Reducing risk factors for coronary heart disease (CHD) is a major objective, especially where CHD is a prevalent cause of morbidity and death . Decreased CHD risk is a feasible benefit flowing from reductions in plasma total cholesterol (Tc)  [which mostly reflects low-density lipoprotein (LDL) cholesterol ] and in triglycerides (TG)  of TG-rich lipoproteins  (which include very low-density lipoprotein and LDL ). An increase in plasma antioxidants , and reduction in resting blood pressure (BP) and clotting  also decrease CHD risk.
Animal studies  demonstrate favourable influences upon CHD risk factors with ingestion of vegetable garlic or garlic powder (GP) or steam-distilled garlic oil (GO). Allicin is formed enzymatically from alliin when fresh garlic is crushed or GP is dissolved in water, and GO sulphides are allicin derivatives produced by the steam-distillation process. Allicin and GO sulphides are the biologically active compounds obtainable from garlic and, according to recent study , they share similar metabolic fates.
Around 50 human trials support a reduction in Tc by garlic and most of these also show reductions in TG. Two meta-analyses [9,10] confirmed significant decreases in plasma Tc of about 10%, but excluded most trials because of poor design or description. Most meta-analysed studies have involved hyperlipidaemic subjects given 0.6–0.9 g GP/day, equivalent to around 1.5–2.3 g fresh garlic. A recent review  has advocated a need for strict implementation and documentation of diet and clear disclosure of appropriate clinical trial design features where, as in garlic studies, ‘expected changes in lipid concentration are relatively small'. Since publication of the meta-analyses, such features of good design have been incorporated in five trials [12–16] yielding negative findings for cholesterol-lowering that have fuelled scepticism [17,18] for the lipid-lowering effect of garlic. Four of these involved recently re-formulated ‘Kwai’ tablets [12–15], and the other tested ‘Tegra’ tablets providing 10 mg GO/day in a cyclodextrin matrix . None of these recent studies provided analyses of the administered doses for alliin-derived components. Moreover, batches of tablets current at the time of these negative studies have been analysed under simulated in vivo conditions, with results indicating low availability of allicin  from ‘Kwai’ GP tablets and of GO sulphides  from ‘Tegra’ GO tablets. A recent meta-analysis  relies substantially upon such studies [eg 13,14,16] which use supplements of low or uncertain activity [19,20]. Thus its’ conclusion that ‘.. use of garlic for hypercholesterolemia is ... of questionable value’ would be more appropriately limited to the garlic supplements used in these studies.
Further controversy associated with ‘Kwai’ tablets  has concerned the authenticity  of angiographic results evidencing reversal of the atherosclerotic process. However, official vindication of these findings has recently been provided . In vitro, aged garlic extract increased the resistance of isolated human LDL to copper-induced oxidation . Also, LDL isolated from persons who ingested aged garlic extract tended to have enhanced oxidation resistance . The latter phenomenon was also observed in an early study of the intake of ‘Kwai’ GP tablets . Later studies [12,28] were negative, but are questionable since they involved samples [12,13] in which allicin was probably of low availability in vivo (see earlier). In 15 studies of hypertensives  given GP, reductions in resting BP averaging 6.7% (systolic) and 7.9% (diastolic) were observed. In addition, garlic intake has been shown to lower blood coagulability by reducing fibrinogen , increasing fibrinolysis  and prothrombin time , and inhibiting platelet aggregation .
Almost exclusively, garlic studies have centred on individuals identified with CHD risk factors and have employed conservative dose regimes . This is surprising in the context of widespread use of garlic by persons lacking patent risk of CHD and the fact that daily ingestion of 15–20 g raw garlic has no observable adverse effects .
No previous controlled human trials have evaluated the influence of GO upon resting BP or blood coagulability, or effects of any garlic material upon LDL lipid, LDL total antioxidant status (TAS), or plasma TAS. Therefore, in the present study, we report on garlic effects upon major CHD risk factors and address a population of trained male runners, at low risk of CHD, who were administered GO that delivers a high dose of bioactive sulphides.
