Primary prevention using statins: to be or not to be?

Key words:

• CVD
• primary prevention
• OTC statins
• NNT
• apoB
• cascade screening
• JUPITER

During the past 30 years big strides have been made in the management of cardiovascular disease (CVD). Mortality from coronary heart disease (CHD) has fallen by half, of which half is owed to improved acute intervention (e.g., thrombolysis, stenting) and half to reduction in risk factors, especially smoking, cholesterol and hypertension1–10. A great impact has been made from the use of statins, although diabetes and obesity probably off-set some of this gain.

Although ‘prevention is better than cure’ some vascular risk factors – such as age and gender – cannot be modified. Other risk-factors that can be changed without pharmacological intervention – such as stopping smoking, achieving optimal weight and taking some exercise – are not easily maintained. Against this background there is a need to decide whether primary prevention by reducing low density lipoprotein cholesterol (LDL-C) pharmacologically is justified. In this context, two recent meta-analyses addressed this issue reaching contradictory conclusions4,5. Furthermore, it has been suggested that statins might accompany fats foods like condiments6. The authors review these analyses and discuss their implications.

The meta-analyses of 2009 and 2010

Recently, the presentation of the outcomes of both the Scandinavian Simvastatin Survival Study (4S; a secondary prevention trial)7 and the West of Scotland Coronary Prevention Study (WOSCOPS; a primary prevention trial)8 as reduction in absolute instead of relative risk was debated and the calculations of the number needed to treat (NNT) have illustrated how much easier it is to achieve a benefit in secondary compared with primary prevention9. The NNT in WOSCOPS to prevent one extra coronary mortality event within 5 years of treatment was 200. This is about three times more than in 4S7. A meta-analysis of seven trials where 80% or more of those enrolled were considered to have no CVD at inclusion also reported that the cost effectiveness of statins in primary prevention varied according to the risk of patients10. The estimated NNT was 133, 61 and 40, for those at low, intermediate and high risk, respectively; which is only about a doubling compared with secondary prevention10.

The debate continues in a recent issue of Archives of Internal Medicine which includes articles and an editorial devoted to the conundrum of whether primary prevention with statins is justified5,11,12,13.

The meta-analysis by Ray and colleagues5 was prompted by another one published in 2009 in the BMJ4. The BMJ article favoured statins in the primary prevention setting while at the same time acknowledging that certain groups were at particular risk (men >65 years with risk factors, or older women with diabetes and risk factors). After removing the Justification for the Use of Statins in Prevention: an Intervention Trial Evaluating Rosuvastatin (JUPITER) study (n = 17 802) from the analysis, the reduction in mortality in the other nine trials remained significant. The meta-analysis by Brugts et al.4 pooled 70 388 people but the absolute mortality risk reduction seems to only be 0.6% (5.7 vs. 5.1%) over 4.1 years. However, treatment with statins significantly reduced the risk of all-cause mortality (odds ratio 0.88, 95% confidence interval, CI, 0.81–0.96), major coronary events (0.70, 95% CI 0.61–0.81) and major cerebrovascular events (0.81, 95% CI 0.71–0.93) without significant heterogeneity in clinical subgroups. There was no evidence of an increased risk of cancer.

Ray and colleagues5 took the view that various recent meta-analyses had defects, which included trials with incomplete randomisation and patients with prevalent CVD. Therefore, they limited their analysis to studies that only enrolled patients without existing CHD or studies from which they were able to obtain raw data. Although Ray and colleagues clearly set out how they arrived at the 11 studies in their analysis, the exclusion of the Heart Protection Study (HPS) ‘no prior CHD’ patients14 may move the upper relative risk boundary away from its current 1.00–1.02 to be either unequivocally positive or negative. This study with simvastatin was conducted in the UK where the authors of the meta-analysis are based. The HPS14 randomised 20 536 patients of whom 7150 (35%) had no history of CHD. This subgroup represents the 5th largest study in the Ray et al. meta-analysis. In the HPS group the event ratio for first major vascular event for ‘no prior CHD’ was 0.75 (574/3575 vs. 744/3575) compared to 0.76 (2033/10 269 vs. 2585/10 267) for all patients. Given that for the whole group there were 1328 vs. 1507 (p = 0.0003) deaths from any cause, of which 781 vs. 937 (p < 0.0001) deaths were due to vascular cause and 587 vs. 707 (p = 0.0005) specified as due to CHD, it is indeed a pity that the results for patients with ‘no prior CHD’ were not reported separately.

