil. Thus, some of the Dietary recommendations to decrease intakes of SFA and, more
treatment effects under the experimental conditions of this study recently, to replace SFA with unsaturated fat have been emphas-
may have been larger than what would be expected when a ised by some health and advisory committees(1,5)and research-
more typical amount of ghee is consumed. Although self- ers(47). In addition, it has been proposed that different foods
reported compliance was good, it was subjective, and the study rich in SFA may have different effects on cardiovascular risk fac-
would have benefited if objective biomarkers of compliance tors, and the dietary recommendation to reduce intake of SFA
 https://doi.org/10.1017/S0007114521004645 Published online by Cambridge University Press
1728 S. Mohammadi Hosseinabadi and J. Nasrollahzadeh

should distinguish between specific food sources of SFA(48). The 10. Meng H, Matthan NR, Wu D, et al. (2019) Comparison of diets
results of this study indicate that a decrease in olive oil intake enriched in stearic, oleic, and palmitic acids on inflammation,
which is predominantly monounsaturated fat, and a concomitant immune response, cardiometabolic risk factors, and fecal bile
acid concentrations in mildly hypercholesterolemic postmeno-
increase with ghee intake, a predominantly saturated fat,
pausal women-randomized crossover trial. Am J Clin Nutr 110,
increases blood apo B and non-HDL-cholesterol levels. Given 305–315.
that plasma apo B concentrations could be regarded as a repre- 11. den Hartigh LJ (2019) Conjugated linoleic acid effects on
sentation of atherogenic particles(30) and regarding a direct rela- cancer, obesity, and atherosclerosis: a review of pre-clinical
tionship between the magnitude of non-HDL-cholesterol levels and human trials with current perspectives. Nutrients 11, 370.
and cardiovascular risk(34), the current trial adds further evidence 12. Chinnadurai K, Kanwal HK, Tyagi AK, et al. (2013) High con-
to emphasise the existing prudent recommendations to replace jugated linoleic acid enriched ghee (clarified butter) increases
the antioxidant and antiatherogenic potency in female Wistar
dietary fats high in SFA with dietary fats high in unsaturated fats rats. Lipids Health Dis 12, 121.
to reduce CVD risk(1,5). 13. Raff M, Tholstrup T, Basu S, et al. (2008) A diet rich in conju-
gated linoleic acid and butter increases lipid peroxidation but
does not affect atherosclerotic, inflammatory, or diabetic risk
markers in healthy young men. J Nutr 138, 509–514.
Acknowledgements
14. Eyjolfson V, Spriet LL & Dyck DJ (2004) Conjugated linoleic acid
The authors would like to thank Elaheh Tajeddin for her assis- improves insulin sensitivity in young, sedentary humans. Med
tance in laboratory measurements and the participants who com- Sci Sport Exerc 36, 814–820.
15. Riserus U, Vessby B, Arnlov J, et al. (2004) Effects of cis-9,trans-
pleted the trial.
11 conjugated linoleic acid supplementation on insulin sensitiv-
This research is funded by a project grant supporting by the ity, lipid peroxidation, and proinflammatory markers in obese
National Nutrition and Food Technology Research Institute, men. Am J Clin Nutr 80, 279–283.
Shahid Beheshti University of Medical Science, Tehran, Iran. 16. Hosseini M & Asgary S (2012) Effects of dietary supplementa-
The authors’ contributions were as follows – J. N. and S. M. H. tion with ghee, hydrogenated oil, or olive oil on lipid profile and
designed the research; S. M. H. conducted the research, includ- fatty streak formation in rabbits. ARYA Atheroscler 8, 119–124.
ing the recruitment of study volunteers and the collection of sam- 17. Sharma H, Zhang X & Dwivedi C (2010) The effect of ghee
(clarified butter) on serum lipid levels and microsomal lipid
ple and data; J. N. and S. M. H. analysed and interpreted data; J.
peroxidation. Ayu 31, 134–140.
N. performed the statistical analysis and wrote the paper. 18. Kumar MV, Sambaiah K & Lokesh BR (1999) Effect of dietary
The authors declare that there is no conflict of interest. ghee – the anhydrous milk fat, on blood and liver lipids in rats.
J Nutr Biochem 10, 96–104.
19. Shankar SR, Yadav RK, Ray RB, et al. (2005) Serum lipid
response to introducing ghee as a partial replacement for mus-
Reference tard oil in the diet of healthy young Indians. Indian J Physiol
1. Sacks FM, Lichtenstein AH, Wu JHY, et al. (2017) Dietary fats Pharmacol 49, 49–56.
and cardiovascular disease: a presidential advisory from the 20. Mohammadifard N, Nazem M, Naderi GA, et al. (2010) Effect of
American Heart Association. Circulation 136, e1–e23. hydrogenated, liquid and ghee oils on serum lipids profile.
