Our pilot study investigated the ability of a machine learning–derived EWS in combination with an algorithm for hemodynamic management in reducing hypotension episodes and duration during major general surgery. Furthermore, we evaluated the clinical impact of intraoperative hypotension on end-organ damage and oxidative stress with a series of biochemical assays.
We found that the intervention group had significantly reduced incidence and duration of intraoperative hypotension, as well as lower time-weighted average of hypotension during surgery. Our results are consistent with previous studies showing that the EWS was able to predict hypotension with good sensitivity and specificity[10, 11]. To the best of our knowledge, our study is the first one demonstrating that the use of an EWS with an hemodynamic algorithm with the consequent reduction in intraoperative hypotension was associated also with a significant reduction in the value of biomarkers of organ injury and oxidative stress.
Our results add knowledge to the growing body of evidence regarding the role of HPI in predicting intraoperative episodes of hypotension, thus allowing a proactive anesthesiological approach. Once an alarm was detected (HPI>85%), the anesthesiologist in charge followed a treatment algorithm based on advanced hemodynamic parameters, which suggested vasopressor, fluid and/or inotrope administration (alone or in combination), or eventually observation. We found a significant reduction in hypotension in the intervention group and this was consistent with all the measures performed (episodes, absolute and relative time of hypotension, and time-weighted average). The time-weighted average of hypotension seems a very promising variable to evaluate the degree of intraoperative hypotension, taking into account both the severity of hypotension and its duration as well as the total surgical time. In this regards, our intervention group found similar values of time-weighted average (0.12 mmHg) as compared to other larger studies (0.10 mmHg[10] and 0.14 mmHg[12]). Indeed, our results seem in line with most of currently published findings. Although a recent study by Maheshwari et al conducted in 214 patients undergoing non-cardiac surgery questioned the role of HPI[12], most of the published evidence supports the value of HPI in predicting hypotension. Indeed, such results have been found in patients undergoing total hip arthroplasty[13], major/general surgery[14, 15] and cardiac surgery[16]. Such studies have demonstrated good sensibility and specificity in predicting hypotension from 5 minutes (both around 85%) to 15 minutes (both around 75%-80%) before the hypotensive episode[14, 16, 17]. Of note, two of these studies evaluated the use of HPI throughout non-invasive arterial pressure waveform, with encouraging findings[15, 17].
Our study has the originality of evaluating also biochemical aspects, thus functioning as a pilot study for the investigation of the impact of intraoperative hypotension on oxidative stress and end-organ injury. Of course, the reliability of our findings are limited by the small sample size, but the correlation between the primary outcome measures and several biomarkers may warrant further investigation.
Firstly, we showed that HIF-1a is not significantly up-regulated in both groups and is not dependent on the number and duration of hypotensive events. We hypothesized that one possible complication of hypoxia could be recurrent hypoxia which causes modifications of gene transcription as well as post-translational protein modification, including the ones regulating metabolism or cardiovascular system. HIF is a heterodimeric complex, which consists of two subunits: α (HIF α) and β (HIF β)[18]. Subunit α is oxygen-sensitive, during normoxia is associated with von Hippel-Lindau (VHL) protein, which is responsible for the induction of its proteasome degradation[19]. Therefore, in normoxia, the half lifetime of HIF1α is greatly shorted than in hypoxia conditions [20], as low pressure of oxygen is responsible for blocking the binding of VHL and HIF1α and its degradation is then inhibited [21]. Recently, Gabryelska A et al. [22] showed that serum HIF-1α was higher in obstructive sleep apnea patients when compared to control. In particular, control groups exhibited HIF-1α serum levels comparable to both groups included in our studies whereas obstructive sleep apnea patients showed much higher values. However, it should be noted that HIF-1α is generally up-regulated during chronic hypoxic conditions and therefore it is conceivable that hypotension, despite causing a possible oxygen supply–demand mismatch, is not sufficient in terms of duration and oxygen delivery to trigger the HIF-1α pathway. This hypothesis is consistent with our data showing non-significant changes in lactate and acetyl-CoA levels in both groups of patients, thus suggesting that cells are not metabolically rewiring toward an