Association of Intraoperative Cerebral and Somatic Tissue Oxygen Saturation with Postoperative Cognitive Dysfunction After Spine Open Surgery in Patients with Hypertension: An Observational Study


 Background: Intraoperative cerebral desaturations have been associated with worse neurological outcomes after supine surgery. However, it is not clear whether intraoperative somatic tissue oxygenation is more associated with postoperative cognitive dysfunction (POCD) than cerebral oxygenation in patients with hypertension after prone spine surgery.Methods: Patients with hypertension scheduled for spine open surgery were included from 2020 to 2021 in a single-center, prospective, observational study. Baseline both cerebral and somatic tissue oxygen saturation were measured in operating room before surgery. Cerebral and somatic tissue oxygen saturations were monitored continuously throughout surgery. The presence of POCD was assessed using the Mini-Mental Status Examination (MMSE). Association with POCD was evaluated with unadjusted analyses and multivariable logistic regression.Results: One hundred and one of 112 patients were included, and 28 (27.8%) developed POCD. None of the investigated SctO2 variables was predictive of POCD. On the contrary, the patients with POCD had a higher decrease in intraoperative absolute SstO2 decrease and relative SstO2 decrease compared with the patients without POCD (4.9%±3.8% vs. 3.6%±2.6%, P=0.037; 7.4%±5.6% vs. 5.3%±3.8%, P=0.036; respectively). Finally, three SstO2 parameters respectively were associated with POCD, including a higher absolute SstO2 decrease (OR, 1.223; 95%CI, 1.031-1.451; P=0.021), a higher absolute SstO2 decrease (OR, 1.138; 95%CI, 1.011-1.281; P=0.032) and falling below 90% of baseline SstO2 (OR, 11.388; 95%CI, 2.367-54.785; P=0.002), independent of ASA III, preoperative platelet and postoperative sepsis. Conclusions: Twenty-eight (27.8%) of 101patients developed POCD. Somatic tissue oxygenation has a stronger association with POCD than cerebral tissue oxygenation after spine open surgery in patients with hypertension.Clinical trial registration: ChiCTR1900028392. Registered 20 December 2019.

and neurological complications. [11][12][13][14][15][16] However, of these studies, patients were in a supine position in most studies, patients were in a prone position in only one study and only SctO2 was monitored, [11] so the relationship between SctO2/SstO2 and POCD in prone surgery remains unclear. A study, in prone spine surgery recently, had shown a good association between postoperative complications and SstO2, but almost no association with SctO2. [17] As such, we hypothesized that in a prone position, SctO2 is affected by position but SstO2 is not. The aim of this prospective observational study was to evaluate the association between intraoperative SctO2/SstO2 and POCD after open prone spine surgery in patients with hypertension.

Methods
This prospective observational study was conducted in a tertiary hospital. The study protocol was approved by the Institutional Committee for Medical Ethics of Zigong Fourth People's Hospital A liated to Southwest Medical University, Zigong, Sichuan, China. Written informed consent was obtained from the patient before surgery.

Study population
Patients undergoing elective posterior lumbar or/and thoracic spine open surgery in a prone position were enrolled between January 2020 and April 2021. Our goal was to recruit 100 patients in this study, as the previous observational studies assessing the association between SctO2/SstO2 and postoperative outcomes had approximately 100 patients in whom monitoring only was implemented. [13,17] The exclusion criteria included: 1) patient refusal; 2) age < 18 years old; 3) American Society of Anesthesiologists (ASA) physical status score > III; 4) combined with other operations; 5) skin condition affecting the monitoring of the oximetry probe; 6) preoperative Mini-Mental State Examination (MMSE) score < 18; 7) literacy problems, language di culties, hearing or visual impairment.

