A Pharmacometric approach for opioid use disorder in Puerto Rico: A research agenda to individualize buprenorphine pharmacotherapy.
Darlene Santiago (1), Victor Mangas(2), Raman Venkataramanan(3), and Jorge Duconge (1)
(1) University of Puerto Rico, Puerto Rico, (2) University of Valencia, Valencia, (3) University of Pittsburgh, United States
Objectives: The objective of this poster is to present our strategy for transforming the current standard of care for opioid use disorder (OUD) buprenorphine (BUP) pharmacotherapy in Puerto Rico (PR) using population modeling. We have reasoned that the OUD emergency in PR is in desperate need of a pharmacometrics approach to estimate the most adequate dose for BUP-maintained OUD patients in PR, and significantly improve the efforts towards combating the OUD crisis. This poster will present a summary of past and ongoing studies that supported the development and refinement of our BUP-model. Challenges associated with the operationalization of our population approach are discussed. Implementation strategies of the BUP population model are proposed as next steps in our agenda of optimizing BUP pharmacotherapy for OUD in PR.
Methods: For the development of our BUP model, we conducted an 8-hour PK study that collected BUP and it’s metabolites plasma levels after dosing from 12 buprenorphine maintained OUD subjects in PR (6 males/6 females; 96 total samples); in steady state dose, no dose adjustment in the last 3 months and negative toxicology results for illicit drug use. Compartmental pharmacokinetic analyses were performed using NONMEM® (v7.3, 2013, ICON plc Development Solutions, Hanover, MD, USA). Plasma concentrations were logarithmically transformed. The population PK parameters were estimated using the Stochastic Approximation of the Expectation Maximization and the Importance Sampling Estimation method. Inter-individual variability (IIV) associated to the PK model parameters was modeled exponentially, preventing negative values of the individual estimates, and residual unexplained variability (RUV) was described with an additive model on the logarithmic scale. The significance of the non-diagonal elements of the Ω variance-covariance matrix and subject specific RUV were also evaluated. Model selection was established through the minimum value of the objective function value (OFV) provided by NONMEM®, which was approximately equal to −2 × log(likelihood) (−2LL), together with the visual inspection of the goodness of fit (GOF) plots. A decrease of 3.84 points of the −2LL value between two nested models differing in one parameter was considered significant at the 5% level. Evaluation of the selected models was performed through prediction-corrected visual predictive checks (pc-VPC). Briefly, one thousand simulated datasets were simulated, and the 2.5th, 50th, and 97.5th percentiles for every simulated study and sampling time period were calculated. Then, the 95% prediction intervals of the above-described percentiles were calculated and displayed graphically together with corresponding percentiles computed from raw data. Precision of the model parameter estimates, defined as the relative standard error (RSE), was calculated from the variance-covariance matrix (when possible) and from the analysis of one thousand simulated bootstrap datasets. For graphical and statistical analysis, R software (http://cran.r-project.org, version 3.5.0, 2018) was used. Pc-VPC and bootstrap analysis were performed using PsN.
Results:
BUP population model: A total of 307 plasma levels of BUP, and metablites (Nor-BUP, BUP-g and Nor-BUP-g) from 10 patients were considered for the population PK analysis. The structural model assumes rapid absorption of BUP (ka = 2.54 h−1) into a buccal compartment, from which a partial metabolism of BUP to BUP-g appear to occur (k14). The distribution of BUP into the central compartment occurs at a higher rate compared to BUP-g, since k12 is significantly higher than k14. Subsequently, BUP is distributed according to a two-compartment model. From the central compartment, BUP is metabolized to BUP-g (k24 =1.28 × 10−1 h−1) and Nor-BUP (k23 = 6.43 × 10−2 h−1), in similar kinetic terms. Nor-BUP is mainly metabolized to Nor-BUP-g (k35 =1.23 × 10−1 h−1) and, to a minor extent, is eliminated unchanged (k30 = 3.81 × 10−3 h−1). The transformation of BUP-g is carried out to Nor-BUP-g, which is eliminated from the body (k50 =1.27 × 10-1 h-1). (preliminary PK model parameter estimates, precision and bootstrap results table will be presented in the poster). The analysis of the inter-individual variability has confirmed its significance in k24, k23, k35, k45 and k50 parameters, showing a moderate-to-high behavior. Non-diagonal elements were explored, but they were not statistically significant (p > 0.05). The developed preliminary BUP popPK model predictions, represented as shaded areas in the concentration vs time curve, demonstrate that although preliminary, this pilot model has the potential to describe the observed behavior of each analyte, both in its mean trend and in the observed variability.
External validation of the BUP model: External validation of the popPK model developed was achieved through the comparison of model-predicted and observed AUC and Cmax values across several dose levels of single-dose regimen of BUP. The exposure values have been obtained by simulating a virtual population of 1000 patients, following the same study conditions. Subsequently, the AUC values were calculated using the trapezoidal rule and the Cmax value directly from the simulated values. Most of the ratios (88% for AUC and 97% for Cmax) for both exposure endpoints of BUP lay between 0.5–1.5, which indicates a good predictive capacity of the developed model.