Also, in view of the low mortality rate in primary prevention trials there is a need to seriously consider whether a meta-analysis should exclusively focus on mortality (as in the Ray et al. meta-analysis5) and not include all vascular events. In this context, an older meta-analysis (2006) showed that statin therapy (after a mean follow-up of 4.3 years) reduced the relative risk of major coronary events, major cerebrovascular events, and revascularizations by 29.2% (95% CI, 16.7–39.8%) (p < 0.001), 14.4% (95% CI, 2.8–24.6%) (p = 0.02) and 33.8% (95% CI, 19.6–45.5%) (p < 0.001), respectively10. The fall in CHD and overall mortality was not significant.

Can studies with such different entry criteria be combined?

Although Ray and colleagues5 selected studies in a predefined formalised manner, it has to be considered whether studies in such varied populations with different inclusion and exclusion criteria can be pooled. For example, the Air Force/Texas Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS)15 addressed low high-density lipoprotein cholesterol (HDL-C) patients; the Management of Elevated Cholesterol in the Primary Prevention Group of Adult Japanese (MEGA)16 was conducted exclusively in Japan; WOSCOPS8,17 was run in a higher-risk’ population (Western Scotland); and Prospective Study of Pravastatin in the Elderly at Risk (PROSPER)18 specifically considered older people. Finally, in the much discussed JUPITER19 study, recruitment was based on raised high sensitivity C-reactive protein (hsCRP) but a ‘low’ LDL-C (<130 mg/dl; 3.4 mmol/l). In contrast to Brugts and coworkers4, Ray et al.5 did not perform an analysis excluding JUPITER probably because it would not add to their conclusion that primary prevention was not worthwhile. Instead they attempted to set JUPITER in context and argued as have others11,12 that JUPITER may represent extreme or exaggerated effects.

Age is a well-recognized and non-modifiable risk factor and age accounted for two-thirds of the variation in mortality across the studies in the meta-analysis by Ray et al.5. They also reported that there was no relationship with the baseline level of LDL-C, or with the absolute/relative reduction in LDL-C. In contrast, an earlier meta-analysis of 2006 reported a significant association with both a larger absolute and also relative change in LDL-C10. As well as age, we also need to consider the effects of other variables such as gender20 and kidney function in primary prevention statin trials21.

Are meta-analysis models based on these data, however weighted, valid?

Possible solutions to the doubts concerning primary prevention statin trials

A new, definitive, study

One way forward is for a definitive study large enough to address the issues that the meta-analyses have raised. However, managing the practical aspects (including the considerable cost and deciding which statin should be evaluated), must not be underestimated. The pharmaceutical industry is unlikely to fund such a study knowing that the population will need to be huge, the outcome doubtful (especially if mortality is the main endpoint), and that by the time the results are published, their drug might already be available as a generic product.

How can these doubts about primary prevention with statins be resolved? Perhaps we should look at a key debate that has emerged in the field of ophthalmology. A head-to-head comparison without industry support is well advanced in ophthalmology where the National Eye Institute (NEI) of the National Institutes of Health (NIH) is conducting a comparison between bevacizumab (Avastin) and ranibizumab (Lucentis) in wet macular degeneration22. The issue about these two drugs is well-described. Suffice it to say that the active ingredient in either drug formulation is a monoclonal antibody directed against VEGF. Bevacizumab was developed and registered for oncological indications while ranibizumab was developed for the eye indication. By splitting a vial of bevacizumab between several patients it is possible to treat wet macular degeneration for $40 rather than $1500 a shot23. Also, bevacizumab is thought to remain in the eye longer than ranibizumab and therefore possibly permit less frequent injections. The NEI study randomised 1200 patients and closed to recruitment earlier this year. Meanwhile, a small randomized study in 131 patients has been reported24. Thus, rather than performing ever shrinking meta-analyses and arguing about what might have happened had specific studies been conducted differently, or whether a certain study is unrepresentative, it behoves the academic CVD community to take a leaf out of the ophthalmologists’ book and conduct a definitive primary prevention statin trial because the issues involved are of major importance. It has been suggested that such studies might be defrayed through a small levy on drug sales25.

