Lidocaine Intraoperative Infusion Pharmacokinetics during Partial Hepatectomy for Living Liver Donation

This prospective study was reviewed and approved by the Colorado Multiple Institutional Review Board (Denver‚ Colorado; approval No. 19-1626). Signed informed consent was obtained from all patients before participation.

Any adult (at least 18 yr) patient scheduled for an open partial hepatectomy for living liver donation at the University of Colorado Hospital (Aurora‚ Colorado) between December 2018 and June 2019 was eligible for this study. Exclusion criteria included refusal to participate in the study, being unable to provide informed consent, or presenting with a plasma hemoglobin concentration less than 8 g/dl.

All study patients undergoing an open partial hepatectomy for liver donation received our standardized multimodal analgesia regimen for living liver donors as described in supplemental table S1 (http://links.lww.com/ALN/C953). These recommendations include the administration of an initial bolus of 1 mg/kg IV lidocaine hydrochloride during induction of general anesthesia, immediately followed by a continuous infusion of 1 mg · kg ^˗1 · h^˗1. The dosing recommendations were adjusted by the anesthesia team at their discretion. The lidocaine infusion was discontinued when the graft was isolated.

Study patients provided informed consent allowing the collection of up to 30 ml of blood, collected at different time points during surgery and 24 h later for research purposes. Blood samples were obtained from a central venous catheter or from an arterial catheter at the following time points: before induction of general anesthesia in the preoperative area (preoperative), immediately after induction of general anesthesia (induction), during hepatectomy (hepatectomy), within 30 min after hepatectomy completion (posthepatectomy), within 30 min after the end of surgery at case end/intensive care unit arrival (surgery end), and 24 h after surgery (24 h postoperative). Central venous and arterial samples sites are roughly equivalent in terms of drug mixing during an IV infusion, and so these data were treated equally for data analyses.^16,17  Of note, our research assistant requested blood samples from the most convenient site that was available during normal clinical care to minimize interference with the anesthetic care of the patient. If the anesthetic care of the patient impeded the collection of study blood samples, the patient would be removed from the study, and no additional study samples would be obtained. Once collected, blood samples were immediately placed on ice and then transported to the laboratory, where the plasma was separated by centrifugation and stored at ˗80°C until analysis.

We collected the following variables from each patient: age on the date of surgery (yr), sex, weight (kg), height (cm), doses of IV lidocaine hydrochloride administered (mg/kg; bolus dose, infusion rate(s), infusion duration, and total dose), graft type (right lobe, left lobe, or left lateral segment), duration of surgery (min), preoperative estimated graft volume (ml), total estimated liver volume (ml) determined by imaging (MeVis AG, Bremen, Germany), and the liver graft weight (g) upon surgical removal. The percentage of liver resected was calculated as the ratio of the actual liver graft weight and the estimated total liver volume.

The samples were thawed immediately before mass spectrometry analysis to measure plasma lidocaine, MEGX, and GX concentrations. The reference compound lidocaine was obtained from Sigma-Aldrich (USA). MEGX and GX, as well as the corresponding stable isotope-labeled internal standards, were from Toronto Research Chemicals (Canada). All calculations were carried out using Sciex Analyst software (version 1.6.2; Sciex, USA). The analytical range was 25 to 10,000 ng/ml for lidocaine, MEGX, and GX. Said liquid chromatography–mass spectrometry analysis was carried out by iC42 Clinical Research and Development (USA) in an American College of Pathologists–accredited and Clinical Laboratory Improvement Amendments–certified laboratory environment. To estimate accuracy, 75% of calibration samples were less than ±15% from nominal, except for the sample with the lowest concentration of analyte that can be quantitatively determined, which was less than ±20%. As a measure of precision, two thirds of the quality control samples in an analytical batch were within 15% of the nominal concentration. The assay and its key performance parameters are described in more detail in the supplemental digital content (http://links.lww.com/ALN/C953).

Categorical variables were summarized using frequency (%). Continuous variables were summarized as mean (SD) or median (first quartile, third quartile) based upon evaluation of normality via boxplots and the Shapiro–Wilk test. Prospective power calculations were not completed, given the lack of existing data on the outcomes. Rather, a minimum sample size of 10 was targeted during the study period, with additional cases collected if the number of procedures was higher than expected. MATLAB (Mathworks GmbH, Germany) and R version 4.1.0 (Austria) were used for all analyses.

