Interaction of Rifampin and Darunavir-Ritonavir or Darunavir-Cobicistat In Vitro

ABSTRACT

Treatment of HIV-infected patients coinfected with Mycobacterium tuberculosis is challenging due to drug-drug interactions (DDIs) between antiretrovirals (ARVs) and antituberculosis (anti-TB) drugs. The aim of this study was to quantify the effect of cobicistat (COBI) or ritonavir (RTV) in modulating DDIs between darunavir (DRV) and rifampin (RIF) in a human hepatocyte-based in vitro model. Human primary hepatocyte cultures were incubated with RIF alone or in combination with either COBI or RTV for 3 days, followed by coincubation with DRV for 1 h. The resultant DRV concentrations were quantified by high-performance liquid chromatography with UV detection, and the apparent intrinsic clearance (CL_int.app.) of DRV was calculated. Both RTV and COBI lowered the RIF-induced increases in CL_int.app. in a concentration-dependent manner. Linear regression analysis showed that log₁₀ RTV and log₁₀ COBI concentrations were associated with the percent inhibition of RIF-induced elevations in DRV CL_int.app., where β was equal to −234 (95% confidence interval [CI] = −275 to −193; P < 0.0001) and −73 (95% CI = −89 to −57; P < 0.0001), respectively. RTV was more effective in lowering 10 μM RIF-induced elevations in DRV CL_int.app. (half-maximal [50%] inhibitory concentration [IC₅₀] = 0.025 μM) than COBI (IC₅₀ = 0.223 μM). Incubation of either RTV or COBI in combination with RIF was sufficient to overcome RIF-induced elevations in DRV CL_int.app., with RTV being more potent than COBI. These data provide the first in vitro experimental insight into DDIs between RIF and COBI-boosted or RTV-boosted DRV and will be useful to inform physiologically based pharmacokinetic (PBPK) models to aid in optimizing dosing regimens for the treatment of patients coinfected with HIV and M. tuberculosis.

INTRODUCTION

Approximately 25% of human immunodeficiency virus (HIV) type 1 (HIV-1)-infected patients worldwide are coinfected with Mycobacterium tuberculosis (1, 2) (referred to here as HIV-tuberculosis [TB] patients), with such coinfections accounting for 390,000 deaths in 2014 (3). The clinical management of HIV-TB patients presents significant challenges, especially in resource-limited settings (2, 4), where virological failure or intolerance to first-line antiretroviral therapy requires the use of HIV protease inhibitors (PIs) (5). PIs largely undergo phase I metabolism by cytochrome P450 3A4 (CYP3A4) and are also substrates of P-glycoprotein (P-GP; ABCB1) (6). Consequently, PIs are commonly administered in combination with pharmacokinetic (PK) boosters, such as ritonavir (RTV) or cobicistat (COBI), which act by inhibiting CYP3A4-mediated PI metabolism and P-GP-mediated PI efflux, thereby improving the PK profile of PIs by prolonging PI half-life and increasing PI bioavailability (7–9).

Rifampin (RIF) is an essential component of short-course anti-TB treatment regimens (2, 10); however, RIF is also a potent inducer of the expression and activity of several metabolic enzymes, including CYP3A4 (11). The coadministration of RIF with PIs can result in clinically significant drug-drug interactions (DDIs), whereby PI bioavailability may be significantly reduced (>75%) (10, 12–14). Consequently, administering standard doses of RTV-boosted PIs to HIV-TB patients receiving RIF is contraindicated under the current World Health Organization (WHO) guidelines (15). The search for effective second-line therapeutic options for the treatment of patients coinfected with HIV and M. tuberculosis is therefore a research priority (16).

Darunavir (DRV) is chiefly metabolized by CYP3A4 (17), and coadministration of a low dose of either RTV or COBI together with DRV increases the systemic bioavailability of DRV (18, 19). In addition, the high barrier to genetic resistance, as well as the tolerability, safety profile, and potency of DRV when it is administered in combination with a low dose of either RTV (DRV/r) or COBI (DRV/c), has made these fixed-dose combinations important options for the treatment of HIV-infected patients (20–22).

Previous studies have demonstrated a markedly reduced exposure to RTV-boosted PIs, including atazanavir (ATV) (12), indinavir (IDV) (13), and lopinavir (LPV) (14), as well as an increased risk of hepatotoxicity, when RIF is coadministered with these drugs in healthy volunteers. For this reason, studies aimed at investigating DDIs between DRV/r and RIF in HIV-negative subjects have not been undertaken. Similarly, the extent of the DDI between DRV/c and RIF remains unknown. A recent population PK (pop-PK) analysis showed that it was possible to offset the effects of RIF on the DRV trough concentration (C_trough) by increasing the dose of DRV/r administered (23), which raises the possibility that RTV may overcome potential DDIs between DRV and RIF in vitro and in vivo. The aim of the present study was to quantify, using an in vitro model, the extent of DDIs arising from coincubation of RIF with either RTV or COBI by specifically measuring the apparent intrinsic clearance (CL_int.app.) of DRV by primary human hepatocytes.