Anesthesia
No premedication was given. After patients entered operating room, radial arterial cannulation, arterial blood gas analysis and the monitoring of bispectral index (BIS) monitoring were performed in advance under the breathing room air. General anesthesia was induced using midazolam (0.05 mg/kg), sufentanil (0.3-0.4 ug/kg), cisatracurium (0.2 mg/kg) and etomidate (0.2 mg/kg). Then propofol (0-1.5 mg/kg) was used according to the BIS before tracheal intubation was performed. At the administration of the same attending anesthesiologist, general anesthesia was maintained with BIS from 40-60% using sevo urane in an oxygen-air mixture (oxygen concentration 40%). Respiratory rate was set to meet the end-tidal CO2 35-40 mmHg. Sufentanil and cisatracurium were added intermittently by the attending anesthesiologist according to the needs of the surgery. Ephedrine was usually used to maintain blood pressure, and when ephedrine did not work well, deoxyepinephrine was used. Tranexamic acid was routinely applied 30 minutes before surgical skin cutting. Cell salvaging was available in all patients. After surgery, analgesia through incision in ltration was routinely performed, and oral analgesia drugs and/or self-controlled electronic analgesia pump were administered as needed.
Tissue oxygenation monitoring Tissue oxygenation was monitored using a tissue oximeter (EGOS-600A, Eqin tissue oxygen saturation monitor, China) based on NIRS. The oximeter had four cables with each cable connected to an NIRS sensor. Two NIRS sensors were placed on the upper forehead, one nger above the eyebrow, to monitor the left and right frontal lobe SctO2, respectively. The other two NIRS sensors were placed on the waist in the midaxillary line, two ngers below the costal margin, to monitor the SstO2 of the left and right waist, respectively. All four NIRS sensors were placed on the patient when entering the operating room. The patient's baseline SctO2 and SstO2 were acquired before anesthetic induction when the patient breathed air. The monitoring and data recording were started when the anesthetic induction was begun, and were stopped when the operation was done. The oximeter was put at the back of anesthesia machine and blinded to anesthesiologists.
The tissue oximeter generated a new data point every two seconds that was continuously recorded by a computer. The data points of SctO2/ SstO2 were used for tissue oxygenation indices derivation. Eight indices of tissue oxygenation were derived that include the baseline, minimum, mean, median, absolute SctO2 decrease, relative SctO2 decrease, area under the curve (AUC) and number of desaturation < 80% or < 90% of baseline. The AUC was the sum of the differences between the special threshold and the data point of every minute that was less than the threshold. For each index, SctO2 and SstO2 were treated independently.

Clinical covariates
Patient characteristics including age, sex, weight, height, current smoke, alcohol consumption and ASA were collected. Past medical history including diagnosis of hypertension, coronary artery disease, previous stroke, diabetes mellitus, COPD, chronic hepatitis B and chronic kidney disease were recorded. Surgical information that were recorded were fusion or not, number of segments, estimated blood loss and surgical time. The usage of ephedrine and phenylephrine, uid infusion volume, lactate when suturing, urine output and blood transfusion were recorded. After surgery, the time of tracheal extubation, length of PACU stay, numerical pain score (NPS, in postoperative1d and 3d), and postoperative 24h drainage volume were also recorded.

Postoperative outcomes
The primary outcome measure was POCD. The testing protocol was the quick assessment scales of the mini-mental state examination (MMSE) that used widely as a screening tool for cognitive dysfunction. [18] MMSE involves testing orientation to time, place, registration, attention, calculations, recall, repetition, and complex commands. [18] It is known to be less sensitive when evaluated repeatedly. [19] Therefore, in this study, cognitive function tests were performed only on day 1 before surgery and day 4 after surgery, Page 6/24 respectively. When administering the MMSE a second time, variation in the order of test elements was used to increase the sensitivity of the test.
The testing of cognitive function was performed by a special clinician with training in cognitive assessment. The testing clinician was blinded to surgery and anesthesia for each patient. In this study, a decrease by 2 points or more from the preoperative score was considered as a signi cant decline in cognitive function, and de ned as a POCD. [11] Primary postoperative complications were identi ed based on chart review of the electronic medical records at the Zigong fourth People's Hospital A liated to Southwest Medical University. Information of length of hospital stay, length of postoperative hospital stay and total medical costs were also recorded.

