Stewart R, White LR, Xue QL, Launer LJ.
Twenty-six-year change in total cholesterol levels and incident dementia: the Honolulu-Asia Aging Study.
Arch Neurol. 2007 Jan;64(1):103-7.
Please login to recommend the paper.
To make a comment you must login or register.
Cholesterol has received a lot of attention as a potential modifiable risk factor for dementia and AD. Interestingly, experimental studies have linked disturbances in cholesterol homeostasis with all major neuropathological features of AD. Some long-term epidemiological studies have indicated that high serum cholesterol at midlife may increase the risk of AD later in life (Notkola et al., 1998; Kivipelto et al., 2002; Whitmer et al., 2005). However, shorter term follow-up studies in older populations have reported controversial results.
In this study, Stewart and colleagues studied changes in total cholesterol (TC) from midlife to late life in the well-described cohort of the Honolulu-Asia Aging Study (HAAS). HAAS has several strengths, including a large population sample, extensive long-term follow-up (26 years), and multiple TC measurements. The study indicated that TC levels in men with AD had declined at least 15 years before the diagnoses. This trend of change remained significant even after adjustments for a large scale of potential confounders. The decline in TC was strongest among the ApoE4 carriers and those with self-reported worse general health. At the baseline examination, TC levels did not differ by later dementia status.
The main results of the HAAS study are in line with our recent results from the Cardiovascular Risk factors, Aging and Dementia (CAIDE) study. Serum TC levels decreased in most individuals, but the decline was more rapid among those who later developed dementia and MCI (Solomon et al., Neurology, in press). A moderate decrease in serum TC from midlife to late life remained significantly associated with the risk of having a more impaired late-life cognitive status even after adjustments for several confounding factors.
However, in contrast to the HAAS study, in the CAIDE study elevated TC at midlife represented a risk factor for more severe impairment in cognitive functioning later in life. This difference between the HAAS study and our results may be partly due to differences in characteristics of the populations, especially regarding TC levels, which are higher in Finland. Nevertheless, similar findings in two populations that are dissimilar with respect to genetics, lifestyle, and basic level of vascular risk factors give further support to the hypothesis that declining cholesterol levels after midlife may be related to early stages in dementia development.
Significance of serum total cholesterol: risk factors vs. risk marker
These recent results support the idea that the relationship between serum TC and dementia may be bidirectional. High midlife serum TC may be a risk factor for subsequent dementia/AD, but decreasing serum TC after midlife may reflect ongoing disease processes (dementia, other diseases, frailty) and may represent a risk marker for late-life cognitive impairment. This may at least partly explain why short-term follow-up studies with older populations at baseline have led to conflicting results concerning the cholesterol-dementia association.
Other factors such as blood pressure and body mass index also seem to decrease before dementia onset. However, the findings from the HAAS indicate a different pattern of changes: additional decline in cholesterol may start much earlier than decline for other factors. Given that the exact onset of AD (before it becomes Alzheimer dementia) cannot be identified with currently available means, it is particularly important to have studies with long follow-up periods, starting at a time when AD is less likely to be present (such as midlife). Otherwise, true risk associations may be masked by reverse causality, and risk factors may even appear protective.
Implications and future directions
The mechanisms behind this pattern of cholesterol change over time need to be further clarified. Regarding the cholesterol-dementia relationship, it is important to keep in mind that serum and brain cholesterol are two separate pools, and their interactions are not entirely understood. Oxysterols (particularly 24- and 27-hydroxycholesterol) may be one of the missing links between cholesterol and AD (Björkhem et al., 2006; Leoni et al., 2006).
From a clinical point of view, the pattern of cholesterol change indicated by both HAAS and CAIDE should not lead clinicians into believing that cholesterol-lowering treatments are dangerous for elderly persons. High TC carries risk even in old age, and results from clinical trials in vascular diseases support the benefit of lipid-lowering treatment in elderly patients (Shepherd et al., 2002). Low TC may be a life-long (“primary”) low TC or secondary to different diseases (including AD). The interpretation of low TC levels in old age needs thus to be done with respect to the patients´ health and cognitive status.
