Blood Pressure During Cardiopulmonary Bypass

Unquestionably, major advancements in the care of cardiac surgical patients have been made in the past half century since cardiopulmonary bypass (CPB) brought about the advent of modern-day cardiac surgery. However, despite improvements in surgical technique, mechanical advances in the conduct of CPB, and an ever-evolving understanding of cardiovascular physiology and pharmacology, several fundamental questions in cardiac anesthesia persist. The target arterial blood pressure during CPB to maintain homeostasis during surgery is one of these questions. There have been several well-intentioned attempts to address this issue, but most studies have been limited by significant patient heterogeneity or other deficiencies in study design and statistical power.1 So, despite decades debating about what is the optimal blood pressure during CPB, the answer remains elusive. In this issue of Anesthesia & Analgesia, Joshi et al.2 have reported a major advancement in our understanding of blood pressure management during CPB, thereby offering some hope that we could soon have a more definitive approach to defining this important physiologic parameter. In this sophisticated study, they utilized near-infrared spectroscopy (NIRS) measurements of regional cerebral oxygen saturation (SctO2) to determine cerebral autoregulatory thresholds. By using cerebral oximetry to examine an individual's SctO2 responses to various changes in blood pressure, they were able to define, in real-time, the specific cerebral lower limit of autoregulation (LLA). By definition, a patient's cerebral autoregulation is intact when there is a poor correlation between perfusion pressure and cerebral blood flow (CBF), and is lost when CBF becomes pressure passive. The LLA is a threshold when the mathematical correlation between CBF and blood pressure transitions from near zero to that approaching one. In their study, the transition was defined when the correlation was ≥0.4, which is a generally well-accepted threshold.3 Although their ability to define the LLA was not perfect, it does provide a major step in moving this field forward. In the absence of any specific monitoring, the usual, albeit arbitrary, LLA has typically been depicted to be approximately 50 mm Hg. This popular concept, however, has previously been argued to be in error.4 Accordingly, it has been contested that this 50 mm Hg LLA should not be used. But, however elegant these arguments may be, without more objective confirmation, we have largely procrastinated in making any meaningful change in our consideration of it. As such, it remains a repeatedly described threshold despite the tenuous data on which it was originally based.5,6 The uncertainty of any specific blood pressure target based on this traditional LLA approach is further accentuated in the cardiac surgery population where hypertension is highly prevalent. In the absence of real-time data on CBF responses to blood pressure, the conventional approach has been to use a higher LLA target in patients with chronic hypertension, predicated by the assumption that these patients have a rightward shift in their autoregulatory curve.7 However, the degree to which the LLA is shifted, if at all, has always been uncertain. As a result, anesthesiologists have been unable to reliably determine when an individual's blood pressure may be too low. The implication of what Joshi et al. describe is that we have largely been inaccurate in our estimation of where the LLA is, with there being a wide range to this threshold in most of our patients, regardless of their baseline blood pressure condition. Even in the absence of significant preexisting hypertension, interpatient heterogeneity alone appears to question the validity of any arbitrary blood pressure above which patients should be maintained. Furthermore, when superimposed on variable degrees of cerebrovascular disease, it is likely even more difficult to define hypotension with any unifying number.8 The question we are then faced with is, Where do we go from here? For example, will actually knowing, and intervening based on, the LLA make any difference to patient outcome? This will require considerable study to determine what outcomes could be expected to be improved. Because this finding intuitively has direct relevance to the brain, a logical principle effect of utilizing the LLA to target hemodynamic management would be to reduce neurologic dysfunction. An example of how this may have an impact has previously been described by these same investigators in an observational study that identified impaired autoregulation during rewarming from hypothermic bypass.9 They highlighted that those with impaired autoregulation may be at risk for other cerebral complications, such as stroke and/or transient ischemic attacks. Thus, using LLA determinations could identify high-risk patients and also open up the possibility of intervening to optimize CBF to attenuate these adverse outcomes. It is not clear, however, as