The GREek Atorvastatin and Coronary-Heart-Disease Evaluation (GREACE) Study

Dimitri P. Mikhailidis and Anthony S. Wierzbicki


Curr Med Res Opin. 2002;18(4) 


The GREek Atorvastatin and Coronary-heart-disease Evaluation (GREACE) Study consisted of 1600 patients with coronary heart disease (CHD) followed up for three years.[1] The patients were randomised to either of two groups:

  1. A specialist hospital clinic and treatment with atorvastatin (10-80 mg/day) to achieve the National Cholesterol Education Program expert panel (Adult Treatment Panel III) (NCEP ATP III) low-density-lipoprotein cholesterol (LDL-C) target (2.6 mmol/l; 100 mg/dl), or,

  2. 'Usual' care by general practitioners.

Atorvastatin was well tolerated and it significantly decreased overall mortality (p = 0.0021) and coronary mortality (p = 0.0017), as well as the risk of non-fatal myocardial infarction (MI) (p = 0.0001) when compared with 'usual' care. There was also a significant benefit when considering subgroups (women and older patients) and other endpoints (unstable angina, need for revascularisation, heart failure and stroke). Only a few patients (26%) in the 'usual' care group received lipid-lowering therapy at any time during the study. Only 3% of these patients achieved the NCEP LDL-C goal.

Because a direct comparison was carried out between the two patient groups, the findings would have been slightly more significant if a placebo group had been used. However, this option would have been unethical.

The GREACE study emphasises the urgent need for the wider use of lipid-lowering drugs in high-risk patients. The poor 'usual' care performance in Greece (only 14% of the patients were given lipid-lowering therapy throughout the study) reflects the under-treatment seen in other European countries.[2,3] In contrast to lipid lowering, aspirin (about 87% of all patients) and antihypertensive drug use (about 85% and 54% of all patients were taking a -blocker and ACE inhibitor, respectively), was high and virtually identical in both patient groups in the GREACE study.[1] Therefore it would appear that lipid-lowering drugs had not achieved the wider acceptance seen with aspirin and antihypertensives, despite an excellent evidence base.

It is surprising that it took an unfunded study to provide what is, in our opinion, convincing evidence that atorvastatin reduces mortality and the risk of vascular events. Several studies assessed the effect of atorvastatin on the risk of atherosclerosis-related events.[4,5,6] However, these studies had limitations. These ranged from confounding endpoints, the absence of a 'placebo' group, borderline or non-significant statistics, short duration, unrealistically high dosages (80 mg/day) for long-term treatment and/or a lack of clinically relevant events.[7,8,9] Nevertheless, these studies[4,5,6] showed that atorvastatin is safe and effective in certain populations and clinical settings, even at the maximal dose (80 mg/day). In particular, MIRACL showed a significant (p < 0.048) 16% reduction in the composite endpoint (all causes of death, non-fatal MI, resuscitated cardiac arrest or hospitalisation for angina pectoris) in patients with acute coronary syndrome, despite the short duration of this trial (16 weeks).[4] In contrast to these studies,[4,5,6] GREACE[1] has highly significant results - the duration of follow-up of the 1600 patients was three years, the average dose of atorvastatin was a clinically realistic 24 mg/day and there was a 'placebo' group. These facts make GREACE a compelling piece of evidence showing that atorvastatin significantly reduces mortality and morbidity in patients with established CHD.

The authors[1] point to the comprehensive correction of three lipid fractions by atorvastatin: LDL-C, high-density lipoprotein cholesterol (HDL-C) and triglycerides (i.e. very-low-density lipoprotein, VLDL). After subtracting the small improvement in the lipid profile in the 'usual' care group, there was a 41% and 28% fall in LDL-C and triglycerides, respectively, together with a rise in HDL-C of 5% in those taking atorvastatin. These changes are considerable when compared with those that occurred in the major endpoint-driven statin trials reviewed by these authors.[1] The modification of each of these lipid fractions may be beneficial in terms of reducing the risk of vascular events, including strokes.[10,11,12,13,14,15,16,17] Therefore this comprehensive improvement of LDL-C, HDL-C and VLDL levels may have contributed to the significant reduction in events seen in the GREACE study.

