Prediction of Blood Pressure Change During Surgical Incision Under Opioid Analgesia Using a New Opioid Sensitivity Index: A Cross-Sectional Study


 Opioid inhibition of nociceptive stimuli varies in individuals and is difficult to titrate. We have reported the vascular stiffness value (K) as a standard monitor to quantify sympathetic response with high accuracy. On the contrary, among individuals, a considerable variation in the rate of change in K for constant pain has been observed. In this study, we proposed a new index, minimum evoked current of K (MECK) and evaluated its accuracy in predicting sympathetic response to nociceptive stimuli under constant opioid administration. Thirty patients undergoing open surgery under general anesthesia were included. After anesthetic induction, remifentanil was administered at a constant concentration of 2 ng/ml at the effect site followed by tetanus stimulation. MECK was defined as the minimal current needed to produce a change in K. MECK significantly (P < 0.001) correlated with the rate of change of systolic blood pressure during skin incision (ROCBP). Bland-Altman plot analysis using the predicted ROCBP calculated from MECK and the measured ROCBP showed that the prediction equation for ROCBP was highly accurate. This study showed the potential of MECK to predict blood pressure change during surgical incision under opioid analgesia.


Introduction
The International Association for the Study of Pain de nes pain as an aversive sensory and emotional experience. Patients under general anesthesia are unconscious and do not have an aversive experience.
However, anesthesiologists commonly use analgesics in addition to sedatives because nociceptive stimuli can cause noxious autonomic re exes. Opioids are the most commonly used analgesics during general anesthesia due to their lack of a ceiling effect. Opioids exert their analgesic effects mainly by inhibiting sensory nerve transmission in the spinal cord and by inhibiting the excitation of pain conduction pathways in the brain. When opioids are administered during general anesthesia, noxious autonomic re exes are suppressed and multiple parameters such as heart rate, blood pressure, electrocardiogram, and respiratory rate are affected. During surgery, anesthesiologists rely on these parameters to estimate pain levels and adjust the opioid dosage. However, this nociceptive stimuliinduced sympathetic response varies among individuals, and therefore, so do individual anesthetic requirements. Adverse events, either as an unexpected increase in blood pressure due to underdosing or delayed arousal due to overdosing, are common. Thus, if the individual's opioid requirement can be accurately quanti ed in advance, more stable general anesthesia can be performed.
Nociceptive stimuli input to the central nervous system is output to effector organs such as the heart and blood vessels via the sympathetic nervous system. Opioids inhibit the input of nociceptive stimuli to the central nervous system. Therefore, the administration of opioids blunts the sympathetic response to nociceptive stimuli. In other words, by accurately quantifying the sympathetic response to a given nociceptive stimulus under opioid administration, we can determine the relationship between the opioid dose and the response to the nociceptive stimulus and quantify opioid sensitivity. Photoplethysmography (PPG), an increasingly popular tool, has recently begun to be used to measure sympathetic cutaneous blood ow responses for quantifying sympathetic nerve activity in peripheral vasculature [1][2][3][4]. However, PPG measures blood ow at the measurement site and does not, in principle, directly indicate the degree of sympathetic response. Therefore, we proposed a method to extract only the degree of vasoconstriction from sympathetic cutaneous blood ow responses to PPG and quantify it as vascular stiffness value (K) [5]. We reported that K could be used to quantify pain and re ect the activity of the sympathetic nervous system [6,7]. Moreover, we reported that K value re ects changes in pain caused by the administration of opioids during general anesthesia and that the response of K to pain was attenuated with increasing doses of opioids [8][9][10].
On the contrary, among individuals, a considerable variation in the rate of change in K for constant pain has been observed [9,10]. We speculated that this may have been due to the fact that the pathway of nociceptive stimuli, which travel from the peripheral nerves through the central nervous system to effector organs such as peripheral blood vessels, is strongly in uenced by autonomic changes due to aging and coexisting diseases. While opioids inhibit the afferent pathway of pain perception, the rate of change of K represents the intensity of the sympathetic response or the efferent pathway. In other words, the rate of change in K includes information from both efferent and afferent pathways, so the measurement results may vary depending on the sensitivity of the effector. Therefore, we hypothesized that the "intensity of nociceptive stimuli" at which sympathetic responses appear is a better indicator of opioid sensitivity than the "intensity of sympathetic responses" to nociceptive stimuli (that is, the rate of change in K) because it exclusively extracts only information regarding afferent pathways.
In the present study, we measured two indices of opioid sensitivity: the minimum stimulus intensity that elicits a change in K and the rate of change in K for speci c stimuli. We then assessed which index predicted the rate of change in blood pressure (ROC BP ) more accurately during a skin incision under a constant effect-site concentration of opioids.