(This work was presented in part at the Conference on Retroviruses and Opportunistic Infections [CROI], Boston, MA, USA, 22 to 25 February 2016 [24].)

RESULTS

Assessment of CL_int.app. of darunavir following incubation of primary human cryopreserved hepatocytes with the combination of ritonavir and rifampin.

Primary human hepatocytes are commonly used as a tool to predict the hepatic metabolic clearance of xenobiotics and DDIs in vitro (25, 26). Using this model system, the CL_int.app. of DRV was initially calculated under control conditions in which hepatocytes (from donors HU1399, HU1587, and HU1621) were incubated with DRV alone. Experiments aimed at determining the degree of protein binding of DRV within Williams’ medium E (WME) incubation medium revealed that the mean fraction unbound (f_u) of DRV was 0.76 ± 0.07 (n = 3) (Table 1). Following correction for DRV protein binding in WME incubation medium, under control conditions, the mean DRV CL_int.app. was 10.5 ± 3.8 μl/min/10⁶ hepatocytes (n = 3). Incubation of human hepatocytes with RIF over 72 h has previously been shown to induce CYP3A4 enzymatic activity (27, 28). Similarly, in this model system, incubation of hepatocytes with RIF was sufficient to markedly increase the CL_int.app. of the CYP3A4 substrate DRV at each concentration of RIF tested (0.5 to 20 μM; Fig. 1). The maximal RIF-induced increase (1.9 ± 0.3-fold; n = 3) in the DRV CL_int.app. was observed with 10 μM RIF (Fig. 1).

TABLE 1.

Fraction unbound (f_u) values in WME incubation medium of each compound used in this study

Compound	fu in WME incubation mediuma
COBI	0.53 ± 0.04
DRV	0.76 ± 0.07
RIF	0.63 ± 0.05
RTV	0.14 ± 0.01

^aThe data are means ± SDs (n = 3).

FIG 1.

Effects of rifampin alone or in combination with ritonavir on the mean DRV CL_int.app. in primary human hepatocytes in vitro. Cryopreserved primary human hepatocytes were incubated with RIF alone (0.5 to 20 μM) (hatched bars) or with RIF (0.5 to 20 μM) together with RTV (0.01 to 10 μM) (gray bars) for 72 h. All cells were then incubated with RIF (0.5 to 20 μM) or RIF (0.5 to 20 μM) together with RTV (0.01 to 10 μM), as described above, together with DRV (5 μM) for 60 min. Control cells were treated with DRV (5 μM) alone for 60 min (black bar). The results shown represent the mean DRV CL_int.app. from three biological replicates measured in hepatocytes from three independent donors (donors HU1399, HU1587, and HU1621). Error bars indicate SDs.

Coincubation of RIF with RTV reduced the 10 μM RIF-induced increases in CL_int.app. in an RTV concentration-dependent manner (Fig. 1). Notably, 1 μM RTV was sufficient to overcome the effect of 10 μM RIF on DRV CL_int.app., reducing the DRV CL_int.app. to 0.78 ± 0.25-fold, equivalent to −22% of the levels observed under control conditions, in which cells were treated with DRV alone (n = 3) (Fig. 1). Increasing the concentrations of RIF above 10 μM (12.5 to 20 μM) did not impact the effectiveness of the ability of RTV to overcome the RIF-induced elevation in DRV CL_int.app. (Fig. 1). Specifically, 1 μM RTV lowered the 12.5 μM RIF-induced, and the 20 μM RIF-induced, increases in DRV CL_int.app. by 55%, and 47%, respectively, to 8.6 ± 3.2 μl/min/10⁶ hepatocytes (n = 3) and 8.8 ± 3.4 μl/min/10⁶ hepatocytes (n = 3).

Assessment of the CL_int.app. of darunavir following incubation of primary human cryopreserved hepatocytes with the combination of cobicistat and rifampin.

