JUPITER - Exploring New Frontiers

Register or Login to View PDF Permissions
Permissions× For commercial reprint enquiries please contact Springer Healthcare:

For permissions and non-commercial reprint enquiries, please visit to start a request.

For author reprints, please email
Average (ratings)
No ratings
Your rating


Over the last two decades statins have become indispensable for managing patients at high risk of cardiovascular disease. Since their introduction in the late 1980s, their role has expanded from drugs to be used in secondary prevention to standard treatment in primary prevention for individuals at high risk of cardiovascular disease. Targets for low-density lipoprotein (LDL) cholesterol, as well as LDL levels at which statin therapy should be started, have gradually gone down. The JUPITER study succeeded in expanding the limits of earlier interventions and aimed for lower plasma LDL cholesterol levels, as well as including high-sensitivity C-reactive protein (hs-CRP) as a biomarker for risk and target for treatment. The investigators included patients who, according to current guidelines, would not qualify for statin therapy. Patients with elevated levels of hs-CRP (>2mg/dl) and low LDL cholesterol (<130mg/dl) were treated with rosuvastatin 20mg or placebo. The trial was terminated after 1.9 years because of unexpected early benefits, whereby the relative risk of developing atherothrombotic complications was reduced by 44%. The implications of this trial are far-reaching, not only because of the public health consequences of treating a larger number of individuals with aggressive statin therapy than current guidelines dictate; it also forces us to re-evaluate current cardiovascular risk calculating tools. The provocative results of this important primary prevention trial have generated new questions on how to intervene earlier in the development of atherosclerosis in patients at risk of cardiovascular disease.

Disclosure:Peter J Lansberg is on the speaker&#039;s bureau for Merck Sharp &amp; Dohme, Schering Plough, Pfizer, AstraZeneca and Genzyme, and is on Pfizer&#039;s advisory board.



Correspondence Details:Peter J Lansberg, Co-ordinator, Durrer Centre for Cardiogenetic Research, Department of Vascular Medicine F4-159.2, Academic Medical Centre, Amsterdam, The Netherlands. E:

Copyright Statement:

The copyright in this work belongs to Radcliffe Medical Media. Only articles clearly marked with the CC BY-NC logo are published with the Creative Commons by Attribution Licence. The CC BY-NC option was not available for Radcliffe journals before 1 January 2019. Articles marked ‘Open Access’ but not marked ‘CC BY-NC’ are made freely accessible at the time of publication but are subject to standard copyright law regarding reproduction and distribution. Permission is required for reuse of this content.

The majority of all heart attacks and strokes are observed in individuals without elevated cholesterol levels, and some are in those with low cholesterol. Risk factors are absent in approximately 20% of patients with vascular events. If traditional risk factors seem to be unreliable for predicting events in a large number of individuals, expanding the classic risk prediction model would be the natural next step.

Inflammation is a basic determinant of atherosclerosis, and inflammatory biomarkers are gaining ground as independent predictors of vascular risk. Over the past 10 years high-sensitivity C-reactive protein (hs-CRP) has evolved into a potential biomarker to be included in cardiovascular risk calculation models. Regardless of whether hs-CRP plays a causal role in atherothrombosis, this biomarker reflects a metabolic disturbance and has shown to be useful for improving cardiovascular disease (CVD) risk prediction.

The Justification for the use of Statins in Primary Prevention: an Intervention trial Evaluating Rosuvastatin (JUPITER) is a large multicentre, international, randomised, placebo-controlled clinical trial that was conducted in 26 countries involving 1,315 physicians. The purpose of this trial was to assess whether statin therapy should be given to apparently healthy individuals with low-density lipoprotein (LDL) cholesterol levels below 130mg/dl but increased hs-CRP concentrations. These were individuals who would not be considered candidates for lipid lowering under the accepted guidelines.1

The rationale for the JUPITER trial was based on several observations:

  • inflammation plays an important role in the development of atherosclerosis;2
  • the inflammatory biomarker hs-CRP can predict vascular events and would improve risk classification regardless of the LDL cholesterol level;3–6
  • statin therapy reduces hs-CRP independently of LDL cholesterol reduction;7–8 and
  • the scope of benefit relates to the achieved levels of hs-CRP in patients with acute coronary syndrome as well as in stable patients.9–11

