42

4. P450 Drug Metabolism DDIs: Inhibition (2012) Factor 4: Inhibition of Metabolism: When a P450 enzyme that is responsible for metabolizing an object drug (R) is inhibited by an interactant drug (I) then the capacity of the liver to clear the object drug from the system is decreased and, absent a dose correction, the blood levels of the object drug will increase. The following points are generally true with respect to inhibition. 1) The magnitude of the inhibition will be dose dependent. Higher inhibitor doses provide higher

blood levels leading higher inhibitor concentrations at the site of inhibition (the enzyme). We normally assume that the concentration of the inhibitor the the enzyme is exposed to is equal to the unbound concentration in plasma. a) The extent of inhibition will be a function of the inhibitor levels in the blood so inhibition of

object drug metabolic clearance depends on the dose and blood levels of [I].

i) Below we see an example where the extent of inhibition, given as the fold decrease in metabolic clearance of an object drug (Cl(control)/Cl(inhibitor) is dependent upon the plasma concentration of the inhibitor (fluvoxamine). In this plot the Y-axis reflects the extent of inhibition (bigger numbers=more inhibition).

2) Mechanisms of Inhibition: Reversible vs Irreversible Inhibition Mechanisms. i) Reversible Inhibition: The inhibitor “competes” for enzyme with the object drug.

I free enzymeinhibitor

complexsubstratecomplex

[P450 : I] R-OH[R : P450]P450

R

KI Km

43 (1) The affinity of the enzyme for the inhibitor is given by the constant Ki. The lower the

Ki the more potent the inhibitor. The Ki is equal to the concentration of inhibitor that is required to bind up half of the available enzyme.

(2) Each enzyme inhibitor pair has a characteristic and constant Ki. This means that we can apply the inhibition information obtained about one inhibitor-enzyme pair to the effect observed on all object drugs metabolised by that enzyme.

(3) Graphs like that shown above of in vivo data allow us to “calibrate” in vivo effect of an

interactant drug.

(4) We can determine a Ki for an inhibitor enzyme pair by doing in vitro experiments in microsomes. If we know the plasma levels of the inhibitor we can predict the magnitude of an interaction from the I/Ki ratio and specific information about the percent contribution of the enzyme to the clearance of the object drug.

(5) Most inhibitory DDI’s are the result of reversible competitive inhibition of a P450

enzyme by an interactant drug.

ii) Irreversible Inhibition (Mechanism based enzyme inactivation (MBI)):

(1) The enzyme attempts to metabolize the inhibitor and produces a reactive intermediate

that binds covalently to the active site amino acids or the heme group to produce an irreversible enzyme adduct that is no longer active.

(2) The body must produce new enzyme molecules before drug metabolism activity can be restored.

(3) The inhibition of the enzyme requires time (hours to days) to occur since it depends on

a catalytic rate. Mechanism based inhibition is is time dependent.

(4) Many alkylamine containing drugs undergo sequential metabolism to produce reactive intermediates the bind directly to the heme iron of P450. These stable adducts are called metabolite intermediate complexes or MICs. Drugs that do this and cause DDIs include troleandomycin (see notes pages 36 and 37), erythromycin and diltiazem.

Efree + I E-IKi

when [I]=Ki [Efree] = [E-I]

The higher the [I]/Ki ratio the greater the inhibition

Efree + I E-I

FeO

EIinactive

irreversible covalent bondbetween enzyme and inhibitorEnzyme is dead forever.Must make new enzyme to get activity back

44

3) Principle: Effect of an interactant drug will be observed on all object drugs metabolised by that enzyme. The magnitude of the effect at an equivalent dose of the inhibitor will depend upon how much of the object drug clearance is due to that enzyme. a) Here fluvoxamine is an inhibitor of CYP1A2. The data are taken from individual clinical

studies.

F3C CN

(CH2)4OCH3(CH2)2NH2

Fluvoxamine

SS Dose Object Drug AUC Half-life Clearance

50 mg x 2 Clozapine ↑ 2.5 fold

100 x 1 Caffeine ↑ 6.1 fold ↓ 5 fold

75 x 1 Theophylline ↑ 2.4 fold ↑ 2.5 fold

50 mg x 1 Tacrine ↓ 6 fold

100 x 1 Imipramine ↑ 2.5 fold ↑ 1.8 fold

X

Nitrone

MDOH

MAOH

MD

MANitroso

+ e

R NO

C H2

+-

R NO

Fe+2R N

OR N

H

HR N

OH

H

R NOH

C H3

R NH

C H3H2O

455 nmN N C H3

C H3S O

OO

C H3O

DTZ

?