2. Carr SS, Hooper AJ, Sullivan DR, et al. (2019) Non-HDL-choles- ARYA Atheroscler 6, 16–22.
terol and apolipoprotein B compared with LDL-cholesterol in 21. Pacheco YM, Bermudez B, Lopez S, et al. (2006) Ratio of oleic
atherosclerotic cardiovascular disease risk assessment. to palmitic acid is a dietary determinant of thrombogenic and
Pathology 51, 148–154. fibrinolytic factors during the postprandial state in men. Am J
3. Lopez S, Bermudez B, Pacheco YM, et al. (2008) Distinctive Clin Nutr 84, 342–349.
postprandial modulation of beta cell function and insulin sen- 22. Masquio DC, de Piano A, Campos RM, et al. (2014) Saturated
sitivity by dietary fats: monounsaturated compared with satu- fatty acid intake can influence increase in plasminogen activa-
rated fatty acids. Am J Clin Nutr 88, 638–644. tor inhibitor-1 in obese adolescents. Horm Metab Res 46, 245–
4. Lopez S, Bermudez B, Ortega A, et al. (2011) Effects of meals 251.
rich in either monounsaturated or saturated fat on lipid concen- 23. Potter van Loon BJ, Kluft C, Radder JK, et al. (1993) The cardio-
trations and on insulin secretion and action in subjects with high vascular risk factor plasminogen activator inhibitor type 1 is
fasting triglyceride concentrations. Am J Clin Nutr 93, 494–499. related to insulin resistance. Metab Clin Exp 42, 945–949.
5. DeSalvo KB, Olson R & Casavale KO (2016) Dietary guidelines 24. Guasch-Ferre M, Liu G, Li Y, et al. (2020) Olive oil consumption
for Americans. JAMA 315, 457–458. and cardiovascular risk in U.S. adults. J Am Coll Cardiol 75,
6. Hooper L, Martin N, Jimoh OF, et al. (2020) Reduction in satu- 1729–1739.
rated fat intake for cardiovascular disease. Cochrane Database 25. Khandouzi N, Zahedmehr A & Nasrollahzadeh J (2021) Effect of
Syst Rev 8, CD011737. polyphenol-rich extra-virgin olive oil on lipid profile and
7. Hamley S (2017) The effect of replacing saturated fat with inflammatory biomarkers in patients undergoing coronary
mostly n-6 polyunsaturated fat on coronary heart disease: a angiography: a randomised, controlled, clinical trial. Int J
meta-analysis of randomised controlled trials. Nutr J 16, 30. Food Sci Nutr 72, 548–558.
8. Khaw KT, Sharp SJ, Finikarides L, et al. (2018) Randomised trial 26. Khandouzi N, Zahedmehr A & Nasrollahzadeh J (2020) Effects
of coconut oil, olive oil or butter on blood lipids and other of canola or olive oil on plasma lipids, lipoprotein-associated
cardiovascular risk factors in healthy men and women. BMJ phospholipase A2 and inflammatory cytokines in patients
Open 8, e020167. referred for coronary angiography. Lipids Health Dis 19, 183.
9. van Rooijen MA & Mensink RP (2020) Palmitic acid versus ste- 27. Mifflin MD, St Jeor ST, Hill LA, et al. (1990) A new predictive
aric acid: effects of interesterification and intakes on cardiome- equation for resting energy expenditure in healthy individuals.
tabolic risk markers – a systematic review. Nutrients 12, 615. Am J Clin Nutr 51, 241–247.
 https://doi.org/10.1017/S0007114521004645 Published online by Cambridge University Press
Ghee consumption and cardiometabolic risk factors 1729

28. Foscolou A, Critselis E & Panagiotakos D (2018) Olive oil con- 39. Lambert EV, Goedecke JH, Bluett K, et al. (2007) Conjugated
sumption and human health: a narrative review. Maturitas 118, linoleic acid versus high-oleic acid sunflower oil: effects on
60–66. energy metabolism, glucose tolerance, blood lipids, appetite
29. Craig CL, Marshall AL, Sjostrom M, et al. (2003) International and body composition in regularly exercising individuals. Br
physical activity questionnaire: 12-country reliability and valid- J Nutr 97, 1001–1011.
ity. Med Sci Sport Exerc 35, 1381–1395. 40. Noone EJ, Roche HM, Nugent AP, et al. (2002) The effect of
30. Sniderman AD, Thanassoulis G, Glavinovic T, et al. (2019) dietary supplementation using isomeric blends of conjugated
Apolipoprotein B Particles and Cardiovascular Disease: A linoleic acid on lipid metabolism in healthy human subjects.