to the scope of what neurologic outcomes (i.e., stroke, encephalopathy, delirium, cognitive dysfunction) might be influenced by using a real-time LLA-guided assessment and intervention. A relationship between lower CPB pressure and increased stroke risk in cardiac surgical patients with severe aortic atherosclerosis has previously been reported.10,11 It was postulated that the lower pressure in those patients exposed to excess atheromatous emboli increased the risk of cerebral ischemia secondary to a reduction in blood flow in the pressure-dependent cerebral collateral vessels.12 However, it is unlikely that using a global cerebral LLA target would be expected to influence regional thromboembolic events per se or that NIRS would be able to identify these relatively small at-risk regions within the brain. However, if aortic atherosclerosis is indicative of severe diffuse cerebrovascular disease, then it could have an impact on overall stroke risk. That is, it would be far more likely to have an impact on the general sequelae of impaired CBF. For example, watershed strokes (secondary to global cerebral hypoperfusion) have previously been shown to occur more frequently in patients managed during CPB with a lower (than their prebypass baseline) blood pressure.13 Maintaining a blood pressure well above the LLA might reduce these types of watershed (hypoperfusion) stroke events. As for more subtle neurologic deficits, there is also evidence that prolonged hypotension could also have an impact on delirium.14 Finding the optimal pressure for patients has the theoretical potential to mitigate all of these subtypes of neurologic complications, but the evidence thus far, is weak. Other non-neurologic outcomes could also be influenced with this information. Murkin et al.,15 in an interventional trial using NIRS to optimize SctO2 (with one intervention being blood pressure manipulation), described an improvement in their major morbidity composite outcome. That study essentially described the brain as a potential sentinel or index organ serving to identify “at-risk” situations when SctO2 was low.16,17 However, it is well known that there is a hierarchy of organ blood flow, and what is an optimal blood pressure (and flow) for the brain may not be the same for other major organs (such as the kidney or other splanchnic organs).18 Because of its inherent autoregulation, as well as the body's preferential perfusion of the brain at the expense of other organs, it may be that despite having a normal brain oxygen saturation, the other organs have significant impairment in their own blood flow and tissue oxygen delivery.19 It is important to understand the context, and limitations, in which the brain can be considered an index organ. First, it is the only organ that is readily accessible to the penetration and reflection of light wavelengths used in NIRS cerebral oximetry technology.20 It is easy to postulate that a more favorable non-neurologic outcome effect might be expected if tissue oximetry signals were used to target interventions by a monitor focused on the kidney or gut, for example. In effect, rather than the brain being an early warning system, it should more appropriately be considered the opposite of a “canary in a coal mine” (in effect, being the last to go) because it is highly likely that once the brain desaturates, the other organs have long since been exposed to compromised blood flow and oxygenation. Thus, using the cerebral LLA as a target that might influence other outcomes is a potential limitation to the data presented by Joshi et al. However, it is not so much a limitation in their study design per se, but in the potential clinical utilization of this information. One further caveat to this study relates to the limitation that the technology and the software that they used to collect and analyze the LAA data are not commercially available. Although commercialization in some related form is likely to occur, it will not be known how this might impact patient care until it is more widely available in a more user-friendly mode. Importantly, however, cerebral oximetry, irrespective of whether specific NIRS-guided determination of autoregulation becomes available, is increasingly being used in the day-to-day management of cardiac and other surgical patients and has shown great promise.19,21 Its ease of application and the intuitive information it provides in reflecting both cerebral oxygen supply and demand issues can be applied to multiple physiologic parameters other than just blood pressure. However, the ultimate impact of this technology will need to be determined in larger-scale prospective controlled trials. The information presented by Joshi et al. shows great promise, but until we have more data related to its impact on patient outcomes, the question of “how low is too low” a blood pressure remains incompletely answered. DISCLOSURES Name: Hilary P. Grocott, MD, FRCPC, FASE. Contribution: Hilary P. Grocott, MD, FRCPC, FASE, wrote and approved this manuscript.