Patients
Prior to the study, we received approval from the Ethics Committee of Hiroshima University ('Hi'-226, 'E'-1523-1) and registered the clinical trial (registry: university hospital medical information network, registration number: UMIN000041816, principal investigator's name: Satoshi Kamiya, Date of registration: July 9th, 2019). This study was conducted in accordance with the Declaration of Helsinki and STROBE statement. The study population consisted of patients over 20 years of age who underwent open surgery under general anesthesia from July 2019 to October 2019. A total of 30 patients gave their written informed consent before the study. All procedures were conducted at Hiroshima University Hospital. We excluded patients with irregular R-R via electrocardiogram (ECG), inability to perform invasive arterial pressure measurements in the radial artery, signi cant hemodynamic or neurological impairment in the upper extremity, and severe stenosis or occlusive lesions in the coronary arteries or cerebral vessels.

Measurement and protocol
Before induction of anesthesia, a photoplethysmography probe (TL-271T, Nihon Kohden, Tokyo, Japan) on the middle nger of the left hand, ECG on the chest, electroencephalogram (EEG; Entropy, GE Healthcare UK Ltd., Buckinghamshire, UK) on the anterior forehead, neuromuscular blockade monitoring device (NMT-Neuromuscular Transmission, GE Healthcare UK Ltd., Buckinghamshire, UK) on the ulnar side of the forearm of the right hand, and a non-invasive blood pressure cuff on the right upper arm were placed. In all patients, a preoperative dosing plan was developed to achieve a predicted effect-site concentration of 2 ng/ml remifentanil. Minto's pharmacokinetic model [11] was used to calculate predicted effect-site concentrations. Administration of remifentanil was initiated as per the dosing plan, and propofol 3 μg/mL was used to induce anesthesia using a target-controlled infusion (TCI) pump with built-in 'Diprifusor' (TE-371, Terumo, Tokyo, Japan). After the patient was unconscious, 50 mg of rocuronium was administered and a 22 G catheter was secured in the left radial artery for measuring arterial blood pressure (ABP). Data from ECG, ABP, and PPG were output to a personal computer from a bedside patient monitor (BSS-9800, Nihon Kohden, Tokyo, Japan) and were used to calculate K values in real-time.
After the predicted effect-site concentration of remifentanil reached a steady state at 2 ng/ml, tetanus stimuli at 50 Hz for 5 s were delivered through a two-pole body surface electrode on the ulnar side of the right hand using the INNERVATOR 252 (Fisher & Paykel Healthcare, Auckland, New Zealand). The current value was initially 10 mA and increased in increments of 10 mA until 80 mA, the maximum output of the INNERVATOR 252, was reached. Thus, a total of eight tetanus stimulation sessions were performed. Once the K value returned to its pre-stimulus value, the next tetanus stimulation was performed. Subsequently, tracheal intubation was performed using a Macintosh laryngoscope or McGRATH TM MAC video laryngoscope (Medtronic, Dublin, Ireland). All endotracheal intubations were successfully executed in a single attempt. Thereafter, the same amount of propofol TCI target blood levels and rate of remifentanil administration were maintained until the skin incision was performed.