For purposes of a combined pharmacokinetic model of lidocaine and its metabolites, MEGX and GX (fig. 1), concentrations of the metabolites were adjusted by their molar ratios relative to lidocaine. The population pharmacokinetics for plasma lidocaine, MEGX, and GX were analyzed in a stepwise manner. First, one, two, and three compartmental models were fitted to lidocaine concentration versus time data from which the most parsimonious model was determined, and then potential covariates were assessed for inclusion in the model. The lidocaine post hoc parameter estimates, including covariates, were carried forward to the modeling of the MEGX and GX data, resulting in proportionate reductions in clearance posthepatectomy for these metabolites. Subsequent modeling of metabolites requires assumptions, as actual production (or dose) of the metabolites is unknown. For modeling purposes, we assumed that all lidocaine is metabolized to MEGX and that all MEGX is metabolized to GX. Thus, lidocaine, MEGX, and GX concentration versus time data were fit to the most parsimonious lidocaine model in which the elimination clearance was assumed to produce MEGX with a volume of distribution equal to that of lidocaine and MEGX elimination clearance producing GX with a volume of distribution set to one half that of lidocaine. Two cell metabolic delay elements were placed between parent and metabolite compartments.

Pharmacokinetic data analysis was performed with Phoenix NMLE 8.3 using the FOCE ELS algorithm (Certara, USA). Model parameters were assumed to be log-normally distributed across the population. Residual error was calculated as relative error. Model selection was based on a significant decrease in the ˗2LogLikelihood (chi-square test, P < 0.01), assessing goodness of fit by visual inspection, and the following goodness-of-fit plots: conditional weighted residuals versus time plots, conditional weighted residual versus predicted plots, and individual predicted versus observed plots. Random effects (ω) were removed for parameters in which shrinkage values were above 0.90. Visual predicted checks were created by simulating 500 data sets based on the model’s fixed and random effects parameters and comparing the simulated 5th, 50th, and 95th quantiles to those of the actual data. The m³ method of Beal¹⁸  for including GX concentration data below the limits of quantitation was applied (all lidocaine and MEGX concentrations were within the quantifiable range).

Potential model covariates were sought via stepwise search in which potential covariates were added when the ˗2LogLikelihood decrease was at the P < 0.05 significance level (χ²-test) and removal did not result in an increase of the ˗2LogLikelihood at the P < 0.01 significance level. We used these relaxed levels of significance as this pilot study had only 15 subjects with only 6 blood samples each, inclusive of baseline. Demographic data considered in the covariate testing were: body weight (BW) as an exponential relationship:

where θ_TV and is the typical value for the parameter, and θ_COV is the parameter that quantifies the covariate effect. The fraction of remaining liver (f_{liver-remaining}) was assessed as a time-varying covariate with a value of 1 at all times before the graft was surgically isolated and a constant value less than 1 at all times thereafter, modeled as a linear relationship:

where θ_TV and is the typical value for the parameter, and θ_COV is the value of the relative covariate effect. Age and sex were not considered for covariate analysis.

A total of 15 consecutive patients were approached and included in this study. No patients refused to participate or were excluded for any reason. The patient characteristics and surgical details are shown in table 1. The majority (11 of 15, 73%) of participants underwent a right lobe hepatectomy, and the resected graft size ranged from 14 to 63% of the total donor’s liver. Preoperative samples (A) were missing from five patients. However, all measured preoperative samples were below the limits of quantification. Therefore, we also assumed concentrations below the level of quantification for the missing samples. No patients were removed from the study once enrolled. No adverse events were noted in any participant throughout the study period.

All patients received an IV lidocaine bolus during induction of anesthesia. This bolus dose ranged from 40 to 100 mg, with a mean of 73.8 mg. When adjusted for individual patient weight, patients received a bolus between 0.47 and 1.64 mg/kg with a mean of 0.97 mg/kg. All patients received a continuous IV lidocaine infusion that was initiated shortly after induction. The infusion rate of lidocaine was between 0.60 and 1.48 mg · kg ^˗1 · h^˗1 with a mean of 0.98 mg · kg ^˗1 · h^˗1. The total lidocaine infusion doses were calculated and ranged from 3.1 to 5.3 mg/kg with a mean of 4.1 mg/kg. Infusions were discontinued at the time of graft isolation, and no further lidocaine was administered. The summary of administered IV lidocaine is shown in table 1.

