Do really statins reduce blood pressure?

The beneficial cardiovascular effect of statins in both the primary and the secondary disease prevention remains indisputable [1]. On the basis of randomized evidence, this effect is proportionally related – by logarithmical means – to the extent of low-density lipoprotein cholesterol (LDL-C)-lowering [2]. Statins use has been also associated with amelioration of endothelial dysfunction, increased nitric oxide bioavailability, antioxidant and antiproliferative properties, and inhibition of inflammation, mechanisms not related to their lipid-lowering effect, and which are usually reported as ‘pleiotropic effects’ of statins [3]. Due to the above direct effects of statins on the vasculature, it has been hypothesized that statins may also act as blood pressure (BP)-lowering agents and may contribute to the BP-related cardiovascular risk reduction. Several studies on which this hypothesis is based have important limitations. The vast majority of the available evidence was rather heterogeneous (Table 1) [4–18], because it was obtained by observational studies being performed without a control arm [6,15,18], the data were retrospectively collected and analyzed [15,18], the number of participants was rather limited [7–11], the intensity or up-titration of the ongoing antihypertensive treatment was variable [16] and the follow-up period to investigate BP changes among prospective studies was small (generally not more than 1 year) [5,7–12]. Another important drawback is related to the methodology of BP measurement as the majority of the studies [4–6,8–13,16–17] used clinic BP measurements and not ambulatory BP, which is of greater prognostic importance. In cases in which randomized data are available or ambulatory BP measurements were performed, statins did not show a clear-cut BP-lowering, but many of these studies were designed to address questions different from the BP-lowering effect of statins (Table 1).TABLE 1: Selected studies providing results about the blood pressure-lowering effect of statinsMore specifically, randomized studies with a cross-over design [8–10,12] yielded conflicting results, and even when a wash-out period was anticipated, the underlying carry-over effect cannot definitively be excluded. Among trials with a parallel design, the Anglo-Scandinavian Cardiac Outcomes Trial-Lipid Lowering Arm (ASCOT-LLT) [13] was a large scale, long-term, randomized placebo-controlled study, in which the addition of either atorvastatin or placebo to an effective ongoing antihypertensive treatment regimen was accompanied by a negligible clinic SBP/DBP reduction of −1.1/−0.7 mmHg. In the Plaque Hypertension Lipid-Lowering Italian Study (PHYLLIS) trial [14], which was a relatively small, long-term, randomized placebo-controlled study, pravastatin or placebo were administered to different antihypertensive regimens of equal BP-lowering intensity and the obtained clinic and ambulatory SBP/DBP difference was slightly worse in those randomized to pravastatin compared with placebo (1/−0.3 and 1.6/1.2 mmHg, respectively). In the Heart Outcomes Prevention Evaluation (HOPE)–3 trial [17], rosuvastatin was tested against placebo in intermediate cardiovascular risk patients being also treated with either a combination of an angiotensin receptor blocker and a diuretic or a BP-lowering double placebo. Treatment with active rosuvastatin compared with rosuvastatin placebo was not accompanied by a BP reduction, as the resulted averaged SBP/DBP- difference was 0.3/0.3 mmHg. Two previous meta-analyses of randomized controlled trials [12,16] both suggested that statin treatment was accompanied by a modest BP reduction and one of them [12] suggested that the extent of BP-lowering was larger the higher was the baseline SBP or DBP. In the present issue of the Journal of Hypertension, Spannella et al.[18] by using a relatively large cohort of 1824 hypertensive patients, investigated whether statin-treated hypertensive patients had different BP levels as compared with their nonstatin-treated hypertensive counterparts. To better determine the extent of BP-lowering, apart from clinic BP measurements, they also used ambulatory BP monitoring. Authors reported that statin users (80% of them under antihypertensive drug treatment), had a lower office DBP, and a lower 24-h SBP/DBP compared with unmatched nonstatin users. Also, statin therapy was associated with a better ambulatory BP control after adjustment for known confounders. Indeed, taking a statin was associated with a 2.1-fold higher 24-h BP control compared with nonstatin users after statistical adjustment for demographic and clinical characteristics including the antihypertensive treatment intensity index. Spannella et al.