In a separate set of experiments, human hepatocytes from three individual donors (donors HU1399, HU1574, and HU1587) were used to determine the effects of incubating RIF together with COBI upon the CL_int.app. of DRV. Under control conditions, where primary human cryopreserved hepatocytes were incubated with DRV alone, the DRV CL_int.app. was 13.2 ± 1.8 μl/min/10⁶ hepatocytes (n = 3). Incubation of hepatocytes with RIF (0.5 to 20 μM) induced a mean increase in the DRV CL_int.app. of 55.8%. In cells treated with 1 μM RIF, coincubation with the lowest concentration of COBI tested (0.42 μM) was effective in lowering the RIF-induced DRV CL_int.app. by 36.9%, yielding a DRV CL_int.app. of 12.2 ± 2.8 μl/min/10⁶ hepatocytes (n = 3). Hepatocytes treated with 10 μM RIF exhibited a DRV CL_int.app. of 21.6 ± 2.6 μl/min/10⁶ hepatocytes (n = 3). COBI induced a concentration-dependent attenuation of the DRV CL_int.app. elicited by 10 μM RIF (Fig. 2), with 1.28 μM COBI being sufficient to lower the DRV CL_int.app to 11.6 ± 2.6 μl/min/10⁶ hepatocytes (n = 3), 13% below the level obtained under control conditions with DRV alone. COBI was also effective at reducing the CL_int.app. elevations induced by higher concentrations of RIF, as coincubation with 1.28 μM COBI reduced the 20 μM RIF-induced elevation in the DRV CL_int.app. by 46% (12.4 ± 3.9 μl/min/10⁶ hepatocytes; n = 3).

FIG 2.

Effects of rifampin alone or in combination with cobicistat on the mean DRV CL_int.app. in primary human hepatocytes in vitro. Cryopreserved primary human hepatocytes were incubated with RIF (0.5 to 20 μM) (hatched bars) or with COBI (0.13 to 12.76 μM) and RIF (0.5 to 20 μM) (gray bars) for 72 h. All cells were then incubated with RIF (0.5 to 20 μM) or RIF (0.5 to 20 μM) together with COBI (0.13 to 12.76 μM), as described above, together with DRV (5 μM) for 60 min. Control cells were treated with DRV (5 μM) alone for 60 min (black bar). The results shown represent the mean DRV CL_int.app. from three biological replicates measured in hepatocytes from three independent donors (donors HU1399, HU1574, and HU1587). Error bars indicate SDs.

Comparison of cobicistat- and ritonavir-mediated reduction of the rifampin-induced increase in the CL_int.app. of darunavir.

To compare the relative effectiveness of RTV and COBI in attenuating RIF-induced increases in DRV CL_int.app., the percent inhibition of the 10 μM RIF-induced elevations in DRV CL_int.app. achieved by coincubation with COBI (0.13 to 12.76 μM), or RTV (0.1 to 10 μM), was determined by comparison to that achieved under control conditions where cells were treated with 10 μM RIF alone (Fig. 3). Following correction for protein binding, the half-maximal (50%) inhibitory concentrations (IC₅₀s) of COBI and RTV, calculated from the percent change in the DRV CL_int.app. obtained under these conditions, were 0.223 μM and 0.025 μM, respectively (Fig. 3). In addition, the maximal inhibition of 10 μM RIF-induced elevations achieved by COBI and RTV were different, with RTV eliciting a 69.5% inhibition of the 10 μM RIF-induced increase in DRV CL_int.app., while the COBI-mediated reduction in the 10 μM RIF-induced increase in the CL_int.app. of DRV was 56.9% (P = 0.05).

FIG 3.

Comparative effectiveness of COBI and RTV at lowering the RIF-induced increase in the CL_int.app. of DRV in human primary hepatocytes in vitro. The graph shows the relative effects of COBI and RTV on the inhibition of 10 μM RIF-induced elevations in DRV CL_int.app. in cryopreserved primary human hepatocytes. Cells were coincubated with RTV (starting total concentrations, 0.1 to 10 μM; donors HU1399, HU1587, and HU1621) or COBI (starting total concentrations, 0.13 to 12.76 μM; donors HU1399, HU1574, and HU1587) in combination with RIF (10 μM) for 72 h prior to coincubation with DRV (5 μM) for 1 h. Each condition was tested in triplicate with cells from each donor. The concentrations of RTV and COBI plotted represent the unbound concentrations present in WME incubation medium following correction for protein binding. The untreated control (blank) was assigned a value of 0.001 μM in each case. Error bars indicate SEMs.

