Age influences on Propofol estimated brain concentration and entropy during maintenance and at return of consciousness during total intravenous anesthesia with target-controlled infusion in unparalyzed patients: An observational prospective trial

Purpose Aging affects pharmacodynamics/pharmacokinetics of anesthetics, but age effects on Entropy-guided total intravenous anesthesia with target-controlled infusions (TIVA-TCI) are not fully characterized. We compared aging effects on effective estimated brain concentration of Propofol (CeP) during TIVA-TCI Entropy-guided anesthesia, without neuromuscular blockade (NMB). Methods We performed an observational, prospective, single-center study enrolling 75 adult women undergoing Entropy-guided Propofol-Remifentanil TIVA-TCI for breast surgery. Primary endpoint was the relationship between age and CeP at maintenance of anesthesia (MA) during Entropy-guided anesthesia. Secondary endpoints were relationships between age and CeP at arousal reaction (AR), return of consciousness (ROC) and explicit recall evenience. We calculated a linear model to evaluate the age’s impact on observational variable and performed pairwise tests to compare old (≥65 years, n = 50) and young (<65 years, n = 25) patients or patients with and without an AR. Results We did not observe age-related differences in CeP during MA, but CeP significantly (p = 0,01) decreased with age at ROC. Entropy values during MA increased with age and were significantly higher in the elderly (RE: median 56[IQR49.3–61] vs 47.5[42–52.5],p = 0.001; SE: 51.6[45–55.5] vs 44[IQR40-50],p = 0.005). 18 patients had an AR, having higher maximum RE (92.5[78–96.3] vs 65[56.5–80.5],p<0.001), SE (79[64.8–84] vs 61[52.5–69],p = 0.03, RE-SE (12.5[9.5–16.5] vs 6 [3–9],p<0.001. Conclusion Older age was associated with lower CeP at ROC, but not during MA in unparalysed patients undergoing breast surgery. Although RE and SE during MA, at comparable CeP, were higher in the elderly, Entropy, and in particular an increasing RE-SE, is a reliable index to detect an AR.


Introduction
Total intravenous anesthesia with target-controlled infusions (TIVA-TCI) allows anesthetists to achieve a stable plasma or estimated brain concentration of Propofol and Remifentanil (CeP and CeR, respectively) and to promptly respond to signs of inappropriate anesthetic/analgesic plans, being considered the "ideal" approach by some clinicians [1]. TIVA-TCI systems use multi-compartment pharmacodynamics/pharmacokinetic models to calculate the necessary infusion rates to reach and maintain CeP and CeR, based on patient age, sex, weight, and height [2][3][4][5]. CeP at loss of consciousness (LOC) and return of consciousness (ROC) during TIVA-TCI are lower in the elderly (age �65 years) than in younger patients (age <65 years) [6,7]. Further, it seems that lower CeP is necessary to maintain a similar anesthetic level in elderly using TIVA-TCI during cardiac surgery [8], suggesting that older patients are more sensitive to Propofol administered for anesthesia maintenance.
Commercial monitors using processed electroencephalographic (pEEG) parameters can be used to target the adequate range of general anesthesia, by titrating the anesthetic to a recommended index range. The Entropy module calculating the State (SE) and Response Entropy (RE) (Entropy Module, GE Healthcare, Chicago, IL, USA) [9] is one of these devices that translates the EEG activity into dimensionless indices inversely related to the anesthetic level by using specially designed algorithms [9]. The use of these indices seems to decrease the risk of Post-Operative Cognitive Dysfunction (POCD) or intraoperative awareness by helping to avoid excessively deep or light levels of anesthesia [10,11], but their usefulness for these purposes is matter of controversial discussion [12][13][14][15]. Previous results describe an influence of age on these indices [16,17], because aging changes intraoperative EEG features [18,19].
Noxious stimulation during general anesthesia can cause arousal events that are accompanied by body movements in the absence of neuromuscular blocking agents. These events may reflect an increased probability of consciousness or insufficient general anesthesia without neuromuscular blockade [20]. To detect arousal reactions, non EEG-based parameters such as the Surgical Pleth Index (SPI; GE Healthcare, Helsinki, Finland) also exist. The SPI measures the nociception/anti-nociception balance during anesthesia and is a function of stimulation intensity and the analgesic component provided by an opiate infusion [21]. An increase in SPI or RE and RE-SE is indicative of an arousal reaction with movement [22]. Here, we present results from our investigation regarding the impact of age on pEEG-guided TIVA-TCI as well as the RE, SE and SPI reaction during intraoperative arousal events in the absence of neuromuscular blockade.