Calculation of K
In this study, the motion of the vessel wall was predicted using a model in which springs and dampers were aligned in parallel to the diameter direction of the vessel. The arterial wall, moved by a force (blood pressure) applied in the direction of the diameter of the vessel, was dampened by springs and dampers that represent the stiffness and viscosity of the vessel wall itself. The behavior of the vessel wall after damping was observed as a change in the PPG waveform ( Figure 1). We used the spring constant in this model equal to K. Details of the calculation of K value have been reported previously [8]. In brief, we rst detected the heartbeats with the R wave of the ECG and cut out the ABP and PPG waveforms for each heartbeat. All wave height data associated with one heartbeat were provided in the following equations: The time at which the change was rst observed was considered t 0 , ABP wave height at time t was Pb(t), and PPG wave height at time t was represented as Pl(t). Pl(t) is equal to the rst derivative of the PPG wave height. The coe cients for vascular stiffness and vascular viscosity values, K and B, respectively, were determined as one value per heartbeat by performing a least-squares t of the equation to all data acquired in each heartbeat. K determined in this way was actually displayed automatically at the bedside in real-time, every heartbeat. If the coe cient of determination was less than 0.95, or if K and/or B were negative, the data were removed from the analysis. Only K was analyzed in the present study because previous studies have estimated that K was a promising candidate as a measure of nociceptive stimulus intensity [8]. Figure 2 shows an example of how the MEC was measured in this study. Each parameter measurement in tetanus stimulation is described in the following text. First, pre-stimulus values were set as the median value determined throughout the 10 s that preceded tetanus stimulation. Post-stimulus values were the maximum values of K, heart rate, systolic blood pressure, and minimum values of PPG amplitude for 20 s after tetanus stimulation. The post-stimulus values were divided by the pre-stimulus values and the percentage change in parameters in response to tetanus stimulation at each stimulus intensity was calculated. MEC was de ned as the minimum stimulus intensity value that produced a change of more than 5% in the parameter during tetanus stimulation. However, once the change was greater than 5%, and if the rate of change was less than 5% at a subsequent stage, the MEC calculated at a lower current value was rejected as noise. In other words, the MEC would eventually be one step stronger than the maximum stimulus intensity at each stimulus intensity, where the rate of change of the parameter was below 5%. If no more than 5% change appeared at 80 mA tetanus stimulation, MEC was classi ed as outside the measurement range. The MEC for K, HR, BP, and PPG were designated as MEC K , MEC HR , MEC BP , and MEC PPG , respectively. The rate of change in K before and after 80mA tetanus stimulation (K R80 ) was calculated as a measure of the rate of change for constant intensity nociceptive stimuli. The median systolic blood pressure of 10 s before the scalpel skin incision was de ned as pre-BP, and the maximum systolic blood pressure between the skin incision and the start of electrocautery use was de ned as post-BP. The ROC BP before and after skin incision was calculated by dividing post-BP by pre-BP. A rst-order regression was performed to con rm the relationship between MEC and ROC BP . Speci cally, a rst-order regression equation was created between cases where MEC and ROC BP could be measured. A Smirnov-Grubbs test was performed using this regression formula, and a measurement in which P was greater than or equal to 0.05 was removed as an outlier. The regression equation was remade for each of the removed outliers. The regression equation at the time outliers were removed was adopted as the nal correlation line. Pearson's correlation coe cients were calculated using measurements at the time at which outliers were eliminated. The signi cance level was set at 0.05. Further, we attempted to predict the ROC BP using MEC K and K R80 . Speci cally, the prediction equation was based on the rst-order approximate equation previously developed, and the predicted ROC BP was calculated from MEC K . For cases in which MEC K was outside the measured range, we tted the mean value of ROC BP in out-of-range cases to the approximate equation and substituted the estimated value. A similar rst-order approximation and outlier treatment was performed for K R80 and ROC BP . From the nal approximate equation, the predicted ROC BP was calculated. A Bland-Altman analysis was performed for both measured and predicted ROC BP , and xed bias and limit of agreement range were calculated for each.

Statistical analysis
No prior studies are available on MEC K , and therefore, the prior number of cases cannot be determined.
Genetic polymorphisms related to opioid sensitivity were measured in the subjects at the same time the data used in this study were collected. In the genetic polymorphism study, the power test calculated the required number of cases as 30, and the present study was conducted in 30 cases with the analysis of genetic polymorphisms. Moreover, the effect size was calculated to ensure a su cient sample size. In Pearson's correlation analysis, the r-value, which is the correlation coe cient, was used as a measure of the effect size. All the r-values of the main results of this study indicate moderate-to-high effect sizes.