Materials and methods
The dose period extended from attendance for baseline measurements until the end of the trial 16 weeks later. During the baseline visit, volunteers were allocated randomly and double-blinded to GO or placebo groups by a person not involved in the trial.
Eligibility criteria were male gender, age range 17–45 years, absence of chronic ailments and regular long-distance running. Recruitment was initiated by contacts with Midlands region (UK) running club coaches. Exclusion criteria were current intake of garlic supplements and the presence of chronic ailments. In accordance with the Helsinki Declaration plus modifications, the candidate group received oral and typed information detailing the study aims and requirements. The study was approved by the Ethics Committee of the University of Wolverhampton. Equal numbers of subjects were present in each group at the commencement of the treatment phase. Timing and reasons for dropout were recorded for subjects lost to follow-up.
Doses consisted of three capsules of GO (Cardiomax; Seven Seas Ltd, Hull, UK) in vegetable oil or three placebo capsules of identical appearance containing vegetable oil only. The capsules contained an average of 4.1 mg GO sulphides. All subjects were asked to take the dose daily at a specific time. Completed 1-week diet diaries were also requested at each attendance to enable consistency in food and drink intakes to be evaluated and encouraged during the study.
Overnight fasts (12–14 h) were followed by collection of venous blood at each attendance. Blood required for biochemical and haemostatic variables was dispensed into appropriate tubes. Blood required for LDL isolation was collected in lithium-heparin tubes, and plasma was separated (6 min × 6000 rpm at 0–5°C) and used for analysis of apoprotein B (apo B) and vitamin E.
Blood pressure measurement
Following blood sampling, BP and pulse were determined on the upper arm using a medical digital BP meter (Model UA-751; Takeda, Aachen, Germany). Measurements were performed over a period of 10–30 min in a quiet standardized environment until repeated low values were obtained. The means of the three lowest pulse rates plus associated systolic and diastolic BP values were utilized.
Laboratory analytical methods
Vitamin E present in plasma samples was quantified [coefficient of variance (CV) = 4%] using a C-18 reverse-phase column (Phenomenex Ltd, Macclesfield, UK) with methanol as mobile phase at a flow rate of 2 ml/min and an ultraviolet detector (Pye-Unicam Ltd, Cambridge, UK) set at 292 nm . Sulphides in GO capsules were analysed by high-performance liquid chromatography . Routine clinical analyses were carried out on plasma samples in the Departments of Chemical Pathology and of Haematology, The Manor Hospital, Walsall, UK. Lipid variables, TG (CV ≤ 2%), Tc (CV ≤ 2%) and high-density lipoprotein cholesterol (CV ≤ 4%), were measured with a Beckman CX7 automated analyzer (Beckman Instruments, High Wycombe, UK). LDL cholesterol was calculated from these values using the Friedewald formula . Plasma and LDL TAS measurement was by the enhanced chemiluminescence method  (CV = 5.2% at 408 mmol Trolox equivalent/l). Plasma apo B was determined turbidometrically (CV < 3%) after reaction with apo B antibody (Sigma-357 Diagnostics Kit, Sigma, Poole, UK). Coagulation variables, prothrombin time (CV < 3%), partial thromboplastin time (CV < 3%) and fibrinogen (CV < 5%), were measured with a Sysmex CA 1000 Coagulation Analyser (Sysmex (TOA) Medical Electronics, Milton Keynes, UK), and platelet numbers (CV < 3.2%) and size used a Coulter STKS full blood count analyser (Beckman–Coulter, High Wycombe, UK).
