Berthold HK, Sudhop T, von Bergmann K. Effect of a Garlic Oil Preparation on Serum Lipoproteins and Cholesterol MetabolismA Randomized Controlled Trial. JAMA. 1998;279(23):1900-1902. doi:10.1001/jama.279.23.1900
From the Department of Clinical Pharmacology, University of Bonn, Bonn, Germany (Drs Berthold, Sudhop, and von Bergmann).
Context.— Garlic-containing drugs have been used in the treatment of hypercholesterolemia
even though their efficacy is not generally established. Little is known about
the mechanisms of action of the possible effects on cholesterol in humans.
Objective.— To estimate the hypocholesterolemic effect of garlic oil and to investigate
the possible mechanism of action.
Design.— Double-blind, randomized, placebo-controlled trial.
Setting.— Outpatient lipid clinic.
Patients.— We investigated 25 patients (mean age, 58 years) with moderate hypercholesterolemia.
Intervention.— Steam-distilled garlic oil preparation (5 mg twice a day) vs placebo
each for 12 weeks with wash-out periods of 4 weeks.
Main Outcome Measures.— Serum lipoprotein concentrations, cholesterol absorption, and cholesterol
Results.— Baseline lipoprotein profiles were (mean [SD]): total cholesterol, 7.53
(0.75) mmol/L (291  mg/dL); low-density lipoprotein cholesterol (LDL-C),
5.35 (0.78) mmol/L (207  mg/dL); high-density lipoprotein cholesterol
(HDL-C), 1.50 (0.41) mmol/L (58  mg/dL); and triglycerides, 1.45 (0.73)
mmol/L (127  mg/dL). Lipoprotein levels were virtually unchanged at the
end of both treatment periods (mean difference [95% confidence interval]):
total cholesterol, 0.085 (−0.201 to 0.372) mmol/L (3.3 [−7.8 to
14.4] mg/dL), P=.54; LDL-C, 0.001 (−0.242 to
0.245) mmol/L (0.04 [−9.4 to 9.5] mg/dL), P=.99;
HDL-C, 0.050 (−0.028 to 0.128) mmol/L (1.9 [−1.1 to 4.9] mg/dL), P=.20; triclycerides, 0.047 (−0.229 to 0.135) mmol/L
(4.2 [−20.3 to 12.0]) mg/dL, P=.60. Cholesterol
absorption (37.5% [10.5%] vs 38.3% [10.7%], P=.58),
cholesterol synthesis (12.7 [6.5] vs 13.4 [6.6] mg/kg of body weight per day, P=.64), mevalonic acid excretion (192  vs 187 
µg/d, P=.78), and changes in the ratio of lathosterol
to cholesterol in serum (4.4% [24.3%] vs 10.6% [21.1%], P=.62) were not different in garlic and placebo treatment.
Conclusions.— The commercial garlic oil preparation investigated had no influence
on serum lipoproteins, cholesterol absorption, or cholesterol synthesis. Garlic
therapy for treatment of hypercholesterolemia cannot be recommended on the
basis of this study.
GARLIC (Allium sativum) has been advocated
as a remedy for the treatment and prevention of a number of diseases. As a
pharmaceutical product, its putative cardioprotective properties, such as
lipid-lowering and blood pressure–lowering, antioxidant, antiplatelet,
and fibrinolytic effects,1,2 seem
interesting. Studies investigating garlic's lipid-lowering effect are sometimes
flawed in design because they lack adequate description of the methods and
patients studied or are overtly subjected to conflicts of interest. Meta-analyses
found overall effects of between 9% and 12% reduction of total cholesterol.3,4 However, the confidence in these data
is limited by the poor quality of the underlying studies and the possibility
of a publication bias in that there are fewer than expected studies reporting
negative results.5 Likewise, meta-analyses
based on published reports rather than on individual patient data may be misleading,6 implying that meta-analyses provide false-positive
test results.7 The value of meta-analyses as
accurate predictors of treatment outcome as compared to prospective randomized
controlled trials has been questioned.8 Well-designed
recent studies7,9 found no lipid-lowering
effects, while 3 other studies reported some efficacy.10- 12
We hypothesized that the modest lipid-lowering effect found in meta-analyses
may be further understood if more were known about the possible mechanisms
of the action of garlic-containing drugs on cholesterol metabolism. We designed
a double-blind, randomized, placebo-controlled, cross-over trial to investigate
possible influences of a garlic preparation on serum lipoproteins and cholesterol
metabolism. We used a steam-distilled garlic-oil preparation.
Patients with moderate hypercholesterolemia (total cholesterol, 6.2-9.0
mmol/L [240-348 mg/dL]; triglycerides, <3.0 mmol/L [<265 mg/dL])
recruited through the local newspaper. None of the 26 unpaid patients randomized
for the study (1 later dropped out because of a scheduling conflict) had taken
any lipid-lowering drugs or drugs that would interfere with lipid metabolism
for 8 weeks, though some were taking antihypertensive medication, hormone
replacement drugs, or thyroid hormones. After ensuring that patients who had
given consent to participate were free of active liver or renal diseases,
diabetes, thyroid dysfunction, a history of coronary heart disease, any pathological
laboratory values in the clinical chemistry or hematological routine parameters,
and alcohol or other drug abuse, they entered the study whose protocol had
been approved by the ethics committee of the faculty of medicine at Bonn University
and performed in accordance with Declaration of Helsinki guidelines.