Of the 15 patients, the majority donated a right lobe graft with the others donating left lobe or left lateral segment grafts (see details in table 1). The graft weight to total estimated liver volume weight was calculated as a percentage. For the 11 patients who had a right lobe graft removed, the graft constituted a mean estimated 55% of liver volume; for a left lobe graft, the graft constituted 23%; and for a left lateral segment graft, the graft constituted 18%.

Summarized plasma concentrations of lidocaine, MEGX, and GX at each time point are shown in figure 2. Samples were drawn at the following time points before the start of the lidocaine infusion, shown as median (minimum, maximum): preoperative, ˗99 min (˗120, ˗75); induction, 13 min (˗10, 27); and hepatectomy, 122 min (15, 235). Samples were also drawn at the following time points after stopping the infusion: posthepatectomy, 8 min (-22, 34); surgery end, 107 min (90, 341); and 24 h postoperative: 1.202 min (1.076, 1.278). Plasma lidocaine, MEGX, and GX concentrations (fig. 2a) were undetectable before induction and progressively increased to reach the highest concentrations during and after completion of the hepatectomy. Plasma lidocaine concentrations decreased at the end of surgery and even further 24 h later. All observed plasma lidocaine concentrations were below the potentially toxic threshold of 5 μg/ml, except for a single measurement from a sample collected at induction of anesthesia. Based on the time point and the administration protocol, this could be related to a larger-than-average lidocaine bolus dose, coupled with an earlier than average collection time, thus reflecting a comparatively higher plasma lidocaine concentration during the early tissue distribution phase. Several metabolite concentrations were below the level of quantification; for analysis purposes, these values were represented by 0. The plasma MEGX concentration (fig. 2b) increased after induction of anesthesia and continued to rise even at the end of surgery, being lower 24 h later. The preoperative plasma concentration of GX (fig. 2c) was also undetectable in all samples. After induction of anesthesia, all plasma GX concentrations except one were undetectable. Plasma GX concentrations rose during hepatectomy, after hepatectomy, and through to the end of surgery, after which they were lower at 24 h.

Individual lidocaine plasma concentrations versus time data are shown in figure 3a, along with a simulation of the plasma lidocaine concentrations versus time relationship for an infusion of median duration using the pharmacokinetics of Rowland et al.¹⁹ Figure 3b displays the observed plasma lidocaine concentration versus time relationship along with the sum of toxicity potential of lidocaine, MEGX, and GX, calculated as the sum of concentrations: [lidocaine] + [MEGX] + 0.25*[GX], versus time with the reference of toxicity threshold at 5.0 μg/ml. Fitted lidocaine population pharmacokinetic models had ˗2LogLikelihoods of 107.4, 65.5, and 64.2 for one-, two-, and three-compartment models, respectively; thus, a two-compartment model as shown in figure 1 was selected as the most parsimonious model. The only significant covariate in the model was the fraction of the liver remaining after the donor graft was surgically isolated with a ˗2LogLikelihood of 60.16. The post hoc parameter estimates for lidocaine were fixed before fitting the MEGX and GX concentration data to the model in figure 1. The final pharmacokinetic model parameter values are given in table 2 along with ω standard errors and shrinkage estimates. Estimates for the typical values of the metabolic delay elements were 89.9 and 37.6 min for CL_lid-MEGX and CL_MEGX-GX, respectively, but the percent coefficients of variation for these parameters were greater than 100%. Thus, these estimates were each fixed to 45 min. It is likely that more blood samples would be needed to adequately estimate the metabolic delays. The typical value (± standard error of the estimate) for lidocaine elimination clearance was 0.55 ± 0.12 l/min, which, on average, was reduced to about one third of the baseline clearance, 0.17 ± 0.02 l/min, once the donor graft was surgically isolated and remained so for 24 h according to the current data and model. In accordance with our modeling assumptions, MEGX and GX clearances were proportionately reduced after hepatectomy as well.

The observed versus final typical model and post hoc individual model predictions for lidocaine concentration are presented in figure 4 (a and b). The relationships between conditional weighted residuals versus predicted concentrations and time are presented in figure 4 (c and d, respectively). Similar graphs are presented in figure 4 (e through h) for MEGX and figure 4 (i through l) for GX data, respectively. The visual predictive checks were performed as described and are presented in figure 5.