45 b). Here we see some other examples abstracted from individual clinical studies where we look at different CYP1A2 inhibitors and various substrates. CYP1A2 Substrate

Interactant Drug or Condition Arrow indicate a significant change in object drug half-life

Object Drug

Quinolones Fluvoxamine Mexiletine Smoking Cimetidine

Caffeine ↑ ↑ ↑ ↓ ↑

Theophylline ↑ ↑ ↑ ↓ ↑

Tacrine ↑ ↓

Imipramine ↑ ↑ ↓ ↑

Clozapine ↑ ↑ ↓

Mexiletine ↑ ↑ ↓

Propranolol ↓

4) The magnitude of the inhibitory effect depends upon the dose of the inhibitor and the fraction of

the drug that is metabolised by the enzyme that is inhibited. Below we see the effect of ciprofloxacin (a quinolone) on the half-life of theophylline. Maximum 2 fold effect is observed. About 60% of theophylline clearance is due to CYP1A2 so a maximum effect is observed even as the plasma levels of ciprofloxacin continue to climb. If the clearance of theophylline was totally due to CYP1A2 we would predict a rising straight line from this plot.

!

Mean T 1/2 Data and Fit

6

7

8

9

10

11

12

13

14

0 3 6 9 12 15

Plasma Ciprofloxacin (!M)

Theo

ph

yllin

e T

1/2

(h

r)

46 Generally I/Ki ratios and the fraction of the object drug clearance that is due to the inhibited enzyme (fm) determines the magnitude of the effect observed. Magnitude of Effect on Object Drug Clearance Inhibitor Concentration fm (% of Object Drug Metabolized by the Inhibited Enzyme)

[I]/Ki Low High Low ↓ ↓↓ High ↓↓ ↓↓↓↓

a) Sulfaphenazole and tolbutamide Below is shown the effect of sulfaphenazole (interactant drug)

on tolbutamide (object drug) levels in the body. Note that the amount of tolbutamide (cleared mostly by CYP2C9) increases substantially when sulfaphenazole is added to the dosing regimen. Note also that the interaction takes time to develop and dissappear.

Corollary: Any drug that is substantially cleared by CYP2C9 will experience a drug/drug interaction with sulfaphenazole. Thus since warfarin is substantially cleared by CYP2C9 it also is subject to an interaction with sulfaphenazole.

b) Another drug that inhibits CYP2C9 drug metabolizing capacity is cimetidine. As expected

from the above corollary the half-lives of (S)-warfarin and tolbutamide are both increased by cimetidine.

!

47

c) A more complex example of a different kind (the parent drug is toxic and a metabolite is active)

(1) Terfenadine (Seldane) was an extremely popular (15 million scrips a year in 1990) non-sedating H1 antagonist used to treat allergies.

(2) When marketed, nothing was known about what enzymes produced it’s metabolites.

Interestingly it was known that the levels of terfenadine in the blood were very low and sometimes undetectable relative to levels of its metabolites. This suggested that the antihistamine effect may be due to a metabolite.

(3) As it turns out, CYP3A4 converts almost all of the dose (>99%) to metabolites before the drug (oral administration) reaches the systemic circulation (a massive first pass effect)

(4) As clinical experience with this drug increased there were reports of cardiotoxicity (increase in Q-T interval on an electrocardiogram: tosardes de pointes) when the drug was given with CYP3A4 inhibitors such as ketoconazole, itraconazole and others. This effect is now known to be due to blockage of the hERG channel, a potassium channel in the heart tissue, by terfenadine. hERG blockade can result in ventricular fibrillation and death.

risk. Unfortunately, screening drugs purely for IKr blockis confounded by the lack of a straightforward relation-ship among IKr block, degree of QT prolongation, andrisk for TdP. For example, the antiarrhythmic agentverapamil blocks IKr but does not prolong the QT inter-val and is not associated with TdP (Yang et al., 2001),probably reflecting its potent block of L-type calciumchannels. Likewise, ranolazine is an IKr blocker but infact prevents experimental TdP despite prolongation ofthe action potential; this is thought to be because thedrug also blocks sodium current during the plateau ofthe action potential (Wu et al., 2006). Thus, IKr blockalone is not sufficient to result in TdP.

Further, clinical studies make it clear that QT prolon-gation alone is also insufficient to generate TdP. Theantiarrhythmic drug amiodarone and the anestheticagent sodium pentobarbital both result in QT prolonga-tion, but TdP caused by amiodarone is exceedingly rare(Lazzara, 1989), and sodium pentobarbital probably re-duces the risk of TdP (Shimizu et al., 1999). In addition,some drugs may cause polymorphic ventricular tachy-cardia without QT prolongation, which, by definition,are not labeled TdP, but the proarrhythmic mechanismsmay be similar and the distinction between these inci-dents and diLQTS is not always clear (Shah andHondeghem, 2005).