In 1998, the Cholesterol And Recurrent Events (CARE) study showed for the first time that coronary heart disease (CHD) patients with a total cholesterol of <240mg/dl and an hs-CRP >2mg/dl derived significant clinical benefit from using pravastatin 40mg.12 In the Air Force/TexCAPS Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS), individuals whose hs-CRP was <2mg/dl did not benefit, despite the fact that lovastatin induced LDL cholesterol reduction.13

Later trials included the Pravastatin or Atorvastatin Evaluation and Infection Therapy—Thrombolysis in Myocardial Infarction (PROVE-IT-TIMI- 22) and Reversal of Atherosclerosis with Aggressive Lipid Lowering (REVERSAL).14,15 Patients who were using high-dose atorvastatin obtained superior reductions of hs-CRP and had lower event rates of atherothrombotic complications and higher rates of arrested atherosclerosis progression than the individuals randomised to less potent statins. However, all of these results were derived from a post hoc analysis and justified the need for a major prospective clinical trial.


The JUPITER study was a randomised, placebo-controlled trial in which a total of 17,802 asymptomatic individuals (men ≥50 years of age and women ≥60 years of age) without a history of coronary heart disease, stroke or diabetes were included. All participants had an initial screening LDL cholesterol of <130mg/dl and an hs-CRP >2.0mg/dl. They were randomised to rosuvastatin 20mg or placebo. The median LDL cholesterol and hs-CRP concentrations were 108 and 4.3mg/dl, respectively. Approximately 40% of the participants had metabolic syndrome at baseline (see Figure 1).

A number of distinguishing features separate this trial from the previous landmark statin trials. First, the included individuals would not have been considered for cholesterol-lowering treatment according to current guideline recommendations because they had low LDL cholesterol levels and a low to intermediate risk based on their Framingham Risk Score. Second, the large sample size included 6,800 women and 5,000 black and Hispanic people, which made it possible for the first time to evaluate statin safety and efficacy in these previoulsy under-represented groups.

Over 89,980 patients had to be screened for enrolment, and the majority were excluded because of too high LDL cholesterol and too low hs-CRP. This indicates that it was not easy to find individuals who would qualify to participate, so one could question whether the results of JUPITER can be generalised. It has been suggested that the omission of a control group without raised CRP levels devalues this study by preventing robust analysis of the value of hs-CRP in an aggressive lipid-lowering trial. However, the beneficial effects of statins in patients with low CRP and low LDL cholesterol levels were undetectable in AFCAPS/TexCAPS, largely due to an extremely low event rate. In order to add this specific subgroup, an additional 50,000–60,000 patients would need to be included. This would have been a prohibitive number that would most likely also have shown a null outcome.

An additional interesting feature of the JUPITER trial was the fourweek placebo run-in period. This enabled the investigators to exclude unwilling or non-compliant participants, but still allowed them to accurately record report side effects. This was in sharp contrast to the Heart Protection Study,16 where an active treatment was used during the run-in phase and patients who did not tolerate statins from the start were excluded before the trial began.

The core scientific hypothesis was based on the role of low-grade inflammation, so patients were excluded if they were previous or current users of lipid-lowering drugs, had severe arthritis, lupus or inflammatory bowel disease or were taking immune-modulating drugs such as cyclosporine, tacrolimus, azathiprine or long-term oral corticosteroids.

The primary end-point was the composite of non-fatal myocardial infarction, non-fatal stroke, hospitalisation for unstable angina and arterial revascularisation procedure or confirmed death from cardiovascular causes. Secondary end-points were the individual components of the primary end-point.

This investigator-initiated trial was expected to last for approximately four years. It was an event-driven trial and was designed to terminate when 520 confirmed primary end-points had been documented. The study was stopped prematurely after approximately two years because the Independent Data Monitoring Board (IDMB) decided it was unethical to withhold rosuvastatin from the placebo group in light of the observed benefits of 44% reduction in the primary end-point in the active treatment group. In retrospect, this decision received mixed comments from the scientific community. Some have suggested that the IDMB acted too quickly and should have continued the trial in order to guarantee the validity of the results. In fact, the IDMB had debated whether to stop the trial six months earlier when formal stopping boundaries were crossed. However, the results from the multiple specific subgroups and data on total mortality convinced them to continue the study until March 2008.

Healthy Population?

The aim of this study was to evaluate a relatively healthy population with elevated hs-CRP as a distinguishing characteristic. Whether this was a truly low-risk group could be debated. The median age was 66 years. Approximately 16% were current smokers and 11% had a positive family history for heart disease.

