Thomas Duflot1, Maxime Riffault2, Phillipe Gomis2, Chloé Hauet2, Claire Lepousé2, Daphné Michelet2, Jean-Marc Malinovsky3, Zoubir Djerada1
1Department Of Pharmacology, University Of Reims Champagne-Ardenne, Ppf Ur 3801, Reims University Hospital, 2Department Of Anesthesiology, Reims University Hospital, 3Department Of Anesthesiology, University Of Reims Champagne-Ardenne, Ppf Ur 3801, Reims University Hospita
Objectives: Spinal anesthesia with levobupivacaine and ropivacaine is widely used due to its favorable safety and efficacy profiles. Although previous studies have characterized the pharmacokinetic-pharmacodynamic (PK-PD) relationships of these agents for sensory block, the relationship between drug exposure and motor block intensity remains poorly defined [1-3]. We aimed to develop a population-based PK-PD model to quantify motor block as a function of intrathecal anesthetic amount, thereby elucidating interindividual variability and informing optimized dosing strategies for improved clinical efficacy. Methods: Sixty patients (aged 18–82 years) scheduled for spinal anesthesia were prospectively enrolled after providing written informed consent. Participants received intrathecal ropivacaine or levobupivacaine in one of three dose groups: 15 mg (low), 20 mg (middle), or 25 mg (high) for ropivacaine; and 10 mg (low), 12.5 mg (middle), or 15 mg (high) for levobupivacaine. Serial plasma samples were collected at predefined time points (0, 5, 15, 30, 60, 120, 240, 480, and 1440 minutes post-administration) to characterize drug exposure. Motor block was assessed using the Bromage scale at high-frequency intervals: every 2.5 minutes from 0 to 30 minutes, every 5 minutes from 30 to 60 minutes, and every 15 minutes from 60 to 330 minutes. A population PK-PD model was constructed using nonlinear mixed-effects modeling (Monolix2024R1). The pharmacokinetic profile was described by a one-compartment model with a depot compartment and an additional effect-site compartment representing spinal drug distribution. Motor block was modeled as an ordered categorical outcome (Bromage scale 0–3) using a cumulative logit function, with the primary predictor being the effect-site drug amount transformed by the square root and scaled by a beta coefficient. Model performance was evaluated via five-fold cross-validation by comparing observed versus individual predicted plasma concentrations and examining the normalized prediction distribution error (npde) for the PD component. Furthermore, the time course of the Bromage scale was modeled using a gamma-shaped function of the form Motor block = a × Time^b × exp(–c × Time). Because the gamma function provides a continuous output while the Bromage scale is discrete, outputs were discretized using cut-offs (0.5, 1.5, and 2.5). Covariate effects on the gamma parameters (a, b, and c) were explored, including age, body mass index (BMI), anesthetic type, and dose. Individual parameter estimates were used to compute clinical anesthesia metrics such as the area under the Bromage block curve (AUC), time to recovery (return to Bromage 0 or 1), and the total duration of each block level. Results: A total of 458 levobupivacaine and 407 ropivacaine plasma samples, along with 2169 Bromage scores, were analyzed. BMI was negatively associated with the elimination rate constant, and ropivacaine was less potent compared to levobupivacaine (lower beta coefficient). Overall, the relative standard errors for all parameters were below 20%. Cross-validation demonstrated robust performance, with mean prediction errors (MPE) below 20% and R² values exceeding 0.85 for the PK model across all folds; npde values were comparable between the original dataset and merged cross-validation subsets. Treatment type and dose significantly predicted all parameters driving the shape of the motor block profile, while BMI significantly influenced block onset and duration. Additionally, anesthesia metrics revealed that age and dose group were positively associated with AUC (area under the Bromage block curve), time to recovery, and total duration of each