Following data normalization and correction for protein binding, linear regression analysis of the effects of RTV, or COBI, in combination with RIF, at each concentration tested on the percent change in DRV CL_int.app. showed an association between the log₁₀ RTV unbound concentrations and the percent inhibition of RIF-induced DRV CL_int.app. and between the log₁₀ COBI unbound concentrations and the percent inhibition of the RIF-induced DRV CL_int.app., where β was equal to −234 (95% confidence interval [CI] = −275 to −193; P < 0.0001) and −73 (95% CI = −89 to −57; P < 0.0001), respectively. Linear regression analysis of the effects of RIF on DRV CL_int.app. revealed that RIF exerted a similar effect on DRV CL_int.app. in the two independent sets of RTV and COBI experiments, with a positive association being observed between the unbound concentration of RIF and DRV CL_int.app., where β was equal to 19 (95% CI = 4 to 34; P = 0.017), and 16 (95% CI = 4 to 29; P = 0.013), in the RTV experiments, and COBI experiments, respectively.

DISCUSSION

RIF strongly induces the expression of metabolic enzymes, such as CYP3A4 (29–31), and can also induce the activity of drug transporters (32). Collectively, this can result in clinically relevant DDIs in patients that receive RIF together with other medications (11, 33). These DDIs present challenges for the treatment of HIV-TB patients, as several therapeutic options are contraindicated due to known DDIs (10), while other potentially viable treatment regimens may be either delayed or avoided completely, due to hypothetical DDIs that are predicted to occur between anti-TB drugs and antiretrovirals (ARVs), such as PIs. For example, the coadministration of the standard dose of any PI with RIF is currently contraindicated under WHO guidelines (15), but the extent of potential DDIs between RIF and PIs has not been determined for all PIs, including DRV. Currently, coadministration of dose-adjusted RTV-boosted LPV or RTV-boosted saquinavir together with RIF is indicated, albeit with the caveat that high levels of toxicity can occur. This raises the possibility that administration of other PIs, such as RTV- or COBI-boosted DRV, together with RIF may also be feasible. The present study addresses this issue by providing the first experimental insight into the effects of the coincubation of either RTV or COBI together with RIF on the DRV CL_int.app. in a human hepatocyte-based in vitro model of drug metabolism.

The utilization of human hepatocytes to predict hepatic metabolic clearance of xenobiotics is well established (25, 26). In this study, incubation of cryopreserved human hepatocytes with RIF increased the CL_int.app. of DRV (Fig. 1 and 2). This was likely due to the induction of CYP3A4 (17, 34), although the effects of RIF on transporters may also be important (26). Uptake transporters, such as organic anion transporting polypeptide isoform 1B1 (OATP1B1) (35), and efflux transporters, such as P-GP (36), have been shown to play a role in PI elimination and therefore may also be relevant in the DDIs between RIF and COBI- or RTV-boosted DRV. Indeed, RIF has been shown to inhibit OATP1B1 (37), and DRV uptake by OATP1B1 and OATP1B3 in transfected CHO cells has also been reported (38). Utilizing a pop-PK model, it has been suggested that OATP3A1 polymorphisms are associated with DRV PK (39). In addition, a recent physiologically based PK (PBPK) modeling-based study that investigated the PK of DRV/r during pregnancy has also suggested a role for hepatic transporters in DRV disposition (40).

Coincubation of human cryopreserved hepatocytes with COBI and RIF or RTV and RIF, using concentrations spanning the in vivo therapeutic range of these compounds, revealed that both RTV and COBI could reduce the RIF-induced elevation in DRV CL_int.app. in a concentration-dependent manner (Fig. 1 and 2). RTV was more effective than COBI at attenuating the RIF-induced increase in DRV CL_int.app., with RTV exhibiting a lower IC₅₀ than COBI, while RTV also achieved greater maximal inhibition of the 10 μM RIF-induced increase in the CL_int.app. of DRV than COBI (Fig. 3). Furthermore, regression analysis revealed a stronger effect of RTV than COBI for their relative contribution in reducing the RIF-induced increases in the CL_int.app. of DRV. Due to the more recent approval of COBI, data regarding potential DDIs between COBI and other medications are more limited than those for RTV. The expected differential DDI profiles of COBI and RTV when they are administered with comedications have recently been reviewed (41, 42). RTV and COBI both serve as mechanism-based inhibitors of CYP3A4 in vivo (43, 44); however, RTV is also known to induce the expression of various metabolic enzymes, including CYP3A4, in primary human hepatocytes in vitro (30). Very few studies aimed at investigating the relative effects of COBI as an inducer of metabolic enzyme expression have thus far been conducted, although it has been suggested that the induction potential of COBI is less than that of RTV (45) and that COBI is not expected to induce CYP3A4 expression (46). It was recently suggested that the hepatic uptake of RTV occurs chiefly by passive diffusion (47). In addition, RTV has been shown to induce the expression of the efflux transporters P-GP (30) and multidrug resistance-associated protein 1 (MRP1; ABCC1) in primary human hepatocytes in vitro (30). DRV is a substrate of P-GP (48) and of OATP1A2 and OATP1B1 (35), while RTV appears to inhibit P-GP (48), as well as OATP1B1 and OATP1B3 (38), in vitro. RTV is also reported to be a substrate of P-GP (49). At the same time, RIF has been described to be both a substrate and an inhibitor of OATP1B1 and OATP1B3 in vitro (50). In addition, chronic exposure to RIF has been shown to exert an inhibitory effect on P-GP in vitro (51), while the RIF-induced induction of P-GP/ABCB1, OATP1B1, and ABCC2 expression has also been reported (52). It remains to be seen, therefore, what the net contribution of transporters such as OATP1B1, OATB1B3, and P-GP may be on the plasma levels of DRV in vivo, especially when DRV is administered in combination with other compounds, such as RIF.