LDL separation and analysis
LDL separation was performed by the method of Graham et al.  using Iodixanol (`OptiPrep'; Nycomed Pharma AS, Oslo, Norway) adapted for use with large diameter tubes (`OptiSeal’ 36.2 ml; Beckman) in VAC50 or VTi50 vertical rotors with Beckman L8-60M or XL-70M ultracentrifuges (50 000 rpm and 16°C for 3 h). The distribution of cholesterol and triglycerides helped to locate the LDL position within the fractions collected. Identification of lipoprotein classes present in fractions was achieved by agarose gel electrophoresis (`Hydragel Lipo+Lp(a)'; Sebia, Issy-les-Moulineaux, France). LDL fractions were pooled and purified by passage down a Sephadex G-25 column (Amersham Pharmacia Biotech, Uppsala, Sweden). Lipids in post-column LDL fractions were measured using a 96-well plate-reader (Labsystems, Helsinki, Finland) with kits from bioMerieux (Basingstoke, UK): ‘Cholesterol Enzymatique PAP’ for cholesterol (CH) (CV < 1%) and ‘Triglycerides Enzymatique PAP’ for TG (CV < 2%). Total protein (TP) was measured (CV < 5%) with the ‘Protein Assay Kit’ (Bio-Rad, Hercules, California, USA) LDL lipids were expressed as ratios of LDL TP (i.e. LDL TG/TP and CH/TP) and LDL TAS was expressed as a ratio of LDL TG (i.e. LDL TAS/TG).
Statistical analytical methods
Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS) (Version 7.5; SPSS Inc., Chicago, Illinois, USA). Baseline means and standard deviations (SDs) of the combined GO and placebo treatment group of subjects completing follow-up and of the small numbers lost to follow-up were compared by visual inspection to assess any bias introduced by such losses. In addition, randomization balance was assessed between treatments for subjects completing the study, and inter-relationships of variables at baseline and at the end of the trial were investigated by correlation analysis using Spearman's Rank test.
The main outcomes concerned the changes in values to the end of the 16-week treatment period. Mean changes with GO were compared with those with placebo using 95% confidence intervals and P values.
This study was the first upon lipid variables of GO given to healthy subjects and, for several other variables measured, was the first with any garlic material. Realistic effect sizes were therefore unavailable so that statistical power calculations were not utilized to decide the numbers of subjects recruited. Numbers were, instead, chosen to approximate earlier GP studies upon plasma lipids in healthy subjects [39,40]. These provided lower bulb equivalent garlic doses than that selected for this study.
Questionnaire and baseline analysis
Of the subjects randomized to each of GO and placebo, three per group failed to complete the baseline part of the study (Table 1). Of the 15 subjects per group that completed baseline analysis, one GO and two placebo group subjects were lost prior to completion of the study. Subjects completing the study were of white Caucasian ethnicity and the mean ± SD age was 28.8 ± 8.2 years.
Questionnaire returns indicated that dose-taking and fasting instructions were well-adhered to in both treatment groups. Few subjects took dietary supplements in the 3 weeks preceding or during the study, and no one in the GO group took antioxidant vitamins. Insufficient numbers were available to subdivide the results for blinding by subject guesses but inspection of the results suggested that blinding was partly effective.
Mean ± standard error (SE) changes in percentage of total energy intakes as carbohydrates (GO, 1.5 ± 5.7% versus PL, 0.7 ± 4.0%), fats (GO, −2.2 ± 3.5% versus PL, 3.0 ± 2.1%) and proteins (GO, 2.9 ± 2.4% versus PL, −4.4 ± 2.8%) were small by the end of the trial, and there were no significant differences between GO and PL subjects. Similarly, there were few differences for a wide range of water-soluble vitamins and minerals. Alcohol consumption patterns were generally moderate, stable and similar for both groups.
There were only small changes in 70% of VO2max test duration during the intervention period (GO, 6 ± 4 min versus PL, 5 ± 10 min), with little difference between the two subject groups. There were similar findings for lung function, indicating comparable levels of cardiorespiratory performance capacity during the trial.
Mean baseline levels were within normal ranges for clinical measurements, and were similar to values reported in previous studies for LDL variables . Generally, higher plasma  and LDL TAS  were observed compared with published values obtained by the enhanced chemiluminescence method.