[18] also raised the hypothesis that hypertensive statin users are less likely to have adverse hypertension phenotypes, such as masked uncontrolled hypertension, and sustained uncontrolled hypertension as compared with the sustained controlled hypertension phenotype. To overcome the shortcoming of nonrandomized data (selection bias), and therefore of unequal clinical characteristics between groups (e.g. different antihypertensive treatment regimens, laboratory parameters, and clinical characteristics) authors performed a propensity score matching, which allowed them to obtain two derivative groups with similar clinical characteristics for known confounders. Following this kind of statistical exercise, statin therapy still mildly reduced the clinic and out-of-clinic DBP, but not the SBP in both cases. Again, after achieving a ‘virtual statistical equity’ between groups through propensity scores, statin users achieved 1.8 times better BP-control than their statistically derived controls. Although propensity score matching is quite popular and many investigators have the false belief that this statistical tool can resolve the drawbacks related to the observational nature of the studies [19], the selection of the variables to use together with the retrospective nature of data collection (i.e. only the collected confounders were available) makes the potential for confounding from factors that were not measured in the study still substantive. To assess how much of a problem unmeasured confounding factors may pose, investigators are generally conducting a sensitivity or bias analysis, which is steadily accompanied by different post hoc assumptions (investigator bias) for unmeasured confounders. The susceptibility to misuse of the traditional sensitivity analysis, may be outreached by an alternative tool, namely the E-value, that avoids making assumptions [20]. The E-value analysis used by Spannella et al.[18] asks the question: how strong would the unmeasured confounding have to be to negate the observed results? The authors [18], calculated that the observed E-value was 2.3, meaning that the residual confounding could explain the observed association between statin therapy (determinant) and 24-h BP control (outcome) if there exists an unmeasured covariate having a relative risk association at least as large as 2.3 with both determinant and outcome. As the authors [18] considered all important (available) traditional factors modulating the association between statin therapy and 24-h BP control, it might be unlikely that an unmeasured or unknown confounder would have a substantially greater effect on the investigated relationship than these known risk factors by having a relative risk exceeding 2.3. However, in the study by Spannella et al.[18], the adherence to statin treatment was not evaluated. Is the poor or good adherence to statin treatment related with better 24-h BP control in statin users? In my opinion, this cannot be excluded, though the magnitude of this association remains unknown. The present study cannot provide answers for this particular question not only because adherence was not only assessed but also because the cross-sectional design does not allow conclusions about the long-term drug adherence before the time of BP measurements. This is particularly important as approximately 50% of patients discontinue statin therapy within 1 year, and adherence decreases over time [21]. The study by Spannella et al.[18] cannot provide definitive answers for the BP-lowering effect of statins in hypertension, but may stimulate future randomized research in the field. Although, the PHYLLIS appears to be a well conducted study for the evaluation of BP-lowering effect of statins, all patients received antihypertensive treatment with the intention to create identical BP levels. After randomization to the tested antihypertensive drugs, patients also received treatment up-titration (second-line agents: open-label nifedipine GITS) to achieve an effective and equal BP-lowering between randomized arms. Thus, in the PHYLLIS [14] any potential BP-lowering of statins, may be neutralized or obscured by antihypertensive treatment adjustments according to the individual-patient ongoing BP levels at different follow-up times. Similar considerations can be extended to the results of the ASCOT-LLA trial [13]. By contrast, the HOPE-3 trial in which randomized treatment to rosuvastatin or placebo was tested on top of either a combined antihypertensive treatment or a double BP-lowering placebo, was not biased by a potentially unequal open-label antihypertensive treatment throughout the follow-up period. Additionally, in the HOPE-3 trial [17] no reduction in clinic BP was observed in patients under active rosuvastatin compared with rosuvastatin placebo independently of the presence or not of an ongoing antihypertensive treatment, a hypothesis that have not been approached by both the PHYLLIS [14] and the ASCOT-LLA [13], in which statins were administered together with effective antihypertensive agents. Although the study by Spannella et al.[18] reopens the issue of a synergistic effect between statins and antihypertensive treatment, the results should be seen in the context of the finding that the initially observed SBP/DBP reduction was restricted to −2/−2 mmHg after applying a propensity score matching and second, that the available evidence from large randomized clinical trials denies any BP-lowering effect of statins. ACKNOWLEDGEMENTS Conflicts of interest There are no conflicts of interest.