Statistical analysis
Statistical analysis was performed using IBM SPSS version 23 (IBM, Chicago, IL). Normal distribution of continuous data was evaluated with the Kolmogorove-Smirnov test. Normally or approximately normally distributed data are presented as mean (± standard deviation), and skewness distributed data are presented as median with rst and third quartiles. Categorical variables are expressed as n (%). Continuous data were compared by independent t-test or Mann-Whitney U-test, and categorical data were compared between groups with and without POCD using chi-square test or Fisher's exact test. Repeated measurements between two groups were compared by paired t-test. Correlation between two continuous variables was analyzed using Spearman or Pearson correlation.
Baseline patient characteristics and perioperative characteristics that were signi cant (at a threshold P < 0.2) in the univariable logistic regression analysis were entered into a forward multivariable logistic regression model. The multicollinearity test was performed on all included variables in advance, and variables with a variance in ation factor > 5 were excluded. The goodness of t of the nal multivariable logistic regression model was evaluated using the Hosmere-Lemeshow test. The predictive power of the nal model was assessed by a receiver operating characteristic (ROC) curve, and the AUC was calculated to evaluate the performance in predicting POCD. Finally, to clarify the relationship between SctO2/SstO2 parameters and POCD, and avoid the in uence of collinearity among parameters, the SctO2/SstO2 parameters (P < 0.2 in the univariable logistic regression analysis) were adjusted one by one using independent risk factors obtained in the nal multivariate Logistic regression model. Statistical signi cance was set at P < 0.05.

Study population
From December 2019 until April 2021, 112 patients scheduled for elective posterior lumbar or/and thoracic spine open surgery were included in this prospective observational study (Fig. 1). Eleven patients were excluded for data analysis. In three patients, there were no postoperative MMSE assessments, and in eight patients, there were more than 5% missing SctO2/SstO2 values. In the end, 101 patients were included in the data analysis. The postoperative MMSE positively correlated with preoperative MMSE (r 2 = 0.695, P < 0.001, Fig. 2), while postoperative MMSE decreased, compared with preoperative MMSE (24.9 ± 2.9 vs. 24.1 ± 3.3, P < 0.001, Fig. 2 and Table 1). No correlation was present between intraoperative SctO2 and intraoperative SstO2 (P = 0.702), nor was there any correlation between their respective absolute declines (P = 0.247) (Fig. 2). Of 101 patients, 28 (27.7%) developed POCD. The number of 1, 2, 3 and 4 operative segments were performed in 62 (61.4%), 34 (33.7%), 4 (4.0%) and 1 (1.0%) patients, respectively (Table 2). Decompression and fusion were performed in 75 (74.3%) and 74 (73.3%) patients, respectively. The perioperative characteristic data in study population are shown in Table 1 and Table 2.  Perioperative characteristic data between two groups The preoperative characteristic data of both patient groups are shown in Table 1. The patients with POCD were older, had a higher rate of chronic kidney disease and ASA III, and had a higher number of platelet compared with the patients without POCD. The length of education and the preoperative MMSE scores were similar in both groups (P = 0.818, P = 0.647; respectively), while the patients with POCD scored lower than the ones without POCD on the postoperative MMSE (P < 0.001; Table 1). Perioperative management and postoperative complications data of both groups are shown in Table 2. POCD patients had a higher proportion of intraoperative phenylephrine use, more postoperative 24h drainage, and a higher proportion of postoperative cerebrospinal leak. However, there was no difference in intraoperative blood loss and proportion of intraoperative ephedrine use between two groups.