In experimental settings, prolongation of repolariza-tion in Purkinje fibers or in some myocardial cells canlead to the development of EADs, oscillations in themembrane potential during repolarization (Fig. 5).EADs represent inward current and are thought to re-sult from reopening of either L-type calcium channels orsodium channels or from current generated via aug-mented sodium-calcium exchange, and they may resultin ectopic beats if the amplitude of the EAD reaches acritical threshold and occurs in a large enough region ofthe heart (January and Riddle, 1989). Triggered up-strokes from EADs in tissue in which repolarization issufficiently disturbed is the likely initiating mechanismfor TdP. Thus, drug-induced proarrhythmia results in-

directly from IKr block, which enables activation or re-activation of inward currents underlying EADs. Thispartly explains why verapamil rarely causes TdP: eventhough it is a potent IKr blocker (Zhang et al., 1999), italso reduces EADs by blocking inward calcium current(Shimizu et al., 1995). Furthermore, verapamil shortensQT interval, reduces EADs, and lowers the incidence ofTdP in a model of acquired LQTS by a combination of IKsand IKr block (Aiba et al., 2005).

Studies in multicellular preparations such as thewhole heart or the wedge described below have de-scribed mechanisms responsible for maintenance of thearrhythmia after its initiation. Studies in the wedgehave emphasized the role of heterogeneity of action po-tential durations as a mechanism contributing to main-tenance of the arrhythmia by promoting re-entrant ex-citation (Antzelevitch, 2008).

D. Reduced Repolarization Reserve

A unifying framework to understand diLQTS is theconcept of “reduced repolarization reserve” (Roden,1998, 2008; Roden and Yang, 2005) (Fig. 6). This conceptsuggests that multiple often-redundant mechanismsmaintain normal repolarization, so minor alterations infunction may not be obvious at baseline. Thus, for ex-ample, a minor reduction in repolarizing current (e.g., asa result of a genetic lesion resulting in a small reductionin an important repolarizing ion channel) may be with-out consequence because other mechanisms come intoplay to maintain a near-normal QT interval. However,such reduced reserve may become obvious when furtherstressors to repolarization, such as drug challenge, slowheart rates, or hypokalemia are superimposed. Thisframework is a specific example of the more generalconcept that complex biologic systems rarely break downbecause of single lesions (genetic or acquired).

The specific nature of repolarization reserve and howit might be reduced has been investigated, and a leadingcandidate mechanism seems to be loss of IKs function.This makes intuitive sense, because reduction in IKsmight be without consequence as long as a robust IKr ispresent; in this situation, however, IKr block may resultin catastrophic failure to maintain normal repolariza-tion and thus TdP. Human ventricular myocytes displayresistance to action prolongation with IKr block when IKsis stimulated (Jost et al., 2005) (Fig. 7). In a simulationstudy, Silva and Rudy (2005) suggested that IKs gatingproperties were ideally suited to meet such a reservefunction. One especially intriguing observation is thatexposure to some QT-prolonging drugs can down-regu-late IKs by stimulating a microRNA (Xiao et al., 2008);this in turn suggests that repolarization reserve may bedynamic and itself regulated by drug exposures. Finally,genomic approaches discussed below implicate IKs al-leles in risk for drug-induced TdP.

ddrug

HERG!Action potential duration, EADs, QT Torsades de HERG

blockduration, EADs, !heterogeneity of

repolarization

QT prolongation pointes

degenerating to VF

FIG. 5. Mechanisms of proarrhythmia associated with human ether-a-go-go-related gene (HERG) channel blockade. Blockade of the HERGchannel produces prolongation of the QT interval (blue) and generates anEAD (red) in the cardiac action potential. These changes, which areheterogeneous across the ventricular wall, create a substrate for TdP. Inthis example, torsades de pointes degenerates into ventricular fibrillation.[Adapted from Roden DM and Viswanathan PC (2005) Genetics of ac-quired long QT syndrome. J Clin Invest 115:2025–2032. Copyright ©2005 American Society for Clinical Investigation. Used with permission.]

DRUG-INDUCED LONG QT SYNDROME 767

at Univ of W

ashington on January 16, 2012pharm

rev.aspetjournals.orgD

ownloaded from

!

48

(5) It turns out that the antihistamine effects of terfenadine were due mostly to a metabolite terfenadine acid, while the cardiotoxic effect was due to the terfenadine itself.

(6) Those individuals that experienced the toxicity were taking other compounds that inhibited CYP3A4 such as troleadomycin, ketoconazole, erythromycin and grapefruit juice.

(7) Terfenadine was removed from the market in the United States. The acid metabolite (fexofenadine: Allegra) has replaced terfenadine.

(8) Many amine containing drugs are now known to occasionally cause QT prolongation by binding to the hERG channel. Thus inhibition of their metabolism can lead to a hightened risk of this reasonably common off-target toxicity. Note the examples below (1,2,3,5,6,17,18,19) for psychiactric inpatients. The object drugs are on the left in the table.