Metabolic syndrome was present in 44% of the participants, the median blood pressure was in the pre-hypertenisve range and 25% had a systolic blood pressure >145mgHg. All of these conditions are associated with elevated hs-CRP.17 The median hs-CRP level of 4.2mg/dl is considered an indication of higher global risk according to the American College of Cardiology/ American Heart Association.18

When comparing the baseline LDL cholesterol, non-high-density lipoprotein (HDL) cholesterol and apolipoprotein B (apoB) levels, a different perspective emerges as well. Using the percentile levels as presented in the fourth cycle of the Framingham Offspring Study as a comparator, a clear difference between the different lipid biomarkers studied emerged (see Table 1). Although the LDL cholesterol levels were below the 23rd and 30th percentile for men and women, respectively, their apoB corresponded to the 60th and 70th percentile, respectively. Based on their apoB level, participants in JUPTER would be considered at moderate to high risk.19

Low-density Lipoprotein Cholesterol and High-sensitivity C-reactive Protein Lowering Effects

After 12 months of rosuvastatin 20mg, LDL cholesterol had fallen by 50% (from 108 to 55mg/dl) and the median hs-CRP was reduced by 37% (from 4.2 to 2.2mg/dl). HDL cholesterol was unchanged when the trial was stopped, but triglycerides were lowered by 17%.1

In 25% of the individuals on rosuvastatin, LDL cholesterol was <45mg/dl and hs-CRP <1.2mg/dl. Compared with previous trials, this was an unprecedented reduction. An hs-CRP of <2.0mg/dl was achieved in almost half of all participants. A combination of the two treatment targets of LDL cholesterol <70mg/dl and hs-CRP <2mg/dl was observed in 30% of the rosuvastatin group and <1% in the placebo group. The most likely explanation for the result in the placebo group was off-study statin use. The lower hs-CRP of <1.0mg/dl target and the standard LDL cholesterol target of <70mg/dl was reached in 10% of individuals in the actively treated group.

Having lower baseline LDL cholesterol and hs-CRP levels was associated with better outcome in terms of reaching the pre-specified goals, as well as reduced event rates.20


Atherothrombotic Complications

Compared with placebo, participants allocated to rosuvastatin who achieved LDL cholesterol <70mg/dl had a 55% reduction in vascular events (p<0.0001), and those achieving hs-CRP <2mg/dl a 62% reduction (p<0.0001). A 65% reduction in vascular events was observed in participants who achieved the dual target of an LDL cholesterol <70mg/dl and hs-CRP <2mg/dl versus a 33% reduction in those who achieved one or neither target (p<0.0001 across treatment groups). Participants who were able to achieve an hs-CRP of <1mg/dl as well as LDL cholesterol <70mg/dl had an impressive 79% reduction (p<0.0001).20

All pre-specified subgroups within JUPITER benefited from rosuvastatin, including those previously considered to be ‘low-risk’, such as women, those with a body mass index (BMI) <25kg/m2, no metabolic syndrome, non-smokers, non-hypertensives and those with a Framingham Risk Score <10%. Even the 6,375 individuals who presented with an elevated hs-CRP as their only risk had a similar risk reduction, as was observed in the overall trial results (see Figure 2). JUPITER is the first statin primary prevention trial to demonstrate significant clinical benefits in women, black and Hispanic patients and the elderly (>70 years of age at baseline).

Venous Thromboembolisms

The occurrence of venous thromboembolisms was added as a protocol-specified secondary end-point in the JUPITER study and showed clear beneficial results. That statins might have antithrombotic effects was not a new concept – it had been observed in earlier prospective observational studies and post hoc analysis of statin trials.21–25 Deep vein thrombosis (DVT) or pulmonary embolism (PE) was classified as unprovoked if it occurred in the absence of any recent trauma, hospitalisation or surgery (occurring within three months before the event) and in the absence of a malignancy diagnosed before or up to three months after the event.

Symptomatic PE or DVT occurred in 94 participants (34 in the rosuvastatin group and 60 in the placebo group). Statin therapy was associated with a 43% reduction in the risk of venous thromboembolisms compared with placebo. This risk reduction was an independent benefit of statin use, beyond the reduction in the risk of arterial thrombosis.26 Among the 94 cases of symptomatic PE or DVT, 44 occurred in patients with cancer or recent trauma, hospitalisation or surgery, whereas a direct cause was not identified in 50 cases. The observed reductions in risk were similar for unprovoked and provoked events.