Clinical endpoints and variables
The aim of this study was to assess the effects of age on Entropy-guided TIVA-TCI anesthesia. Primary end-point was the relationship between age and TIVA-TCI's effective estimated brain concentrations of Propofol (CeP) and Remifentanil (CeR) during anesthesia maintenance.
Secondary endpoints were the relationship between age and CeP at arousal, ROC and the occurrence of explicit recall.
For the primary endpoint, the following potential confounder variables were considered: body mass index (BMI), ASA physical status, number of previous general anesthetic expousure, years of schooling, surgery duration, fentanyl use (total quantity).
An anesthetist among the authors, who was not involved in delivering anesthesia to patients, recorded variables on a paper data-collection form and was in charge to record any arousal events, defined as any involuntary movement, inadequate ventilation because of vocal cord closure, or significant hemodynamic response [24]; these patients were defined arousable. Another author, blinded to the arousal events during anesthesia, assessed the Brice modified questionnaire [25] for explicit recall 15 minutes after return of consciousness (in the Post-Anesthesia Care unit) and on the first post-operative day, in both cases in the quietest and most comfortable condition as possible.

Statistical analysis
We based the sample size calculation on the difference in CeP at ROC. Therefore, we used a pilot sample of 10 patients, five younger (age 18-64 years) and five older (age �65 years). We used following parameters: 0.2 μg ml -1 difference in median CeP ROC between younger and older groups; 0.25 μg ml -1 SD; a type I error probability of 0.05 and a type II error probability of 0.2. At least 25 patients were required for each group. We recruited patients consecutively until the necessary number in both groups was reached.
We used the Shapiro-Wilk test to test for normality. We report continuous, normally distributed variables as mean ± SD and continuous non-normally distributed variables as median [IQR] or minimum to maximum. We tested for differences between groups using the twotailed Student t-test or two-tailed Mann-Whitney U test for normally and non-normally distributed variables, respectively. We report categorical variables as number (percentage), and tested for differences between groups using the Chi-square test. We supplement the findings from the t-test/Mann-Whitney U test with the area under the curve as effect size, together with 10k-fold bootstrapped 95% confidence intervals (CI). Therefore, we used the MATLAB-based MES toolbox [26].
We calculated a linear model to assess the relationships between age and selected variables with the MATLAB fitlm function. The strength and direction of association between two continuous variables were determined using the Pearson's correlation.
We calculated the Wilcoxon signed rank test to evaluate possible differences between CeP/ CeR at ROC and arousal.

Demographics
During the study period, 81 of 89 patients undergoing breast surgery provided written informed consent to join the study. Of these, 5 patients were excluded because of not meeting inclusion criteria, and 1 patient was excluded due to incomplete data. Data from 75 patients were therefore analyzed. Their median age, weight, height, BMI and schooling years were 60 years (IQR:50-70), 65 kg (IQR:57-78), 161 cm (IQR:158-168), 25.1 kg m -2 (IQR:20,6-29,6) and 13 years (IQR:8-13) respectively; ASA I/II was 37/38. Median surgery duration was 68.6 min (IQR:40-90). Table 1 contains the patient characteristics for the young and old patients. Older patients had significantly fewer schooling years and a higher number of previous general anesthesia.

Age effects on CeP and CeR
During anesthesia maintenance, we did not find any age-related difference or linear trend in median CeP or CeR concentrations: CeP was 3[IQR:2-3]μg ml -1 in the old vs 2.5[IQR:2-3]μg ml -1 in the young (p = 0,332) and CeR was 2[IQR:1.5-2.2]ng ml -1 in the old vs 1.8[IQR:1.5-2.0] ng ml -1 in the young (p = 0,384). The fentanyl doses did not differ significantly between the age groups. Fig 1A and 1B contain the linear model plots and the young vs old box plots. Table 1 contains the statistical details of the young versus old comparisons and Table 2 contains the details of the linear models.