Effects upon plasma cholesterol variables including (calculated) LDL cholesterol were small and failed to approach significance (Table 2). On average, plasma TG in subjects taking GO fell, by 0.06 mmol/l, in contrast with a rise of 0.14 mmol/l in subjects taking placebo. The effect of GO on TG was estimated by the difference between these two means, −0.06 −0.14 = −0.20 mmol/l, i.e. the effect of GO was to reduce TG levels by 0.20 mmol/l, on average. This effect was null statistically, although well correlated (r = 0.80, n = 18, P ≤ 0.001) with LDL TG/TP, which showed significant reduction for GO over placebo. A slight trend towards reduction in LDL CH/TP was also observed.
After the baseline variable was forced into multiple regression analyses to allow for baseline differences, with the dependent being the change variable, the estimate of GO effect upon plasma TG (P = 0.35), and LDL TG/TP (P = 0.20) diminished. Nonetheless, when split by median baseline levels, the descriptive trends in these variables remained consistent with a GO effect. LDL lipid reduction was also supported by the concurrence of negative correlation at baseline between LDL TG/TP and LDL density (r = −0.44, n = 25, P = 0.03) with a trend to lower LDL density in the GO group.
Plasma apo B and LDL TP were well correlated (r = 0.66, n = 18, P = 0.002), supporting the absence of significant plasma protein contamination of the LDL samples and the correspondence of LDL TP to apo B. The falls in both LDL TP and plasma apo B with GO compared with placebo are consistent with a slight reduction in the number of LDL particles.
No effects of GO upon the possible plasma antioxidants , albumin, urate or bilirubin were observed, and the positive trend in plasma TAS with GO was small and far from significant. In the combined GO and placebo group, strong correlations of plasma TAS levels with plasma urate were observed at baseline (r = 0.77, n = 26, P < 0.001), and of changes in TAS with those in plasma urate at the end of the study (r = 0.40, n = 26, P = 0.044). Good correlation was observed for plasma vitamin E with plasma Tc (r = 0.49, n = 26, P = 0.014), high-density lipoprotein cholesterol (r = 0.59, n = 23, P = 0.003) and apo B (r = 0.50, n = 26, P = 0.009), and with the ratios LDL TAS/TP (r = 0.54, n = 26, P = 0.005) and LDL TAS/TG (r = 0.49, n = 26, P = 0.011). These correlations appropriately reflect the established importance of vitamin E as a plasma lipoprotein antioxidant. LDL TAS/TP changes were similar between groups (data not shown), indicating no GO effect upon TAS per LDL particle.
Strong negative correlations at baseline of LDL TAS/CH with LDL CH/TP (r = −0.59, n = 27, P = 0.001) and of LDL TAS/TG with LDL TG/TP (r = −0.75, n = 27, P < 0.001) were observed, indicating an association of greater LDL TAS/TG with less fatty LDL particles. Therefore, the trend to greater LDL TAS/TG with GO treatment may reflect the relative fall in LDL TG per LDL particle in the GO group.
For BP, effects of GO were null statistically, although trends were favourable. Prothrombin and partial thromboplastin times were greater by 0.5–1.0 SD of baseline values with GO than placebo and, although null statistically, differences approached significance. Effects of garlic upon platelet count and volume have not previously been reported but, as for fibrinogen, these variables were unaffected by GO treatment.
Overall, our results are similar to those of recent negative placebo-controlled trials [12–16] that also studied the effects of garlic upon plasma lipids. However, these trials used products likely to provide low levels of active garlic components in vivo (see Introduction), whereas good in vivo availability of GO was expected for the type of capsules utilized in our study . This may help explain why positive trends were less evident in these trials than in the present study. The mainly small and non-significant effects of GO in our study may reflect the favourable initial values of the CHD risk factors of the healthy study group involved, which contrasts with the hyperlipidaemia of subjects tested in other recent negative studies. This concept is illustrated by statin treatment where blood lipid-lowering is proportional to initial lipid levels  and the findings of earlier trials of garlic materials involving normal subjects, where null  or modest (−6.5% and −4.4%) plasma Tc responses were observed.
The possible reductions of plasma TG, LDL TG/TP and LDL density that suggest a lipid-lowering effect of GO in our study are comparable with effects reported in another largely negative garlic trial . Thus, in the latter study, GP intake resulted in a trend to lower plasma TG, especially in LDL subclass A subjects among whom this coincided with a significant reduction of mean peak LDL particle diameter relative to the placebo group.
