Do Let the Blood Pressure Fall

In this issue of the journal, Aronson et al.1 present their investigation of the relationship of systemic blood pressure (BP) variability with perioperative outcomes after cardiac surgery. They, along with others, have previously analyzed the implications of either preoperative or preinduction systolic or pulse pressure with outcome, reporting the superiority or inferiority of one over another (or versus diastolic BP).2–8 This topic has long been of interest to cardiac anesthesiologists, with most well-publicized studies focusing primarily on hemodynamic predictors of ischemia.9,10 However, it was not until adoption of electronic anesthesia record systems, facilitating study of large cohorts of patients, that it gained much needed “traction.” The result has been important studies of intraoperative hemodynamic patterns.11,12 Despite these studies, closer attention has been devoted to the implications of heart rate and its control, given the high-profile studies by Slogoff and Keats of intraoperative ischemia in the 1980s and the perioperative β-blocker boom of the 1990s (often with minimal consideration of the systolic BP as long as it is >100 mm Hg).13–15 The predictive value of either heart rate or BP parameters remains controversial with little consensus on precise ranges for an individual patient. Thus, further data are crucial. Most recently, Aronson et al.7 have entered new territory, reporting on the relation of intraoperative variability in systolic BP with 30-day mortality in 7504 patients undergoing coronary artery bypass graft (CABG) surgery at Duke University using continuous intraarterial measurements. They reported that the mean duration of systolic excursion outside of a range of 105 to 130 mm Hg was predictive of 30-day mortality with an odds ratio of 1.03 per minute (outside the range). In addition to their work on observational associations of BP with outcome, Aronson and colleagues have also assessed the safety and efficacy of a new ultrashort-acting antihypertensive agent, clevidipine, in comparison to nitroglycerin, sodium nitroprusside, and nicardipine in prospective, randomized, open-label, parallel comparison studies involving 1912 patients undergoing primarily CABG surgery.4,6 Given that these pooled data constitute a large prospectively and carefully acquired resource, mining it for interesting associations would appear to be of interest to clinicians, although appropriate caution must be entertained when dealing with post hoc analyses.16 In this issue of the journal, they present a retrospective, post hoc analysis of these patients, specifically evaluating variability of systolic BP on 30-day mortality.1 When approaching this type of study design, the reader should be clear about where the data came from and understand the study methods and results. Thus, with regard to the ECLIPSE studies, the researchers were investigating therapy to reduce BP outside of locally established thresholds during the intraoperative period and for 24 hours postoperatively. The ECLIPSE study did not investigate how to raise BP in hypotensive patients. BP was limited to systolic measurement at intervals ranging from every 15 minutes to 1 hour in the postoperative period. In the ECLIPSE study, 23% of the enrolled patients never met the locally established criteria for antihypertensive treatment and thus were excluded from this analysis. How this sizable percentage of patients fits into the implications of the statistical model is unknown. The ECLIPSE study failed to demonstrate superiority of one drug over another for its primary prespecified composite outcome, although when considering mortality alone, sodium nitroprusside was associated with higher mortality. Clevidipine appeared to be superior to the other agents for the secondary outcome of “adequacy and precision” of BP control, a benefit linked to one of the present study's findings (e.g., that it is beneficial to treat hypertension above a threshold hazardous to longer-term outcome). The analysis looks at outcome as a function of variability in systolic BP. Systolic BP variability is a complex mathematical concept that could be modeled in many different ways. For this analysis, the authors utilized a relatively straightforward “area under the curve” to integrate the time and magnitude that the systolic pressure was either above or below “bandwidths” derived from the ECLIPSE dataset. The use of all-cause mortality is an important caveat to consider given that certain outcomes are more likely than others to either be caused by, or themselves cause, abnormality in BP (e.g., hypertension causing a hemorrhagic stroke or septic shock from infection causing hypotension). The overall event rate was relatively low. As noted in the discussion, the review process addressed how many candidate variables per outcome event are acceptable in a logistic model. Using several statistical approaches, the primary conclusion is that variability in systolic BP, proportional to the extent of excursion outside of a bandwidth of 75 to 135 mm Hg intraoperatively and 85 to 145 mm Hg postoperatively, is associated with increased 30-day mortality, and that this increase is considerably greater in high-risk patients. As the authors note, classification of patient risk was based on available demographic and risk variables collected in the ECLIPSE study, rather than more widely used risk models (e.g., EUROSCORE, STS risk model).17 Despite this, the analysis discriminated well between extremes of risk, and higher-risk patients had greater variability in BP and, as expected, higher mortality. What should the interested reader look for in potentially applying the results of this analysis to their own practice? As the authors note, the results are hypothesis generating, and require prospective validation before widespread clinical implementation or mandating that hemodynamics be managed precisely within these bandwidths. At first glance, this appears just a bit circular, as one could argue that the initial ECLIPSE study failed to validate the utility of BP control. However, the ECLIPSE study was specifically designed as a safety and efficacy trial, and thus conclusions drawn in a retrospective analysis mandate prospective validation. Segregating the treatment of hypotension from hypertension (2 very different physiologic events) in a meaningful manner is intellectually complex. There is a burgeoning literature on variability in arterial characteristics with aging, the effects of different antihypertensive therapies on them, and on prognostic implications of ambulatory systolic BP variability.18–26 Experienced clinicians are well aware that there are a few mandatory hemodynamic derangements commonly accepted during on-pump (e.g., lowering of systolic BP for aortic cannulation) or off-pump (e.g., hypotension of varying degrees during verticalization of the heart) CABG surgery. There are also a variety of local surgeon or institutional practices that can influence BP (e.g., using high levels of positive end-expiratory pressure postoperatively to reduce mediastinal drainage). Thus, it will be very difficult to develop a convincing prospective trial that encompasses the many sources of BP variability. In 1976, my former chairman, Dr. William Hamilton, provided the provocative admonition “Do let the BP drop and do use myocardial depressants!” as the title of an editorial in response to a canine study of Bland and Lowenstein evaluating the response of ST segment changes to halothane administration during coronary artery occlusion wherein reduction in systemic arterial pressure was associated with improvement in ST segment changes.27,28 The primary “take home” message was that indices of myocardial oxygen supply and demand were favorably influenced by halothane despite reduction of systemic pressure. Dr. Hamilton made no bones about his intellectual differences with the esteemed senior author of that report, whom he playfully accused of “worship at the temple of morphine anesthesia” (in contrast to his own avowed affection for halothane). However, it was his admonition that the effects of this study “demonstrate that the time honored cliché about maintenance of blood pressure is subject to serious question” that remains particularly relevant today, after 35 years of investigation. Aronson and coinvestigators are surfing a wave of multiple potentially intersecting lines of investigation being reported by many cardiovascular researchers evaluating different facets of BP and heart rate from the laboratory setting to national populations. Although these investigators are riding the same wave (studying the implications of BP dynamics in human physiology), most are still “feeling different parts of the elephant.”a What parameter will be most predictive and clinically useful? Static or dynamic? Preoperative, intraoperative, or postoperative? Systolic, diastolic, pulse pressure, or even pulse wave velocity? And what about our most commonly accepted “culprit” for bad outcome: heart rate? The crest of this wave is tall and the face steep! Whether or not Aronson and coinvestigators are ultimately able to “hang ten” and ride the wave to a photo finish victory with style points awarded for well-validated clinical applicability to individual patients remains to be seen. The cameras are on the beach and the spectators waiting. In the meantime, I'll follow the sage advice of Dr. Hamilton ………. but maybe not so much or so often.