OTC statins as an option for primary prevention

The UK regulatory agency (MHRA) broke new ground in 2004 when it approved simvastatin for pharmacy sale without a prescription26–28. The reclassification report relies on a 2003 systematic review comparing simvastatin 10 mg with placebo, which showed a 27% fall in LDL-C29. This was associated with a reduction in risk of a major coronary event (death or non-fatal myocardial infarction) by about one-third after 3 years30. Briefly, the justification of the MHRA was that ‘the current evidence suggests that, for adults in Western societies, it can be beneficial to reduce cholesterol levels whatever the starting point’ – and – that ‘statins are now recognized to be so safe so that there is no routine need for tests to check for safety once people have started taking them’. It should be noted that in the UK simvastatin from pharmacists (Zocor Heart-Pro) is limited to men aged 45–70 years or women aged 55–70 years who have gone through the menopause, and that the sale is subject to approval by a pharmacist after a review of the level of risk27,28.

The approved simvastatin OTC dose is at the lowest level (10 mg/day). Although this may, as the MHRA has concluded, make a substantial difference in public health terms, one has to wonder, in the light of evidence from studies, whether this dose is enough31 and what long-term compliance will be. There is evidence that patients who are compliant with statins are generally health conscious32, so those that pay for their own primary prevention therapy may be those who are less likely to need it because of their satisfactory lifestyle.

About 6 months after simvastatin became available through pharmacies, a questionnaire-based survey of a random sample of 2000 registered community pharmacy premises in the UK found that, of the 58% who responded, 83% reported no sales of simvastatin in the previous 14 days, 7% had sold one pack, 4% two packs and 2% three packs or more (4–5% did not answer)33. Thus, the uptake was not great. This was supported by a telephone survey of 165 pharmacies in six different geographic regions in the UK during April and May 2005 to assess OTC simvastatin sales34. A total of 141 (86%) of the 165 pharmacies surveyed stocked OTC simvastatin. However, only 13 pharmacies (9%) had sold a total of 18 packs during the prior 2-month period. Sales were independent of socioeconomic status of the region and urban or rural status. There were more requests for OTC simvastatin than sales, primarily due to requestors not satisfying the required cardiovascular risk level or who were not willing to pay for it34,35. A further cross sectional questionnaire survey of all primary care general practices in Scotland found that the majority did not support the supply of OTC simvastatin by community pharmacists, being particularly concerned by the lack of cholesterol and blood pressure data in the CHD risk assessment36. The results from these surveys may not be representative of the entire country. Nevertheless, simvastatin has been available through pharmacists in the UK for 6 years now, so the UK regulatory agency or an academic group should assess its success/failure.

Although the MHRA considered that simvastatin at 10 mg/day was safe enough to be available OTC, one has to wonder if this is realistic. The primary safety issue with the statins has been myositis. The risk of myositis increases in the elderly and in the presence of certain other drugs, impaired renal function or hypothyroidism37–40. Populations at greatest risk of developing hypothyroidism include women older than 60 years of age, patients with a history of radiation to the neck, head, or chest, persons with diabetes, and patients with pernicious anemia41. In a Swedish study of 601 women and 285 men aged 85 years, the prevalence of previously undetected hypothyroidism was estimated to be 4.0% in women and 2.5% in men42. A UK study of 1210 patients aged over 60 years found 18 patients who were hypothyroid and another 13 developed hypothyroidism over 1 year and were started on thyroxine (i.e., overall approximately 2.5%)43. An American study of 968 persons in an urban population aged above 55 years found a much higher rate of 6.9% with hypothyroidism with dysfunction commoner in women, and in subjects older than 75 years as compared with the 55- to 64-year age group44. It is somewhat surprising and a concern that after 6 years not enough data are easily accessible to show if OTC simvastatin is a worthwhile option. Does the concept that the individual pays for his or her own prevention work? In this context, no other EU country has so far followed the UK and the FDA has rejected OTC statins three times45,46. Notwithstanding, a UK based group has just proposed that statins should be offered like condiments along with fast food because in their statistical modelling (not quite a full-fledged meta-analysis) statin therapy can off-set the CVD danger from eating poor diets6. The reason why the UK has experimented with pharmacy availability of statins is illustrated in an editorial which questioned, not only the risk-benefit, but the cost effectiveness (admittedly from an US standpoint)47.

