Introduction

The advent of highly active antiretroviral therapy (HAART) has not only reduced the mortality rate of human immunodeficiency virus (HIV)-positive individuals, but also vastly improved their quality of life. However, treatments of acquired immunodeficiency syndrome (AIDS) and its associated conditions remain very complex. Several protease inhibitors, drug combinations (antibacterial and antifungal) indicated for opportunistic infections (OIs), medicines for depression, cardiovascular problems, and other associated conditions such as hyperlipidemia are recommended for the treatment of AIDS. These medications are generally taken orally and during combinations are absorbed from the gastrointestinal tract. Any agent that alters the absorption and transport of these medications can affect therapy. Therefore, multidrug regimens and corresponding drug–drug interactions may cause potential adverse drug reactions and ultimately treatment failures. Absorption of many oral drugs involves efflux and metabolism in the enterocytes and liver. Drug efflux proteins [P-glycoprotein (P-gp) and multiple drug resistance associated proteins (MRPs)] and metabolizing enzyme cytochrome P450 3A4 (CYP3A4) are the main factors leading to such interactions (Patel and Mitra 2001). Overlapping substrate specificities of these proteins result in complex and sometimes perplexing pharmacokinetic profiles of multidrug regimens. In this report, we discuss the mechanisms by which drugs in selected drug combinations interact with these efflux proteins (P-gp, MRP-1, and MRP-2) and metabolizing enzymes, CYP450. The entire discussion is based on clinical reports and some in vitro results from our laboratory. Our primary hypothesis is that the kinetics of these drug classes—protease inhibitors (PIs), azole antifungals, macrolide antibiotics, lipid lowering drugs, anticonvulsing agents will be altered when given in combinations due to efflux, and/or metabolism at cellular targets. Therefore, these drug combinations may cause unexpected and unexplainable therapeutic outcomes. Moreover, botanicals in combination therapy can also potentially change clinical outcomes (Pal and Mitra 2006). These hypotheses are based on the following observations. Saquinavir undergoes efflux by MRP-2 and P-gp, because PIs are substrates of both P-gp and MRPs. Erythromycin and ketoconazole are found to interact with MRPs and/or P-gp. Our studies, along with other published reports, indicate that herbal products also interact with P-gp and possibly with MRPs. These therapeutic agents in the presence of herbals such as St John's Wort and garlic) can undergo P-gp- and/or CYP3A4-mediated interactions (Pal and Mitra 2006). Because multidrugs resistance (MDR) and CYP-mediated drug–drug interactions involve many agents, it is not possible here to provide comprehensive listing of all possible drug interactions; rather, we restrict our discussion to antiretroviral agents and their interactions with other combination therapy in AIDS treatment. This review attempts to mechanistically predict drug–drug interactions that may lower the risk of treatment failure in AIDS patients. It will also allow clinicians and pharmacists to recommend doses in binary/ternary combinations of drug therapy to achieve maximum effectiveness with minimum toxicity.

General basis of drug interactions

Drug–drug interactions comprise two broad categories: pharmacodynamic and pharmacokinetic. Many drug interactions involve the effect of one drug on the absorption or disposition of another. Pharmacodynamic interactions involve two or more drugs having additive, synergistic, or inhibitory effects. In Kaletra, lopinavir and ritonavir act together to produce maximum efficacy in reducing viral load. Ritonavir causes higher area under curve (AUC) of lopinavir, possibly through the inhibition of efflux protein, P-gp, and metabolizing enzyme, CYP3A4. For treating OIs, saquinavir and ketoconazole administered together can elevate AUC of saquinavir as a result of inhibition of both P-gp and CYP3A4. Several antimycobacterial agents, i.e., clarithromycin and refabutin, inhibit CYP3A4, but rifampicin induces CYP3A4. In both cases, any combination therapy with PIs requires dosage adjustment.

