Effect of sex and polymorphisms of CYP2B6 and UGT1A9 on the difference between the target-controlled infusion predicted and measured plasma propofol concentration

Introduction To examine whether sex and polymorphisms of cytochrome P450 (CYP) 2B6 and UDP-glucuronosyltransferase (UGT) 1A9 affect the difference between predicted and measured plasma propofol concentration during continuous infusion by target-controlled infusion. Results Blood samples of 69 patients (48 men and 21 women) were obtained at 4 h after initial propofol infusion. Percentage performance error (PE) was calculated to assess the difference between measured and predicted propofol concentration. Regression coefficients (β) and 95% confidence intervals (CI) of sex and the polymorphisms of CYP2B6 and UGT1A9 for PE were, separately and mutually, estimated with linear regression. Covariates included age and body mass index in the minimal adjusted model, and additionally included clinical factors (mean blood pressure, heart rate, volume of intravenous fluid, surgical site, surgical position, and pneumoperitoneum) in the full adjusted model. PE was higher in men than in women (28.7% versus 10.5%, p = 0.015). Female sex was inversely associated with PE: the minimal adjusted β = − 8.84 (95% CI, − 16.26 to − 1.43); however, the fully adjusted β with clinical factors became not significant. The average of PE did not differ between polymorphisms of CYP2B6 and UGT1A9, and β of CYP2B6 516G>T polymorphisms mutually adjusted with female sex was not significant. Mean blood pressure, heart rate, and volume of intravenous fluid were independently associated with PE in the full adjusted model. Conclusions Under 4 h anesthesia with propofol target-controlled infusion in our population, sex differences appeared to exist in the propofol concentration, which might be largely mediated by clinical factors, such as hemodynamic status. Trial registration UMIN-CTR UMIN000009015, Registered 1 October 2012 Electronic supplementary material The online version of this article (10.1186/s40981-018-0196-8) contains supplementary material, which is available to authorized users.


Background
Propofol (2,6-diisopropyl-phenol) is commonly used in induction and maintenance of general anesthesia due to its rapid onset, relatively short emergence time, and favorable safety profile. The target-controlled infusion (TCI) system is widely used to administer propofol. Diprifusor™ is a TCI system that automatically regulates propofol dose using a pharmacokinetic model to achieve a target blood concentration [1]. Body weight is used as an input parameter for Diprifusor™. Predicted plasma propofol concentration (Cp) and measured plasma propofol concentration (Cm) were found to be correlated, in spite of the slightly greater values of Cm compared to Cp [1][2][3][4]. If Cm is higher than Cp, arousal delay may occur; if Cm is lower than Cp, it may lead to intraoperative body movement and awakening during surgery, which may result in poor prognosis for the patient.
Regarding the pharmacokinetics of propofol, sex has been reported to influence propofol metabolism and the effect of propofol, in addition to age, height, and weight [5][6][7][8][9][10][11][12][13][14][15]: for example, propofol metabolism in women is faster than that in men, and women awake faster than men. For clinical factors, hemodynamic state might also affect the accuracy of the TCI system [16], because the clearance of propofol depends on the hepatic blood flow [17][18][19].
Herein, we examined whether sex and polymorphisms of CYP2B6 and UGT1A9 affect the difference between the predicted concentration and actual plasma propofol concentration in the perioperative period with continuous propofol infusion of 4 h.

Main outcome and patient selection
The main outcome of the study was the difference between Cp and Cm around the maintenance concentration of propofol in surgery, which was evaluated using percentage performance error (PE) as follows: PE (%) = (Cm -Cp)/Cp × 100. We included patients aged ≥ 20 years, underwent anesthesia (expected anesthesia time ≥ 4 h) in spine and lateral positions by administration of propofol by TCI, and monitored with direct arterial blood pressure. We excluded those with anemia (hemoglobin < 10 g/dl), liver dysfunction (Child-Pugh B or C), renal dysfunction (eGFR < 30 ml/min/ 1.73m 2 ), American Society of Anesthesiologists-physical status (ASA-PS) class III/IV, hepatic or renal surgery, and psychoneurotic disorders or psychiatric pharmacotherapy.
Of 70 eligible patients, we excluded one outlier of 39-year-old woman with a great discrepancy in Cp and Cm (PE = 267.9%) due to a possible technical error; 69 patients (48 men and 21 women) comprised the study subject. A prior analysis included the excluded patient showed the similar trend (data not shown).

