Model based evaluation of higher doses of rifampicin using a semi-mechanistic model incorporating auto-induction and saturation of first-pass hepatic extraction
Maxwell Chirehwa (1), Roxana Rustomjee (2), Thuli Mthiyane (3), Philip Onyebujoh (4), Peter Smith (1), Helen McIlleron (1), Paolo Denti (1)
(1) Division of Clinical Pharmacology, Department of Medicine, University of Cape Town, South Africa; (2) Strategic Health Innovation Partnerships (SHIP), South African Medical Research Council, Cape Town, South Africa; (3) South African Medical Research Council, Durban, South Africa; (4) Intercountry Support Team for East and Southern Africa, World Health Organization, Regional Office for Africa, Harare, Zimbabwe
Introduction: Rifampicin is a key drug in the treatment of tuberculosis. It is hepatically cleared and it undergoes extensive first-pass metabolism [1], whose saturation with larger doses has been reported since early pharmacokinetic (PK) studies [2]. Rifampicin also induces its own metabolism via Pregnane X Receptor (PXR) [3]. WHO currently recommends 8-12 mg/kg doses of rifampicin. However, higher doses which are likely to be more effective are being investigated.
Objectives: To develop a population model to describe rifampicin PK among tuberculosis patients accounting for the auto-induction of rifampicin clearance and saturation of first-pass hepatic extraction.
To explore changes in rifampicin exposures when doses are increased beyond the currently recommended range.
Methods: As previously described, blood samples were collected from 61 HIV/TB co-infected patients in South Africa who had a median age of 32 years (range: 18-47 years) and weight of 55 kg (range: 34-99 kg)[4]. Intensive PK sampling was performed on four occasions after initiation of treatment: day 1, 8, 15, and 29. On each occasion, samples were collected just prior to the dose and at 1, 2, 4, 6, 8, and 12 h after dose. 41 of the patients received efavirenz-based antiretroviral therapy starting on day 15. Weight band-based dosing for antituberculosis treatment was implemented according to WHO guidelines [5]. Plasma rifampicin was quantified using LC-MS/MS. The assay was validated for concentrations between 0.1-30 mg/L.
Data were analysed using nonlinear mixed effects modelling in NONMEM VII using first-order conditional estimation with eta-epsilon interaction (FOCE-I). For absorption, the models evaluated during model building included first-order absorption with and without delay with lag time or a series of transit compartments [6]. With regards to disposition, a one compartment with first-order elimination and a semi-mechanistic well-stirred hepatic model similar to the one used by Gordi et al. [7] were assessed. Auto-induction of clearance over time was described using an exponential model. To adjust for body size, allometric scaling [8] was applied to clearance, volume of central and liver compartments, and hepatic blood flow. The M6 method was applied to handle concentrations below the limit of quantification [9]. Model development was guided by change in objective function value (OFV) and diagnostic plots.
The final model was employed to simulate exposures at increasing doses using the demographic data of 870 tuberculosis patients from South Africa and West Africa (200 repetitions). Based on the current weight-bands, exposures were evaluated using doses equivalent to 1.5, 2, and 3.5 times the current dose. Twenty-four hour areas under the curve at steady-state (AUC0-24) were predicted and dose/exposure proportionality was assessed.
Results: Rifampicin PK was best described using a transit compartment absorption and a well-stirred liver model with saturation of (first-pass) hepatic extraction. For a typical individual, volume of the liver compartment was assumed to be 1 L, and the liver plasma flow 50 L/h. Free fraction of unbound rifampicin was fixed to 20% (2). Saturation of first-pass metabolism was parameterized using a Michaelis-Menten saturation model and hepatic metabolism was found to have a maximum CLint of 98 L/h at baseline with a Km of 3.2 mg/L.
Large between-occasion variability was detected for ka and absorption mean transit time, while between-subject variability was relatively low for both clearance and volume of distribution (<25%). Fat free mass was the best descriptor of body size for all clearance and volume parameters. Increase of hepatic CLint over time was characterised in the model and it was estimated CLint approximately doubles from baseline to steady state with a half-life of the auto-induction of around 4 days.
As expected, the current model predicts that increases in dose of rifampicin result in more-than-linear increases in drug exposures as measured by AUC0-24. Our simulations show that giving patients 20 mg/kg rather than 10 mg/kg results in 3.2 times higher AUC and further increasing the dose to 35 mg/kg (3.5 times), the AUC becomes 8.4 times higher. Simulation results from this study are predictive of the observed exposures recently reported in studies using higher doses of rifampicin by Aarnoutse et al. [10]. They reported a 3.1 times higher AUC when giving patients 1200 mg compared to 600 mg dose, closely in line with our model predictions.
Conclusion: We developed a model for rifampicin PK that characterises auto-induction of clearance and saturation of metabolism, which is evident on first-pass extraction even at the current doses. The model correctly predicts that increasing the dose of rifampicin results in a more than proportional increase in drug exposure. Our simulations with a 20 mg/kg doses produce results closely in line with those from recent clinical trials.
References:
[1] Loos U, Musch E, Jensen JC, Mikus G, Schwabe HK, Eichelbaum M. 1985. Pharmacokinetics of oral and intravenous rifampicin during chronic administration. Klin. Wochenschr. 63:1205–1211.
[2] Acocella G. 1978. Clinical pharmacokinetics of rifampicin. Clin. Pharmacokinet. 3:108–27.
[3] Chen J, Raymond K. 2006. Roles of rifampicin in drug-drug interactions: underlying molecular mechanisms involving the nuclear pregnane X receptor. Ann. Clin. Microbiol. Antimicrob. 5:3.
[4] McIlleron H, Rustomjee R, Vahedi M, Mthiyane T, Denti P, Connolly C, Rida W, Pym A, Smith PJ, Onyebujoh PC. 2012. Reduced antituberculosis drug concentrations in HIV-infected patients who are men or have low weight: implications for international dosing guidelines. Antimicrob. Agents Chemother. 56:3232–8.
[5] World Health Organization. 2003. Treatment of tuberculosis: Guidelines for national programmes (3rd edition). WHO/CDS/TB/2003.313. Gevena, Switzerland.
[6] Savic R, Jonker D, Kerbusch T, Karlsson M. 2007. Implementation of a transit compartment model for describing drug absorption in pharmacokinetic studies. J Pharmacokinet Pharmacodyn 34:711–726.
[7] Gordi T, Xie R, Huong N V, Huong DX, Karlsson MO, Ashton M. 2005. A semiphysiological pharmacokinetic model for artemisinin in healthy subjects incorporating autoinduction of metabolism and saturable first-pass hepatic extraction. Br. J. Clin. Pharmacol. 59:189–98.
[8] Anderson BJ, Holford NHG. 2008. Mechanism-based concepts of size and maturity in pharmacokinetics. Annu. Rev. Pharmacol. Toxicol. 48:303–32.
[9] Beal SL. 2001. Ways to fit a PK model with some data below the quantification limit. J. Pharmacokinet. Pharmacodyn. 28:481–504.
[10] Aarnoutse RE, Kibiki GS, Reither K, Phillips P, Hoelscher M, Gillespie SH, Plemper van Balen G, Boeree MJ. 2014. HIGHRIF2: a phase II trial comparing 10, 15 and 20 mg/kg rifampicin for two months. 7th International Workshop on Clinical Pharmacology of TB Drugs.