Role of efflux proteins and metabolizing enzymes

Both active efflux and metabolism in the small intestine result in poor drug absorption. CYP450 enzymes are highly expressed on enterocytes that can metabolize xenobiotics during their transit across intestinal epithelium (Shen et al. 1997). The most abundant isozyme of CYP present in intestinal enterocytes is CYP3A4 (Kolars et al. 1992; Watkins et al. 1987). In addition to these metabolic enzymes, MDR-1 gene product P-gp and MRPs are also highly expressed on the brush border membrane of enterocytes. MDR gene products are “primary active” drug efflux pumps that bind and transport drug molecules against a concentration gradient by an energy-dependent process. This process plays an important role in the efflux of xenobiotics back to intestinal lumen, thereby exposing the drug to further metabolism (Gottesman and Pastan 1993; Ambudkar et al. 1999). A recent review by Katragadda et al. (2005) from our laboratory has described in details the expression of efflux proteins and their role in oral drug delivery. CYP450 enzymes and P-gp/MRPs proteins act synergistically to reduce the oral bioavailability of their substrates (Parkinson 1996).

Efflux proteins

P-gp, a product of MDR gene, was first characterized as the ATP-dependent transporter responsible for the efflux of chemotherapeutic agents from resistant cancer cells (Endicottaland and Ling 1989). P-gp is an integral membrane protein (170 kDa) having two homologous halves, each consisting of one hydrophobic domain with six transmembrane segments and one hydrophilic nucleotide-binding domain. Two halves are connected by a short flexible linker polypeptide to form the functionally active 1280-amino-acid protein. P-gp encoding genes are found in hamster, mice, human, and other species. P-gp is a transporter protein, encoded by a small multigene family, described by MDR I, II, and III. P-gps from all three MDR genes are present in rodents, whereas human cells express P-gp belonging to class I and III only. P-gp was the first ATP-dependent transporter found in the liver, and represents the most studied member of the ATP binding cassette (ABC) family of transporters. MDR proteins in humans and other species play an important role in protecting cells from cytotoxic drugs (Borst et al. 2000). These efflux pumps exhibit a versatile ability to efflux xenobiotics from cells. These proteins cover a broad range of substrates such as cyclosporin A, taxol, dexamethasone, lidocaine, erythromycin, protease inhibitors, and many anticancer agents (Borst et al. 1993; Gottesman and Pastan 1993; Hofsliand and Nissen-Meyer 1989; Ueda et al. 1987, 1992; Cole et al. 1992; Deeley and Cole 1997). Nearly 200 proteins are involved in the active transport of substances across biological membranes. Most of these proteins belong to an ABC superfamily. One or more ATP binding sites characterizes this family of transmembrane proteins. This includes P-gp, MRPs, breast cancer resistance protein (BCRP), bile salt export pump, sodium bile acid cotransporter, and the organic anion transporter (Cvetkovic et al. 1999). Localization of P-gp, MRPs and BCRP in the intestinal epithelium is depicted in Fig. 1.

Figure 1.
figure 1

Localization of efflux transporters in the intestine. Arrows indicate the direction of efflux.

Full size image

P-gp is expressed in a broad spectrum of tissues including the adrenals, cells of the blood–brain barrier, kidney, liver, lungs, and pancreas (Gottesman and Pastan 1993; Borst et al. 1993; Gottesman and Pastan 1993). Mammalian P-gps display approximately 60–65% homology with most P-gps from other species, suggesting that their role in drug trafficking is highly conserved throughout evolution. Transfection studies have shown that class I and II isoforms confer MDR, whereas class III isoform represents phosphatidylcholine translocation responsible for the secretion of phospholipids into bile (Gottesman et al. 1996; Sharom 1997). Evidence accumulated to date suggests that the transporter interacts directly with nonpolar substrates within the membrane environment, and may act as a drug flippase, transferring drugs from the inner to the outer leaflet of the bilayer, although a clear mechanism of translocation has not yet been established. Chemosensitizers that block the action of P-gp/MRPs appear to act as alternate substrates (Gottesman et al. 1996; Sharom 1997). MRPs, glycophosphoproteins belonging to the same ABC superfamily, are 190-kDa proteins and responsible for transport of drugs across lipid membranes from inside to outside the cell (Zaman et al. 1994; Krishnamachary and Center 1993). Although the role of P-gp in this drug interaction process has received some attention, the actual participation of MRPs and BCRP in this cascade also deserves further consideration. Unlike P-gp, MRPs primarily transport conjugated organic anions including cysteinyl leukotriene (LTC) (Xiao et al. 2005; Larkin et al. 2004). So far, eight different MRPs have been reported for the ABC proteins (Table 1). Human MRP-1, MRP-2, and MRP-3 are involved in the efflux of many therapeutic agents, i.e., antineoplastic drugs and their conjugated metabolites, floroquinoles, PIs (Letourneau et al. 2005; Bakos et al. 2000; Naruhashi et al. 2002; Hulot et al. 2005; Huisman et al. 2002). BCRP (ABCG2) is considered as ABC half transporter having only six transmembrane domains, extrudes a variety of unrelated compounds (anthracyclines, donarubicins, doxorubicins, campothecins, eporubicin) (Litman et al. 2000, 2001). A recent report suggests that BCRP imparts resistance to HIV-1 NRTI (zidovudine). Figure 2 diagrammatically represents the structural differences of P-gp, MRP, and BCRP.