PLOS ONE
Age, Entropy, and TIVA-TCI  . Fig 2A and 2B and Tables 1 and 2 contain the detailed information. We did not observe significant differences in the entropies with age at ROC (Fig 2C and 2D and Tables 1 and  2). Further, there was no significant age effect on the mean SPI during maintenance (S2A Fig).

PK/PD and processed EEG differences between arousers and nonarousers
No patients exhibited explicit recall 15 minutes after ROC or at the first post-operative day after the interview with the Brice questionnaire. An arousal reaction maintenance occurred in 18 of 75 patients (24%), with no significant difference (p = 0.566) in occurrence rate between the young and old patient group. We did not observe a significant difference in CeP and CeR in patients with and without arousal reaction (Fig 3A and 3B

Discussion
Our results confirm previous findings about age-influence on CeP at ROC: older patients regain consciousness at lower CeP [6,7]. We couldn't find significant differences in the median CeP during Entropy-guided TIVA-TCI maintenance described by previous trials [8], but meanwhile we observed an age-induced change in the pEEG parameters: RE, SE, and RE-SE were significantly higher during maintenance in the elderly. Further, we couldn't find agerelated differences in CeP at arousal reactions. Arousable patients expressed significantly higher maximum RE-SE difference values. Hence, we could reveal an age-influence of the pEEG-guidance on anesthesia navigation. Without targeting the anesthetic level to the recommended index range, CeP of our older Notes. Data are median (IQR), minimum to maximum unless otherwise indicated. Arousal reaction was defined as "light anesthesia" associated with involuntary movement, inadequate ventilation because of vocal cord closure, or a definite hemodynamic response (Sanders RD 2012). patients might have been lower [3]. In our study, targeting the level to a SE of 40-60, it wasn't. This is the case, because age-induced changes in the EEG seem to influence the pEEG indices.

Characteristic of patients according to age
Pharmacodynamics and pharmacokinetics of anesthetics are affected by age [2][3][4][5][6]16], and an increased Propofol-sensitivity in older patients may explain the lower CeP at ROC observed in our and previous studies [6,7]. However, we could not find the expected age-related difference in CeP during maintenance, probably because of the use of SE to target the anesthetic level. EEG amplitude under general anesthesia decreases with age and composition of recorded EEG changes towards a stronger contribution of higher frequencies [18,19]. EEG architecture's changes could explain our higher Entropy values observation in elders during anesthesia maintenance, even in the case of actively targeting the range. Higher indices with age during maintenance were described for the Bispectral Index (BIS) and for the spectral Entropy [18], the underlying method for SE and RE calculation [9]. Here, we can confirm the impact of age for SE and RE. Further, the increase in RE-SE with age during maintenance is indicative of a stronger, age-related, contribution of higher frequencies in the 32 Hz to 47 Hz range processed by RE [9]. Age influence on SE and RE probably is the reason for our older patients having a higher CeP during maintenance, because anesthesiologists navigated anesthesia by the index. Hence, our results highlight a possible conflict between TCI and EEG-based indices for elders. Consequently, more hypnotics than required were possibly given to older patients, increasing the risk of hypnotics overdosage in this population. The case of one 82-year-old patient with an adequate CeP during maintenance (2.5 μg ml -1 ) showing median RE and SE of 95 and 88 during maintenance most probably highlights another issue: undetected EEG burst suppression (BSupp). BSupp indicate severe brain neuronal activity and metabolic rate reduction that may increase postoperative delirium and POCD risk [27,28]. pEEG indices seem to have limited ability to reliably detect BSupp patterns, underestimating their occurrence [29]. If undetected, the suppressed EEG low-amplitude and high frequency characteristics can cause high index values [30]. This may have been our patient's case since she showed no arousal reaction during surgery and had no awareness with explicit recall. In order to generally identify BSupp, even if the indices do not, a visual identification of the raw EEG or its spectral representation was suggested [29]. This case also highlights that Propofol concentrations that are usually appropriate can cause persisting BSupp in certain cases. Hence, EEG-based patient monitoring can help to identify these cases and to individually adjust the anesthetic level. But, using processed EEG indices only to guide TCI anesthesia with Propofol may expose older patients to higher than necessary doses that could lead to a higher risk of adverse outcomes such as postoperative neurocognitive disorders [10].