Measurement of propofol concentration and clinical parameters
Anesthesia was induced and maintained with continuous infusion of remifentanil and propofol. A propofol TCI system (TE-371, TERUMO, Tokyo, Japan) was used to administer propofol. The infusion rates of propofol and remifentanil were adjusted by the anesthesiologists in charge according to the patients' condition. Direct arterial blood pressure, heart rate (HR), ECG, SpO 2 , central core temperature, and end tidal CO 2 were recorded throughout all operations. Bispectral index (BIS, QE-910P, Nihon Kohden, Tokyo, Japan) were applied unless it did not disturb the procedure of the surgeries.
Blood samples were collected from the radial artery at 4 h after initial propofol infusion after matching predicted blood concentration and effect-site concentration displayed on the TCI devices. When the duration of propofol infusion was < 4 h, the sample was collected before changing the target blood concentration. Mean blood pressure (mBP) and HR were recorded at the time of sample collection. Total volume of intravenous fluid was measured from initial propofol infusion to blood sample collection. Blood samples were used for gene polymorphism analysis and measurement of plasma propofol concentration. Although the time of blood pressure before blood sample collection might affect the concentration of propofol, a prior analysis using blood pressure of 10 min before blood sampling showed the same results (data not shown).
The plasma concentration of propofol was determined by a modified method of a previous report [25] by a commercial laboratory, BML, Inc. (Tokyo, Japan), using a reverse phase high-performance liquid chromatography system (Hitachi High-Technologies Corporation, Tokyo, Japan, and Shimadzu Corporation, Kyoto, Japan) with a Hypersil C18 reversed-phase column (3 μm particle size, 100 × 5.0 mm I.D.). The excitation and emission wavelength were 276 and 310 nm, respectively. Blood samples were centrifuged (1150 g for 10 min) and stored at 4°C. A calibration graph was created by plotting the ratios of the areas for propofol to those for thymol (internal standard) from 0.2 to 5 μg/ml. The limit of quantitation was 0.1 μg/ml.

Statistical analysis
To assess the contribution of sex and polymorphisms for PE separately, crude regression coefficients (β) and 95% confidence intervals (CI) of female sex and CYP2B6 516G > T polymorphisms were, respectively, estimated with linear regression (model 1). In addition, we minimally adjusted for baseline characteristics (age and BMI) in model 2 and fully adjusted for baseline characteristics (age and BMI) and clinical factors (mBP, HR, volume of intravenous fluid, surgical site, head down position, pneumoperitoneum) in model 3. Those covariates were adjusted as mediators for the association; other clinical factors, such as BIS scores and body temperature, were not included for covariates because those did not affect sex differences of PE in a prior analysis (Additional file 1: Table S1).
In mutual adjusted models, we included explanatory variables of female sex and CYP2B6 516G > T polymorphisms simultaneously (model 4). Additionally, we minimally adjusted for baseline characteristics (model 5) and fully adjusted for baseline characteristics and clinical factors (model 6). In the mutual adjusted models, due to the lack of female patients with CYP2B6 516TT polymorphisms, we excluded three male patients with CYP2B6 516TT polymorphisms.
Alpha was set at 0.05, and all p values were two-sided. Data were analyzed using JMP Pro 11.0.0 (SAS Institute Inc., Tokyo, Japan).