Table 1 Human MRP gene family
Full size table
Figure 2.
figure 2

Diagrammatic representation of ABC transporters: (a) P-glycoprotein, (b) MRP, and (c) BCRP. NBD: nucleotide binding domain.

Full size image

Metabolizing enzymes

Role of CYP450 in drug–drug interactions

The most versatile enzyme system involved in the metabolism of xenobiotics is cytochrome P450. This enzyme system is responsible for the oxidative metabolism of a wide variety of compounds including PIs, nonnucleoside reverse transcriptase inhibitors (NNRTIs), cholesterol-lowering drugs, cardiovascular agents, steroids, psychotropic agents, antibiotics, azoles, and toxins. Enhanced metabolism and biliary clearance may affect drug efficacy, but delayed and lowered metabolism can lead to toxicity and other therapeutic interactions. These drugs can act as CYP inducers or inhibitors, or simply substrates. Depending on how the drugs interact with this enzyme system, it can be beneficial by enhancing blood levels or detrimental leading to therapeutic failure. Drug–drug interactions can become very complex in multidrug regimen when all three combinations (inducers, inhibitors, and substrates) are administered such as Kaletra + atorvastatin, Kaletra + ketoconazole, or Kaletra + rifampin. A recent review from our laboratory has discussed this enzyme system and its role in drug delivery (Katragadda et al. 2005). The CYP3A family of enzymes constitutes the most predominant phase I drug metabolizing enzymes and accounts for approximately 30% of hepatic CYP and more than 70% of intestinal CYP activity. CYP3A4 is a major congener of the CYP family known to metabolize more than 50% of currently administered drugs (Shen et al. 1997). This CYP3A4 enzyme is abundant in the hepatocytes and enterocytes (Guengerich et al. 1986; Katragadda et al. 2005; Parkinson 1996). A recent study indicates the expression of the polymorphic form of CYP3A5 in the small intestine, which significantly contributes to drug metabolism (Lown et al. 1994). Although hepatic biotransformation can make a major contribution to systemic drug elimination, a combination of hepatic and intestinal drug metabolism may cause significant presystemic or first-pass drug loss.

Because HIV protease inhibitors, macrolide antibiotics, and azole antifungals along with many therapeutic agents are substrates of CYP3A4, these compounds can alter the oral bioavailability of one another upon coadministration. Depending on the mechanisms of drug–drug interactions, multidrug regimens could lower blood levels of anti-HIV drugs, thus possibly placing people at risk for the development of drug resistance.

A significant amount of CYP3A4 is expressed in the enterocytes to metabolize xenobiotics during their transit across the intestinal epithelium (Shen et al. 1997). Although hepatic metabolism makes a major contribution to systemic drug elimination, the combination of hepatic and intestinal drug metabolism appears to have a significant effect on presystemic or first-pass drug loss. Saquinavir undergoes extensive first-pass metabolism by CYP3A4. Ketoconazole (selective CYP3A4 inhibitor) can inhibit the formation of all saquinavir metabolites. Also, saquinavir inhibits the metabolism of terfenadine and the formation of 6-β-hydroxylation products of testosterone (Vella and Floridia 1998). Metabolism of ritonavir, on the other hand, is caused by CYP3A4 and CYP2D6. It significantly inhibits the metabolism of CYP3A4 substrates such as nifedipine and CYP2D6 substrates such as dextromethorphan (Gottesman and Pastan 1993; Hsu et al. 1998). The major isozyme responsible for indinavir metabolism is CYP3A4, whereas the metabolism of nelfinavir is caused by several isozymes including CYP3A4, CYP2C19, CYP2D6, and possibly CYP2C9 and CYP2E1 (Williams and Sinco 1999; Malaty and Kuper 1999; Li and Chan 1999). In addition to oxidative metabolism, conjugation reactions may play an important role in detoxification of xenobiotics in the small intestine. Transporters responsible for cellular efflux of the conjugated metabolites and organic anions has been characterized recently (Leslie et al. 2001).