Comparison of SctO2 and SstO2 between two groups
The univariable comparison of intraoperative NIRS (both SctO2 and SstO2) parameters between the patients with and without POCD is shown in Table 3. The baseline SctO2, lowest intraoperative SctO2, mean SctO2 and median SctO2 were similar between the patients with and without POCD (P = 0.593, 0.947, 0.728 and 0.722, respectively). Likewise, intraoperative absolute SctO2 decrease, relative SctO2 decrease, the AUC values of postoperative SctO2 values below 60%, 55% and 50%, the times below 80% of baseline SctO2 and the times below 90% of baseline SctO2 did not differ between both groups (P = 0.649, 0.730, 0.296, 0.239, 0.201, 0.682 and 0.561, respectively). Generally, none of the investigated SctO2 variables was predictive of POCD. In contrast to SctO2 measured from the forehead, the patients with POCD had a higher decrease in intraoperative absolute SstO2 decrease and relative SstO2 decrease compared with the patients without POCD (4.9%±3.8% vs. 3.6%±2.6%, P = 0.037; 7.4%±5.6% vs. 5.3% ±3.8%, P = 0.036; respectively; Fig. 3). The proportion below 90% of baseline SstO2 in the patients with POCD were signi cantly higher than patients without (32.1% vs. 5.5%, P < 0.001), but 19 (67.9%) of the 28 patients who developed POCD did not experience intraoperative SstO2 desaturations below 90% of baseline (Fig. 3). Unfortunately, the times below 80% of baseline SstO2 did not differ between both groups, as the frequency of below 80% baseline SctO2 was 2, in 1 case in each group (Table 1). In addition, other investigated SstO2 variables such as baseline SstO2, lowest intraoperative SstO2, mean SstO2 and median SstO2 did not differ signi cantly between both groups (Table 1).  (Table 4). Likewise, A negative correlation was present between the parameters of SctO2 (e.g., the absolute decrease) and the length of postoperative hospital stay (r 2 = 0.082, P = 0.004, Fig. 3), while the parameters of SctO2 were not.