hypoxic phenotype. To the best of our knowledge this is the first report evaluating the possible activation of HIF-1α pathway activation during hypotension. On the other hand, we observed a significant reduction of the reduced form of GSH in the controls when compared to the intervention group, thus suggesting that hypotensive episodes increased oxidative stress and that the application of a machine learning–derived EWS for pending intraoperative hypotension in combination with a hemodynamic diagnostic guidance may be sufficient in preventing such condition. These results are also consistent with previous reports showing that oxidative stress is a variable and common condition occurring during surgical procedures [22]; however, other variables (i.e. ischemia/reperfusion, surgical procedures, anesthesia protocols) rather than hypotension have been advocate in order to explain intraoperative oxidative stress. Interestingly, non-significant changes were observed for lipid peroxidation markers. Two non-mutually exclusive hypotheses may be responsible to explain such result. The first is that the reduced GSH and its enzymatic biosynthetic machinery is sufficient to counteract the production of lipid hydroperoxides. To this regard, we should note that we excluded from the study those patients affected by chronic diseases which are also responsible for imbalanced glutathione synthesis (i.e. liver diseases). The second possibility is that other more instable adducts are produced downstream the lipid peroxidation pathway (i.e. hydroxynonenal, malonyldhyaldheyde, etc) and are therefore not detected by the used analysis in the present study. Furthermore, NGAL and troponin were not significant in the intervention group when compared to controls. Interestingly, we also observed a significant reduction of NSE in the intervention group when compared to control. This result is of particular relevance since no previous clinical studies reported neurological complications following intraoperative hypotensive events. Furthermore, increased circulating NSE levels seems to be specific of intraoperative hypotensive events since previous reports showed that NSE was not significantly changed following hypotension in cardiac arrest patients [23]. However, it should be noted that hypotensive events during intraoperative procedures are clinically different from those occurring during cardiac arrest in terms of number of episodes and duration. On the other hand, increased NSE levels, but not S100B, were observed in patients undergoing controlled hypotension during skull base procedures [24]. These results are also consistent with our observations regarding S100B levels. To this regard, even though S100B did not reach statistical significance it showed a trend similar to NSE. It is possible that an increased sample size would clarify this issue. Finally, previous reports showed that both biomarkers of brain injury can cross blood-brain barrier with different intensity, irrespective of differences in their molecular weight (NSE>S100B) [25]. However, it remains to be determined whether increased NSE levels are dependent on increased troponin levels following cardiac impairment or it is a direct consequence of increased oxidative stress mediators formed outside the central nervous system. Studies using near infrared spectroscopy associated with intraoperative events monitoring are currently running in our unit in order to further elucidate this point. Finally, our study did not include a long-term neurological examination of enrolled patients and therefore it is not possible to determine whether increased NSE levels are also associated to the clinical outcome in the postoperative period.
Taken all together, our study confirms the importance of monitoring and preventing intraoperative hypotension, reinforcing the clinically meaningful impact that its occurrence has on end-organ damage and oxidative stress. Future multicenter studies on a larger cohort of patients are now warranted in order to fully elucidate the clinical effectiveness of hypotension prevention on organ injury.
Limitations
Our study has several limitations that should be considered. First, it is a single center study in a relatively small sample size. Larger and multicenter studies are warranted to support or not our findings. Second, the study was not powered enough to investigate the correlation of intraoperative hypotension with short and long-term postoperative complications. Third, it remains to be studied if the higher values of several biomarkers of organ injury and oxidative stress truly correlates with postoperative organ damage and with patient’s outcome. Forth, we included a heterogeneous sample of patients undergoing major non-cardiac surgery and with different perioperative risk stratification. Considering the volume of surgery at our Institution it would have not been feasible to enroll patients undergoing the same surgical operation.