Results
The baseline characteristics and clinical factors did not differ between men and women (Table 1). Cp and Cm were statistically correlated: the correlation coefficients were, respectively, 0.73 in men (p < 0.001, Fig. 1a) and 0.50 in women (p < 0.001, Fig. 1b). The average of PE in men was significantly higher than that in women (28.7% versus 10.5%, p = 0.015, Table 1 and Fig. 1c).
The results of CYP2B6 and UGT1A9 SNPs analyses are shown in Table 2 and Additional file 2: Table S2. Although the allele frequencies for any polymorphisms did not statistically differ between men and women (Additional file 2: Table S2), the average of PE tended to differ between CYP2B6 516G > T polymorphisms (p = 0.050, Table 2).
In regression analysis, female sex was inversely associated with PE (Table 3): the minimal adjusted β = − 8.84 (95% CI, − 16.26 to − 1.43). However, the association was substantially attenuated and became not significant on the full-adjustment with clinical factors (model 3, Table 3). Likewise, compared with CYP2B6 516GG, CYP2B6 516G > T polymorphisms was inversely associated with PE in the crude and minimally adjusted models (Table 3). However, the association was substantially attenuated and became not significant on the full-adjustment with clinical factors (model 3, Table 3). CYP2B6 516TT polymorphisms were not associated with PE.
In the mutual model, β of female sex was significant, but β of CYP2B6 516G > T polymorphisms was not significant (model 4, Table 4). The association of female sex was marginally significant on the minimal adjustment with baseline characteristics (model 5, Table 4); however, the association was substantially attenuated and became not significant in the full-adjusted model with clinical factors (model 6, Table 4). The clinical factors, mBP, HR, and volume of intravenous fluid were independently associated with PE (model 3 in Table 3 and model 6 in Table 4).

Discussion
This is the first study to investigate PE associated with sex and polymorphisms of CYP2B6 and UGT1A9, with clinical factors, after long-term propofol infusion. We found that PE in women was closer to 0 than that in men, which might indicate that the propofol TCI system is more accurate in women than in men.
A number of reports have suggested that the plasma concentration of propofol decreases more rapidly in women than in men, and that women tend to recover faster from propofol anesthesia than men [7][8][9][11][12][13][14][15]. Loryan et al. suggested this sex difference was due to the fact that women's livers have 1.9-fold greater CYP2B6 protein levels than men [7]. In this study, the mean PE in the perioperative period was significantly lower in women than in men, which might be caused by the sex difference of propofol pharmacokinetics.
In our study, it should be noted that clinical factors of hemodynamic status (HR and mBP), which could influence the hepatic blood flow, may substantially mediate the sex difference of PE. A higher tendency of HR in women (Table 1, p = 0.096) might partly explain the mediation in our study, given the fact that propofol is mostly metabolized in the liver, and the hepatic clearance of propofol depends on the hepatic blood flow [17][18][19]. Volume of intravenous fluid was inversely associated with PE, which could be explained by the decreased concentration of propofol due to increased plasma volume and the increased propofol metabolism rate due to increased hepatic blood flow.
This study had some limitations. First, we collected only one sample for each subject. More samples are usually collected for the external evaluation of TCI system. Second, more hemodynamic information, such as cardiac output, which can enhance liver blood flow is required. We only have the data of blood pressure and HR at the time of sample collection. Third, the target concentration of propofol was determined by the anesthesiologist in charge according to the patients' condition because this was an observational study. Fourth, blood laboratory test was not conducted at the time of sample collection; therefore, serum albumin or hemoglobin values were not obtained. Lastly, due to the small sample size with insufficient statistical power, we could not fully assess CYP2B6 and UGT1A9 polymorphisms, in relation to sex and clinical factors. Further studies with a large number of patients are needed to clarify the influence of CYP2B6 and UGT1A9 polymorphisms on the accuracy of propofol TCI system.

Conclusions
The accuracy of estimated propofol concentration by Diprifusor™ TCI system appeared to differ between men and women in our population. The PE in men was higher than in women which might be mainly caused by hemodynamic status. Although this study did not show significant influence of CYP2B6 and UGT1A9 polymorphisms, further studies are expected to elucidate the effect of those polymorphisms.