- Email: [email protected]
Statins for people at low risk of cardiovascular disease
Correspondence
be used to prove that this assumption is indeed correct. It would be valid and informative to subgroup individual trial results by observed LDL cholesterol reduction, then compare the reductions of risk seen across trials in each subgroup. Thus graphs such as figure 5 could reflect real data rather than what seems to be a display of values calculated using a linear regression model, and as such could show nothing other than straight lines. I (and my employer) have received speakers’ honoraria from AstraZeneca, Merck Sharp and Dohme, and Schering Plough. I have received hospitality in connection with attendance at international scientific meetings from AstraZeneca, Genzyme, and Roche. My employer has received unrestricted educational grants from Schering Plough.
William G Simpson [email protected] NHS Grampian, Aberdeen AB25 2ZD, UK 1
Cholesterol Treatment Trialists’ (CTT) Collaborators. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012; 380: 581–90.
The conclusions of the analysis of the effect of statins at different levels of risk1 are misleading and do not answer relevant questions facing clinicians, patients, and policy makers. The statistical models derived are based on data from populations with previous cardiovascular disease (CVD) or those enriched for risk-factors predisposing to CVD, and thus are not applicable to general primary prevention populations; risk prediction does not guide statin therapy in secondary prevention. Model 1 (derived from 22 mixed trials) included previous history of coronary heart disease and trial-specific covariants, and thus when applied to the same trials individually, observed and expected risk are unsurprisingly similar. Current clinical risk-prediction tools derived solely from general populations without prevalent disease or enrichment for risk factors are unlikely to do as well in the self-selected CTT dataset. Hence the predicted risks from Framingham2 1816
and the CTT Collaborators are not equivalent and the benefits noted by the CTT Collaborators are likely to be a gross overstatement and should not be used to change guidelines or policy. Furthermore, the absolute benefits are modest, with only two vascular deaths prevented per 10 000 people per year at less than 5% risk versus 46 prevented at 10–20% risk. We favour an approach whereby the CTT Collaborators recalculate the baseline 10-year Framingham risk and plot the relation between a 1 mmol/L lowering of LDL cholesterol and CVD outcomes at different levels of Framingham risk. The absolute benefit from such analyses, rather than the proportional reduction, is what will guide any change to the present guidelines. KKR has received honoraria for consulting and lectures and research support from Pfizer, AstraZeneca, Merck, and Bristol-Myers Squibb. RFR declares that she has no conflicts of interest.
*Kausik K Ray, Rita F Redberg [email protected] St Georges University of London, London SW17 0RE, UK (KKR); and University of California at San Francisco, San Francisco, CA, USA (RFR) 1
2
Cholesterol Treatment Trialists’ (CTT) Collaborators. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012; 380: 581–90. Pencina M, D’Agostino RB, Larson MG, et al. Predicting the 30 year risk of cardiovascular disease. Circulation 2009; 119: 3078–84.
targets. Furthermore, cardiovascular risk in statin-treated individuals remains high—the so-called residual risk.2 These findings clearly suggest that the “reduction” of major vascular events by statin therapy in fact represents only a delay in their clinical manifestation. Indeed, in people at high risk (as defined as those with a 5-year risk of major vascular events higher than 25%), the average delay of a major cardiovascular event has been calculated at 0·09 years (33 days) over 5 years with a statininduced 1·1 mmol/L reduction in LDL cholesterol.3 Although it is difficult to make a similar calculation for the whole CTT Collaborators’ meta-analysis, it seems unlikely that the delay would be longer. Therefore, what is the point in treating a healthy individual with lifelong statin therapy with the hope of a few months’ delay in the clinical manifestation of a vascular event, when side-effects—easily dismissed in randomised trials and not trivial in real life4,5—are factored in? LM was a reviewer of the paper in question, recommending rejection. MRG declares that he has no conflicts of interest.
*Luca Mascitelli, Mark R Goldstein [email protected] Comando Brigata Alpina “Julia”, Medical Service, 33100 Udine, Italy (LM); and NCH Healthcare Group, Naples, FL, USA (MRG) 1
In a meta-analysis of essentially industry-driven trials in selected individuals, the Cholesterol Treatment Trialists’ (CTT) Collaborators1 found that, in people at low risk of vascular disease (defined as those with a 5-year risk of major vascular events lower than 10%), each statin-induced 1 mmol/L reduction in LDL cholesterol produced an absolute reduction in major vascular events of about 11 per 1000 over 5 years. We feel that translating these results into clinical practice might not be so straightforward. Patients can present with major vascular events despite having LDL cholesterol concentrations that fall well within guideline-recommended
2
3
4
5
Cholesterol Treatment Trialists’ (CTT) Collaborators. The effect of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012; 380: 581–90. Mora S, Wenger NK, DeMicco DA, et al. Determinants of residual risk in secondary prevention patients treated with high- versus low-dose statin therapy: the Treating to New Targets (TNT) Study. Circulation 2012; 125: 1979–87. Bassan M, Panush N. Treating hypercholesterolemia—without the hype. Am J Cardiol 1997; 79: 1001–03. Mascitelli L, Goldstein MR. Statins in dermatology: is nonmelanoma skin cancer the dark side of the moon? Int J Dermatol 2012; published online April 18. http://dx.doi. org/10.1111/j.1365-4632.2011.05067.x. Rosenbaum D, Dallongeville J, Sabouret P, Bruckert E. Discontinuation of statin therapy due to muscular side effects: a survey in real life. Nutr Metab Cardiovasc Dis 2012; published online June 28. http://dx.doi.org/10.1016/ j.numecd.2012.04.012.
www.thelancet.com Vol 380 November 24, 2012
be used to prove that this assumption is indeed correct. It would be valid and informative to subgroup individual trial results by observed LDL cholesterol reduction, then compare the reductions of risk seen across trials in each subgroup. Thus graphs such as figure 5 could reflect real data rather than what seems to be a display of values calculated using a linear regression model, and as such could show nothing other than straight lines. I (and my employer) have received speakers’ honoraria from AstraZeneca, Merck Sharp and Dohme, and Schering Plough. I have received hospitality in connection with attendance at international scientific meetings from AstraZeneca, Genzyme, and Roche. My employer has received unrestricted educational grants from Schering Plough.