592 VOLUME 90 NUMBER 4 | OCTOBER 2011 | www.nature.com/cpt

ARTICLES

allowed us to identify and further evaluate potential drug inter-actions under natural conditions in a large psychiatric inpatient population with a previously unachieved degree of e!ciency. At the same time, we also provide a methodological solution for large-scale evaluation of CDSS using real-life prescription data. We used a new interaction classi"cation system to evalu-ate clinical relevance, management implications, and associated ADEs.

MediQ contained data for the active substances of 97.9% of all drug prescriptions in our study population, including many combinations that are not found even in Hansten and Horn’s collection. #is re$ects the comprehensiveness of MediQ’s data collection, particularly with respect to central nervous system drugs, combined with the fact that our data set was limited to psychiatric patients. Also, our data set covered 8,840 (44%) of the approximately 20,000 combinations in the MediQ database. Consequently, MediQ generated a very large number of alerts, suggesting high sensitivity for the task. At the same time, an average of 4.6 alerts or comments per patient, with a maximum of 87 alerts for a single patient, raises the issue of overalerting. #is can lead physicians to override even severe alerts without discrimination, thereby compromising the e!cacy of CDSS to

improve medication safety in clinical practice.15,22 Although the display can be limited to high-danger alerts, our results, and those of others, suggest that this is not a reliable measure to identify interactions that require active management.18 #e ORCA classi"cation was designed with this issue in mind, and we have now extended this classi"cation and applied the result-ing ZHIAS system to a real-life population. First, reclassi"cation according to ORCA suggests that ~35% of distinct high-danger MediQ alerts (Figure 2), or 30% of those occurring in our study population (Table 4), relate only to a conditionally increased risk and that combinations with absolute contraindications are very rare. Second, extended evaluation according to the ZHIAS pro-vides categorical management recommendations and suggests that in many instances a combination may have been prescribed deliberately because the bene"ts outweigh the theoretical risks of a drug interaction. In many cases, appropriate monitoring or reduction in dosage may be e!cacious measures to control a potential risk. #e additional dichotomous information that the ZHIAS provides regarding mechanisms and speci"c possi-ble ADEs may then be of high practical value because it can be readily displayed in appropriately designed CDSS. For example, a warning about increased drug concentration levels in plasma

Table 3 The top 20 drug combinations most frequently associated with high-danger interaction alerts in the study population, according to MediQ

RankDrug combination with high-danger interaction, according to MediQ

Frequency of interactions

Adverse event for which there is increased riskn (%) Cum. %

Total 2,308 (100)

1 Clozapine–fluvoxamine 402 (17.4) 17.4 Seizures and other CNS effects, QTc/TdP and arrhythmias, bone marrow toxicity

2 Lithium–hydrochlorothiazide 190 (8.2) 25.6 Seizures and other CNS effects, QTc/TdP and arrhythmias (lithium intoxication)

3 Olanzapine–clozapine 177 (7.7) 33.3 QTc/TdP and arrhythmias, hypotension, and metabolic and endocrine disorders, including disruption in blood glucose levels

4 Metoprolol–paroxetine 114 (4.9) 38.3 Hypotension

5 Clozapine–pimozide 68 (2.9) 41.2 EPS and other CNS effects, QTc/TdP and arrhythmias

6 Torasemide–lithium 60 (2.6) 43.8 Seizures and other CNS effects, QTc/TdP and arrhythmias (lithium intoxication)

7 Levomepromazine–doxepin 58 (2.5) 46.3 Sedation, hypotension

8 Metoprolol–fluoxetine 48 (2.1) 48.4 Hypotension

9 Ramipril–spironolactone 42 (1.8) 50.2 Hypotension, hyperkalemia

10 Phenytoine–valproate 40 (1.7) 51.9 Hepatotoxicity

11 Amitriptyline–tranylcypromine 39 (1.7) 53.6 Serotonin syndrome

12 Amitriptyline–moclobemide 38 (1.6) 55.3 Serotonin syndrome

13 Moclobemide–trimipramine 38 (1.6) 56.9 Serotonin syndrome

14 Duloxetine–tramadol 35 (1.5) 58.4 Serotonin syndrome

15 Tranylcypromine–carbamazepine 31 (1.3) 59.8 Seizures and other CNS effects

16 Spironolactone–potassium 30 (1.3) 61.1 Hyperkalemia

17 Trimipramine–thioridazine 28 (1.2) 62.3 QTc/TdP and arrhythmias

18 Amitriptyline–thioridazine 22 (1.0) 63.3 QTc/TdP and arrhythmias

19 Doxepin–thioridazine 21 (0.9) 64.2 QTc/TdP and arrhythmias

20 Nefazodone–carbamazepine 21 (0.9) 65.1 Loss of nefazodone efficacy

CNS, central nervous system; EPS, extrapyramidal symptoms.