Results
A total of 30 patients (male, 15) were included in the study and the mean age of participants was 62 years. Thirteen patients underwent an upper abdominal incision, and 17 underwent a lower abdominal incision. None of the patients met the exclusion criteria. A summary of patient characteristics is shown in Table 1. ROC BP was 18.3 ± 11.8% (mean ± SD). Figure 3 shows the values of K, HR, sBP, and PPG amplitude after tetanus stimulation of each intensity. K and sBP values tended to increase with the intensity of the stimulus current. PPG amplitude tended to decrease with increasing stimulus intensity. HR did not show consistent changes. Even for K, which produced the least individual differences, actual measured values had a large degree of individual variability. Therefore, it was not possible to detect the presence of a change in K by setting a threshold for the actual measured value of K. Next, MEC was determined based on the rate of change between each measured value and its pre-stimulus value. Figure 4 shows a histogram of the four MEC. MEC could be measured in 27 cases for K, 8 for HR, 21 for BP, and 25 for PPG. The frequency peaks were generally in the vicinity of 40-50 mA, except for HR, where MEC was very often outside the measurement range.
Then, in addition to the MEC of each parameter determined above, K R80 , the rate of change in K at the maximum current value (80 mA), was calculated. A scatter plot of each MEC and K R80 with ROC BP , respectively, is depicted in Figure 5 Among the out-of-measurement range groups, there was no statistically signi cant difference in the ROC BP of MEC K and MEC PPG measurement range groups. However, out-of-range cases of MEC PPG had a large degree of variability and included cases with large ROC BP . Therefore, we determined that MEC K was the index with the highest degree of correlation with ROC BP among MEC K , MEC HR , MEC BP , and MEC PPG .
We then used MEC K and K R80 to predict ROC BP . First, for MEC K , a rst-order regression equation was applied after outlier treatment in the scatter plot in Figure 5a, as follows: Predicted ROC BP (%) -0.28 × MEC K (mA) + 31.26 (r = 0.723, P < 0.001), (4) Where the mean value of ROC BP in cases outside the MEC K measurement range was 0.7%. This rate of change was substituted into the temporal regression equation, which yielded a value of 107 (mA). This value was used as a provisional MEC K value for cases outside the MEC K measurement range. Next, for K R80 , the rst-order regression equation after outlier treatment in the scatter plot in Figure 5e was applied, as follows: Predicted ROC BP ( ) 0.05×K R80 (%) + 13.25 (r = 0.441, P = 0.017). (5) The above prediction equation was used to calculate the predicted ROC BP . The results of the Bland-Altman plot of measured and predicted values are shown in Figure 6. The difference between measured and predicted ROC BP is shown as the vertical axis and the mean as the horizontal axis. For MEC K , xed bias was small (-0.17%), and the Pearson correlation coe cient in the scatter plot was R = 0.322 (P = 0.088), and thus, no signi cant proportional bias was found. In contrast, for K R80 , the xed bias was small (-0.01%); however, the Pearson correlation coe cient of the scatter plot contained signi cant proportional bias (R = 0.714, P < 0.001).

Discussion
Although opioids are essential drugs for general anesthesia, inappropriate administration may lead to unpleasant effects on patients. Estimation of appropriate dosage is di cult because of large individual differences in opioid sensitivity. In this study, a prediction formula using MEC K as an indicator was able to predict ROC BP with high accuracy.
MEC K is the threshold for the appearance of a sympathetic response when blood concentrations of opioids are 2 ng/ml and can be regarded as an intrinsic value that indicates individual opioid sensitivity.
In other words, for patients with a low MEC K , increasing the dose of opioids prior to skin incision can reduce circulatory uctuations. The relationship between the MEC K and ROC BP at different opioid concentrations remains to be determined; however, the present results indicate that the MEC K may be a good indicator of opioid sensitivity.