A combination of metabolic enzymes and efflux proteins play an important role in determining the bioavailability of xenobiotics from the intestinal lumen into the systemic circulation (Katoh et al. 2001). It is also true that most compounds, which are substrates for P-gp, are also substrates for CYP3A4. Because of a similarity in substrate specificity for both CYP3A4 and P-gp, these proteins appear to act synergistically (Chiou et al. 2000). Such coordinated function of active efflux and metabolism in the small intestine result in poor absorption of xenobiotics (Wacher et al. 1995, 2001; Watkins 1997).

Antiretroviral agents

Antiretroviral agents may be classified into three broad groups: (A) nucleoside and nucleotide reverse transcriptase inhibitors (NRTIs), (B) NNRTIs, and (C) PIs. Their affinities for efflux protein, P-gp, and metabolizing enzyme CYP3A4 are depicted in Table 2. This classification is based on available information relating to substrate specificities of antiretroviral agents. Accordingly, NRTIs belong to class A, which indicates that these drugs do not posses affinity for either P-gp or CYP3A4. NNRTIs are placed in class B, because these compounds are metabolized by CYP3A4 enzymes. PIs, which belong to class C, are substrates for both P-gp and CYP3A4. Thus, in multidrug regimens, drugs from class A may not influence the absorption of any drugs from other two groups, but it can influence the pharmacodynamics of drug action. However, interaction of NRTIs can occur as a result of competition for nucleoside transporter during influx and competitive inhibition of efflux by efflux proteins (MRPs) (Patel et al. 2004). Drugs belonging to class B (NNRTIs) and C (PIs) will interact, possibly as a result of competition for metabolic enzyme, CYP3A4. Therefore, combining drugs from class B and C may result in higher bioavailability because of metabolic inhibition leading to toxicity. Similar results are also expected if two drugs are selected from class C in multidrug therapy such as Kaletra where ritonavir and lopinavir both compete for P-gp and CYP3A4, thereby altering plasma concentrations. Appropriate addition of modulating agents, which are substrates for either CYP3A4 or, efflux proteins, or both, can result in higher oral bioavailability. In fact, interaction between two protease inhibitors is not always nonproductive; instead, it may be beneficial in some cases. Thus, ritonavir's unique interaction allows the drug to be a boosting master. Lopinavir's antiviral property is profoundly enhanced after a low dose combination with ritonavir in Kaletra, which inhibits lopinavir's efflux and metabolism. Currently, ritonavir is combined with all other PIs for its boosting effect, which resulted in the production of new Kaletra (ritonavir + tipranavir)—the most potent anti-HIV drug. Enfuvirtide (Fuzeon) is a new class of anti-HIV agent, known as fusion inhibitor. This drug needs to be administered by injection because it is degraded in the low stomach pH. So far, enfuvirtide has not produced any interaction with other antiretroviral medications.

Table 2 Absorption–metabolism classification of antiretroviral agents
Full size table

In general, NRTIs have mostly pharmacodynamic interactions because of their additive/synergistic effects. Because NRTIs are not substrates for CYP3A4 and are eliminated by kidney, these drugs have little kinetic impact on PIs and NNRTIs. In contrast, all NNRTIs are either inducer or inhibitors of CYP3A4. Nevirapine is a CYP3A4 inducer, whereas delavirdine serves as an inhibitor for the same enzyme and efavirenz has both inducer/inhibitor properties. Therefore, selection of NNRTIs in HAART regiment requires more attention because some combinations can lead to low bioavailability and others may exert toxicity of PIs. Dosages of PIs need to be carefully considered, because these drugs can induce/inhibit both P-gp and CYP3A4. Therefore, PIs cause a maximum number of drug–drug interactions, particularly if the other drugs are also substrate for P-gp and/or CYP3A4.