Subjects were advised to adhere to their usual diet during the study,
but they were prohibited from taking additional garlic or other food supplements.
Food intake was assessed at the end of 2 treatment periods, using 7-day food
records that were evaluated by computerized nutrient analysis.
The study was a single-center, double-blind, randomized, placebo-controlled,
cross-over trial. A marketed enteric coated preparation (Tegra, Hermes Arzneimittel
GmbH, Munich, Germany) containing 5 mg of steam-distilled garlic oil bound
to a matrix of beta cyclodextrin or matching placebos whose coating tasted
like garlic, was used. The daily dosage corresponds to about 4 g to 5 g of
fresh garlic cloves or 4000 units of allicin-equivalents per day.13 The active ingredients are the stable sulfur compounds
diallyl disulfide (>30%) and diallyl trisulfide (>25%), which are formed from
alliin and allicin.
The patients were randomly assigned to the treatment sequences placebo-garlic
or garlic-placebo in blocks of 10 for the first 20 patients and in blocks
of 2 for the remaining patients. After randomization, the patients were given
placebo for 4 weeks in a single-blind fashion. Thereafter, they received the
garlic preparation or placebo for 12 weeks in a double-blind fashion. Then
a 4-week, single-blind placebo wash-out was performed followed by the 12-week,
double-blind cross-over phase. Lipoprotein concentrations from blood drawn
at the beginning and end of each phase were measured enzymatically using standard
laboratory procedures. High-density lipoprotein cholesterol (HDL-C) was determined
after anionic precipitation of apolipoprotein B–containing lipoproteins.
Low-density lipoprotein cholesterol (LDL-C) was calculated according to the
method of Friedewald.14
During the last week of each 12-week treatment period, cholesterol absorption
and endogenous cholesterol synthesis was measured by the double-isotope continuous
feeding method as described by Lütjohann et al.15
For this purpose the patients took 1 capsule containing [D6]cholesterol
and [D4]sitostanol 3 times a day for 1 week. The disappearance
of deuterated cholesterol and its intestinal bacterial products (coprostanol
and coprostanone) relative to deuterated sitostanol were measured in fecal
samples by gas chromatography-mass spectrometry. The capsules also contained
unlabeled sitostanol as a nonabsorbable fecal flow and recovery marker for
the measurement of fecal excretion of neutral and acidic sterols. The patients
kept a 7-day dietary protocol to determine their intake of nutrients and cholesterol.
Cholesterol synthesis was calculated by subtracting the amount of dietary
cholesterol intake from the sum of neutral and acidic sterols excreted in
feces.16 As an additional indicator of short-term
changes in endogenous cholesterol synthesis, 24-hour urinary excretion of
mevalonic acid was measured by gas chromatography–mass spectrometry
as described by Lindenthal et al.17 The measurement
of the cholesterol precursor lathosterol in serum was determined by gas chromatography.18
Statistical analysis between lipoprotein concentrations at the end of
both treatment periods was performed using t statistical
tests for cross-over designs (in the case of triglycerides after log transformation
of the data), after excluding carryover effects.19
Correlation between the change in the primary study parameter, low-density
lipoproteins (LDL), and the parameters of cholesterol metabolism was analyzed
using a simple linear regression model. For all tests a significance level
of P<.05 was defined. The study was powered at
a level of greater than 95% to detect differences between treatment periods
of 10% LDL-C lowering (or −0.52 mmol/L [−20 mg/dL]). Statistical
analyses were performed using StatView 4.1 for the Macintosh (Abacus Concepts
Inc, Berkeley, Calif) and Microsoft Excel 5.0a for the Macintosh (Microsoft
Inc, Redmond, Wash). All lipoprotein measurements were performed twice on
2 separate days. The average of the 2 values was used for calculations. Within-individual
coefficients of variation of the measures were 4.5% (total cholesterol), 6.2%
(LDL-C), 6.8% (HDL-C), and 17.4 (triglycerides), respectively. The laboratory's
precision in measurement of lipoproteins (day-to-day coefficient of variation)
was 0.99% (total cholesterol), 2.64% (LDL-C), 2.22% (HDL-C), and 1.14% (triglycerides).
Twenty-five subjects completed the study and 1 subject had to be excluded
from the fecal balance calculations because of incomplete intake of the marker
capsules in 1 test period. The baseline characteristics and the serum lipoprotein
profiles of the 25 patients are listed in Table 1.
The drug was generally well tolerated. Except for garlic odor and slight
abdominal discomfort in a few cases, caused by both pills, no serious adverse
events occurred. Laboratory safety parameters remained in the normal range.
Compliance as measured by pill count was excellent and averaged 98.4%±6.3%
(mean±SD) during all phases. During active-drug treatment phase, none
of the subjects had a medication intake of less than 88%.