Narrative Review. JAMA Cardiol 4, 1287–1295. Br J Nutr 88, 243–251.
31. Robinson JG, Wang S & Jacobson TA (2012) Meta-analysis of 41. Tricon S, Burdge GC, Kew S, et al. (2004) Opposing effects of
comparison of effectiveness of lowering apolipoprotein B v. cis-9,trans-11 and trans-10,cis-12 conjugated linoleic acid on
low-density lipoprotein cholesterol and nonhigh-density lipo- blood lipids in healthy humans. Am J Clin Nutr 80, 614–620.
protein cholesterol for cardiovascular risk reduction in random- 42. Riserus U, Vessby B, Arner P, et al. (2004) Supplementation
ized trials. Am J Cardiol 110, 1468–1476. with trans10cis12-conjugated linoleic acid induces hyperproin-
32. Nordestgaard BG, Langlois MR, Langsted A, et al. (2020) sulinaemia in obese men: close association with impaired insu-
Quantifying atherogenic lipoproteins for lipid-lowering strate- lin sensitivity. Diabetologia 47, 1016–1019.
gies: consensus-based recommendations from EAS and EFLM. 43. Teng KT, Chang LF, Vethakkan SR, et al. (2017) Effects of
Atherosclerosis 294, 46–61. exchanging carbohydrate or monounsaturated fat with satu-
33. Robinson JG, Wang S, Smith BJ, et al. (2009) Meta-analysis of rated fat on inflammatory and thrombogenic responses in sub-
the relationship between non-high-density lipoprotein choles- jects with abdominal obesity: a randomized controlled trial.
terol reduction and coronary heart disease risk. J Am Coll Clin Nutr 36, 1250–1258.
Cardiol 53, 316–322. 44. Tholstrup T, Miller GJ, Bysted A, et al. (2003) Effect of individual
34. Zhang Y, Wu NQ, Li S, Zhu CG, et al. (2016) Non-HDL-C is a dietary fatty acids on postprandial activation of blood coagula-
better predictor for the severity of coronary atherosclerosis tion factor VII and fibrinolysis in healthy young men. Am J Clin
compared with LDL-C. Heart Lung Circ 25, 975–981. Nutr 77, 1125–1132.
35. Chang LF, Vethakkan SR, Nesaretnam K, et al. (2016) Adverse 45. Oakley FR, Sanders TA & Miller GJ (1998) Postprandial effects
effects on insulin secretion of replacing saturated fat with of an oleic acid-rich oil compared with butter on clotting factor
refined carbohydrate but not with monounsaturated fat: a ran- VII and fibrinolysis in healthy men. Am J Clin Nutr 68, 1202–
domized controlled trial in centrally obese subjects. J Clin 1207.
Lipidol 10, 1431–1441.e1431. 46. Stonehouse W, Brinkworth GD & Noakes M (2015) Palmolein
36. Roche HM, Zampelas A, Jackson KG, et al. (1998) The effect of and olive oil consumed within a high protein test meal have
test meal monounsaturated fatty acid: saturated fatty acid ratio similar effects on postprandial endothelial function in over-
on postprandial lipid metabolism. Br J Nutr 79, 419–424. weight and obese men: a randomized controlled trial.
37. Itoh K, Moriguchi R, Yamada Y, Fujita M, et al. (2014) High satu- Atherosclerosis 239, 178–185.
rated fatty acid intake induces insulin secretion by elevating 47. Kris-Etherton PM & Krauss RM (2020) Public health guidelines
gastric inhibitory polypeptide levels in healthy individuals. should recommend reducing saturated fat consumption as
Nutr Res 34, 653–660. much as possible: YES. Am J Clin Nutr 112, 13–18.
38. Christiansen E, Schnider S, Palmvig B, et al. (1997) Intake of a 48. Astrup A, Magkos F, Bier DM, et al. (2020) Saturated fats and
diet high in trans monounsaturated fatty acids or saturated fatty health: a reassessment and proposal for food-based recommen-
acids. Effects on postprandial insulinemia and glycemia in dations: JACC state-of-the-art review. J Am Coll Cardiol 76,
obese patients with NIDDM. Diabetes Care 20, 881–887. 844–857.