Simultaneous Assessment of Time-to-positivity and Colony-forming Unit in tuberculosis patients under high-dose rifampicin therapy
Rami Ayoun Alsoud1, Robin J Svensson1, Elin M Svensson1,2, Stephen H Gillespie3 Martin J Boeree4,5, Andreas H Diacon6, Rodney Dawson7,8, Rob E Aarnoutse2, and Ulrika SH Simonsson1
1Department of Pharmaceutical Biosciences, Uppsala University, Sweden; 2Department of Pharmacy, Radboud Institute for Health Sciences, the Netherlands; 3School of Medicine, University of St. Andrews, United Kingdom; 4Department of Lung Diseases, Radboud University Medical Center, Nijmegen, and 5University Center for Chronic Diseases Dekkerswald, Groesbeek, the Netherlands; 6TASK Foundation, 7Division of Pulmonology, Department of Medicine, University of Cape Town, and 8University of Cape Town Lung Institute, Cape Town, South Africa
Objectives:
Colony forming unit (CFU) counting on solid agar has been used for more than three decades to measure anti-tuberculosis (TB) drug activity in sputum samples before, during, and after treatment. Mycobacterium tuberculosis has been hypothesized to exist in three states: fast-, slow- and non-multiplying i.e. persisters. Studies have shown that CFU only detects fast- and slow-multiplying subpopulations [1]. An alternative biomarker, time-to-positivity (TTP) in liquid medium (Mycobacterial growth indicator tube; MGIT), has been developed. In comparison to CFU, TTP is more sensitive as it is known to be reflective of more TB subpopulations than only the fast and slow states. The relationship between the CFU and TTP is not well characterized. Describing this relationship could allow the prediction of one biomarker using prior knowledge of the other. This is especially useful as TTP is a cheaper and less labour-intensive biomarker when compared to CFU. Furthermore, taking into account the three mycobacterial subpopulations enables the capture of the non-multiplying ‘persistent’ population thought to be responsible for late TB relapse.
The objective of this paper was to identify the relationship between TTP and CFU in rifampicin-treated TB patients and to develop a framework for predicting one of the biomarkers using the data from the other.
Methods:
Clinical trial data was obtained from the PanACEA HIGHRIF1 trial where an escalating dosing of rifampicin starting with 10 mg/kg and up to 40 mg/kg was used [2]. The trial included 83 previously untreated subjects with uncomplicated pulmonary tuberculosis and treatment lasted for 14 days, of which the first 7 days with rifampicin monotherapy were used. Initially, a CFU sub-model was developed by applying the Multistate Tuberculosis Pharmacometric (MTP) model [3,4], which was linked to the PK model from Svensson et al [5] in order to incorporate drug exposure response. All MTP disease model parameters were fixed except for a patient-specific Bmax, which controls the initial bacterial load and allows for the adjustment of the individual CFU baseline. Exposure-response relationships were evaluated on bacterial growth and killing of different bacterial sub-states. Subsequently, the CFU sub-model was linked to the TTP sub-model by Svensson et al [6] to generate the final CFU-TTP model. Initially, all parameters for this sub-model were fixed except the TTP Bmax, which is a liquid culture-specific parameter that controls the bacterial growth in the MGIT tube, and a hazard scale parameter. Model selection and evaluation were done using the objective function value (OFV), parameter uncertainties and diagnostic plots e.g. visual predictive checks (VPC). All modeling and simulation were performed in NONMEM version 7.4 (ICON, Hanover, MD, USA) along with R statistical software version 3.5.1.
Results:
During model building, the CFU baseline parameter (patient-specific Bmax) was estimated along with the inter-individual variability (IIV) in Bmax. Moreover, drug effect on the fast-multiplying subpopulation was best described using an on-off effect for growth with no effect on its death rate. On the other hand, an Emax model was the best descriptor of the kill rate of the slow-multiplying (semi-dormant) subpopulation as well as the inclusion of IIV in the Emax parameter. As for the non-multiplying (persisters) subpopulation, a linear kill model with IIV in slope was selected. The liquid culture-specific TTP Bmax and hazard scale parameters were also estimated during the CFU-TTP model building. Based on the aforementioned model evaluation tools, the final CFU-TTP model was shown to successfully predict the data.
Conclusions:
The CFU-TTP model was successfully developed using human clinical data. This model serves to link the two biomarkers CFU and TTP into one framework which further serves to predict one of the two biomarkers from the other in clinical trial simulations.
References:
[1] Dhillon, J., Fourie, P.B., Mitchison, D.A. Persister populations of Mycobacterium tuberculosis in sputum that grow in liquid but not on solid culture media. Journal of Antimicrobial Chemotherapy. 69(2): 437-440 (2013)
[2] Boeree, M.J., Diacon, A.H., Dawson, R., Narunsky, K., du Bois, J., Venter, A., Phillips, P.P., Gillespie, S.H., McHugh, T.D., Hoelscher, M., Heinrich, N., Rehal, S., van Soolingen, D., van Ingen, J., Magis-Escurra, C., Burger, D., Plemper van Balen, G., Aarnoutse, R.E. A dose-ranging trial to optimize the dose of rifampin in the treatment of tuberculosis. American Journal of Respiratory and Critical Care Medicine. 191(9): 1058-1065 (2015).
[3] Clewe, O., Aulin, L., Hu, Y., Coates, A.R.M. and Simonsson, U.S.H. A multistate tuberculosis pharmacometric model: a framework for studying anti-tubercular drug effects in vitro. Journal of Antimicrobial Chemotherapy. 71(4): 964-974 (2016).
[4] Svensson, R.J., Simonsson, U. Application of the Multistate Tuberculosis Pharmacometric Model in Patients With Rifampicin-Treated Pulmonary Tuberculosis. CPT: Pharmacometrics & Systems Pharmacology. 5(5): 264-273 (2016).
[5] Svensson, R.J., Aarnoutse, R.E., Diacon, A.H., Dawson, R., Gillespie, S.H., Boeree, M.J., and Simonsson, U.S.H. A Population Pharmacokinetic Model Incorporating Saturable Pharmacokinetics and Autoinduction for High Rifampicin Doses. Clinical Pharmacology & Therapeutics. 103(4): 674-683 (2018).
[6] Svensson, R.J., Svensson, E.M., Aarnoutse, R.E., Diacon, A.H., Dawson, R., Gillespie, S.H., Moodley, M., Boeree, M.J. and Simonsson, U.S.H. Greater Early Bactericidal Activity at Higher Rifampicin Doses Revealed by Modeling and Clinical Trial Simulations. The Journal of Infectious Diseases. 218(6): 991-999 (2018).