BUP population model initial estimates: We have performed a stochastic simulation using our BUP model in steady state conditions after the administration of 4, 8 and 16 mg of BUP. This initial evaluation may estimate and quantify the proportion of patients that would show a trough level above 3 ng/mL in order to preliminarily propose the optimal dose level to be administered. The 3ng/mL threshold has been proposed as adequate BUP plasma levels for saturation of >70% of opioid receptors for accurate suppression of withdrawal symptoms in patients. According to these initial results, 0%, 0% and 35% of the simulated patients taking 4, 8 and 16 mg, respectively, would show a steady state concentration (Css) above the threshold limit of 3 ng/mL. These preliminary simulations suggest a high probability of withdrawal symptoms of patients receiving 4 and 8 mg of BUP.
BUP population model proposed dosing schedule: Two exploratory yet optimized BUP dosing schedules have been proposed able to achieve a 80% proportion of patients with Css ≥ 3 ng/mL: 8 mg of BUP three times a day (TID) or 16 mg of BUP twice-daily (BID) predict a probability of 81 and 93%, respectively, of patients with Css levels equal of greater than 3 ng/mL. Simulations of a once-daily schedule provide a wider range of concentrations for each analyte, whereas the twice-daily schedule guarantees a lesser fluctuation in concentration levels for each analyte, which may be relevant to avoid any safety concern and to promote adherence to dosing.
Next steps and proposed implementation strategies: we are currently conducting model refinement studies, expanding the number of recruited subjects for intense sampling, and increasing the dose range of recruited subjects. Patients’ response to the administered dose as a PD measure is also being added to the model.
We are proposing a randomized control trial as an initial implementation evaluation of the model in a real world setting in PR (OUD clinic) after model refinement activities are completed. We anticipate using steady state BUP plasma levels as an input to the model and propose a dosing strategy based on the threshold of 3ng/mL at t=0, 30, 60 and 90 days in the controlled arm; patients’ response to the proposed dose will be assessed also. Treatment arm participants will receive the current BUP pharmacotherapy standard of care for OUD in PR. A comparison will be approached in terms of adherence, illicit drug use, patient response to administered dose, among other clinical outcome variables.
Conclusions: This clinical implementation of a BUP dosing model as an optimization strategy for OUD in PR represents a first and unique description of the pharmacometrics of BUP. Despite the limitations, our study provides insight into specific BUP pharmacokinetic characteristics and the metabolic processes involved after its administration. The population PK model, which has been externally validated using published data across several studies, is a preliminary framework able to describe the time-course of BUP and metabolites (BUP-g, Nor-BUP and Nor-BUP-g). This study also proposed a model-informed dose selection strategy that suggests dose frequency adjustments to decrease the risk of withdrawal symptoms in the PR population. Implementation studies and current research efforts were presented as a roadmap to clinical uptake and sustainability of the model in OUD treatment protocols in PR.
References:
[1] Greenwald, M.; Johanson, C.-E.; Bueller, J.; Chang, Y.; Moody, D.E.; Kilbourn, M.; Koeppe, R.; Zubieta, J.-K. Buprenorphine duration of action: Mu-opioid receptor availability and pharmacokinetic and behavioral indices. Biol. Psychiatry 2007, 61, 101–110.
[2] Greenwald, M.K.; Johanson, C.-E.; Moody, D.E.; Woods, J.H.; Kilbourn, M.R.; Koeppe, R.A.; Schuster, C.R.; Zubieta, J.-K. Effects of buprenorphine maintenance dose on mu-opioid receptor availability, plasma concentrations, and antagonist blockade in heroin-dependent volunteers. Neuropsychopharmacology 2003, 28, 2000–2009.
[3] Greenwald, M.K.; Comer, S.D.; Fiellin, D.A. Buprenorphine maintenance and mu-opioid receptor availability in the treatment of opioid use disorder: Implications for clinical use and policy. Drug Alcohol Depend. 2014, 144, 1–11.
[4] Ng, C.M.; Dombrowsky, E.; Lin, H.; Erlich, M.E.; Moody, D.E.; Barrett, J.S.; Kraft, W.K. Population Pharmacokinetic Model of Sublingual Buprenorphine in Neonatal Abstinence Syndrome. Pharmacotherapy 2015, 35, 670–680.
[5] Nasser, A.F.; Heidbreder, C.; Gomeni, R.; Fudala, P.J.; Zheng, B.; Greenwald, M.K. A population pharmacokinetic and pharmacodynamic modelling approach to support the clinical development of RBP-6000, a new, subcutaneously injectable, long-acting, sustained-release formulation of buprenorphine, for the treatment of opioid dependence. Clin. Pharmacokinet. 2014, 53, 813–824.
[6] Yassen, A.; Olofsen, E.; Romberg, R.; Sarton, E.; Danhof, M.; Dahan, A. Mechanism-based pharmacokinetic-pharmacodynamic modeling of the antinociceptive effect of buprenorphine in healthy volunteers. Anesthesiology 2006, 104, 1232–1242.
[7] Yassen, A.; Olofsen, E.; Romberg, R.; Sarton, E.; Teppema, L.; Danhof, M.; Dahan, A. Mechanism-based PK/PD modeling of the respiratory depressant effect of buprenorphine and fentanyl in healthy volunteers. Clin. Pharmacol. Ther. 2007, 81, 50–58.