William G Simpson [email protected] NHS Grampian, Aberdeen AB25 2ZD, UK 1
Cholesterol Treatment Trialists’ (CTT) Collaborators. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012; 380: 581–90.
The conclusions of the analysis of the effect of statins at different levels of risk1 are misleading and do not answer relevant questions facing clinicians, patients, and policy makers. The statistical models derived are based on data from populations with previous cardiovascular disease (CVD) or those enriched for risk-factors predisposing to CVD, and thus are not applicable to general primary prevention populations; risk prediction does not guide statin therapy in secondary prevention. Model 1 (derived from 22 mixed trials) included previous history of coronary heart disease and trial-specific covariants, and thus when applied to the same trials individually, observed and expected risk are unsurprisingly similar. Current clinical risk-prediction tools derived solely from general populations without prevalent disease or enrichment for risk factors are unlikely to do as well in the self-selected CTT dataset. Hence the predicted risks from Framingham2 1816
and the CTT Collaborators are not equivalent and the benefits noted by the CTT Collaborators are likely to be a gross overstatement and should not be used to change guidelines or policy. Furthermore, the absolute benefits are modest, with only two vascular deaths prevented per 10 000 people per year at less than 5% risk versus 46 prevented at 10–20% risk. We favour an approach whereby the CTT Collaborators recalculate the baseline 10-year Framingham risk and plot the relation between a 1 mmol/L lowering of LDL cholesterol and CVD outcomes at different levels of Framingham risk. The absolute benefit from such analyses, rather than the proportional reduction, is what will guide any change to the present guidelines. KKR has received honoraria for consulting and lectures and research support from Pfizer, AstraZeneca, Merck, and Bristol-Myers Squibb. RFR declares that she has no conflicts of interest.
*Kausik K Ray, Rita F Redberg [email protected] St Georges University of London, London SW17 0RE, UK (KKR); and University of California at San Francisco, San Francisco, CA, USA (RFR) 1
2
Cholesterol Treatment Trialists’ (CTT) Collaborators. The effects of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012; 380: 581–90. Pencina M, D’Agostino RB, Larson MG, et al. Predicting the 30 year risk of cardiovascular disease. Circulation 2009; 119: 3078–84.
targets. Furthermore, cardiovascular risk in statin-treated individuals remains high—the so-called residual risk.2 These findings clearly suggest that the “reduction” of major vascular events by statin therapy in fact represents only a delay in their clinical manifestation. Indeed, in people at high risk (as defined as those with a 5-year risk of major vascular events higher than 25%), the average delay of a major cardiovascular event has been calculated at 0·09 years (33 days) over 5 years with a statininduced 1·1 mmol/L reduction in LDL cholesterol.3 Although it is difficult to make a similar calculation for the whole CTT Collaborators’ meta-analysis, it seems unlikely that the delay would be longer. Therefore, what is the point in treating a healthy individual with lifelong statin therapy with the hope of a few months’ delay in the clinical manifestation of a vascular event, when side-effects—easily dismissed in randomised trials and not trivial in real life4,5—are factored in? LM was a reviewer of the paper in question, recommending rejection. MRG declares that he has no conflicts of interest.
*Luca Mascitelli, Mark R Goldstein [email protected] Comando Brigata Alpina “Julia”, Medical Service, 33100 Udine, Italy (LM); and NCH Healthcare Group, Naples, FL, USA (MRG) 1
In a meta-analysis of essentially industry-driven trials in selected individuals, the Cholesterol Treatment Trialists’ (CTT) Collaborators1 found that, in people at low risk of vascular disease (defined as those with a 5-year risk of major vascular events lower than 10%), each statin-induced 1 mmol/L reduction in LDL cholesterol produced an absolute reduction in major vascular events of about 11 per 1000 over 5 years. We feel that translating these results into clinical practice might not be so straightforward. Patients can present with major vascular events despite having LDL cholesterol concentrations that fall well within guideline-recommended
2
3
4
5
Cholesterol Treatment Trialists’ (CTT) Collaborators. The effect of lowering LDL cholesterol with statin therapy in people at low risk of vascular disease: meta-analysis of individual data from 27 randomised trials. Lancet 2012; 380: 581–90. Mora S, Wenger NK, DeMicco DA, et al. Determinants of residual risk in secondary prevention patients treated with high- versus low-dose statin therapy: the Treating to New Targets (TNT) Study. Circulation 2012; 125: 1979–87. Bassan M, Panush N. Treating hypercholesterolemia—without the hype. Am J Cardiol 1997; 79: 1001–03. Mascitelli L, Goldstein MR. Statins in dermatology: is nonmelanoma skin cancer the dark side of the moon? Int J Dermatol 2012; published online April 18. http://dx.doi. org/10.1111/j.1365-4632.2011.05067.x. Rosenbaum D, Dallongeville J, Sabouret P, Bruckert E. Discontinuation of statin therapy due to muscular side effects: a survey in real life. Nutr Metab Cardiovasc Dis 2012; published online June 28. http://dx.doi.org/10.1016/ j.numecd.2012.04.012.
www.thelancet.com Vol 380 November 24, 2012