Evaluation of the two 7-day food records showed that macronutrients,
cholesterol, fiber, and alcohol were consumed similarly during both phases.
Body weights remained constant during the entire course of the study (Table 2). Lipoprotein concentrations were
virtually unchanged between placebo and active-drug treatment (Table 1). There was a slight increase in all lipoprotein fractions
during active-drug treatment compared with placebo, none statistically significant.
The post hoc calculated power of the study of 93.8% would have been able to
detect differences in the primary study parameter of LDL-C of greater than
or equal to −0.429 mmol/L (−16 mg/dL) between the 2 pills.
There were virtually no effects of garlic drug on the parameters of
cholesterol metabolism (placebo vs active drug; mean [SD] values): cholesterol
absorption (38.3% [10.7] vs 37.5% [10.5%], P=.58),
cholesterol synthesis (13.4 [6.6] vs 12.7 [6.5] mg/kg of body weight per day, P=.64), or mevalonic acid excretion in urine (187 
µg/d vs 192  µg/d, P=.78). Changes
in the ratio of lathosterol to cholesterol were not statistically different
during either treatment (garlic, 4.4% [24.3%]; baseline, 1.30 [0.42] µg/mg;
placebo, 10.6% [21.1%]; baseline, 1.18 [0.35] µg/mg, P=.62). Simple linear regression analyses between changes in serum
LDL-C and cholesterol absorption, cholesterol synthesis, mevalonic acid excretion,
or the ratio of lathosterol to cholesterol revealed no significant correlations
(cholesterol absorption, r=0.26, P=.22; cholesterol synthesis, r=0.17, P=.43; mevalonic acid excretion, r=0.11, P=.61; ratio of lathosterol to cholesterol, r=0.05, P=.81).
We evaluated the effects of a commercially available garlic preparation
using a double-blind, randomized, placebo-controlled study design. No changes
in serum lipoprotein levels in patients with moderate hypercholesterolemia
were found. Although 2 meta-analyses and a recent study had shown small but
significant effects of garlic on serum lipoprotein levels,3,4,11
there were 2 other well-designed studies that found no influence.7,9 Thus, the overall evidence for a positive
effect of garlic on serum lipid levels is questionable.
We have addressed some new questions yet to be elucidated during treatment
with garlic preparations. During trials with lipid-lowering substances, it
is important to exclude changes in body weight or dietary habits, especially
total calories, fat, and cholesterol content of the diet.
Earlier studies were often criticized for dosage, duration of treatment,
and baseline cholesterol values. Most studies used dried garlic powder preparations
in doses from 600 to 900 mg/d, the equivalent of 1.8 to 2.7 g/d of fresh garlic.
Nonpowder preparations, however, seem according to the literature to have
a stronger lipid-lowering effect than powder preparations, although their
effects showed also a greater heterogeneity.4
Few studies have used steam-distilled garlic oils or oil-macerated garlic.20 These preparations contain only polysulfides and
other volatile thioallyls. Based on comparisons of the content of active ingredients,
the dosage of our study medication would be relatively high. The duration
of treatment is assumed to be sufficient to document changes in serum lipoproteins.
To circumvent the notion that garlic lowers only elevated cholesterol levels,21 our baseline levels were high enough (cholesterol,
7.53±0.75 mmol/L [291±29 mg/dL] and LDL-C, 5.35±0.78
mmol/L [207±30 mg/dL]). Although the effects of garlic on serum lipoprotein
levels have been studied extensively, very little is known about its possible
mechanism of action. In vitro data in rat hepatocytes suggest that allicin
and ajoene inhibit cholesterol synthesis at various steps or inhibit acetate
uptake into liver cells, respectively.22,23
Validated in vivo methods for measurement of cholesterol synthesis in humans
include the sterol balance technique, the determination of mevalonic acid
in 24-hour urine, and the measurement of sterol precursors in serum.24 It has been shown that the ratio of the cholesterol
precursor lathosterol to cholesterol in serum is a reliable indicator of cholesterol
synthesis because it closely reflects the activity of hepatic 3-hydroxy-3-methylglutaryl
coenzyme A reductase.18 In this study an influence
of garlic on cholesterol synthesis, based on determinations of 3 indicators,
could be excluded. Furthermore, it could be shown that absorption of cholesterol
was not affected by garlic. On the basis of these findings, it can be concluded
that cholesterol metabolism at multiple metabolic sites is not influenced
by garlic at least with the pharmaceutical formulation used in this study.
Moreover, individual changes in the effect of garlic on LDL-C concentrations
did not correlate with any of the parameters of cholesterol metabolism, so
that the conjecture of possible specific effects that are overridden by counterbalancing
effects seems to be excluded.
Based on a meta-analysis from 1994, the total patient experience in
randomized trials amounted to only 1365 individuals until then. This is a
surprisingly low number, assuming that garlic may be effective in reducing
elevated lipid levels without harmful side effects. Based on the results of
the present study, however, there is no evidence to recommend garlic therapy
for lowering serum lipid levels.