The debate over whether cholesterol-lowering statins have the potential to ward off dementia has taken a new turn. A review of US data suggests not only that the claimed benefits may be significant but also that they vary markedly according to the type of statin used.

An international team led by Dr Benjamin Wolozin from Boston University School of Medicine used the Decision Support System (DSS) of the US Veteran Affairs database, which includes diagnostic, medication and demographic information on some 4.5 million subjects, to analyse the relationship between statins and dementia.

The researchers employed three types of Cox proportional hazard model to examine associations between lovastatin (Mevacor), simvastatin (Zocor) or atorvastatin (Lipitor) and dementia in subjects aged over 64 years who did not have a prior diagnosis of Alzheimer’s disease or Parkinson’s disease.

These cases, which included more than 700,000 people taking simvastatin and over 50,000 taking atorvastatin, were compared with subjects on cardiovascular medication other than statins, after adjusting for co-variates associated with dementia or Parkinson’s disease.

The first hazard model incorporated adjustments for age, the second for three known risk factors in dementia (hypertension, cardiovascular disease and diabetes), and the third for the Charlson index, which provides a broad assessment of chronic disease. The results were published online in the open-access journal BMC Medicine.

These showed a strong reduction in the hazard ratio (HR) or relative risk for incident dementia with simvastatin that was significant in each of the models. In model 3 the HR for simvastatin was 0.46. Atorvastatin was associated with a reduction in HR for incident dementia that was significant in model 1, borderline significant in model 2 and not significant (0.91) in model 3. Lovastatin was not associated with a significant reduction in HR in any of the three models.

Looking at Parkinson’s disease, the researchers did not find any significant link between use of lovastatin or atorvastatin and reduced incidence of the disease in any of the proportional hazard models. However, simvastatin exhibited a reduced hazard ratio for newly acquired Parkinson’s in all three models, with a HR of 0.51 in model 3.

Mixed results

As the researchers noted, previous studies exploring a possible association between statins and reduced prevalence of dementia have generated mixed results. This led them to hypothesise that the ambiguity might be down to differences in efficacy among the different statins, variations in response to treatment and insufficient power in the sample sizes to detect these modifying factors.

The strength of the reduction in dementia incidence seen with simvastatin in the latest study, one based on a very large population database, was “striking”, the authors commented. “Further studies are required to determine whether this effect represents a biological action of simvastatin or a statistical bias skewing results obtained from the DSS database,” they said.