More specifically, in relation to stroke, it is relevant to consider the following evidence:

  • Several meta-analyses, as well as individual trials, have shown that statins prevent thrombotic stroke in high-risk patients.[12,17,18,19,20] One secondary prevention trial showed a reduction in strokes after using a fibrate.[12] This suggests that correcting HDL-C (and possibly triglyceride) levels may be relevant in stroke prevention.[17]

  • The age-standardised mortality from stroke relative to that from CHD is greater in Greece than in the northern European populations that formed the bulk of the major statin trials (see Table 1).[18,21] Therefore the GREACE study suggests that statins are effective in countries with a high mortality from cerebrovascular disease.

  • Statins can beneficially affect the atherosclerotic process in the aortic arch and carotids.[18,22] Atorvastatin (20 mg/day) can rapidly (within 8 weeks) reduce the carotid intima media thickness (IMT).[23,24] This effect has also been documented in a more extensive study involving patients with familial hypercholesterolaemia treated over a period of two years with 80 mg/day of atorvastatin.[5] The carotid IMT is a good predictor of stroke and CHD.[5,22,23,24]

  • Statins can lower blood pressure.[25,26,27] Hypertension is a major predictor of stroke. We do not know if this effect is more evident in some ethnic or patient groups or environmental conditions (e.g. Mediterranean diet).

  • The placebo-controlled MIRACL study[4] showed a reduction (p = 0.045) in strokes in patients with acute coronary syndrome treated with atorvastatin (80 mg/day). However, the number of events was small (12 vs. 24), possibly as a result of the short duration of the study (16 weeks).

  • The recently presented Heart Protection Study (HPS)[19] included 20 536 high-risk patients and it showed that treatment with simvastatin (40 mg/day) significantly reduced the risk of stroke.

It is crucial that clinicians are well aware that statins prevent strokes, at least in high-risk patients.

How low should LDL-C levels be in high-risk patients? This remains an unresolved issue; the answer will come from ongoing trials[1,28] such as TNT, IDEAL and SEARCH. However, as the GREACE authors[1] point out, the HPS results[19] showed that high-risk patients with baseline LDL-C levels at or below 2.6 mmol/l (100 mg/dl) benefited from treatment with simvastatin (40 mg/day). Therefore, we might need to aim for a lower LDL-C level than that currently indicated in the guidelines [3.0 mmol/l (115 mg/dl) in the UK[29] and 2.6 mmol/l (100 mg/dl) in the USA[30]]. If, indeed, we need to lower LDL-C levels below 2.6 mmol/l (100 mg/dl), then it is likely that statins with a lesser LDL-C-lowering capacity will no longer be widely used, except in patients with LDL-Cs close to target.[16] For example, data presented at the American College of Cardiology (Atlanta, 2002) from the Lescol Intervention Prevention Study (LIPS) showed that fluvastatin 80 mg/day reduced cardiovascular events by 22% in patients post-angioplasty with an initial LDL-C of 3.4 mmol/l (130 mg/dl).

The HPS findings[19] and other studies[31] raise the question as to whether potential pleiotropic or non-LDL-C-related effects of statins are clinically relevant. These actions are difficult to compare between statins. For example, the effect of statins on plasma fibrinogen (a powerful and independent predictor of vascular risk) levels remains undefined.[32,33,34,35,36]

An earlier study reported that after about 150-200 days' treatment with atorvastatin (10-20 mg/day), the serum LDL-C level 'rebounds', thus weakening the lipid-lowering effect.[37] This 'rebound' phenomenon (tachyphylaxis) was not observed in the GREACE study. This discrepancy may reflect the fact that the original report[37] was based on a retrospective analysis. The authors themselves[37] acknowledged this limitation and pointed to the need for a prospective study. It was not clear how many patients were on atorvastatin in that initial report[37] since a total of 254 patients were taking atorvastatin, fluvastatin, lovastatin, pravastatin or simvastatin; follow-up was for about 300 days. In contrast, the GREACE study[1] had the advantage of a prospective design as well as considerably more power (800 patients on atorvastatin and a three-year follow-up) to detect any tachyphylaxis. Furthermore, to our knowledge, the findings of the initial study[37] have not been reproduced.

