6th April 2009
In this analysis of the previously reported JUPITER study, rosuvastatin 20mg daily compared with placebo reduced the risk of symptomatic VTE in people with few elevated CV risk factors apart from a raised high-sensitivity C-Reactive Protein, in addition to the previously reported small absolute reduction in CV events. There was an increased risk of physician-reported diabetes in people taking rosuvastatin. The MHRA have previously advised caution in initiating this dose of rosuvastatin.
Health professionals who look after hospitalised patients should follow the recommendations of the DH expert working group on VTE by assessing a person’s risk of VTE and taking appropriate steps to manage that. This study does not seem to justify the use of statins for prophylaxis of VTE in the general population and insufficient information is presented to evaluate their usefulness in patients at higher risk.
What is the background to this?
We have previously blogged the JUPITER study, which primarily investigated the effect of rosuvastatin 20mg on the risk of major cardiovascular (CV) events such as myocardial infarction. JUPITER was a large randomised controlled trial (RCT) in 17,802 men and women who would probably not have been assessed as being at high risk of CV events based on conventional risk factors (see trial details, below). However, they all had elevated high sensitivity C-reactive protein (hs-CRP) levels – the median level was 4.3mg/L. Patients were randomised to rosuvastatin 20mg daily or placebo. Observational studies have come to conflicting conclusions regarding the effects of statins on VTE, and the JUPITER authors decided to specify VTE as a secondary endpoint of their study.
What does this study claim?
Over a median follow-up period of 1.9 years, rosuvastatin 20mg daily reduced the incidence of symptomatic VTE by 43% compared to placebo (0.38% vs 0.67%, hazard ratio [HR] 0.57 95% CI 0.37 to 0.86, P=0.007). However, because the baseline risk of VTE in the control group was low the number needed to treat (NNT) was high: 342 people would need to take rosuvastatin 20mg for 1.9 years for one to benefit.
Rosuvastatin did not statistically significantly reduce the incidence of pulmonary embolism (PE): HR 0.77, 95%CI 0.41 to 1.45, P=0.42, although it did reduce the incidence of deep vein thrombosis (DVT): 0.19% vs 0.43%, HR 0.45, 95%CI 0.25 to 0.79, P=0.004, NNT=424
Rosuvastatin reduced the risk of VTE associated with cancer, recent trauma, hospitalisation (including for CV events) or surgery (HR 0.52, 95%CI 0.28 to 0.96, P=0.03, but not in other circumstances (HR 0.61, 95%CI 0.35 to 1.09, P=0.09). However, the study may not have been sufficiently powered for this level of analysis, given the very small number of events. Confidence intervals on results by other subgroups (eg smoking status, obesity etc) were very wide and render attempts at further subgroup analysis futile.
Harms other than bleeding rates were not reported in this paper, but from the main JUPITER report, we know that there were no significant differences between the groups with regard to muscle, hepatic or renal adverse effects, bleeding events or newly diagnosed cancer, but there was a 25% relative increase in the risk of physician-diagnosed diabetes in the rosuvastatin group: 3.0% vs 2.4% in the control group (RR 1.25, P=0.01; number needed to harm [NNH] =165)
The reductions in HR look impressive. However, when absolute benefits and harms are examined, for every 1000 people like those in JUPITER treated with rosuvastatin 20mg daily for 1.9 years, 3 people would not develop VTE who would have done otherwise and 4 people would develop VTE, just as they would have done without rosuvastatin. Combined with the results from the main JUPITER report, this means that 15 people in 1000 would avoid VTE or a major CV event, but 19 people would still have either a VTE or a major CV event or both.
However, on rosuvastatin an extra 6 people would be diagnosed with diabetes in addition to the 24 who would have been diagnosed with diabetes anyway. Moreover, rosuvastatin 20mg did not reduce the risk of PE (the presentation of VTE most associated with death). It is not possible to say from this report how long in advance of surgery or hospitalisation, etc rosuvastatin would need to be taken to reduce the risk of VTE, nor the extent to which rosuvastatin is additive to or compares with the effects of standard treatment in these circumstances or in people with higher baseline risk of VTE..
The MHRA have advised caution in initiating the dose of rosuvastatin used in this trial: patient should start on 10mg (5mg for Asian patients and those with pre-disposing factors for myopathy), including patients switched from other statins, and the dose should be titrated up only if necessary and after a four week trial.
As with the main JUPITER trial, this study may stimulate research in use of hs-CRP as a marker for risk of VTE. However, this study did not evaluate this use of hs-CRP estimation .
The authors speculate on the mechanism by which rosuvastatin might have exerted the observed effects, and highlight the pleiotropic effects of statins (e.g. anti-inflammatory and other effects, independent of their lipid lowering effects). It is possible that other statins might have beneficial effects on VTE risk, but this is has not been tested in an RCT.
Patients: 17,802 men and women (median 66 years, 38% women, median LDL 2.8 mmol/L, median HDL 1.3 mmol/L, median BP 134/80 mmHg, 84% non-smokers, high sensitivity C-reactive protein (hs-CRP) 4.2mg/L.
Intervention and comparison: Rosuvastatin 20mg a day, or placebo, for a median of 1.9 years (maximal follow-up 5.0 years)
Outcome: Over a median of 1.9 years, the incidence of of symptomatic VTE (DVT and PE) was reduced by 43% (0.38% vs 0.67%, HR 0.57 95% CI 0.37 to 0.86, P=0.007, NNT=342). More specifically, the incidence of DVT was reduced to 0.19% from 0.43%, HR 0.45, 95%CI 0.25 to 0.79, P=0.004, NNT=424. However, rosuvastatin did not statistically significantly reduce the incidence of PE: HR 0.77, 95%CI 0.41 to 1.45, P=0.42. The HR reduction in HR for symptomatic VTE was statistically significant only for “provoked” VTE (ie in patients with cancer or recent trauma, hospitalization, or surgery): 0.17% vs 0.33%, HR 0.52 (0.28 to 0.96) P=0.03, NNT=636). In patients in whom the VTE was not associated with a proximate event, the difference in rates was not significant (HR 0.61, 95%CI 0.35 to 1.09, P=0.09).
According to the original JUPITER paper, there were no differences between the groups with regard to muscle, hepatic or renal adverse effects or cancer, but there was a 25% increased risk of physician-diagnosed diabetes in the rosuvastatin group compared to the placebo group: 3.0% vs 2.4% respectively,(RR 1.25, P=0.01; NNH=165;). Median glycated haemoglobin was 5.7% in both groups at baseline and increased to 5.9% in the rosuvastatin group and 5.8% in the placebo group (P=0.001).