Reducing high blood pressure (BP) improves cardiovascular mortality and morbidity. It has been estimated that about 14% of all deaths worldwide are directly attributable to elevated BP, so its effective monitoring and lowering in patients with hypertension is fundamental. Nevertheless, over the past 2 decades, several reports have suggested that excessive BP lowering may increase morbidity and mortality, especially in patients with coronary artery disease (CAD) and when diastolic BP falls below 80 to 85 mm Hg. This so-called J-curve relation has been debated ever since, and its relevance was recently acknowledged by the European Society of Hypertension in its reappraisal of the European guidelines on hypertension management.

Part of the evidence for the J-curve comes from a secondary analysis of the International Verapamil-Trandolapril Study (INVEST). In brief, a J-shaped relation between diastolic BP, with a nadir at 119/84 mm Hg, and all-cause mortality and myocardial infarction (MI) was observed in patients from INVEST, which included >22,000 patients with stable CAD and hypertension who were treated with either sustained-release verapamil or atenolol. Nevertheless, in addition to being a post hoc study with the reduced strength of evidence that such analyses invariably carry, the trial’s subjects did not have their left ventricular (LV) systolic function evaluated. In our opinion, this is a significant shortcoming in the study design, and here we discuss why this confounding factor may have led to biased study results.

First, epidemiologic studies indicate that 4.8% to 8.5% of patients with CAD present with asymptomatic LV systolic dysfunction. Limited data on the prevalence of patients with asymptomatic LV systolic dysfunction and previous MI come from the Trandolapril Cardiac Evaluation (TRACE) study group, in which of 6,676 patients with MIs, a total of 677 (10.1%) were found to have asymptomatic LV systolic dysfunction. Hence, about 1,500 patients with asymptomatic LV systolic dysfunction could have been included in INVEST, because all had known CAD and 32% reported histories of MI at baseline (10% of the 32% of patients with MI plus 5% of the 68% of patients with CAD). Second, patients with asymptomatic LV systolic dysfunction are at high risk for adverse outcomes, independently of their BP. In the Survival and Ventricular Enlargement (SAVE) trial, which included patients with asymptomatic LV systolic dysfunction and recent MIs treated with captopril or placebo, the mortality rate in the placebo arm at an average of 42 months was 25%. Thus, INVEST probably contains an unidentified, but significant, subgroup of patients with increased risk for adverse outcomes. Finally, given the clinical characteristics of these high-risk patients (asymptomatic LV systolic dysfunction), it is possible that they had lower than average BP at baseline and that after intervention were found to be in the lower BP range, contributing to the increased morbidity and mortality observed. Indeed, patients with asymptomatic LV systolic dysfunction from the SAVE trial had a relatively low BP of 112/70 mm Hg at baseline. In the INVEST post hoc analysis, the J-shaped relation observed was, as expected, weakened when baseline symptomatic heart failure, MI, and risk factors for cardiovascular disease were adjusted for in the multivariate analysis. Therefore, if LV systolic dysfunction had been evaluated and adjusted for, it is possible that the association between low diastolic BP and increased morbidity and mortality would have been further attenuated or even abolished.

There is no doubt that the J-curve exists, because there must be a point at which BP becomes too low to provide sufficient perfusion to vital organs. The question is whether the nadir of this J-curve lies within the BP range that treated patients with hypertension exhibit. Data from trials including patients with heart failure treated with β blockers or angiotensin-converting enzyme inhibitors show that the aforementioned drugs are safe and beneficial, even when BP is low. For instance, in a trial with patients with known heart disease and asymptomatic LV systolic dysfunction treated with enalapril or placebo, the average systolic and diastolic BPs were significantly lower in the enalapril group than in the placebo group (5.2 and 3.2 mm Hg, respectively). The baseline BP was relatively low (125/78 mm Hg), yet significant reductions in the incidence of heart failure and the rate of related hospitalizations were reported in the treatment arm. With this in mind, it seems as if the clinical relevance of the J-curve is perhaps not as important as some believe, at least not in some patients. Moreover, we believe that promoting the message that low BP is bad, on the basis of post hoc analyses such as INVEST with potential confounders, will result in clinicians’ undertreating patients with hypertension. Trials with extensive patient characteristics, designed to study the relation between low BP and clinical outcomes, are much needed to provide more insight into the J-curve question.