49

d) An example of an intentional DDI (AIDs, ritonavir, PI boosting strategy) i) Ritonavir (Norvir) is a first generation HIV-protease inhibitor (PI) that is now widely used

as a boosting agent in the treatment of human immunodeficiency virus infection due to its potent off-target inhibitory effect on cytochrome P450 3A4 dependent metabolism of other protease inhibitors such as lopinavir and saquinavir. For example, systemic exposure to saquinavir is increased 17 fold by co-administration of RTV. Currently two of the four preferred drug regimens recommended by DHHS for use in treatment-naïve AIDS patients employ RTV as an adjuvant.

Ritonavir (potent CYP3A4 MBI) Effect of Ritonavir on Saquinavir PK.

ii) This highly successful boosting strategy also features a co-formulated dosage form

(Kaletra; (lopinavir-ritonavir)) and has been applied to the development of other co-formulations as well as an alternate boosting agent. Lopinavir/ritonavir (Kaletra) is the only PI-enhanced regimen co-formulated into a single pill. Multiple-dose pharmacokinetic studies in both healthy volunteers and HIV-infected patients suggest that the optimal combination of lopinavir/ritonavir is 400/ 100mg. (Clinical Pharmacokinetics (2004) 43 p 291-310.

iii) RTV causes time dependent loss of enzyme activity in vitro and clinical studies have

established that otherwise sub-therapeutic RTV doses of 100 mg/day are sufficient to “knock out” P450 3A4 activity in vivo. Look at the table to see the effects of RTV on the PK of other coadministered drugs in HAART therapies.

Enhanced Protease Inhibitor Therapy 293

absorption process, PIs encounter intestinal P-gly- influence Cmax, Cmin, elimination half-life (t1/2!) andthe area under the concentration-time curve (AUC)coprotein, a multidrug-resistance transport protein.of the coadministered PI. For example, low-dosageP-glycoprotein acts as a transmembrane pump thatritonavir improves oral bioavailability of saquinaviractively removes drugs from the cell membrane. Aand lopinavir, which undergo high-first pass meta-P-glycoprotein inhibitor will block transport of PIsbolism.[9,10] Saquinavir and lopinavir Cmax, Cminout of the cell, thus increasing oral bioavailability.and AUC increase, whereas t1/2! is prolonged.[11,12]Once PIs reach the liver, further metabolism occursThe Cmax increase is possibly achieved by ritonavirby the CYP enzyme system, which can also beinhibiting P-glycoprotein transport and/or first-passinhibited or induced by substrates. Hepatic CYPmetabolism. Ritonavir also increases the t1/2!, Cminisoenzymes metabolise approximately 80–95% ofand AUC of indinavir and amprenavir, having only athe currently available PIs.[7] The main drug-small effect on the absorption phase.[13-16] For thesemetabolising isoenzymes are CYP3A4, 2C9, 2C19,PIs, Cmax is similar between ritonavir-enhanced and2D6, 1A2, 2E1, 2B6 and 2A6, with 3A4 being thenon-enhanced regimens.most abundant. PIs also inhibit hepatic CYP3A4

The ability of ritonavir to increase plasma Cminactivity, with ritonavir being the most potent agent,of concomitantly administered PIs is perhaps thefollowed by indinavir, nelfinavir and amprenavir.[3]

greatest clinical benefit of dual PI therapy. Inade-Hepatic enzyme inhibition increases the bioavaila-quate concentrations of antiretrovirals may promotebility of PIs[8] and may also compensate for the CYPthe emergence of viral gene mutations and subse-enzyme induction caused by concurrent use of thequently, support long-term antiretroviral resistance.non-nucleoside reverse transcriptase inhibitorsThe rate of viral resistance is dependent upon the(NNRTIs) nevirapine (NVP) and efavirenz (EFV).number of virus mutations required to decrease drug

Alterations in the metabolism of PIs when efficacy, and the potency of the antiretroviral regi-coadministered with a CYP enzyme inhibitor are men. Combined use of PIs that do not have overlap-reflected by impressive changes in pharmacokinetic ping mutation profiles would be expected to delayparameters. Figure 1 illustrates the effect of low- the emergence of resistance.[17,18] Furthermore, thedosage ritonavir on saquinavir plasma concentra- selection of resistance to two PIs simultaneouslytions. Currently, many dual PI regimens contain may require a greater number of mutations than thatritonavir, the most potent PI CYP enzyme inhibitor, required to overcome a single PI. Dual PI therapyin addition to another PI. Ritonavir is administered increases plasma concentrations throughout the ad-at a dosage of 100 or 200mg twice daily, which is ministration interval and reduces the variability ofmuch lower than the approved dosage of 600mg these concentrations. High plasma concentrationstwice daily. These ritonavir dosages dramatically may overcome moderate phenotypic resistance, and

hence prolong suppression of viral replication. In-creased potency of dual PI therapy as well as anincrease in the genetic barrier are virologically bene-ficial.

Table I summarises pharmacokinetic data for PIsand dual PI combinations.