Interactions between PIs and drugs for opportunistic infections

HIV patients are highly susceptible to OIs caused by both bacteria and fungi and require antimycobacterial antibiotics and azole antifungal agents. Several antimycobacterials (clarithromycin, erythromycin, rifabutin, and rifampin) have been reported to interact with PIs (Table 3). In most cases, coadministration of clarithromycin with PIs caused the inhibition of CYP3A4 and possibly P-gp, resulting in higher AUC of PIs and clarithromycin (Table 3). Similarly, the AUC of saquinavir was elevated by 99% in the presence of erythromycin (Grub et al. 2001). In contrast, another macrolide, azithromycin (zithromax), has shown minimal effect on CYP3A4 and did not alter the blood level of indinavir (Foulds et al. 1999), and may serve as a better alternative to erythromycin. Rifabutin interacted with PI in a different manner (Table 3). Ritonavir caused a 3-fold elevation in rifabutin blood level due to inhibition of CYP3A4 (Cato et al. 1998; Norvir 2005) and possibly P-gp. In general, PIs elevated blood concentration of rifabutin with simultaneous alteration in PI levels. Coadministration of rifabutin and PI caused the reduced blood level of amprenavir, fosamprenavir, indinavir, nelfinavir, and saquinavir possibly as a result of the induction of CYP3A4 causing rapid metabolism. The possibility of higher efflux due to induction of P-gp cannot be ruled out. Interactions between PIs and antituberculosis drug rifampin are a growing concern, because rifampin is an inducer of both P-gp and CYP3A4 (Katragadda et al. 2005). Consequently, coadministration of rifampin with PIs constitutively decreased the blood level of PIs because of higher efflux and metabolism (Table 3). Thus, coadministration of PIs with rifampin may lead to subtherapeutic level of PIs.

Table 3 Drug–drug interactions among antimycobacterials and protease inhibitors
Full size table

Several azole antifungals are administered to treat OIs, and these agents are potent inhibitors of CYP3A4 as well as P-gp. Therefore, coadministration of azole with PIs will result in higher AUC of PIs due to decreased efflux and metabolism. This combination may cause drug toxicity unless the dosage regimen is properly adjusted. Table 4 summarizes the clinical interactions between PIs and ketoconazole, and itraconazole and fluconazole, respectively. An elevated blood concentration of PIs (amprenavir, indinavir, nelfinavir, ritonavir, and saquinavir) was evident, possibly as a result of inhibition of both efflux and metabolism. In contrast, coadministration of ketoconazole with Kaletra resulted in lower blood concentration of lopinavir, whereas ketoconazole level was elevated. As a result, Kaletra effect was diminished, but ketoconazole effect was enhanced significantly (Table 4) (Kaletra [package insert] 2005; CDC 2000a, b). Mechanism of interactions between Kaletra and ketoconazole is unknown. In the case of itraconazole, the blood levels of indinavir and ritonavir were elevated possibly as a result of inhibition of CYP3A4 (Albengres et al. 1998, Crixivan [package insert] 2005). When saquinavir was administered with itraconazole, saquinavir level remained unchanged but the half-life of itraconazole was prolonged (Table 4). However, interaction between fluconazole and PIs are similar to that of ketoconazole. Although blood concentration of indinavir was depressed, levels of ritonavir, saquinavir, and tipranavir were elevated possibly as a result of inhibition of CYP3A4 (Table 4). The role of efflux protein in these interactions is unknown.

Table 4 Drug–drug interactions among antifungals and protease inhibitors
Full size table

Cholesterol lowering drugs and PIs

Although HAART extends the life of HIV-infected population, controlling elevated blood cholesterol levels associated with PIs is a challenging tusk. Because high cholesterol and triglyceride levels are connected to cardiovascular problem, patients are instructed to take prescription for cholesterol-lowering drugs, i.e., the statins. This group of drugs is also metabolized by CYP3A4 and may therefore cause higher PIs levels in the blood. However, all of these medicines do not interact with all the PIs. Atorvastatin (Lipitor) has no significant effect on the blood level of fosamprenavir (Lexiva [package insert] 2005), lopinavir/ritonavir (Kaletra [package insert] 2005; Carr et al. 2000), and tipranavir(Aptivus [package insert] 2005). Similarly, pravastatin (Pravachol) and fluvastatin (Lescol) appears to have lower interactions with most PIs (Kaletra [package insert] 2005; Carr et al. 2000). However, PIs can have adverse effect on these statins. For example, blood levels of simvastatin were dangerously elevated when concurrently taken with indinavir, Kaletra, nelfinavir, and saquinavir (Kaletra [package insert] 2005; Hsyu et al. 2001; Viracept [package insert] 2004; Fortovase [package insert] 2003).