In this prospective study, we evaluated the plasma concentrations of lidocaine from patients during and after open partial hepatectomy for liver donation receiving intraoperative IV lidocaine infusions. In this patient cohort receiving lidocaine infusions following our institutional protocol, we did not observe any toxic plasma concentrations of lidocaine and its metabolites (MEGX and GX) during or after completion of the liver resection. The pharmacokinetic analysis of lidocaine from these patients showed the expected reduced clearance proportional to the resected liver graft, but the reduced clearance did not result in unexpectedly high plasma lidocaine concentrations as the infusions were terminated well before significant fractions of the eventual steady-state concentrations could be attained (i.e., 3 to 5 elimination half-lives).

Previously, patients receiving epidural lidocaine infusions have been investigated to determine the effect of epidural infusion and surgical procedure on lidocaine concentration.²⁰  It was found that neither the catheter insertion site nor the procedure type had any significant effects on the lidocaine concentration with the exception of patients undergoing hepatectomy.²⁰  Not surprisingly, patients undergoing hepatectomy while receiving continuous epidural lidocaine infusions developed markedly elevated plasma concentrations at several time points throughout the operative period without reaching toxic concentrations.²⁰  However, these investigators did not evaluate concentrations during the administration of an IV infusion, nor did they estimate the pharmacokinetic changes associated with partial hepatectomy.

To date, there is insufficient data supporting the safety of lidocaine infusions continued beyond completion of the hepatectomy. The purpose of this study was to ensure that our institutional regimen had an acceptable safety profile and to determine the pharmacokinetics in this patient population to guide lidocaine dosing in the future. Other than the single outlier concentration of plasma lidocaine greater than 5 μg/ml (shown in fig. 2), which was measured at anesthesia induction, no lidocaine concentrations neared toxic levels. The metabolite concentrations followed a similar pattern to lidocaine, showing concentration decreases that began postoperatively. There were no instances of metabolite concentrations that failed to decrease as expected.

Patients undergoing living donor hepatectomy are a unique population as they lack confounding comorbid medical conditions. They have no underlying degree of liver dysfunction, and thus all pharmacokinetic changes are attributed to the conduct of anesthesia, laparotomy, and isolation of hepatic blood vessels for the liver resection. These patients undergo extensive screening to rule out any preexisting medical conditions to ensure that the procedure risk is minimized. It is likely that performing the same study in patients undergoing hepatectomy for liver disease would yield different results; one would expect a higher degree of hepatic dysfunction and reduced elimination clearance perioperatively. However, performing this study in patients who are otherwise healthy without any preexisting comorbidities allowed the establishment of a baseline safety profile and an expected minimally perturbed pharmacokinetic model of lidocaine for hepatectomy.

In comparison to the lidocaine pharmacokinetics estimated by Rowland et al.¹⁹  in healthy volunteers (fig. 3a), we found that V_SS was larger (table 2, 1.1 l/kg versus 2.5 l/kg) but was well within V_SS estimates of (0.6 to 4.5 l/kg) reported in more comprehensive work by Rowland’s group.²¹  More significantly, the baseline estimate of lidocaine clearance, determined during surgery but before hepatic resection, in these healthy adults undergoing laparotomy was 70% of the estimated clearance in nonsurgical healthy volunteers (0.55 l/min versus 0.79 l/min, respectively).¹⁹  This 30% reduction in clearance compares to the 62 to 68% reduction in hepatic blood flow measured in 88 patients undergoing open cholecystectomy and open gastrectomy operation under either halothane or ether anesthesia,²²  suggesting that such surgical interventions have major effects on liver blood flow and, perhaps, commensurate effects on the clearance of drugs such as lidocaine, known to have flow-limited metabolic clearance. Since we do not have independent data regarding nonsurgical baseline, hepatic blood flow, and only limited timed blood samples during this time frame, we cannot rule out other causes for our lower estimates of clearance, including study design. Additionally, there was a further reduction in clearance at the time of surgical isolation of the donor liver graft, demonstrated in the covariate analysis, in direct proportion to the fraction of the liver removed (equation 2 and table 2) that persisted for the 24 h of our sampling schedule. In this pilot study, we did not explore potential nonlinearities involving such factors as loss of liver mass, metabolic capacity, changes in hepatic blood flow and its autoregulation, or further deterioration in these factors in the postoperative period or their recovery over time.

Lidocaine pharmacokinetics share characteristics of other drugs used in the perioperative period. Specifically, lidocaine distributes rapidly to a volume of distribution roughly corresponding to a multiple of body weight, and its metabolism by CYP1A2 and CYP3A4 has high capacity, which means that clearance is determined by liver blood flow. Since these pharmacokinetic characteristics are similar for ketamine, propofol, and fentanyl, it is not unreasonable to expect similar findings for these drugs administered perioperatively in patients undergoing donor hepatectomy. Thus, caution and further studies in this unique patient population are warranted.