2. Dual Protease Inhibitor Regimens

2.1 Saquinavir and Ritonavir

2.1.1 PharmacokineticsThe pharmacokinetics of saquinavir and ritona-

vir, at various doses, have been reported in serone-gative and HIV-infected subjects. Data exist for

0

1

2

3

4

5

6

7

Time (h)

Saq

uina

vir

conc

entra

tion

(mg/

L)

121086420

SQV/RTVSQV

Fig. 1. Effect of low-dosage ritonavir on the pharmacokinetics ofsaquinavir in two HIV-infected pregnant women during gestation.The solid line is saquinavir/ritonavir (SQV/RTV) 800/100mg twicedaily and the broken line is saquinavir (SQV) 1200mg three timesdaily.

" 2004 Adis Data Information BV. All rights reserved. Clin Pharmacokinet 2004; 43 (5)

50

iv) P450 3A4 is responsible for the clearance of more than 50% of commonly administered drugs. Thus the large population of RTV-boosted patients (millions world wide) is at a high risk of unintended but profound drug drug interactions with object drugs significantly metabolized by CYP3A4. Let’s look at a couple of them.

(1) Flonase Interaction:

(a) There were 25 cases (15 adult and 10 paediatric) of significant adrenal suppression

secondary to an interaction between ritonavir and inhaled fluticasone, and three cases involving ritonavir and intranasal fluticasone. There has also been a

!

!

294 King et al.

Table I. Pharmacokinetic parameters of protease inhibitors (PIs) and various dual PI combinations. Values are listed as means ± SD ormedian (range), as available. For combinations, values are reported for the first PI named

Drug or combination Dose (mg) Cmax (mg/L) Cmin (mg/L) AUC! (mg • h/L) t1/2" (h) Reference

Amprenavir 1200 bid 5.4 0.28 18.5 7.1–10.6 19

Amprenavir/ritonavir 600/100 bid 6.0 1.7 29.9 20

Amprenavir/ritonavir 1200/200 od 8.2 0.9 49.8 20

Atazanavir 400 od 3.2 ± 2.2 0.27 ± 0.3 22.3 ± 20.2 6.5 ± 2.6 21

Atazanavir/ritonavir 400/100 od 7.8 ± 1.1 1.0 ± 0.3 70.3 ± 10.5 7.0 ± 1.8 22

Indinavir 800 tid 7.7 ± 2.5 0.15 ± 0.1 18.8 ± 7.0 1.8 ± 0.4 23

Indinavir/ritonavir 800/100 bid 8.7 0.99 44 3.2 24

Indinavir/ritonavir 800/200 bid 6.3 (3.2–15.2) 0.7 (0.1–4.6) 38.2 (18.2–86.8) 25

Indinavir/ritonavir 400/400 bid 3.6 (1.5–7) 0.6 (0.1–1.6) 23 (9.3–45.7) 25

Indinavir/nelfinavir 1200/1250 bid 7.8 0.18 63.2 26

Lopinavir/ritonavir 400/100 bid 8.5 5.3 76.5 27

Nelfinavir 750 tid 3.0 ± 1.6 1.0 ± 0.5 28

Nelfinavir 1250 bid 3.5 ± 0.8 0.87 ± 0.31 24.7 ± 5.6 5.7 ± 2.7 29

Nelfinavir/saquinavir 1250/1000 bid 4.2 1.0 29.2 30

Nelfinavir/indinavir 1250/1200 bid 4.5 1.5 71a 26

Nelfinavir/ritonavir 1250/100 bid 4.5 ± 1.2 1.9 ± 0.8 37.7 ± 12.8 8.2 ± 2.4 29

Ritonavir 600 bid 11.2 ± 3.6 3.0 ± 2.1 60.8 ± 23.4 3.2 31

Saquinavir-HGC 600 tid 0.25 0.78 32

Saquinavir/nelfinavir 1000/1250 bid 1.5 0.1 7.3 30

Saquinavir-SGC 1200 tid 2.5 7.3 ± 6.2 32

Saquinavir/ritonavir 400/400 bid 2.5 0.5 16 ± 8 33

Saquinavir/ritonavir 1000/100 bid 3.9 0.5 23.4 3.2 34a AUC over 24 hours.

AUC! = area under the concentration-time curve over the administration interval for each regimen at steady state; bid = twice daily; Cmax =peak plasma concentration; Cmin = trough plasma concentration; h = hours; HGC = hard gel capsules; od = once daily; SGC = soft gelcapsules; tid = three times daily; t1/2" = elimination half-life.

combinations of ritonavir with both saquinavir soft ues were 3.0 (1.3–18.7), 30.9 (14.7–70.1) and 52.2gel (saquinavir-SGC) and hard gel (saquinavir- (29.6–105.2) mg • h/L, respectively. SaquinavirHGC) capsules. Cmin for the three regimens were 0.04 (0.02–0.12),