References:Stewart R, White LR, Xue Q-L, Launer LJ. Twenty-six-year change in total cholesterol levels and incident dementia. Arch Neurol 2007; 64:103-7. Abstract
Notkola IL, Sulkava R, Pekkanen J, Erkinjuntti T, Ehnholm C, Kivinen P, Tuomilehto J, Nissinen A. Serum total cholesterol, apolipoprotein E epsilon
4 allele, and Alzheimer's disease. Neuroepidemiology 1998;17(1):14-20. Abstract
Kivipelto M, Helkala EL, Laakso MP, Hanninen T, Hallikainen M, Alhainen K, Iivonen S, Mannermaa A, Tuomilehto J, Nissinen A, Soininen H. Apolipoprotein E epsilon4 allele, elevated midlife total cholesterol level, and high midlife systolic blood pressure are independent risk factors for late-life Alzheimer disease.
Ann Intern Med. 2002 Aug 6;137(3):149-55. Summary for patients in: Ann Intern Med. 2002 Aug 6;137(3):I-18.
Whitmer RA, Sidney S, Selby J, Johnston SC, Yaffe K. Midlife cardiovascular risk factors and risk of dementia in late-life. Neurology. 2005; 64(2):277-281. Abstract
Solomon A, Kåreholt I, Ngandu T, Winblad B, Nissinen A, Tuomilehto J, Soininen H, Kivipelto M. Serum cholesterol changes after midlife and late-life cognition: 21-year follow-up study. Neurology, in press.
Björkhem I, Heverin M, Leoni V, Meaney S, Diczfalusy U. Oxysterols and Alzheimer´s disease. Acta Neurol Scand 2006; 114 (Suppl 185):43-49. Abstract
Leoni V, Shaafti M, Solomon A, Kivipelto M, Björkhem I, Wahlund L-O. Are the CSF levels of 24S-hydroxycholesterol a sensitive biomarker for early Alzheimer’s disease? Neurosci Lett. 2006 Apr 10-17; 397(1-2):83-7. Abstract
Shepherd J, Blauw GJ, Murphy MB. PROSPER Study Group. PROspective Study of Pravastatin in the Elderly at Risk. Pravastatin in elderly individuals at risk of vascular disease (PROSPER): a randomised controlled trial. Lancet 2002; 360:1623-1630. Abstract
The relative importance of risk factors for cardiovascular disease in the risk of dementia has gained increasing attention over the past decade. Cholesterol, hypertension, and diabetes have all been suggested to be associated with an increased risk of dementia. The potential role of cholesterol in the pathophysiology of dementia and Alzheimer disease (AD) is particularly interesting. Late-life cholesterol does not appear to be a risk factor for AD, but Kivipelto and colleagues have shown that elevated mid-life cholesterol is a risk factor for dementia and AD . Many groups, including Kivipelto’s, report that cholesterol levels appear to decline prior to the onset of dementia, which might account for why late-life elevated cholesterol is not a risk factor for dementia or AD [1-4]. Two major questions in the field are to understand how important these changes in cholesterol are to the pathophysiology of AD and the extent to which these changes in cholesterol generalize across populations.
The current study, by Stewart and colleagues, examines these questions using 26 years of data from the Honolulu Aging Study . The group observes a general decrease in cholesterol with aging beginning up to 15 years before the onset of dementia, but the decrease in cholesterol that occurred in men who subsequently developed dementia was larger than in those who did not develop dementia. Subjects carrying the Apoe4 allele had the largest effect size, although this was not statistically significant due to the small number of people with this allele in the study cohort. The observed decrease in cholesterol preceding the onset of dementia agrees with observations of other groups. Stewart and colleagues note that the tendency of cholesterol to decrease well before any signs of cognitive loss makes the decrease in cholesterol one of the earliest biological changes associated with the onset of dementia.