The absence of correlation between plasma TAS and LDL TAS indicates that plasma TAS did not provide a measure of LDL TAS status. This can be explained by our confirmatory data (see Results) in support of published evidence for the dependence of plasma TAS upon urate levels  and of LDL TAS upon vitamin E levels . Trends suggestive of beneficial effects of GO upon LDL TAS/TG observed in this study could reflect GO effects upon LDL lipid composition rather than upon LDL antioxidants as such.
The effects of GO upon blood pressure did not reach significance, although the size of the reductions observed were similar to those reported in previous garlic studies utilizing GP . The small trends suggestive of reduced coagulability are consistent with earlier evidence for increased prothrombin time  with bulb garlic intake. These findings were attributed to phytic acid; however, GO consists exclusively of garlic sulphides  and we cannot discount the possibility that these allicin derivatives may be capable of such effects. Confirmation of these results could indicate that GO acts to prevent slight increases in levels of clotting factor X and/or factor V and/or prothrombin . The reduction of platelet aggregation and adhesiveness , fibrinogen levels  and increased fibrinolysis  reported for other garlic materials was not observed here with GO.
Most people eating garlic lack overt CHD risk and our results give no indication via statistical significance of much benefit to such persons. The favourable trends observed would need greatly increased numbers of subjects to be substantiated, but even then such a study would not reflect the real-life situation in which garlic consumption may occur over many years. Such a study would be a large undertaking, but our trial suggests that it might be worthwhile. It would, anyway, allow clear testing of the possible beneficial effects for several CHD risk factors suggested by the small but non-significant trends observed.
In conclusion, unlike most previous trials, the present study delivered levels of GO equivalent to high bulb garlic doses and focused on individuals with low coronary risk profile who are more likely to reflect the situation of the majority of the garlic-ingesting public. Favourable trends were observed for a number of CHD risk factors and a trial of approximately 150 subjects is required to confirm or refute these findings (at the 5% significance level with 80% power).
Seven Seas Ltd (Hull, UK) are thanked for part-funding the project and providing the ‘Cardiomax’ and placebo capsules. The authors thank Dr LD Lawson, Research Scientist, Nature's Way Products, Murdock Madaus Schwabe Group (Springville, Utah, USA) for high-performance liquid chromatography analysis of GO in ‘Cardiomax’ capsules. The authors also thank David A. Roser MCIM, Director of the Garlic Research Bureau, Bury St Edmunds, UK for encouraging them to carry out this study.
1. Pyorala K. CHD prevention in clinical practice. Lancet 1996, 348: S26 –S28.
2. Castelli WP. Lipids, risk factors and ischaemic heart disease. Atherosclerosis 1996, 124: S1 –S9
3. Rosengren A. Cholesterol: how low is low enough? BMJ 1998, 317: 425 –426.
4. Hokanson JE, Austin MA. Plasma triglyceride level is a risk factor for cardiovascular disease independent of high-density lipoprotein cholesterol level: a meta-analysis of population-based prospective studies. J Cardiovasc Risk 1996, 3: 213 –219.
5. Grundy SM. Hypertriglyceridemia, atherogenic dyslipidemia, and the metabolic syndrome. Am J Cardiol 1998, 81: 18B –25B.
6. Steinberg D. Clinical trials of antioxidants
in atherosclerosis: are we doing the right thing? Lancet 1995, 346: 36 –38.
7. Reuter HD, Koch HP, Lawson LD. Therapeutic effects and applications of garlic and its preparations. In: Koch HP, Lawson LD (editors):Garlic: the science and therapeutic application of
Allium sativum L. and related species
2nd ed. Baltimore, MD: Williams & Wilkins, 1996. pp. 135 –212.
8. Lawson LD. Garlic: a review of its medicinal effects and indicated active compounds. In: Lawson LD, Bauer R (editors):Phytomedicines of Europe: their chemistry and biological activity. ACS Symposium Series 691.