0.84 (0.3–1.7) and 0.8 (0.4–2.7) mg/L, respectively.Pharmacokinetic studies evaluating ritonavir andData also indicated that the addition of ritonavir tosaquinavir in healthy volunteers indicate that ritona-saquinavir decreased intersubject variability. Thevir increases saquinavir exposure in a dose-depen-coefficients of variation (CV%) for saquinavirdent manner.[35,36] Sixty-six healthy subjects were800mg twice daily AUC24 and Cmax were approxi-randomised to receive one of eight treatment regi-mately 105 and 124% compared with 46 and 32%mens: saquinavir 800mg, ritonavir 400mg or sa-for 800/200mg twice daily, respectively. Saquinavirquinavir/ritonavir 400/400mg, 600/300mg, 800/had no effect on the pharmacokinetics of ritonavir.200mg, 800/300mg, 800/400mg and 600/400mg, allSimilar results were seen in 42 healthy subjectsgiven twice daily. Subjects received a single dose onreceiving escalating doses of saquinavir-SGCday 1, followed by twice daily administration for an1200–1800mg with or without ritonaviradditional 13 days. Blood samples were collected on100–200mg.[9] A 24-hour pharmacokinetic evalua-day 14 of therapy. Geometric mean (range) sa-tion was performed after the evening dose on dayquinavir Cmax for saquinavir 800mg, saquinavir/13. Saquinavir 1200mg three times daily produced aritonavir 400/400mg and 800/200mg were 0.5median (range) Cmax of 0.86 (0.4–4)) mg/L, Cmin of(0.2–4), 2.9 (1.8–4.7) and 4.5 (2.8–7.3) mg/L, re-

spectively. Saquinavir 24-hour AUC (AUC24) val- 0.08 (0.05–0.4) mg/L and AUC24 of 8.2 (3.8–45)

# 2004 Adis Data Information BV. All rights reserved. Clin Pharmacokinet 2004; 43 (5)

!

51 pharmacokinetic interaction study performed which assessed the magnitude of the interaction. Eighteen healthy volunteers were given oral ritonavir 100 mg every 12 h for 7 days and fluticasone 200 mg daily intranasally. The serum fluticasone area under the concentration–time curve (AUC) was increased 350-fold. (HIV Medicine (2008) 9 pp 389-396).

(2) Midazolam Interaction (note the same data plotted on a linear plot (left) and a semilog plot (right). These results are expected why?

PK parameters (ignore ALT-2074 data)

clearance to 88% of control. Differences in mean valuesbetween ALT-2074 and placebo co-treatments were notsignificant.

When the ANOVA was done on rank-transformed vari-ables, all conclusions were identical.

Table 2 summarizes the analysis of ratios for the phar-macokinetic variables. Based on untransformed ratios ofuntransformed values, all ritonavir/placebo ratios were dif-ferent from 1.0 with a high level of statistical significance.The ALT-2074/placebo ratio for Cmax and clearance did notdiffer significantly from 1.0; the AUC ratio was significantlygreater than 1.0 (P < 0.05, two-tailed test).

Geometric mean ratios underestimated the arithmeticmeans. For the ritonavir/placebo ratios, the 90% CIs fellentirely outside the arbitrary 80–125% boundary. For ALT-2074/placebo, one extreme of the 90% CI fell outside the80–125% boundaries.

Plasma ritonavir concentrationsFigure 2 shows mean plasma ritonavir concentrations atspecific time points during the ritonavir co-administrationtrial. The results demonstrate systemic exposure toritonavir consistent with the ritonavir dosage. However,there was no apparent relationship between the netexposure to ritonavir over the 0–10-h dosage interval(expressed as mean concentration over that interval) andthe extent of midazolam clearance impairment relative tothe control trial (expressed as midazolam total AUC ratiofor Trial 3 divided by Trial 1) (Figure 3).

Discussion

In this study we evaluated the capacity of ALT-2074 andboosting doses of ritonavir to reduce the apparent oral

Hours after dose

Pla

sma

mid

azol

am (

ng/m

L)

00

2

2

4

4

6

6

8

8

10

10 0 2 4 6 8

0 4 8 12 16 20 24

10

12

With placebo

With ALT-2074

Hours after dose

Pla

sma

mid

azol

am (

ng/m

L)

0.1

0.2

0.5

1

2

5

10With placebo

With ALT-2074

Hours after dose

Pla

sma

mid

azol

am (

ng/m

L)

5

10

15

20

25

30

35

With placebo

With RITONAVIR

Hours after dose

Pla

sma

mid

azol

am (

ng/m

L)0.2

0.5

1

2

5

10

20

40

With placebo

With RITONAVIR

0 4 8 12 16 20 24

Figure 1Mean (!SE) plasma midazolam concentrations at corresponding times.Top: placebo and ALT-2074 co-treatments (left: linear scale; right: logarithmic scale).Bottom: placebo and ritonavir co-treatments (left: linear scale; right: logarithmic scale)

Ritonavir and CYP3A

Br J Clin Pharmacol / 68:6 / 923

dose was administered, followed by blood sampling at0.25, 0.5, 1.0, 2, 4, 6, 8, 10 and 12 h after dosage. The thirddose of co-treatment was given at 18.00 h. The final bloodsample was taken at 08.00 h on the following morning,24 h after midazolam dosage. Subjects were then dis-charged from the study unit.