Stewart and colleagues did not observe elevated cholesterol earlier in life in subjects who ultimately developed dementia. This is surprising and contrasts with the observations of Kivipelto and colleagues. Analysis of the study suggests two important differences between the Honolulu cohort and the Finnish cohort studied by Kivipelto. One such difference is age. Although the Honolulu study extends back for 26 years, the average age of the people in the current study is 80 years old, meaning that their average age at the start of the study was already 54. It is conceivable that differences in cholesterol that might have been present earlier in their lives might have become less significant by age 54. However, a potentially more important difference is ethnicity. The Honolulu cohort is mainly composed of subjects of Japanese origin. Stewart and colleagues note the distinct ethnicity, but feel it is unlikely to be a cause of the difference. I disagree with this assessment. A growing number of studies indicate ethnic differences in morbidity resulting from cardiovascular risk factors. Allison and colleagues note that the risk of peripheral artery disease among people with cardiovascular disease risk factors is lower in Chinese than in Caucasians . Hall and colleagues note that the risk of end-stage renal disease shows a rank order of African American>Asian>Caucasian, with up to sixfold differences in odds ratios . Genetic factors that might impact on morbidity resulting from cardiovascular disease also show distinct ethnic tendencies. For instance, Wollmer and colleagues note the rs908832 polymorphism in ABCA2 is a risk factor for AD in some populations, but is not even present as a polymorphism in the Japanese population . This suggests that ethnicity could exert a strong effect on the relationship between cardiovascular biomarkers and incident dementia/AD.
My interpretation of this study is that it strengthens the conclusion that cholesterol levels decrease preceding dementia, but leaves open the question of whether elevated mid-life cholesterol is a risk factor for dementia/AD. The reason for decreased cholesterol preceding dementia/AD remains unclear. An important question is whether any changes in cholesterol represent a cause or an effect. The simplest explanation, which is favored by Stewart and colleagues, is that the changes in cholesterol reflect a general result of cardiovascular disease. Another possibility is that early degenerative changes in the brain alter the body’s cholesterol metabolism, which leads to the gradual reduction in cholesterol levels. One could also invoke the amyloid cascade hypothesis. Aβ has been shown to modulate cholesterol metabolism , but the increases in Aβ preceding AD are central rather than in the plasma, so it is difficult to understand how changes in Aβ in the brain could directly modulate cholesterol metabolism in the liver. The response to these questions is the classic academic one, which is that more studies are needed. We really need to focus attention on the role of ethnicity in risk factors for AD, and we need to understand the relationship between cholesterol and AD. Our ability to modify cardiovascular risk factors through pharmacological intervention behooves us to examine this field carefully.
References:1. Kivipelto M, Helkala EL, Hanninen T, Laakso MP, Hallikainen M, Alhainen K, Soininen H, Tuomilehto J, Nissinen A. Midlife vascular risk factors and late-life mild cognitive impairment: A population-based study.
Neurology. 2001 Jun 26;56(12):1683-9.
2. Kivipelto M, Helkala EL, Laakso MP, Hanninen T, Hallikainen M, Alhainen K, Iivonen S, Mannermaa A, Tuomilehto J, Nissinen A, Soininen H. Apolipoprotein E epsilon4 allele, elevated midlife total cholesterol level, and high midlife systolic blood pressure are independent risk factors for late-life Alzheimer disease.
Ann Intern Med. 2002 Aug 6;137(3):149-55. Summary for patients in: Ann Intern Med. 2002 Aug 6;137(3):I-18.
3. Breteler MM, Bots ML, Ott A, Hofman A. Risk factors for vascular disease and dementia.
Haemostasis. 1998 May-Aug;28(3-4):167-73. Review.
4. Sjogren M, Mielke M, Gustafson D, Zandi P, Skoog I. Cholesterol and Alzheimer's disease--is there a relation?
Mech Ageing Dev. 2006 Feb;127(2):138-47. Epub 2005 Dec 5. Review.
5. Stewart R, White LR, Xue QL, Launer LJ. Twenty-six-year change in total cholesterol levels and incident dementia: the Honolulu-Asia Aging Study.
Arch Neurol. 2007 Jan;64(1):103-7.
6. Allison MA, Criqui MH, McClelland RL, Scott JM, McDermott MM, Liu K, Folsom AR, Bertoni AG, Sharrett AR, Homma S, Kori S. The effect of novel cardiovascular risk factors on the ethnic-specific odds for peripheral arterial disease in the Multi-Ethnic Study of Atherosclerosis (MESA).
J Am Coll Cardiol. 2006 Sep 19;48(6):1190-7. Epub 2006 Aug 28.
7. Hall YN, Hsu CY, Iribarren C, Darbinian J, McCulloch CE, Go AS. The conundrum of increased burden of end-stage renal disease in Asians.
Kidney Int. 2005 Nov;68(5):2310-6.