The PK profiles of DRV/r (800/100 mg once a day [q.d.]) and DRV/c (800/150 mg q.d.) in HIV-infected patients are broadly similar (53, 54). However, in a study conducted in healthy volunteers, it has been reported that the minimum concentration (C_min) of DRV is 30% lower in individuals treated with DRV/c than individuals treated with DRV/r (55). In addition, PK analysis of the PI tipranavir (TPV), when administered in combination with COBI or RTV in healthy volunteers, showed that the TPV area under the concentration-time curve, maximum concentration in plasma, and C_tau (concentration in plasma at the end of the dosing interval) were significantly lower with COBI than with RTV (56). Collectively, these studies suggest that the pharmacoenhancment achieved with COBI is not always equal to that achieved with RTV.

While no studies investigating the effects of the coadministration of either DRV/r or DRV/c with RIF on DRV bioavailability have been conducted, it has recently been shown, using a pop-PK modeling approach, that the administration of dose-adjusted DRV/r (1,600/200 mg q.d., 800/100 mg twice a day [b.i.d.], or 1,200/150 mg b.i.d.) can potentially overcome the effects of RIF on the DRV trough concentration, albeit with the caveat that RTV-related side effects may occur and that a higher pill burden would be required (23). These in silico findings are in general agreement with the in vitro outcomes of the present study. However, extrapolation of the in vivo significance of in vitro data presents multiple challenges (57, 58), and it is difficult to directly infer how the results of the current study may translate in vivo. For example, increasing the dose of RTV when it is used in combination with a given PI is not always sufficient to overcome the effects of RIF. Indeed, a study of the effects of RIF on the steady-state PK of ATV with RTV in healthy volunteers showed that the administration of ATV-RTV at 300/100 mg, ATV-RTV at 300/200 mg, and ATV-RTV at 400/200 mg was insufficient to completely overcome the inductive potential of RIF at 600 mg (12). In an effort to better understand the absorption, distribution, metabolism, and elimination of various compounds, the use of PBPK models has recently gained popularity (59). Various PBPK models have been developed and have proved useful in predicting the effects of the administration of ARVs in HIV-infected patients with comorbidities (60). Indeed, a recent study described the development of a PBPK model for predicting clinical DDIs from RIF-based in vitro human hepatocyte data (61), and it is therefore hoped that the data presented herein will be of use in the development of PBPK models to predict the effects of coadministering boosted PIs with anti-TB drugs.

In conclusion, the results presented here provide insight into the relative effects of RTV and COBI as pharmacoenhancers of DRV when coincubated with RIF in an in vitro model of drug metabolism. This information can be used in conjunction with PBPK models to rationalize future strategies aimed at optimizing treatment regimens for HIV-TB patients. Further work should aim to elucidate the mechanisms that give rise to the differential inhibitory potential of COBI and RTV demonstrated here, as well as to validate these results in vivo. Future studies should also aim to further evaluate the effects of COBI and RTV on the expression of genes encoding metabolic enzymes, as well as the effects of these compounds on the expression and activity of various drug transporters in vitro. Finally, it would also be of interest to use this model system to evaluate potential DDIs that may occur between RIF and RTV or COBI in combination with other PIs or with other comedications.

MATERIALS AND METHODS

Chemicals.

DRV (catalog no. S1620) and COBI (catalog no. S2900) were purchased from Selleckchem (Munich, Germany). RIF (catalog no. R3501), RTV (catalog no. SML0491), potassium phosphate monobasic (catalog no. P0662), methanol (catalog no. 34860), and acetonitrile (catalog no. 34967) were purchased from Sigma-Aldrich (Poole, UK). Orthophosphoric acid (catalog no. 153154D) was purchased from VWR (Lutterworth, UK). High-performance liquid chromatography (HPLC)-grade water was produced by the use of an Elga PureLab system (Veolia Water Technologies, High Wycombe, UK).

Primary hepatocytes.