Washington DC: American Chemical Society, 1998. pp. 176 –209.
9. Warshafsky S, Kamer RS, Sivak SL. Effect of garlic on total serum cholesterol. : A meta-analysis. Ann Intern Med 1993, 119: 599 –605.
10. Silagy C, Neil A. Garlic as a lipid lowering agent — a meta-analysis. J R Coll Phys London 1994, 28: 39 –45.
11. Berthold HK, Sudhop T, von Bergmann K. Effect of garlic oil
preparation on serum lipoproteins and cholesterol metabolism. JAMA 1998, 279: 1900 –1902.
12. Simons LA, Balasubramaniam S, von Konigsmark M, Parfitt A, Simons J, Peters W. On the effect of garlic on plasma lipids
and lipoproteins in mild hypercholesterolaemia. Atherosclerosis 1995, 113: 219 –225.
13. Neil HAW, Silagy CA, Lancaster T, Hodgeman J, Vos K, Moore JW. et al. Garlic powder in the treatment of moderate hyperlipidaemia: a controlled trial and a meta-analysis. J R Coll Phys London 1996, 30: 329 –334.
14. Isaacsohn JL, Moser M, Stein EA, Dudley K, Davey JA, Liskov E. et al. Garlic powder and plasma lipids
and lipoproteins: a multicenter, randomized, placebo-controlled trial. Arch Intern Med 1998, 158: 1189 –1194.
15. Superko HR, Krauss RM. Garlic powder, effect on plasma lipids
, post-prandial lipemia, low-density lipoprotein
subclass distribution and lipoprotein (a). J Am Coll Cardiol 2000, 35: 321 –326.
16. Berthold HK, Sudhop T, von Bergmann K. Effect of garlic oil
preparation on serum lipoproteins and cholesterol metabolism. J Am Med Assoc 1998, 279: 1900 –1192.
17. Beaglehole R. Garlic for flavour, not cardioprotection. Lancet 1996, 348: 1186 –1187
18. Wunsch H. Garlic-oil effects debunked in new study. Lancet 1998, 351: 1864. 1864.
19. Lawson LD. Garlic powder for hyperlipidemia — analysis of recent negative results. Q Rev Nat Med
1998, Fall: 187–189.
20. Lawson LD. Garlic Oil
for hypercholesterolemia — negative results. Q Rev Nat Med
1998, Fall: 185–186.
21. Stevinson C, Pittler MH, Ernst E. Garlic for treating hypercholesterolemia. : A metaanalysis of randomised clinical trials. Ann Intern Med 2000, 133: 420 -429.
22. Koscielny J, Latza R, Schmitt R, Radtke H, Siegel G. et al. The antiatherosclerotic effect of Allium sativum
. Atherosclerosis 1999, 144: 237 –249.
23. Schiermeler Q. German garlic study under scrutiny. Nature 1999, 401: 629. 629.
24. Koscielny J, Schmitt R, Radtke H, Latza R, Kiesewetter H. Garlic study vindicated by official investigation. Nature 2000, 404: 542. 542.
25. Ide N, Nelson AB, Lau BHS. Aged garlic extract and its constituents inhibit Cu2+
-induced oxidative modification of low density lipoprotein. Planta Med 1997, 63: 263 –264.
26. Steiner M, Lin RS. Changes in platelet function and susceptibility of lipoproteins to oxidation associated with administration of aged garlic extract. J Cardiovasc Pharmacol 1998, 38: 904 –908.
27. Phelps S, Harris WS. Garlic supplementation reduces the susceptibility to oxidation of apoprotein B-containing lipoproteins. Lipids 1993, 28: 475 –477.
28. Byrne DJ, Neil HA, Vallance DT, Winder AF. A pilot study of garlic consumption shows no significant effect on markers of oxidation or sub-fraction composition of low-density lipoprotein
including lipoprotein(a) after allowance for non-compliance and the placebo effect. Clin Chim Acta 1999, 285: 21 –33.