8. Wollmer MA, Kapaki E, Hersberger M, Muntwyler J, Brunner F, Tsolaki M, Akatsu H, Kosaka K, Michikawa M, Molyva D, Paraskevas GP, Lutjohann D, von Eckardstein A, Hock C, Nitsch RM, Papassotiropoulos A. Ethnicity-dependent genetic association of ABCA2 with sporadic Alzheimer's disease.
Am J Med Genet B Neuropsychiatr Genet. 2006 Jul 5;141(5):534-6.
9. Grimm MO, Grimm HS, Patzold AJ, Zinser EG, Halonen R, Duering M, Tschape JA, De Strooper B, Muller U, Shen J, Hartmann T. Regulation of cholesterol and sphingomyelin metabolism by amyloid-beta and presenilin.
Nat Cell Biol. 2005 Nov;7(11):1118-23. Erratum in: Nat Cell Biol. 2006 Apr;8(4):424.
The article by Stewart et al. relating 26-year change in cholesterol to incident dementia adds considerably to a complex and seemingly inconsistent body of literature. A key indication that cholesterol plays a central role in Alzheimer disease is the increase in disease risk among persons with the apolipoprotein E4 allele, the primary genetic risk factor in late-onset AD. Apolipoprotein E is the major cholesterol transporter in the brain. There is emerging evidence of the role of cholesterol metabolism in AD (1), but it is not clear whether dyshomeostasis in cholesterol may be a cause or effect of the disease process, or both.
The Stewart et al. study does not find differences in blood cholesterol levels at midlife or late life by AD status, but finds greater rate of decline in cholesterol levels with age among those who eventually develop AD. Similar complex associations have been demonstrated between dementia and other physiologic parameters, such as blood pressure (2) and weight (3). Decline in these risk factors as many as 15 years before clinical manifestation of the disease makes interpretation of late-life relationships difficult.
A more important question to the public health, however, is whether midlife levels of these modifiable risk factors influence the development of dementia. Studies on this topic appear inconsistent until one focuses on the range or level of the risk factor. For example, as we point out in a previous study of blood pressure and incident AD (4), reported associations between high midlife blood pressure and increased risk of late-life dementia have been restricted to levels greater than 160 mmHg systolic or 95 mmHg diastolic.
A similar pattern could be emerging among the limited number of cholesterol studies, and may explain the absence of association with midlife cholesterol level in the Stewart et al. study. Of four studies (5-8) that report on midlife blood cholesterol level in relation to late-life dementia, the three (5-7) that observed increased risk targeted only the highest levels of blood cholesterol that would be considered hypercholesterolemic. Kivipelto et al. (6) and Notkola et al. (5) observed approximately three times the risk of late-life dementia among persons whose midlife blood cholesterol levels were >251 mg/dL or >6.5 mmol/L. In the study by Whitmer et al. (7), high cholesterol was defined at a somewhat lower level (>240 mg/dL) and the increase in risk was smaller (OR = 2.4) but statistically significant.
The Stewart et al. and the Framingham studies (8) did not find an association between midlife blood cholesterol and late-life dementia; however, both measured cholesterol level as a continuous variable. Thus, alternative explanations for these null studies are that the relation between midlife cholesterol and dementia is not linear, and/or that the number of persons in the null studies who had hypercholesterolemia was small and the studies, therefore, were inadequately powered to observe the relation at this high level. The possibility that only hypercholesterolemic levels of cholesterol increase the risk of dementia highlights the importance of considering the range of cholesterol level in the interpretation of findings in future studies. Analytic methods should always include investigations of associations at hypercholesterolemic levels as well as non-linear associations. In addition, longitudinal analyses of blood pressure level and dementia in the Framingham study reminds us of the importance of considering treatment on the observed associations (9).
References:1. Ledesma MD, Dotti CG. Amyloid excess in Alzheimer's disease: what is cholesterol to be blamed for?. FEBS Lett 2006; 580(23):5525-5532. Abstract
2. Skoog I, Lernfelt B, Landahl S, Palmertz B, Andreasson LA, Nilsson L, Persson G, Oden A, Svanborg A. 15-year longitudinal study of blood pressure and dementia.
Lancet. 1996 Apr 27;347(9009):1141-5.