When confirmed in appropriately designed clinical trials or retrospective analysis, a new potential indication for statins could be possible: providing benefit for patients with conditions that would predispose them to thromboembolic complications.

Number Needed to Treat

To address the issue of cost-effectiveness, the number needed to treat (NNT) encompasses both absolute and relative risk reduction and is considered a very useful metric for comparing public health interventions. In the JUPITER trial the two-, three-, four- and five-year NNT values are 95, 49, 31 and 25, respectively, for the primary end-point. When calculating these numbers for the ‘hard’ end-points of myocardial infarction, stroke or deaths, the NNTs were 98, 59, 39 and 32, respectively.27

In terms of net clinical benefit of the combined end-point of first cardiovascular event, venous thromboembolisms or death from any cause, the five-year NNT was 18.20 By comparison, in earlier primary prevention trials such as the West of Scotland Coronary Prevention Study (WOSCOPS) or AFCAPS/TexCAPS, where pravastatin and lovastatin were used, the five-year NNTs ranged between 44 and 63.17,28

Safety and Side Effects

Despite the relatively potent dosage of 20mg rosuvastatin and a median on-treatment LDL cholesterol level of 55mg/dl, there were no significant differences between reported side effects in the patients allocated to rosuvastatin and those allocated to placebo (15.2 versus 15.5%; p=0.6). Serious side effects and complaints, such as muscle weakness, myopathy and hepatic or renal dysfunction, were identical in both groups. After the trial there was a single report of a trauma-induced rhabdomyolysis in a 90-year-old elderly female allocated to rosuvastatin.

Although the treatment duration was too short to draw definite conclusions, it is reassuring that there were no indications of increased incidence of mortality due to cancer. In fact, there was a small reduction in fatal cancers (25 versus 58; p=0.002), but that was likely to be due to chance. On a similar note, renal function did not deteriorate. There was even a somewhat better glomerular filtration rate in the active treatment group, refuting some of the earlier warnings of renal problems in patients on rosuvastatin. The safety profile of 20mg of rosuvastatin in JUPITER was similar to that of other statins currently on the market.

Insulin Resistance and Diabetes

Development of insulin resistance and diabetes was a point of concern. Although overall there was no increase in plasma glucose or glucosuria, glycated haemoglobin (HbA1c) and physician-reported type 2 diabetes were higher in the rosuvastatin arm: 5.9 versus 5.8% (p=0.001) and 3 versus 2.4% (p=0.01), respectively. Although chance might have played a part in these observations, they are in fact similar to earlier reports in the Heart Protection Study, Anglo- Scandinavian Cardiac Outcomes Trial, Lipid Lowering Arm (ASCOT-LLA), Pravastatin or Atorvastatin Evaluation and Infection Therapy (PROVE-IT) and Controlled Rosuvastatin Multinational Trial in Heart Failure (CORONA).16,28,30,31

Whether statin therapy had a negative clinical impact on the outcome of patients who developed diabetes remains uncertain, partly because all patients with diabetes are recommended to take statins to reduce atherothrombotic complications. In this study almost 80% of the patients who developed diabetes had impaired fasting glucose at study entry. This subgroup showed a significantly improved outcome, as defined in the primary end-point, suggesting that the overall risk–benefit ratio was favourable.

Lifestyle and Behaviour

Aggressive lifestyle modification interventions emphasising a balanced diet and routine exercise aimed at maintaining an ideal bodyweight, as well as smoking cessation, are and should remain in the focus of patient management, for both primary and secondary prevention of CVD. Behavioural preventative strategies remain the best investment for the prevention of CVD and its risk factors in predominantly healthy individuals. Healthcare professionals can follow this with statin therapy when the evidence supports its use as there is now type 1B randomised trial evidence for this.


So how should one interpret and perhaps implement the results of this new ‘landmark’ statin trial? One can only acknowledge that it was an impressive study that was executed diligently and with determination by the steering committee and the more than 1,300 participating physicians.

JUPITER was successful in establishing the paradigm of ‘the lower the LDL cholesterol the better’ in primary prevention, not only by reducing atherothrombotic events but also by showing an unexpected reduction in all-cause mortality by 26%. It seems that the unfavourable prognosis of a patient with ‘minor’ CRP increases points to a variety of conditions beyond just cardio-metabolic risk. The next challenge will be the assimilation and proper placement of the outcome data within the framework of current treatment protocols and guidelines.