Venous blood samples were collected into heparinizedtubes and stored on ice until centrifuged. The plasma wasseparated and frozen at -18°C until assay.

Analysis of samplesPlasma concentrations of midazolam in all samples weredetermined by liquid chromatography-mass spectroscopy,having a sensitivity limit of 0.5 ng ml-1 [15]. All samplesfrom a given subject’s set of three trials were extracted andanalysed together on the same day using the same calibra-tion standards.

Plasma concentrations of ritonavir during the ritonavirco-treatment trial were determined by high-performanceliquid chromatography [42]. Methods were not availablefor determination of ALT-2074 concentrations.

Pharmacokinetic analysisThe following pharmacokinetic parameters for midazolamwere determined using standard model-independent(‘noncompartmental’) methods: maximum plasma con-centration (Cmax), elimination rate constant (b), eliminationhalf-life (T1/2), total AUC, and apparent oral clearance (CL).

Statistical analysisA 50% difference in mean values of midazolam clearancebetween placebo or ALT-2074 co-treatments, or betweenplacebo and ritonavir co-treatments, was assumed to be ofpotential clinical importance. Based on prior studies [13,15, 23], the standard deviation of the difference betweenmean values was assumed to be 35% of the differenceitself. Under these conditions, a sample size of n = 12 allowsthis difference to be detected witha= 0.05 and power of atleast 0.8.

Arithmetic mean and SD/SE of untransformed pharma-cokinetic variables were calculated and presented [43].Differences among the three co-treatments (placebo,ALT-2074, or ritonavir) were evaluated using analysis of

variance (ANOVA) for repeated measures, followed by Dun-nett’s test to compare ALT-2074 vs. placebo and ritonavirvs. placebo individually. These analyses were done bothwithout and with rank transformation (nonparametricanalysis).

Ratios of pharmacokinetic variables with ALT-2074 orritonavir co-treatment divided by the placebo value werealso calculated. These were aggregated as arithmeticmeans and standard deviations, or geometric means and90% confidence intervals (CIs).

Kinetic and statistical analyses were performed usingMicrosoft Excel or Statistical Analysis Systems (SAS Insti-tute Inc., Cary, NC, USA).

Results

SubjectsFifteen subjects initiated participation in the study, andcompleted the first of three trials. Two of these individualsdid not complete subsequent trials for administrativereasons. Pharmacokinetic analysis was based on the 13subjects that completed all three trials. Age ranges were21–50 years, and weight ranged from 52 to 97 kg. Theracial/ethnic distribution was eight White, five African-American.

Studies were completed with no adverse eventsreported.

Pharmacokinetics of midazolamFigure 1 shows the mean plasma midazolam concentra-tions at corresponding times, comparing placebo withALT-2074 and placebo with ritonavir.

Analysis of variance for repeated measures showedhighly significant differences among the three treatmentsin all pharmacokinetic variables for midazolam (Table 1).Dunnett’s test showed significant differences betweenritonavir and placebo co-treatments. Ritonavir increasedmidazolam AUC by a factor of approximately 25,and correspondingly reduced clearance to about 4% ofcontrol values. Co-treatment with ALT-2074 increasedmidazolam AUC by a factor of about 1.25, and reduced

Table 1Statistical analysis of midazolam pharmacokinetic variables*

Mean (!SE, n = 13) for treatment conditions:Repeated measures ANOVA

Dunnett’s testPlacebo ALT-2074 Ritonavir ALT-2074 vs. placebo Ritonavir vs. placebo

Cmax (ng ml-1) 9.0 (!1.0) 10.6 (!0.9) 35.6 (!2.8) F = 81.8, P < 0.001 NS P < 0.05

t1/2 (h) 2.06 (!0.15) 2.08 (!0.17) 18.07 (!2.25) F = 49.4, P < 0.001 NS P < 0.05Total AUC (ng ml-1 h-1) 24.65 (!1.93) 29.87 (!2.62) 651 (!77) F = 64.9, P < 0.001 NS P < 0.05

Clearance (ml min-1) 2157 (!142) 1844 (!167) 85.9 (!6.9) F = 92.2, P < 0.001 NS P < 0.05Clearance (ml min-1 kg-1) 28.84 (!2.84) 24.15 (!2.38) 1.13 (!0.11) F = 64.1, P < 0.001 NS P < 0.05

*Analysis of actual values without transformation.

D. J. Greenblatt et al.

922 / 68:6 / Br J Clin Pharmacol