There is some evidence that, at higher doses of atorvastatin (40 mg/day and 80 mg/day), the beneficial effect on raising HDL-C levels is decreased.[14,15,16,38] This phenomenon was not observed in the GREACE study,[1] possibly because the average dose of atorvastatin was only 24 mg/day. However, the other side of this argument is that any adverse effect on the HDL-C level would not be relevant in the majority of CHD patients treated with atorvastatin. For example, only 11% and 3% of the patients in the GREACE study[1] needed atorvastatin 40 mg and 80 mg, respectively, to achieve NCEP LDL-C target levels.

The NCEP ATP III LDL-C goal [2.6 mmol/l (100 mg/dl)][30] was achieved in 95% of the hospital-based patients in the GREACE study with a mean dose of 24 mg/day of atorvastatin.[1] In GOALLS,[39] 87% of high-risk patients achieved the NCEP ATP III LDL-C target with simvastatin 20-80 mg/day. However, the inclusion criteria for GOALLS[39] were an LDL-C of 115-180 mg/dl (3.0-4.7 mmol/l). In contrast, the mean LDL-C value for GREACE[1] was 180 mg/dl (4.7 mmol/l). Pravastatin was also assessed in terms of achieving NCEP targets.[40,41] When the top dose of pravastatin (40 mg/day) was compared with 20 mg of simvastatin, only 2% of patients with CHD achieved an LDL-C of 2.6 mmol/l (100 mg/dl) or less with pravastatin, as opposed to 20% of those taking simvastatin.[40] In this context, it is relevant that there are two higher doses of simvastatin available (i.e. 40 mg/day and 80 mg/day), but pravastatin was administered at what is currently its maximal dose (40 mg/day).[40] In another secondary prevention study, the following statins were evaluated: atorvastatin (10-80 mg), fluvastatin (20-80 mg), pravastatin (20-40 mg) and simvastatin (10-40 mg). The statins could be 'topped up' by cholestyramine so as to achieve an LDL-C target value of 2.84 mmol/l (110 mg/dl). The percentages of patients achieving this LDL-C target were 88.6, 60.3, 48.6 and 78.8 with atorvastatin, fluvastatin, pravastatin and simvastatin, respectively.[41] The percentages of those requiring the cholestyramine 'top up' were 5, 47, 60 and 23 with atorvastatin, fluvastatin, pravastatin and simvastatin, respectively.[41] Because this study was carried out before the 80 mg dose of simvastatin became available, we would now expect better results for this statin.[41] It is clear, therefore, that at least one of the evidence-based statins (pravastatin) will often not achieve the currently recommended LDL-C targets. A head-to-head comparison (atorvastatin vs. simvastatin) was also carried out in 1732 patients with and without CHD.[42] Atorvastatin 10 mg and 80 mg lowered LDL-C levels significantly more than simvastatin 20 mg and 80 mg (p = 0.0097 and < 0.0001, respectively, for the 10 vs. 20 mg and 80 vs. 80 mg comparisons).[42] Nevertheless, the evidence supporting the use of simvastatin or pravastatin remains strong because of the extensive clinical experience and the placebo-controlled trials. However, after the GREACE study,[1] the atorvastatin option has also become very convincing.

The findings of the GREACE study suggest that units specialising in vascular disease prevention provide better care than that available in general practice. This may be true if there is a 'lag phase' before the universal implementation of new and effective preventive measures. Therefore, general practice physicians now need to become aware of the GREACE study results so as to ensure optimal treatment of their patients.

Finally, we support the need for trials with clinically relevant endpoints.[14,15,16] These trials should present their findings before, or at least soon after, a drug is licensed. Based on the GREACE findings, a three-year study of this size and involving high-risk patients, would provide sufficient events to at least suggest parity between two drugs (e.g. statins). If one drug had been previously evaluated against a placebo, then the use of the newcomer would, to some extent, be validated.[16]