Discussion
In this prospective observational study, 28 (27.7%) of 101 patients developed POCD, which was slightly higher than the result (23%) in the study of Jeongmin Kim on spinal surgery. [11] This may be because Jeongmin Kim's study was on elderly patients, while our study was on patients with hypertension in whom the average age was 64.8. Hypertension can raise the brain's threshold for self-regulation of blood pressure, cause blood-brain barrier breakdown, and frequently encounter labile hemodynamics during surgery, which is more likely to lead to cognitive decline. [1,20,21] The results of this study showed that POCD was strongly associated with the absolute decrease in intraoperative SstO2, but had little correlation with rScO2, in patients with hypertension undergoing open posterior spine surgery.
As the brain is known to consume a lot of oxygen that is supplied to the body, the cerebral function is very sensitive and vulnerable to hypoxemia. In the setting of open posterior spine surgery, the importance of avoiding cerebral oxygen de ciency has been increasingly recognized, and cerebral oximetry, as a noninvasive tool to assess the adequacy of cerebral oxygen delivery, is increasingly used in the clinic. However, there were few studies to monitor cerebral oxygen saturation and somatic oxygen saturation simultaneously. The results of this study showed that there was no correlation between intraoperative cerebral oxygen saturation and somatic oxygen saturation, nor was there any correlation between their respective declines. These results may suggest that there may be signi cant differences in the correlation between the oxygen saturation of different sites and the clinical outcome and prognosis. The strategy of multi-site monitoring may be bene cial at present, and identifying the best monitoring site is important to further improve patient outcomes. [22] In the clinic, the purpose of monitoring is to simply and effectively prevent and reduce the occurrence of postoperative complications. So far, there have been few comparisons of the correlation between different sites monitoring tissue oxygen and POCD, and it is not clear which site is better. In this study, there was no correlation between any indicator of intraoperative SctO2 in the forehead and POCD. This result was similar with the result of previous study that they found no correlation between intraoperative SctO2 and postoperative complications after spine surgery. [17] But, this result was completely inconsistent with other studies showing a good correlation between SctO2 and POCD in non-cardiac surgery. Although many preventive measures had been taken, the speci city of prone position might still affect the measurement of SctO2 in this study for two reasons: First, all patients were in prone position, which might result in cephalic and facial venous pooling. Second, patients in a prone position often have edema formation in the lower part of the body, including the forehead. Therefore, the prevention of neurological impairment by cerebral oxygen saturation may be seriously interfered with and weakened undergoing surgery in a prone position. Unfortunately, recent ndings in prone surgery are con icting, and postural factors may not be able to explain all causes. [11,17] Further studies are needed to clarify the relationship between intraoperative SctO2 and postoperative complications in prone surgery.
In contrast, the present study found that intraoperative somatic tissue desaturations during spine surgery were signi cantly associated with POCD. Thus, we proved the previous hypothesis that SctO2 in the forehead is affected by position in prone position, while SstO2 is not. This also implied a potentially modi able causal link, thereby concluding that a persisting somatic desaturation should be avoided. This result is similar to that of Meng L et al. [17] They found that muscular tissue desaturations were signi cantly associated with postoperative complications. However, in this study, the monitoring site was the midaxillary line of the waist at the same height as the operation, while in their study, the lower leg was monitored. The main reasons why we did not choose the lower leg and the more common arm (especially the thenar muscle) for SstO2 measurement are as follows: First, bilateral thenar muscles were usually used for evoked potential monitoring in our patients. Second, the monitoring locations we selected were almost at the same level as the vital organs and the brain, where hemodynamics might be more similarly affected by relative position. Third, the sites we monitored may involve muscle tissue or even splanchnic organs such as small intestine, colon or liver that are more sensitive to ischemia and hypoxia. Nearly all splanchnic organs possess weak autoregulation than the brain and heart. [23] Intuitively, during periods of hemodynamic instability, sites with the lowest priority may have a better correlation with postoperative complications. Therefore, our results further con rmed that different monitoring sites of local tissue oxygen saturation have different effects on the same outcome index, and it is important to study the best monitoring sites of speci c prognostic index and its speci c clinical practical value in the future. Unfortunately, even among the intraoperative SstO2 variables we studied, none completely distinguished POCD patients from non-POCD patients. Correspondingly, neither SctO2 nor SstO2 measured preoperatively was predictive for POCD, in contrast to previous reports. In addition, our study found that SstO2 desaturation did not occur in some POCD patients.
This current study indicated that the patients with POCD were more likely to have lower SstO2 values during surgery, and more likely to experience somatic desaturations below 90% of preoperative baseline.
What's more, a higher absolute/relative decrease in postoperative SstO2 was associated with POCD, independent of other risk factors, such as ASA III, preoperative platelet and postoperative sepsis.
This study had at least six limitations as follows: First, this was a single-center study with a small sample size, and the sample size was not pre-determined before the study. Hence, there is a chance that the association of intraoperative somatic desaturation and cerebral desaturation with POCD after spine surgery may be underestimated. Larger multicenter studies are needed to con rm these associations.
Second, this was an observational study, and therefore the somatic desaturation might be a manifestation or consequence of POCD rather than a causal mediator thereof. Third, different NIRS monitors are commercially available, which allow monitoring tissue oxygenation either cerebrally or peripherally. The differences in the methodological approaches may lead to the bias and variability between different devices. [24] Therefore, study results obtained with different NIRS devices might be hardly applicable and comparable with each other, which might partially explain the con icting of the results from published studies. Fourth, the patients with hypertension included in this study were usually elderly, and the elderly generally had low education due to historical reasons (Table 1). Since education level may have a great in uence on MMSE, the preoperative MMSE score in this study was relatively low.
To ensure the representativeness of the study, we excluded the patients with preoperative MMSE < 18 rather than preoperative MMSE < 24 (Fig. 1). Fifth, as postoperative MMSE assessment was conducted only on the fourth day after surgery, it was not possible to determine the exact point in time when the patients developed POCD. This may lead to an underestimation of the incidence of cognitive dysfunction after spine surgery in patients with hypertension. Sixth, other studies were also using NIRS devices, so not all eligible patients were included. This could have increased the selection bias of the study.

Conclusion And Outlook
In summary, POCD can occur in 28% of patients with hypertension undergoing open posterior spine surgery. Intraoperative somatic tissue oxygen saturation has an association with POCD after surgery, while cerebral tissue oxygenation has little association with POCD in this patient population. At present, a combination of both cerebral and somatic tissue oxygenation monitoring might therefore offer a more complete assessment of the mismatches between oxygen supply and consumption. It is necessary to further clarify the best monitoring locations, the sensitive monitoring indicators, and their relationship to clinical outcomes, especially in high-risk patients undergoing high-risk non-cardiac surgery, for which few studies have been conducted.