Cryopreserved primary human hepatocytes were purchased from Thermo Fisher Scientific (catalog no. HMCPIS; Inchinnan, Scotland). Hepatocytes from a total of four donors were used (Table 2). All cells used in this study were used in compliance with the Human Tissue Act (2004), as regulated by the Human Tissue Authority.

TABLE 2.

Donor information for cryopreserved primary human hepatocytes used

Donora	Sex	Age (yr)	Medicationsb	Drug or substance use
HU1399	Female	72	Insulin glargine, 10 units q.d.; metoprolol, 100 mg q.d.; lisinopril hydrochlorothiazide, 20/12.5 mg q.d.; calcium and vitamin D, 500 mg q.d.; multivitamin, q.d.; aspirin, 81 mg q.d.	Historic long-term tobacco use
HU1574	Male	70	Atorvastatin, 80 mg q.d.; lisinopril, 5 mg q.d.; aspirin, 81 mg q.d.; tamsulosin, 4 mg q.d.	None reported
HU1587	Female	43	Vitamin D oral; multivitamin oral; calcium, vitamin D, and vitamin K	None reported
HU1621	Male	66	Pazopanib, 800 mg q.d.	Rare alcohol use, historic tobacco use

^aAll donors were Caucasian.

^bInformation regarding medications used by donors was obtained from certificates of analysis provided by ThermoFisher Scientific, which are available upon request at https://www.thermofisher.com/uk/en/home/support/certificate-of-analysis-request-form.html.

Stock solutions.

Stock solutions of COBI, DRV, RIF, and RTV at concentrations 6,443, 1,684.3, 15,000, and 6,935.4 μM, respectively, were freshly prepared in 100% (vol/vol) methanol. Prior to use in experiments, all stock solutions were sterile filtered through a Millex 0.22-μm-pore-size polyethersulfone membrane (catalog no. SLGP033RS; Millipore, Watford, UK) and either were used immediately or were stored at −20°C for up to 5 days prior to use.

Concentrations of drugs used in this study.

Primary cryopreserved human hepatocytes were treated with a range of concentrations of the test compounds: COBI at 0.13 to 12.76 μM, RIF at 0.50 to 20.00 μM, and RTV at 0.01 to 10.00 μM, spanning their respective therapeutic concentration ranges in the plasma of humans, as determined from clinical PK data (62, 63). The concentration of DRV used in the experiments (5 μM) was selected from a value within the therapeutic range and close to the C_min of DRV (for DRV/r at 600/100, the C_min is 3.58 ± 1.15 μg/ml, which is equal to 6.03 ± 1.94 μM [34]; for DRV/c at 800/150, the C_min is 2.40 ± 1.22 μg/ml, which is equal to 4.04 ± 2.05 μM [46]), as obtained from PK data supplied on package inserts (34, 46). Unless otherwise stated, the starting drug concentrations quoted within this study refer to the starting total drug concentration present in each case, without adjustment for protein binding. After adjustment for protein binding in Williams' medium E (WME) incubation medium (Table 1), the starting unbound concentrations of the test compounds used were as follows: for COBI, 0.068 to 6.761 μM; for DRV, 3.800 μM; for RIF, 0.315 to 12.600 μM; and for RTV, 0.001 to 1.400 μM.

Culture of primary human hepatocytes.

Primary cryopreserved human hepatocytes were thawed in cryopreserved hepatocyte recovery medium (CHRM; catalog no. CM7000; Thermo Fisher Scientific) and were resuspended in WME plating medium (catalog no. A1217601; Thermo Fisher Scientific) supplemented with a hepatocyte plating supplement pack (catalog no. CM3000; Thermo Fisher Scientific). Cell viability was determined using a NucleoCounter NC-100 apparatus (Sartorius Ltd., Epsom, UK). Viable cells were plated on collagen-coated 96-well cell culture plates (catalog no. CM1096; Thermo Fisher Scientific) at a density of 6.5 × 10⁴ cells per well in 110 μl of WME plating medium. Hepatocytes were incubated at 37°C in a humidified incubator containing 5% (vol/vol) CO₂ for 5 h, prior to removal of the WME plating medium and overlaying of the hepatocyte monolayer with 70 μl per well of Geltrex lactose dehydrogenase-elevating virus-free reduced growth factor basement membrane matrix (catalog no. A1413202; Thermo Fisher Scientific) diluted in WME incubation medium (catalog no. A1217601; Thermo Fisher Scientific) supplemented with an hepatocyte maintenance supplement pack (catalog no. CM4000; Thermo Fisher Scientific) to a final concentration of 0.35 mg/ml. The cells were then incubated in a humidified incubator at 37°C containing 5% (vol/vol) CO₂ for 24 h, prior to removal of the WME incubation medium and replacement with 110 μl of fresh WME incubation medium containing test compounds: COBI (0.128 to 12.76 μM) together with RIF (0.5 to 20 μM) or RTV (0.01 to 10 μM) together with RIF (0.5 to 20 μM). As a control, hepatocytes were incubated with methanol (0.3%, vol/vol) in WME incubation medium. At 24 h and 48 h after the initial treatment, the WME incubation medium containing the test compounds was removed and replaced with fresh WME incubation medium containing the test compounds. At 72 h after the initial treatment, all cells were incubated with the test compounds together with DRV (5 μM) in WME incubation medium for 60 min.