3. Barrett-Connor E, Edelstein SL, Corey-Bloom J, Wiederholt WC. Weight loss precedes dementia in community-dwelling older adults. J Am Geriatr Soc 1996; 44(10):1147-1152. Abstract
4. Morris MC, Scherr PA, Hebert LE, Glynn RJ, Bennett DA, Evans DA. Association of incident Alzheimer disease and blood pressure measured from 13 years before to 2 years after diagnosis in a large community study. Arch Neurol 2001; 58(10):1640-1646. Abstract
5. Notkola IL, Sulkava R, Pekkanen J, Erkinjuntti T, Ehnholm C, Kivinen P, Tuomilehto J, Nissinen A. Serum total cholesterol, apolipoprotein E epsilon 4 allele, and Alzheimer's disease.
6. Kivipelto M, Helkala EL, Laakso MP, Hanninen T, Hallikainen M, Alhainen K, Iivonen S, Mannermaa A, Tuomilehto J, Nissinen A, Soininen H. Apolipoprotein E epsilon4 allele, elevated midlife total cholesterol level, and high midlife systolic blood pressure are independent risk factors for late-life Alzheimer disease.
Ann Intern Med. 2002 Aug 6;137(3):149-55. Summary for patients in: Ann Intern Med. 2002 Aug 6;137(3):I-18.
7. Whitmer RA, Sidney S, Selby J, Johnston SC, Yaffe K. Midlife cardiovascular risk factors and risk of dementia in late life. Neurology 2005; 64(2):277-281. Abstract
8. Tan ZS, Seshadri S, Beiser A, Wilson PW, Kiel DP, Tocco M, D'Agostino RB, Wolf PA. Plasma total cholesterol level as a risk factor for Alzheimer disease: the Framingham Study.
Arch Intern Med. 2003 May 12;163(9):1053-7.
9. Elias MF, Wolf PA, D'Agostino RB, Cobb J, White LR. Untreated blood pressure level is inversely related to cognitive functioning: the Framingham Study. Am J Epidemiol 1993; 138(6):353-364. Abstract
The study by Stewart et al. fails to show differences in cholesterol levels at midlife or late life and AD diagnosis. This is (apparently) in contrast with several epidemiological studies (1-3). Ben Wolozin in his comment above astutely noticed that a substantial difference in the Stewart et al. study is age and stated “…Although the Honolulu study extends back for 26 years, the average age of the people in the current study is 80 years old, meaning that their average age at the start of the study was already 54.”
This is of major importance in explaining some of these apparent disparities. In a prior neuropathological study we conducted (4), cholesterolemia correlated with presence of amyloid deposition only in the youngest subjects (40 to 55 years) with amyloid deposition (p = 0.000 for all ApoE isoforms; p = 0.009 for ApoE3/3 subjects). In this group, increases in cholesterolemia from 181 to 200 almost tripled the odds for developing amyloid, independent of ApoE isoform. In our study, the difference in mean total cholesterol between subjects with and without amyloid disappeared as the age of the sample increased (>55 years: p = 0.491). A logistic regression model showed consistent results, and furthermore, it showed that mild cholesterol elevations occurring early in life, rather than severe increases, may maximally enhance amyloid deposition (4). (This is relevant when considered within the context of the declines in total cholesterol levels that occur before development of AD. Such declines are of sufficient magnitude to completely obscure mild elevations of cholesterol occurring earlier in life.)
If one assumes that amyloid deposition is one prerequisite for development of AD, one can also extend these observations to explain, in part, some of the controversies regarding the effects of statins. Most studies negating a protective role of statins for AD have been conducted in individuals older than 65, overlooking the fact that cholesterol is an early, not a late, risk factor for AD. In the CRISP study, there was no effect of pharmacologic lipid lowering on cognition (5). However, the study was conducted in 431 subjects aged 65 years or older randomized to lovastatin or placebo for 6 months. Likewise, the PROSPER trial randomized 5,804 high-risk elderly adults (aged 70-82 years) to pravastatin or placebo for 3 years but found no effect of treatment on cognitive outcomes (6). Similarly, a community-based prospective cohort study of 2,356 cognitively intact persons, aged 65 and older, found no significant association between statin use and incident dementia or probable AD (7). The Cache County study (8) was also negative in terms of identifying a protective effect of statins but was conducted in individuals older than 65 years. None of these studies took into consideration the evidence that indicated that cholesterol is an early risk factor for AD.