Characteristic of patients according to arousal reaction
Response surface models revealed prominent synergistic effects between Propofol and Remifentanil in blunting the response to nociceptive stimuli and to avoid arousal reactions [21]. In our study, arousable patients had adequate CeP and CeR values during maintenance [1]. However, they received less fentanyl during surgery, and they had a higher median SPI during maintenance. This suggests that, even within the suggested TCI interval [23], analgesic management may have been inadequate, and SPI threshold of 30 suggested for inhalational anesthesia [21] should be adopted also for TCI anesthesia in order to prevent an arousal reaction's risk.
Entropy monitoring may reduce the likelihood of a reaction to intraoperative nociceptive stimulus [31]. In our study, arousable patients had higher maximum RE-SE RE, SE, and SPI values than non-arousable patients, suggesting monitoring devices' detection of arousal reaction. Because we didn't see differences in median values between arousable and non-arousable patients, the predictive power for a possible arousal may be low.
Painful stimulation increases cardiovascular activity, detected by SPI [21]. However, muscle tension increases, another facet of the response to painful stimulation [32], also occur because of proprioceptive inputs to the reticulo-thalamic activating system [33]. SE, computed over the EEG-dominant frequency range (0.8-32 Hz), mainly reflects the patient's cortical state. By contrast, RE is computed over the EEG-and EMG-dominant frequency ranges (0.8-47 Hz) and thereby at least partly serves as an indicator of upper facial EMG activation, which has been reported to represent nociception or impending awakening [31]. Increment in RE-SE, even if it occurs only transiently because of a subsequent increase in SE, may be useful for identifying an inadequate anesthetic/analgesic plan with arousal reaction's risk following painful stimulation. So in case of increasing EMG the SE and RE will start to drift apart because of an increased influence of high frequency oscillatory activity picked up by RE, but not SE. Another case, more important to our investigations is the age-related flattening of the Power spectral density under general anesthesia [18]. This flattening also leads to an increased influence of higher frequencies that will also increase the difference between the RE and SE. So we are confident that the observed increase in the RE-SE with age under general anesthesia is not caused by increase EMG with age, but by the age-induced increased of higher EEG frequencies. Hence, the indices seem to appropriately respond to an arousal reaction after noxious stimulation independent of age.

Limitations
This study has some limitations. Firstly, it was conducted only on female patients undergoing breast surgery: future trials should confirm our observations also on males and other surgerytypes; in addition, although all of our patients were in the menopausal period, we did not account for hormonal contribution in our analysis. Secondly, SE and RE trends, and the raw EEG tracings, weren't electronically stored, so we couldn't analyze RE and SE values during arousal reaction: we can only speculate that the observed maximum RE and SE in arousers were its consequence. Analysis of raw EEG data could have provided additional insights into the arousal reaction's nature. Raw EEG data could have provided additional informationa also on BSupp, that can alter SE and RE values, however because the vast majority of our patients had indices in the index interval reflecting "adequate anesthesia", i.e., between 40 and 60, we are very confident that BSupp is not an issue in our data.

Conclusion
Estimated concentrations of Propofol during anesthesia maintenance seem not to be agerelated in patients during TIVA-TCI for breast surgery when anesthesia is navigated by the processed EEG index, highlighting the impact on age on general anesthesia. The SE and RE seem to increase with age during adequate levels of anesthesia.
The occurrence of an arousal reaction seems independent of age, but lower CeR during anesthesia maintenance might lead to an increased risk of an arousal reaction. Finally, additional information as for instance provided by the SPI could help to improve intraoperative patient monitoring and to optimize the hypnotic and analgesic component of anesthesia.