The metabolic profile of a large number of the participants reflected the average 55+-year-old male and 65+-year-old female American: overweight with characteristics of the metabolic syndrome and presenting with subclinical risk factors, such as hs-CRP elevation. Today over 30% of Americans would fulfil the criteria for metabolic syndrome and, based on the outcome of JUPITER, would benefit from early intervention. The surprising results of the trial not only emerged within a short period of time, but also showed an impressive relative risk reduction.

In retrospect there were earlier trials that showed the impact of statins in patients with metabolic syndrome characteristics. Patients with CHD were followed in the Scandinavian Simvastatin Survival Study (4S), a post hoc analysis of more than 1,000 patients with low HDL cholesterol/high triglycerides or isolated high LDL cholesterol, and showed impressive results.32 The clinical characteristics of the individuals in the low HDL cholesterol/high triglyceride group reflected the classic features of the metabolic syndrome. They had a higher apoB and higher BMI, and there were more patients with hypertension, diabetes and impaired fasting glucose. In this analysis, the relative risk reduction in the high triglyceride, low HDL cholesterol group was an impressive 52% (p<0.0001) with a five-year NNT of six versus a 14% relative risk reduction (p=0.44) and a five-year NNT of 36 in the isolated high LDL cholesterol group. Measurement of hs-CRP was not performed but would most likely have been elevated.

When carrying out a more critical appraisal, one has to realise that the JUPITER results only apply to men 50 years of age and over and women 60 years of age and over. Extrapolation to the general population was criticised because prior screening eliminated almost 80% of the people considered for this study.

Finally, in terms of absolute risk the results were disappointing. The number of patients with cardiac death, myocardial infarction or stroke was reduced from 1.8% in the placebo group to 0.9% in the rosuvastatin group (157 versus 83), which equals a 0.9% absolute risk reduction. However, what should be recognised is that the sum of small incremental steps in risk factor management will add up to a giant leap of reduced risk of CVD – and even more so if patients at risk are exposed to optimal, controlled risk factors early in the progression of their atherothrombotic disease. One major hurdle between the impressive benefits of long-term treatment and the reality of patient behaviour is compliance. Patients and doctors will have to realise that risk factor management is not a quick fix to guarantee a long and cardiovascular disease-free life but a lifelong commitment to comply with the drug and lifestyle

Practical Implications of JUPITER

Clearly, hs-CRP serves as a non-lipid maker to identify those who may benefit from statin therapy. There is no need to use it in patients who are already candidates for statin therapy based on their risk and higher LDL cholesterol levels. Measuring hs-CRP and a standard lipid panel is helpful when in doubt as to whether lifelong statin therapy should be recommended in apparently healthy men 50 years of age and over and women 60 years of age and over. Statin therapy would be indicated in individuals with LDL cholesterol levels <130mg/dl and with an elevated hs-CRP >2mg/dl. Almost half of all CVD events and one in five all-cause deaths can thus be prevented.

In younger persons not many data are available and elevated hs-CRP is uncommon. Exceptions could be made in younger patients with a positive family history of premature heart attack and stroke. In JUPITER, rosuvastatin-use individuals who had a first-degree relative with premature coronary heart disease (males ≥55 years of age and females ≥65 years of age) had a relative risk reduction for cardiovascular disease equal to 65%.

The fact that rosuvastatin showed similar benefits in patients with a Framingham Risk Score of ≥ or ≤10% raises the provocative question of whether global risk assessment is reliable in patients where traditional risk factors are within normal boundaries.

Individuals with increased hs-CRP may also benefit from low-dose aspirin therapy. Men with an hs-CRP >2mg/dl in the Physicians Health Study had a 50% reduction in myocardial infarction.33 In the Women’s Health Study, women 65 years of age and over benefited from low-dose aspirin, but high levels of hs-CRP did not modify the outcome.34

The JUPITER study is a truly landmark primary prevention trial, not only for its unexpected findings but also for its impact on future guidelines for primary prevention and its challenge to current risk assessment. Although JUPITER has expanded the knowledge and understanding of CVD prevention, the results have generated new and important dilemmas that need to be addressed in future trials and novel treatment and risk calculation guidelines. Today medicine is expected to look ahead and evolve by exploring new innovative ways to improve the care of future patients. Yet there are ancient medical texts that even after 4,500 years are still considered provocative: “superior doctors prevent the disease, mediocre doctors treat the disease before evident, inferior doctors treat the full-blown disease” (Huang Dee, Nai Ching; see Figure 3).