MEC K and K R80
MEC K correlated better with ROC BP than K R80 ( Figure 5) and precision was also small ( Figure 6). This implies that MEC K was more accurate when predicting ROC BP than K R80 . We previously reported that the rate of change in K with constant pain was inversely related to opioid concentration. However, there were large individual differences in the relationship between opioid concentration and the rate of change in K [9]. This may be due to individual differences in sympathetic response to noxious stimuli. When the primary afferent sensory nerve was depolarized by nociceptive stimulation, the signal was transmitted to the hypothalamus by the depolarization of the secondary afferent sensory nerve. A nociceptive signal input to the hypothalamus was transmitted to the sympathetic nerve center, and an efferent signal caused an autonomic nervous reaction such as blood vessel contraction, blood pressure increase, and heart rate change. Opioids suppress autonomic nervous responses by various actions such as transmitter release of primary afferent sensory nerves, suppression of depolarization of secondary afferent sensory nerves, and activation of the descending pain inhibitory system by disinhibition of the hypothalamus. The index of the response strength of K itself, such as K R80 , measures the strength of the response when the noxious stimulus is transmitted to the sympathetic nerve center via the afferent and further to the peripheral effector through the centrifugal tract. Naturally, the presence or absence of afferent stimuli alters K R80 ; however, at the same time differences in the responsiveness of the effector also alter K R80 . In contrast, MEC K is independent of effector responsiveness because it measures the presence or absence of sympathetic responses to nociceptive stimuli. Therefore, MEC K could predict blood pressure uctuations associated with skin incision more accurately than the K R80 .
Correlation with ROC BP during skin incision MEC K and MEC PPG correlated better with ROC BP versus MEC HR and MEC BP , which suggests that MEC K and MEC PPG may capture sympathetic responses to nociceptive stimuli more accurately. In contrast, the number of cases outside the measurement range of MEC PPG was higher than that of the MEC K (6 versus 3 for MEC PPG and MEC K , respectively), and ROC BP variability of MEC PPG was also higher. This may be because PPG amplitude re ects changes in blood ow and is in uenced by multiple parameters such as cardiac output and vascular tone. The PPG amplitude was reduced, probably by vasoconstriction with nociceptive stimuli; however, it may have been increased despite nociceptive stimuli as a consequence of an augmentation factor that was a result of increased cardiac output that occurred at the same time [12].
In contrast, K was not affected by blood ow changes due to changes in cardiac output because it measured vascular compliance. This means that in cases where PPG cannot detect or shows excessive changes, K can accurately detect the presence or absence of a sympathetic response. For these reasons, we believe that the MEC K was the most accurate measure of MEC.

Limitations
To determine if MEC K can be an indicator of opioid sensitivity, ndings produced while adjusting the opioid dose based on MEC K must be compared with ndings determined using a constant opioid dose. A study to further evaluate this is in the planning stage.
Remifentanil concentrations were examined solely at 2 ng/ml. Changes in MEC and ROC BP at other concentrations require additional study. However, at 2 ng/ml, MEC varied appropriately between 10-80 mA. If the concentration of remifentanil was greater than 2 ng/ml, the number of cases outside the measurement range was expected to increase. In contrast, when tested at concentrations less than 2 ng/ml, MEC values below 10 mA may have been uniformly regarded as MEC values of 10 mA. These facts suggest that the concentration of 2 ng/ml used in this study was reasonable regarding the amount of opioids needed to inhibit tetanus stimulation.
Sympathetic response also occurs in muscle contraction induced by tetanus stimulation. Therefore, tetanus stimulation is not considered to be a pure Aδ or C ber stimulus. However, Funcke et al. stated that the sympathetic responses to tetanus and intradermal electrical stimulation might be considered as a similar stimulus [13]. In this study, the tetanus stimulus was highly speci c to Aδ and C bers in the inhibition of muscle contraction due to the administration of muscle relaxants.

Conclusions
Compared with K R80 , MEC HR , MEC BP , and MEC PPG , MEC K correlated best with blood pressure changes during skin incisions. The MEC K -indexed prediction equation accurately predicted the increase in blood pressure that occurred during skin incision at a remifentanil effect-site concentration of 2 ng/ml. MEC K was shown to be a potentially more accurate indicator of opioid sensitivity than the rate of change in K.

Declarations
Funding: Institutional and/or departmental Competing interests: The authors have no competing interests to declare.
Author contributions: SK helped design the study, analyze, and interpret the data, and draft and approved the nal version of the manuscript. RN helped conceive and design the study, analyze, and interpret the data, draft the manuscript, and approved the nal version. NS helped design the study, analyze, and interpret data and approved the nal manuscript. TK was involved in recruitment and data collection and approved the nal manuscript. HM assisted with recruitment and data collection and approved the nal manuscript. SN assisted with recruitment and data collection and approved the nal manuscript. AM assisted with recruitment and data collection and approved the nal manuscript. MK assisted with the initial design of the study and approved the nal manuscript. HH assisted with the initial design of the Results are presented as mean±SD (minimum-maximum). Figure 1 Schematic diagram of the mechanical impedance model. The model used in this study consists of a parallel arrangement of springs and dampers across the diameter of the arterial wall. When a force (blood pressure) is applied across the diameter of the artery, the arterial wall is dampened by springs and dampers that represent the rigidity and viscosity of the arterial wall itself. The behavior of the arterial wall after damping is observed as a change in the photoplethysmogram waveform.