There are limitations to our study, the most obvious of which is the small sample size and the implementation at a single institution. There were some missing data in our study: preoperative blood samples were missing from five patients, one missing a blood sample from the case end (E), and one patient did not have a calculated liver volume noted within their surgical procedure note. Thus, we were unable to calculate the proportion of liver resected for this patient (noted in table 1); instead, the fraction planned for resection from preoperative imaging was used. The missing preoperative samples were not felt to be prohibitive given that all remaining patients had preoperative undetectable concentrations for lidocaine and metabolites. Future evaluation of lidocaine clearance in patients undergoing hepatectomy for reasons other than living liver donation may produce different results. This is important given the frequency with which lidocaine infusions are utilized in Enhanced Recovery After Surgery protocols for several types of surgical procedures, including those in patients with perioperative hepatic dysfunction.

Our findings suggest that lidocaine infusions can be continued into the postoperative period after donor hepatectomy. For instance, to maintain a plasma lidocaine concentration of 1.0 μg/ml after resection of the hepatic graft, the lidocaine infusion should be reduced to a rate calculated according to the liver lobe and, thus, the approximate percentage of the total liver removed. In the case of the right lobe (44.5% remaining after graft resection), the calculation from table 2 is as follows: C_SS × Cle = 1.0 μg/ml × 0.55 l/min × (1 + (0.445 - 1) × 1.230) = 1.0 mg/l × 0.175 l/min × 60 min/h = 10.5 mg/h. Similarly, the highest MEGX and GX concentrations would occur at steady-state, so their maximum contributions to potential drug effects and/or toxicities can be calculated for MEGX as the ratio of its production and elimination clearances (i.e., 0.55 l/min/1.69 l/min = 0.325) and for GX as the ratio of its production and elimination clearances, corrected for the steady-state MEGX concentration relative to lidocaine or 0.325 (i.e., 1.69 l/min/0.84 l/min × 0.325 = 0.625, which should be further corrected by 0.25 to account for the lower toxicity of GX, relative to the equipotent lidocaine and MEGX. Thus, at a steady-state plasma lidocaine concentration of 1.0 μg/mL, the total plasma lidocaine concentration in terms of toxic potential would be: 1.0 μg/ml + 1.0 μg/ml × 0.325 + 1.0 μg/ml × 0.625 × 0.25 = 1.48 μg/ml. If different target concentrations are desired, these relationships should be proportional if the clearances are, indeed, first order and stationary.²³  Caveats to these dosing recommendations include using these parameter estimates beyond the 12 h of data collection, beyond the fractions of remaining liver studied in this sample of patients (0.26 to 0.84) and the preliminary nature of these results, based on only 15 patients with 5 plasma lidocaine/metabolite measurements per patient. Follow-up studies to correlate lidocaine plasma concentrations with postoperative pain scores should be combined with these pharmacokinetic results to implement a safe and more rational lidocaine dosing regimen for reduction of postoperative pain after donor hepatectomy. Additionally, these results support the potentially substantial effect of major open abdominal surgery and resection of a significant proportion of the liver on lidocaine elimination clearance, which justify further prospective clinical trials to better elucidate these pharmacokinetic effects.

In conclusion, clearance of lidocaine is reduced during hepatectomy proportionally to the liver mass resected. Plasma concentrations of lidocaine and its metabolites (MEGX and GX) are expected to accumulate up until 24 h after liver resection. However, IV lidocaine infusions do not lead to unsafe accumulation of plasma lidocaine or its metabolites and can be used for intraoperative multimodal pain management during hepatectomy surgery for donation if the lidocaine infusion is stopped at completion of the liver resection.

Support was provided solely from institutional and/or departmental sources.

Dr. Crouch discloses funding from Medtronic (Minneapolis‚ Minnesota) for past consulting work unrelated to this project. Dr. Fernandez-Bustamante discloses funding from the National Heart Lung and Blood Institute (Bethesda‚ Maryland); National Institutes of Health (Bethesda‚ Maryland); the U.S. Department of Defense (Falls Church‚ Virginia); the Merck Investigator-initiated Studies Program (Rahway‚ New Jersey); and the Institute for Healthcare Quality, Safety and Efficiency for research projects unrelated to this work. The other authors declare no competing interests.

Supplemental table and materials, http://links.lww.com/ALN/C953