Quantification of darunavir by HPLC-UV.

Following 60 min of incubation of the hepatocytes with the test compounds together with 5 μM DRV, 100 μl of WME incubation medium was removed from each well and was transferred to Corning Pyrex borosilicate glass tubes (75 by 12 mm; catalog no. KC350; Appleton-Woods) containing 300 μl of 100% acetonitrile. Standards and quality control samples were prepared in WME incubation medium and were treated in the same way. All samples were then vortexed for 5 s and were dried in a Jouan RC10.22 vacuum centrifuge for 6 h at room temperature (18 to 25°C). After the samples were dried, they were reconstituted in 330 μl of 20% (vol/vol) acetonitrile in H₂O. One hundred microliters of the resultant suspension was used to quantify DRV by HPLC with UV detection (HPLC-UV).

Chromatographic separation of DRV was achieved using a Waters Atlantis T3 column (4.6 by 100 mm; particle size, 3 μm; Waters, Elstree, UK) equipped with a Fortis C₁₈ guard (10 by 4 mm; particle size, 3 μm; Fortis Technologies Ltd., Chester, UK). A Dionex P680 HPLC pump, a Dionex ASI-100 automated sample injector, and a Dionex UVD170U UV detector (Thermo-Fisher Ltd., Hemel-Hempstead, UK) were used. Mobile phases C (25 mM KH₂PO₄, pH 3.3, orthophosphoric acid) and D (100% acetonitrile) were used in a step-gradient elution, as follows: 70% mobile phase C and 30% mobile phase D from 0.0 to 1.5 min, 35% mobile phase C and 65% mobile phase D from 1.5 to 7.0 min, 20% mobile phase C and 80% mobile phase D from 7.0 to 9.5 min, and 70% mobile phase C and 30% mobile phase D from 9.5 to 12.5 min. Elution was carried out at room temperature (18 to 25°C), and the flow rate was maintained at 1.00 ml/min. DRV was quantified at 267 nm, and chromatograms were analyzed using Chromeleon software (version 6.8; Thermo-Fisher Ltd.). Each experimental condition was assessed in triplicate. The lower limit of quantification (LOQ) of DRV was determined to be 0.156 μM. The assay was linear between 0.156 μM and 10 μM (upper LOQ). The mean coefficient of variability (CV) of the intraday precision was 2.6%, while the mean CV of the intraday accuracy was 2.0%. The mean CV of the interday precision was 2.2%, and the mean CV of the interday accuracy was 1.2%. The mean recovery of DRV from WME was 96.1%.

Measurement of protein binding of drugs in Williams' medium E incubation medium.