From time to time questions have been raised of possible roles of statins in cancer. Recently, Bardou et al.48 concluded from an extensive review of both randomised clinical trials and observational studies that long-term statin use could not be recommended for the prevention of colorectal cancer because any impact there might result in numbers needed to treat (NNT) that were between 35 and 250 higher than those for CVD prevention (illustrating the uncertainty of their estimate). Any randomised trial would need to enrol hundreds of thousands of people to demonstrate any protective effect. In a separate study Hippisley-Cox and Coupland49 found no support for unintended benefits of statins, except possibly for oesophageal cancer, although the NNT with any statin to prevent oesophageal cancer was on average 35-fold greater (range 30–50, possibly more) than for CVD for both men and women. In addition to myopathy, they identified cataracts, renal failure and liver dysfunction as dangers associated with the use of statins. Statins were not significantly associated with risk of Parkinson's disease, rheumatoid arthritis, venous thromboembolism, dementia, osteoporotic fracture, gastric cancer, colon cancer, lung cancer, melanoma, renal cancer, breast cancer or prostate cancer. There had been speculation that these conditions may be favourably affected by statin use. If at some later date such benefits are confirmed and become recognised indications for statin use then the cost effectiveness of statins will need to be recalculated.

Identifying vascular risk

How do we tackle the problem of identifying those who will benefit from primary prevention in the real world of family (general) practice? This may include selective case finding (e.g., cascade screening of heterozygous familial hypercholesterolaemia)50,51. One option would be to invite those seeking advice to have a random blood sample and measure either apolipoprotein B (ApoB) on its own, the ApoB:ApoA1 ratio or non-fasting non-HDL-C52–56. Those at increased risk can be brought back for a full work-up or referred to a lipid clinic. The use of ApoB:ApoA1 – or simply ApoB – as a screening test is attractive because it can be performed on a random sample and it can be used to monitor how effective lipid lowering treatment is52–56. The apolipoproteins are now included in the American Diabetic Association and American College of Cardiology guidelines. It is also necessary to establish whether detecting ‘advanced for age’ vascular disease (e.g., by imaging techniques and biochemical markers) will help select populations with a smaller NNT. There may also be groups where there is no clinical evidence of vascular disease but the future risk is great, such as rheumatoid arthritis and similar disorders57,58.

After JUPITER, the value of measuring hsCRP levels in identifying vascular risk will also need to be established11–13,19,59. However, this topic is the subject of considerable debate and is beyond the scope of this editorial59.

Conclusions

Although the evidence is far from convincing, it is likely that certain groups need statins as part of primary prevention. Making statins available from pharmacists (as in the UK) is probably not optimal but a detailed assessment (including safety) must be published. The use of markers of risk that can be assessed in the non-fasting state such as ApoB and the ApoB:ApoA1 ratio should be evaluated. The academic community should unite in a final lipid-oriented primary prevention trial (if necessary with 50 000 or more participants); one of its primary questions could be to ApoB or not to ApoB?

Surprisingly, there is still a need to develop realistic and practical algorithms of how to implement a primary prevention strategy in terms of NNT. The public and healthcare professionals deserve better.

Transparency

Declaration of funding

This editorial was written independently; no company or institution supported the authors financially or by providing a professional writer.

Declaration of financial/other relationships

D.P.M. is Editor-in-Chief of this journal. The authors have given talks, attended conferences and participated in trials and advisory boards sponsored by various pharmaceutical companies.