Cardiovascular medicines and antiretrovirals agents

Cardiovascular drugs (i.e., amiodarone, bepridil, diltazem, digoxin, dofetilide, ergotamine, and nifedipine) interact with PIs, and NNRTIs possibly result from inhibition of CYP3A4 by PIs. Digoxin is known to interact with efflux proteins. Profound drug–drug interactions are evident upon concomitant therapy with digoxin and rifampin. Plasma concentration of digoxin after oral administration was significantly diminished in combination with rifampin. Rifampin is a potent CYP3A4 inducer (Kolars et al. 1992). This antibiotic appeared to induce intestinal P-gp by 3.5-fold and CYP3A4 by 4.4-fold (Greiner et al. 1999; Schuetz et al. 1996). Therefore, rifampin caused a higher efflux of digoxin because of induction of P-gp expression in the intestine and higher metabolism by induced CYP3A4 both in the intestine and liver. Both effects reduced the oral bioavailability of digoxin. However, blood concentration of digoxin was elevated only by 29% following the coadministration of ritonavir possibly as a result of inhibition of P-gp-mediated efflux and CYP3A4-mediated metabolism (Norvir [package insert] 2005; Penzak et al. 2003).

Immune modulators and antiretrovirals agents

The immune modulator drugs are not free from adverse interaction when coadministered with antiretroviral agents. Several immune modulators such as cyclosporin, interleukin-2, mycophenolate, rapamycin, sirolimus, and tacrolimus are prescribed in conjunction with anti-HIV medications. Among these, the most interacting drug is cyclosporine. The simultaneous administration of saquinavir with cyclosporin resulted in a 300% increase in blood levels of cyclosporin with increased immunomodulating effect and high renal toxicity (Brinkman et al. 1998). Similar elevation in cyclosporin have been reported when it was coadministered with Kaletra or nelfinavir (Kaletra [package insert] 2005; Viracept [package insert] 2004). Cyclosporin is a well-known substrate for both P-gp and CYP3A4. The enhanced immunomodulatory effect and flux of cyclosporin are attributable to competitive inhibition of intestinal P-gp and CYP3A4 and hepatic CYP3A4 by the antiretrovirals.

Recreational drugs and antiretroviral agents

Interactions between recreational drugs [e.g., benzodiazepines, cocaine, lysergic acid dithylamide (LSD), marijuana, amphetamine (Meth), 3,4-methylenedioxymethamphetamine (MDMA), and opiates] and anti-HIV medications can cause drug toxicity, and influence clinical outcomes of HAART, mostly as a result of altered metabolism. In a recent a review, Wynn et al. (2005) have nicely described the interactions of antiretrovirals and drugs of abuse. Many benzodiazepines (midazolam, trizolam, and alprazolam) are substrates and inhibitor for P-gp and metabolized by CYP3A4. During coadministration of ritonavir with alprazolam, initial clearance of alprazolam diminished leading to enhancement of alprazolam toxicity (Greenblatt et al. 2000). Such effect was reversed following prolonged coadministration (12 days) of these two drugs. In the later case, alprazolam AUC diminished significantly (Venkatakrishnan et al. 1998). Such discrepancy is possibly attributable to ritonavir's initial inhibitory effect on CYP3A4 followed by induction of CYP3A4 with chronic dosing. Such effects can result from both short-term inhibition and long-term induction of P-gp by ritonavir. Inhibition of CYP3A4 and/or P-gp may cause alprazolam, midazolam, and trizolam levels to rise, and will therefore cause toxicity, whereas induction of CYP3A4 and/or P-gp will lead to withdrawal symptoms and therapeutic failure. Flunitrazepam is primarily metabolized by CYP3A4 and 2C19. Ritonavir coadministration may result in flunitrazepam toxicity including hypotension, confusion, and aggressive behavior (Smith et al. 2002). Other benzodiazepines, such as lorazepam, oxazepam, temazepam, and diazepam, are metabolized by non-CYP3A4 enzymes such as uridine 5′-diphosphate glucuronosyltransferase (UGT), UGT1A1, UGT1A3, and UGT2B7, but can also undergo drug interactions with antiretrovirals.