The degree of binding of COBI, DRV, RIF, or RTV to WME incubation medium was determined using a rapid equilibrium dialysis (RED) base plate (catalog no. 90004; Thermo Fisher Scientific) fitted with RED device inserts (catalog no. 89810; Thermo Fisher Scientific). Five hundred microliters of WME incubation medium (catalog no. A1217601; Thermo Fisher Scientific) supplemented with a hepatocyte maintenance supplement pack (catalog no. CM4000; Thermo Fisher Scientific) alone or WME incubation medium containing either COBI (5 μM), DRV (5 μM), RIF (5 μM), or RTV (5 μM) was placed into separate sample chambers, while 750 μl of nonsupplemented WME (catalog no. A1217601; Thermo Fisher Scientific) alone was placed into the corresponding buffer chambers. Each experimental condition was tested in triplicate. Following sealing with Parafilm M (Sigma-Aldrich), the RED device containing these samples was incubated for 5 h at 37°C with orbital shaking (200 rpm). Following incubation, a 450-μl aliquot was removed from the buffer chamber within each RED device insert and was vortexed for 10 s with 112 μl (20% of the total final volume) of acetonitrile in a 1.5-ml microcentrifuge tube, prior to transfer to a 300 μl Chromacol fixed insert vial (Thermo Fisher Scientific), from which 100 μl of the suspension was analyzed directly by HPLC-UV, as described below. For WME incubation medium samples, a 450-μl aliquot was removed from each sample chamber within each RED device insert and was transferred to a 2.0-ml microcentrifuge tube containing 1,350 μl of 100% acetonitrile. The samples were then vortexed for 5 s, prior to centrifugation at 13,100 × g for 10 min at room temperature. The resultant supernatants were transferred to Corning Pyrex borosilicate glass tubes (75 by 12 mm) and were dried in a Jouan RC10.22 vacuum centrifuge at room temperature (18 to 25°C). After the samples were dried, they were reconstituted in 400 μl of 20% (vol/vol) acetonitrile in H₂O, and 100 μl of the resultant suspension was used to quantify COBI, DRV, RIF, or RTV by HPLC-UV. The chromatographic separation of COBI, RIF, and RTV was achieved using a Waters Atlantis T3 column (4.6 by 100 mm; particle size, 3 μm) equipped with a Fortis C18 guard (10 by 4 mm; particle size, 3 μm). A Dionex P680 HPLC pump, a Dionex ASI-100 automated sample injector, and a Dionex UVD170U UV detector were used. Mobile phases A (25 mM KH₂PO₄, pH 3.3, orthophosphoric acid) and B (100% acetonitrile) were used in a step-gradient elution as follows: 70% mobile phase A and 30% mobile phase B from 0.0 to 2.0 min, 52.5% mobile phase A and 47.5% mobile phase B from 2.0 to 4.0 min, 35% mobile phase A and 65% mobile phase B from 4.0 to 6.0 min, 20% mobile phase A and 80% mobile phase B from 6.0 to 9.0 min, and 70% mobile phase A and 30% mobile phase B from 9.0 to 12.5 min. Elution was carried out at room temperature (18 to 25°C), and the flow rate was maintained at 1.00 ml/min. Chromatograms were analyzed with COBI and RTV quantified at 220 nm and RIF quantified at 267 nm using Chromeleon software (version 6.8). Each experimental condition was assessed in triplicate. Standards and quality control samples for each drug were prepared and extracted from WME incubation medium to analyze the corresponding sample chamber samples or were prepared in WME medium containing 20% (vol/vol) acetonitrile for analysis of buffer chamber sample dialysates. The fraction unbound (f_u) of each drug was calculated by dividing the drug concentration quantified in the buffer chamber dialysate by the concentration of drug quantified in sample chamber aliquots. Results are presented as the mean f_u ± standard deviation (SD; n = 3).

Calculation of the CL_int.app. of darunavir by hepatocytes.

The apparent intrinsic clearance (CL_int.app.) of DRV was calculated on the basis of a previously described method (64). This is summarized in Equation 1:

$${\text{CL}}_{\mathrm{int}\mathrm{.}\text{app}\mathrm{.}}=(\ln \hspace{0.17em}2/in\hspace{0.17em}vitro\hspace{0.17em}{t}_{\mathrm{1/2}})\times (\text{incubation}\hspace{0.17em}\text{volume/}{10}^6\text{\hspace{0.17em}hepatocytes})$$ (1)

where t_1/2 is the half-life and the incubation volume is in microliters.

The results are expressed as the mean number of microliters per minute per 10⁶ hepatocytes ± SD for a total of three donors per condition tested. Three biological replicates were quantified per condition tested, using hepatocytes obtained from three separate donors in each case. All DRV CL_int.app. values were calculated using DRV concentrations corrected for DRV protein binding in WME incubation medium (Table 1).

Statistical analysis.

Statistical analyses were carried out using IBM SPSS Statistics software (version 22; IBM Corporation, Armonk, NY, USA). All data were assessed for normality using a Shapiro-Wilk test, and data were compared using a Mann-Whitney U statistical test. Univariate and stepwise-elimination multivariate linear regression analyses (significance [P] threshold < 0.2, α = 0.05) were conducted to characterize the influence of coincubation of primary human hepatocytes with various concentrations of RTV or COBI together with RIF on the CL_int.app. of DRV. Calculation of the half-maximal (50%) inhibitory concentration (IC₅₀) of RTV and COBI required to inhibit the DRV CL_int.app. maximally induced by 10 μM RIF was completed using the DRV CL_int.app. data obtained from experiments performed with COBI (donors HU1399, HU1574, and HU1587) and RTV (donors HU1399, HU1587, and HU1621). Data were first normalized by defining the mean maximal elevation in the DRV CL_int.app. induced by 10 μM RIF alone in each respective data set as 100% and plotting the remaining values relative to this value. GraphPad Prism software (version 5; GraphPad Software, Inc., La Jolla, CA, USA) was used to plot the data using the log(inhibitor concentration)-versus-response equation and a least-squares fitting method.