This old text seems to be a directive on how to practise medicine in the 21st century. One could debate whether the doctors who participated in JUPITER were superior or mediocre doctors; however, they did manage to avoid being categorised as inferior doctors in the eyes of our ancient Chinese colleagues.


  1. Ridker PM, et al., N Engl J Med, 2008;359:2195–2207.
    Crossref | PubMed
  2. Hansson GK, Libby P, Nat Rev Immunol, 2006;6:508–19.
    Crossref | PubMed
  3. Ridker PM, et al., N Engl J Med, 1997;336:973–9.
    Crossref | PubMed
  4. Ridker PM, et al., N Engl J Med, 2000;342:836–43.
    Crossref | PubMed
  5. Danesh J, et al., N Engl J Med, 2004;350:1387–97.
    Crossref | PubMed
  6. Ridker PM, et al., N Engl J Med, 2002;347:1557–65.
    Crossref | PubMed
  7. Ridker PM, et al., Circulation, 1999;100:230–35.
    Crossref | PubMed
  8. Albert MA, et al., JAMA, 2001;286:64–70.
    Crossref | PubMed
  9. Ridker PM, et al., N Engl J Med, 2005;352:20–28.
    Crossref | PubMed
  10. Ridker PM, et al., Circulation, 1998;98:839–44.
    Crossref | PubMed
  11. Morrow DA, et al., Circulation, 2006;114:281–8.
    Crossref | PubMed
  12. Ridker PM, et al., Circulation, 1998;98:839–44.
    Crossref | PubMed
  13. Ridker PM, et al., N Engl J Med, 2001;344:1959–65.
    Crossref | PubMed
  14. Ridker PM, et al., N Engl J Med, 2005;352:20–28.
    Crossref | PubMed
  15. Nissen SE, et al., N Engl J Med, 2005;352:29–38.
    Crossref | PubMed
  16. Heart Protection Study Collaborative Group, Lancet, 2002;360:7–22.
    Crossref | PubMed
  17. Ridker PM, et al., Circulation, 2003;107:391–7.
    Crossref | PubMed
  18. Pearson TA, et al., Circulation, 2003;107:499–511.
    Crossref | PubMed
  19. Sniderman A, Curr Opin Lipidol, 2009;20:282–7.
    Crossref | PubMed
  20. Ridker PM, et al., Lancet, 2009;373;1175–83.
    Crossref | PubMed
  21. Grady D, et al., Ann Intern Med, 2000;132:689–96.
    Crossref | PubMed
  22. Ray JG, et al., Arch Intern Med, 2001;161:1405–10.
    Crossref | PubMed
  23. Ramcharan AS, et al., J Thromb Haemost, 2009;7: 514–20.
    Crossref | PubMed
  24. Doggen CJ, et al., J Thromb Haemost, 2004;2:700–701.
    Crossref | PubMed
  25. Lacut K, et al., Fundam Clin Pharmacol, 2004;18:477–82.
    Crossref | PubMed
  26. Glynn RJ, et al., N Engl J Med, 2009;360:1851–61.
    Crossref | PubMed
  27. Ridker PM, Circ Cardiovasc Qual Outcomes, 2009;2: 279–85.
    Crossref | PubMed
  28. Kumana CR, et al., JAMA, 1999;282:1899–1901.
    Crossref | PubMed
  29. ASCOT Investigators Group, Lancet, 2003;361:1149–58.
    Crossref | PubMed
  30. Cannon CP, et al., N Engl J Med, 2004;350:1495–504.
    Crossref | PubMed
  31. Kjekshus J, et al., N Engl J Med, 2007;357:2248–61.
    Crossref | PubMed
  32. Ballantyne CM, et al., Circulation, 2001;104:3046–51.
    Crossref | PubMed
  33. Petretta M, et al., Int J Cardiol, 2008 Sep 14 (Epub ahead of print).
  34. Ridker PM, et al., N Engl J Med, 1997;336:973–9.
    Crossref | PubMed
  35. Ridker PM, et al., N Engl J Med, 2005;352:1293–1304.
    Crossref | PubMed