Cocaine is metabolized to nor-cocaine primarily by CYP3A4 (Ladona et al. 2000; LeDuc et al. 1993). Therefore, concomitant administration of cocaine with any PIs, particularly ritonavir, will result in overdose effect (acute response) or withdrawal symptom (prolonged response) due to inhibition and induction of CYP3A4 by ritonavir, respectively. Coadministration of cocaine with PIs, efavirenz, ketoconazole, nefazodone, erythromycin, and clarithromycin, i.e., the CYP3A4 inhibitors, will result in cocaine toxicity. In contrast, drugs (nevirapine, rifampin) that induce CYP3A4 may shift the metabolism from hydroxylation to N-demethylation, thereby producing higher levels of toxic metabolites (Bornheim 1998; Pellinen et al. 1994). Many HIV-infected patients use gamma hydroxybutyrate (GHB). However, its metabolism is not established in human. Harrington et al. (1999) reported GHB toxicity in HIV-positive individuals on ritonavir and saquinavir therapy, also consuming MDMA regularly. The authors concluded that PI-mediated inhibition of CYP3A4 was likely the cause of such toxicity.

Both amphetamine and MDMA have potential interactions with ritonavir and delaviridine ( Pau 2002; Henry and Hill 1998). MDMA is demethylated to 3,4- dihydroxymethamphetamine primarily by CYP2D6 with minor contributions from CYP1A2, CYP2B6, and CYP3A4 (Lin et al. 1997; Maurer et al. 2000). Besides CYP3A4, ritonavir can inhibit CYP2D6, CYP2C9, and CYP2C19 (Harrington et al. 1999). Fetal drug–drug interactions were observed in HIV patients receiving MDMA with zidovudine, lamivudine, and ritonavir in their treatment regimen (Henry and Hill 1998). Inhibition of CYP2D6 by ritonavir is likely the cause of enhanced MDMA toxicity. A nonlinear MDMA pharmacokinetic pattern was observed in a study on 14 patients with CYP2D6 genotype (de la Torre et al. 2000). Metabolism of amphetamine and MDMA, although not completely understood, CYP2D6 genetic variability, and possibly P-gp effect may cause adverse toxicity, taking CYP2D6 inhibitors such as ritonavir, bupropion, fluxetine, and quinidine (Ketabi-Kiyanvash et al. 2003; Wynn et al. 2005). Although marijuana, with its active ingredient identified as tetrahydrocannabinol (THC), is metabolized by CYP3A4 and CYP2C9 (Matsunaga et al. 2000), no clinically significant drug interaction was observed in two separate studies (Kosel et al. 2002; Abrams et al. 2003).

A detailed review of opiate interactions with other drugs has been reported by Armstrong and Cozza (2003a, b). Narcotic analgesics [phenylpiperidines, meperidines, fentanyl, pseudopiperidines (methadone), and propoxyphen] and alkaloids (natural: heroin, morphine, and codeine; semisynthetics: hydromorphone, oxymorphone, hydrocodone, oxycodone, dihydrooxycodeine, and buprenorphine) are widely indicated in AIDS patients. Many of these opiates are metabolized by CYP3A4 and substrates/inhibitors/inducers for P-gp (Table 5), posing a serious risk of causing potential drug–drug interactions with multidrug regimens in HAART. Ritonavir dosing in healthy subjects caused a decrease in AUC of meperidine along with a rise in normeperidine levels indicating enhanced metabolism (Piscitelli et al. 2000). Lopinavir + ritonavir (Kaletra), ritonavir, nevirapine, and efavirenz were found to cause opiate withdrawal when given in combination with methadone (Altice et al. 1999; Beauverie et al. 1998; Pinzani et al. 2000; McCance-Katz et al. 2000; Bart et al. 2001). Methadone is primarily metabolized by CYP3A4 with minor contributions from CYP2D6, CYP2C9, and CYP2E1. It is a substrate and inducer of P-gp. The mechanisms behind this opiate withdrawal may involve drug–drug interaction due to induction of P-gp and CYP3A4 by antiretrovirals resulting in enhanced efflux and metabolism of methadone. Metabolisms of heroin, morphine, and codeine occur independently of CYP 450 system. These alkaloids are first metabolized by liver and plasma esterases, and then by UGT2B7 and UGT1A3. Also, several sedatives (hydromorphone, oxymorphone, and dehydrocodeine) are mainly metabolized by UGT2B7 and UGT1A3 with a minor contribution from CYP2D6. Therefore, these alkaloids are unlikely to interact with antiretrovirals. In contrast, semisynthetic opiates (hydrocodone, oxycodone, and buprenorphine) are mainly metabolized by CYP3A4 and CYP2D6. Therefore, inhibition or induction of CYP2D6 and CYP3A4 by retrovirals is likely to produce variable analgesic effects of these agents.