block level. To translate these findings into clinical practice, two dosing algorithms were developed and implemented in Shiny applications: 1) One-Compartment + Effect-Site Algorithm : Description: This algorithm integrates a PK model with a depot (CSF) and effect-site compartment, linking the effect-site drug amount to the Bromage scale via cumulative logit functions. Below is the raw code (with inline comments) used to implement this algorithm: DESCRIPTION: [LONGITUDINAL] input={V1,ke,th1,th2,th3,beta,ka1,ka2,kae} PK: depot(adm=1,target=Acsf) EQUATION: odeType = stiff Acsf_0 = 0 Ac_0 = 0 Ae_0= 0 ddt_Acsf = – ka1*Acsf – ka2*Acsf ddt_Ae = ka1*Acsf – kae*Ae ddt_Ac = ka2*Acsf – ke*Ac lgp0 = th1 – beta * sqrt(Ae) lgp1 = (th1 + th2) – beta * sqrt(Ae) lgp2 = (th1 + th2 + th3) – beta * sqrt(Ae) Ropipl = Ac/V1 ddt_AUCRopipl = Ropipl DEFINITION: blocklevel = {type = categorical, categories = {0, 1, 2, 3}, logit(P(blocklevel<=0)) = lgp0 logit(P(blocklevel<=1)) = lgp1 logit(P(blocklevel<=2)) = lgp2 } OUTPUT: output = {Ropipl, blocklevel,Ae} with: V1_pop = 141.25, # Volume of distribution (L) ke_pop = 0.0023, # Base elimination rate constant (min^-1) beta_ke_logtIMC = -1.24, # BMI effect on elimination rate (log-scale) th1_pop = 12.13, # Logit threshold for Bromage 0 th2_pop = 3.52, # Increment for Bromage 1 th3_pop = 5.56, # Increment for Bromage 2 beta_pop = 1.23, # Slope linking sqrt(effect-site amount) to logit beta_beta_TTT_Ropivacaine = -0.24, # Adjustment for ropivacaine potency ka1_pop = 0.0031, # Rate constant from depot to effect-site (min^-1) ka2_pop = 0.013, # Rate constant from depot to central (min^-1) kae_pop = 0.034, # Elimination rate from effect-site (min^-1) omega_V1 = 0.36, # Variability in V1 omega_beta = 0.19, # Variability in beta omega_ka2 = 0.48, # Variability in ka2 omega_kae = 0.46, # Variability in kae omega_ke = 0.29, # Variability in ke omega_th2 = 0.6, # Variability in th2 omega_th3 = 0.57, # Variability in th3 a1 = 1.87, # Additional model parameter b1 = 0.2 # Additional model parameter 2) Gamma-Shaped Algorithm: Description: This algorithm models the time course of motor block directly with a gamma-shaped function. The continuous output is discretized into Bromage levels (0–3) using predefined cut-offs. Below is the raw code (with inline comments): A simplified approach modeling the time course of motor block directly, ignoring separate PK compartments. The predicted motor block intensity is computed for each time as follows and discretized using 0.5, 1.5 and 2.5 cut-offs : Motor block <- (a0 + a1 * AMOUNT + a2 * TTT_num + a3 * BMI) * TIME^(b0 + b1 * AMOUNT + b2 * TTT_num + b3 * BMI) * exp(-(c0 + c1 * AMOUNT + c2 * TTT_num) * TIME) with: a0 <- 2.620e-01 # Intercept for scaling factor ‘a’ a1 <- 4.091e-05 # Dose effect on ‘a’ a2 <- -2.721e-01 # Anesthetic type effect on ‘a’ (TTT_num) a3 <- -1.114e-02 # BMI effect on ‘a’ b0 <- 7.477e-01 # Intercept for exponent ‘b’ b1 <- -2.173e-05 # Dose effect on ‘b’ b2 <- 1.484e-01 # Anesthetic type effect on ‘b’ b3 <- 7.063e-03 # BMI effect on ‘b’ c0 <- 2.264e-02 # Intercept for decay ‘c’ c1 <- -6.589e-07 # Dose effect on ‘c’ c2 <- 6.080e-03 # Anesthetic type effect on ‘c’ AMOUNT <- Dose of the anesthetic BMI <- Body Mass Index TTT_num <- 0 if Levobupivacaine and 1 if Ropivacaine Conclusion: This study presents a robust population PK-PD model that characterizes the motor block induced by levobupivacaine and ropivacaine in spinal anesthesia. By integrating pharmacokinetic data with a cumulative logit model for motor block and employing a gamma-shaped function to describe block evolution over time, our findings pave the way for personalized dosing strategies. Such an approach could allow clinicians to predict the duration and depth of motor block and optimize recovery times based on individual patient characteristics.
Malinovsky JM. et al. Anesth. Analg. 2000;91:1457-602 Djerada Z.et al. Clin. Pharmacokinet. 2018;57(9) :1135-47 Olofsen E. et al. Anesthesiology.2008;109(4):664-74
Reference: PAGE 33 (2025) Abstr 11414 [www.page-meeting.org/?abstract=11414]
Poster: Clinical Applications