The only study that included younger individuals was the Heart Protection Study (HPS), which randomized 20,536 high-risk adults aged 40-80 years to simvastatin or placebo for approximately 5 years (9). Although this study found no differences in performance on a telephone assessment of cognitive abilities at the participants’ final visits, the HPS trial was not designed to unveil the impact of statin therapy on the young individuals’ risk to go on to develop late-onset AD.
The relationship between cholesterolemia and AD risk is complex. Many more reasons beyond those mentioned in this comment can add to the existing controversies. The available data also suggest that the extent of amyloid deposition may be influenced by events that regulate removal of amyloid peptides from the brain. Therefore, better clearance mechanisms in certain individuals may preclude clinically significant accumulations of amyloid, despite increased cholesterol-mediated amyloid accumulation. This, of course, is assuming that the amyloid hypothesis is correct. From the available data, it is apparent that cholesterolemia is only one of many factors likely to influence the risk and progression of cognitive abnormalities to full-blown AD.
References:1. Notkola IL, Sulkava R, Pekkanen J, Erkinjuntti T, Ehnholm C, Kivinen P, Tuomilehto J, Nissinen A. Serum total cholesterol, apolipoprotein E epsilon 4 allele, and Alzheimer's disease.
Neuroepidemiology. 1998;17(1):14-20. Abstract
2. Kivipelto M, Helkala EL, Laakso MP, Hanninen T, Hallikainen M, Alhainen K, Iivonen S, Mannermaa A, Tuomilehto J, Nissinen A, Soininen H. Apolipoprotein E epsilon4 allele, elevated midlife total cholesterol level, and high midlife systolic blood pressure are independent risk factors for late-life Alzheimer disease. Ann Intern Med. 2002 Aug 6;137(3):149-55. Summary for patients in: Ann Intern Med. 2002 Aug 6;137(3):I-18. Abstract
3. Whitmer RA, Sidney S, Selby J, Johnston SC, Yaffe K. Midlife cardiovascular risk factors and risk of dementia in late-life. Neurology. 2005; 64(2):277-281. Abstract
4. Pappolla MA, Bryant-Thomas TK, Herbert D, Pacheco J, Fabra Garcia M, Manjon M,Girones X, Henry TL, Matsubara E, Zambon D, Wolozin B, Sano M, Cruz-Sanchez FF,Thal LJ, Petanceska SS, Refolo LM. Mild Hypercholesterolemia is an early risk factor for the development of Alzheimer amyloid pathology.Neurology. 2003 Jul 22;61(2):199-205. Abstract
5. Santangello NC, Barber BL, Applegate WB, Elam J, Curtis C, Hunninghake DB, Gordon DJ. Effect of pharmacologic lipid lowering on health-related quality of life in older persons: results from the Cholesterol Reduction in Seniors Program (CRISP) Pilot Study. J Am Geriatr Soc. 1997;45:8-14. Abstract
6. Shepherd J, Blauw GJ, Murphy MB, Bollen EL, Buckley BM, Cobbe SM, Ford I, Gaw A, Hyland M, Jukema JW, Kamper AM, Macfarlane PW, Meinders AE, Norrie J, Packard CJ, Perry IJ, Stott DJ, Sweeney BJ, Twomey C, Westendorp RG; PROSPER (Prospective Study of Pravastatin in the Elderly at Risk) Study Group. Pravastatin in elderly individuals at risk of vascular disease (PROSPER): a randomized controlled trial. Lancet. 2002;360:1623-1630. Abstract
7. Li G, Higdon R, Kukull WA, Peskind E, Van Valen Moore K, Tsuang D, van Belle G, McCormick W, Bowen JD, Teri L, Schellenberg GD, Larson EB. Statin therapy and risk of dementia in the elderly: a community-based prospective cohort study.
Neurology. 2004 Nov 9;63(9):1624-8. Abstract
8. Zandi PP, Sparks DL, Khachaturian AS, Tschanz J, Norton M, Steinberg M, Welsh-Bohmer KA, Breitner JC; Cache County Study investigators. Do statins reduce risk of incident dementia and Alzheimer disease? The Cache County Study.
Arch Gen Psychiatry. 2005 Feb;62(2):217-24. Abstract
9. Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: a randomised placebo-controlled trial.
Lancet. 2002 Jul 6;360(9326):7-22. Summary for patients in: Curr Cardiol Rep. 2002 Nov;4(6):486-7. Abstract