29. Bordia A, Bansal HC, Arora SK, Rathore AS, Ranawat RVS, Singh SV. Effect of the essential oil (active principle) of garlic on serum cholesterol, plasma fibrinogen, whole blood coagulation time, and fibrinolytic activity in alimentary lipaemia. J Assoc Phys India 1974, 22: 267 –270.
30. Bordia A, Bansal HC, Arora SK, Rathore AS, Ranawat RVS, Singh SV. Effect of the essential oils of garlic and onion on alimentary hyperlipemia. Atherosclerosis 1975, 21: 15 –19.
31. Song CS, Kim JH, Kim ES, Lee PH. A blood anticoagulant substance from garlic (Allium sativum
). : I. Its preparation and studies on its anticoagulant effect. Yonsei Med J 1963, 4: 17 –20.
32. Makheja AN, Vanderhoek JY, Bailey JM. Inhibition of platelet aggregation and thromboxane synthesis by onion and garlic. Lancet 1979, 1: 781. 781.
33. Koch HP. Toxicity, side effects, and unwanted effects of garlic. In: Koch HP, Lawson LD (editors):Garlic: the science and therapeutic application of
Allium sativum L. and related species
, 2nd ed. Baltimore, MD: Williams & Wilkins; 1996. pp. 135 –212.
34. Nierenberg DW, Lester DC. Determination of vitamins A and E in serum and plasma using a simplified clarification method and high-performance liquid chromatography. J Chromatogr 1985, 345: 2752 –2784.
35. Lawson LD, Wang ZJ, Hughes BG. Identification and HPLC quantitation of the sulfides and dialk(en)yl thiosulphinates in commercial garlic products. Planta Med 1991, 57: 363 –370.
36. Mackness MI, Durrington PN. Lipoprotein separation and analysis for clinical studies. In: Converse CA, Skinner ER (editors):Lipoprotein analysis: a practical approach
. Oxford: IRL Press, 1992. pp. 1 –42.
37. Whitehead TP, Thorpe GH, Maxwell SRJ. Enhanced chemiluminescent assay for antioxidant capacity in biological fluids. Anal Chim Acta 1992, 266: 265 –277.
38. Graham JM, Higgins JA, Gillott T, Taylor T, Wilkinson J, Ford T. et al. A novel method for the rapid separation of plasma lipoproteins using self-generating gradients of Iodixanol. Atherosclerosis 1996, 124: 125 –135.
39. Saradeth T, Seidl S, Resch KL, Ernst E. Does garlic alter the lipid pattern in normal volunteers? Phytomedicine 1994, 1: 183 –185.
40. Barrie SA, Wright JV, Pizzorno JE. Effects of garlic oil
on platelet aggregation, serum lipids and blood pressure
in humans. J Orthomolec Med 1987, 2: 15 –21.
41. Marinetti GV. Disorders of lipid metabolism
. New York: Plenum; 1990.
42. Maxwell S, Holm G, Bondjers G, Wiklund O. Comparison of antioxidant activity in lipoprotein fractions from insulin-dependent diabetics and healthy controls. Atherosclerosis 1997, 129: 89 –96.
43. Maxwell SRJ, Wiklund O, Bondjers G. Measurement of antioxidant activity in lipoproteins using enhanced chemiluminescence. Atherosclerosis 1994, 111: 79 –89.
44. Frei B, Stocker R, Ames BN. Antioxidant defenses and lipid peroxidation in human blood plasma. Proc Natl Acad Sci USA 1988, 85: 9748 –9752.
45. Cullen P, Assmann G. Treatment goals for low-density lipoprotein
cholesterol in the secondary prevention of coronary heart disease: absolute levels or extent of lowering? Am J Cardiol 1997, 80: 1287 –1294.
46. Lau BHS, Lam F, Wang-Chang R. Effect of an odor-modified garlic preparation on blood lipids. Nutr Res 1987, 7: 139 –149.
47. Silagy CA, Neil HAW. A meta-analysis of the effect of garlic on blood pressure
. J Hypertens 1994, 12: 463 –468.
48. Dacie JV, Lewis SM. Practical haematology
, 8th ed. Edinburgh: Churchill Livingstone; 1995.