Dose Escalation Study to Determine the Pharmacokinetics of Atazanavir Administered With RIfampicin to HIV Positive Adults on sEcond-line ART Regimen With Suppressed HIV-1 Viral Load

Background and Justification

Both HIV and TB can successfully be managed using the currently available medications with the latter being curable. However, due to the detrimental drug-drug interaction (DDI) between RIF and the bPIs, there is a challenge in the management of the two conditions concurrently in persons who need both classes of drugs. RIF is a strong inducer of the hepatic enzyme cytochrome P450 (CYP) 34A which metabolizes several agents including the PI drugs thereby reducing their bioavailability. While rifabutin, an alternative rifamycin is associated with less DDI with bPI and maybe a better alternative to RIF in such circumstances, its cost, toxicity and lack of co-formulated preparations renders it to be a less likely sustainable option especially in LRS.

Though data published in 2008 showed a high incidence of adverse events when higher adjusted doses of LPV/r were used to overcome the DDI effect with RIF among healthy volunteers, data in 2012 among HIV-infected adults treated with adjusted doses of LPV/r co-administered with RIF-based TB treatment showed the drug was well tolerated with no significant adverse events. The major difference is that the 2012 study was conducted among HIV infected patients, that were already on normal doses of LPV/r and RIF at the time of enrolment who underwent PI dose escalation unlike the 2008 study among healthy volunteers who received the RIF and shortly after higher than normal doses of LPV/r in short succession with no step-wise phase to allow compensation. Similarly to LPV, ATV is affected by the inducing effect of RIF on the drug metabolizing enzyme cytochrome P450 (CYP) 34A, thereby significantly reducing its bioavailability to sub-therapeutic levels when co-administered in standard doses with RIF.

Several studies exploring the interaction between ATV/r and RIF have reached similar conclusions to the experience with LPV/r and RIF: A study in healthy volunteers given RIF first, then ATV/r first at standard dose with proposed dose escalation was halted when the first three participants developed hepatotoxicity. It was noted that the pre-induction of giving RIF first could have caused this problem. This argument is supported by a study conducted by Burger and colleagues where participants were first given ATV/r before the addition of RIF and where no hepatotoxicity occurred; in that study, the dose escalation evaluated was insufficient to overcome the interaction, but concluded that further evaluation was safe and appropriate. Similarly, Acosta and colleagues concluded that increased unboosted ATV was safe for co-administration with RIF, but that higher doses are required to overcome the DDI. No studies have yet evaluated this interaction in HIV-infected individuals.

It is notable that all of the currently available PIs have been associated with transient and usually asymptomatic elevations in serum aminotransferase levels, and ATV has been associated with mild-to-moderate elevations in indirect and total bilirubin concentration. All studied PIs are rare causes of clinically apparent, acute liver injury. The protease inhibitors are heterogeneous molecules with little structural similarity, most of which are peptide-like and resemble the short peptide that is cleaved by the viral protease; for this reason, the patterns of liver toxicity will differ between drugs of this class.

A recent study seeking to evaluate an increased dose of the bPI ritonavir boosted darunavir (DRV/r) in combination with rifampicin resulted in rates of hepatotoxicity which necessitated the trial to be halted on the advice of the IDSMB (CROI 2019 Abstract 81). These trial participants had been stable on treatment with LPV/r and were transitioned to DRV/r for the trial; the alteration in liver enzymes became apparent early on, and may have resulted in part from the change in bPI administered to the patient.

For the reasons detailed in the paragraphs above in consideration of the previous studies investigating either LPV/r, ATV/r or DRV/r in conjunction with RIF, we have chosen the study population of healthy HIV-infected patients who are already stable on treatment with standard dose of ATV/r, and who will have RIF introduced at a time when they will be at equilibrium as regards the ATV/r. Continuation on ATV/r at standard dose together with RIF for a two week period will enable a further equilibrium to have been established prior to escalation of the bPI dose. Both of these considerations may reduce the likelihood of hepatotoxicity, although frequent monitoring of safety bloods will be undertaken.

RIF also affects the bioavailability of DTG, however, this is effectively overcome by administering DTG twice daily as opposed to the standard once-a-day dose; the elevated dose of DTG has not been associated with concerns relating to liver toxicity.

As part of the VirTUAL consortium, physiologically-Based Pharmacokinetic (PBPK) modelling was done at the University of Liverpool to characterize bPI and RIF DDIs. Dosing strategies were identified / predicted to overcome the DDI between bPI and RIF that informed the ATV doses that shall be used in this trial to predict the most appropriate dose of ATV given with a standard dose RIF. From this work, it is predicted that increasing the dose of boosted ATV from ATV/r 300/100mg once daily to 300/100mg twice daily will achieve sufficient plasma concentrations when ATV/r is to be given in combination with RIF. The PBPK model utilised for this prediction was developed considering a mathematical description of the molecular and physiological processes underpinning the distribution of ATV/r and RIF. Experimental and clinical data were integrated into a computational framework developed using MATLAB and Simbiology and validated following international guidelines.

The RIF induction potential on Phase I enzyme and transporters expression, the RIF inhibitory effect on uptake transporters and the RTV inhibitory effect on CYP3A4 were all mathematically described for the simulation of the DDI. The PBPK model was qualified considering available clinical data for CYP3A4 probe (midazolam) and ATV ± RTV alone at different doses (400mg q.d unboosted, 300/100 mg q.d boosted) or in combination with RIF as previously described. The qualification was based on the calculation of the absolute average fold error (AAFE) and root mean squared error (RMSE) where appropriate. AAFE is a useful parameter to assess over or under-prediction of the model, values closer to 1 indicates a closer similarity with observed values. RMSE calculates the error between the predicted value and the observed value. RMSE is particularly sensitive to outliers and values closer to zero indicate a reliable prediction. The model was assumed to be qualified if the simulated values were within 2-fold of the mean reported values, AAFE <2 and RMSE <1 as per convention for the approach. Dose adjustment to overcome the effect of RIF on ATV/r was simulated considering available formulations with the overall of achieving ATV exposure comparable to established regimens (e.g. 300/100mg q.d) without RIF.

This study therefore aims to evaluate the steady state of the PBPK modelled adjusted doses of ATV given in different doses in HIV-1 infected adults on second line therapy with suppressed viral load so as to determine the most appropriate dose to be used concurrently with RIF. The study will evaluate baseline ATV/r concentrations in participants who have been treated with this agent for many months, then evaluate the impact of the addition of RIF, before evaluating ATV/r at the full modelled dose (300/100 twice daily) with a final step to consider the impact of increased doses of RIF (from standard dosing of 600 mg once daily to 1200 mg once daily).