Table 5 Drugs interact with P-glycoprotein
Full size table

PIs and NNRTIs are either substrate/inducer/inhibitor for CYP3A4 or P-gp or both and any drug that interacts with this efflux proteins and metabolizing enzyme will generate drug–drug interactions resulting in altered bioavailability and lead to synergism or antagonism. A list of these drugs was presented in our earlier review (Katragadda et al. 2005).

In vitro testing for drug–drug interactions

The effect of MRPs in drug efflux is yet to be established. In clinical settings, it is difficult to determine the interactions of efflux proteins with multidrug regimens. Therefore, we performed in vitro experiments to determine the role of efflux pump in drug absorption. By using MDR-transfected MDCK cell line, several experiments were designed to evaluate such drug–drug interactions particularly for their effects on absorption. Uptake of ritonavir was examined in the presence of ketoconazole, a potent inhibitor of P-gp. Such uptake of ritonavir was challenged with FK571, which is potent inhibitor of MRP. A significant elevation in ritonavir uptake was evident in the presence of both Ketoconazole and MK571 (Fig. 3). This higher uptake was probably caused by inhibition of P-gp and MRP with ketoconazole and MK571, respectively. When both ketoconazole and MK571 were added, uptake of ritonavir was further enhanced compared to ketoconazole alone indicating that ritonavir is effluxed by both P-gp and MRP. Once such efflux was stopped by the inhibitors, higher intracellular uptake of ritonavir was observed. Another experiment was conducted to establish P-gp-mediated ritonavir efflux. This study was conducted in a Transwell plate, where cells were grown on semipermeable membrane. Polarized MDCK-MDR-1 cells express P-gp on the apical surface. During a transport run, ritonavir was added either onto the apical or basolateral surface, and samples were collected from the opposite side. Figure 4 shows that ritonavir transport was significantly enhanced in the presence of ketoconazole. Similar profile is expected in clinincal setting, if ketoconazole is administered with ritonavir, it will inhibit the P-gp-mediated efflux of PI resulting in higher transport. We next examined the transport interaction between erythromycin and saquinavir. Both erythromycin and saquinavir are substrates for P-gp and MRP. Thus, uptake of erythromycin was augmented by 3-fold in the presence of saquinavir possibly as a result of competitive inhibition of both P-gp and MRP (Fig. 5). A recent study from our laboratory revealed that erythromycin flux was enhanced in the presence of some fluoroquinolones (sparfloxacin, levofloxacin, and grepafloxacin) possibly as a result of competitive inhibition of P-gp and MRP (Fig. 6). This result led us to propose that the simultaneous administration of fluoroquinolones and macrolides could minimize the efflux and metabolism of both drugs. Such interaction would result in reduction of the required dose, thus preventing dose-related side effects and development of resistance by gram-positive bacteria such as Streptococcus pneumoniae and Legionella pneumoniae (Sikri et al. 2004). Several fluoroquinolones, such as norfloxacin, ofloxacin, grepafloxacin, enoxacin, and lomefloxacin, were evaluated for P-gp-mediated efflux of cyclosporin (Fig. 7). These in vitro studies indicate that grepafloxacin, levofloxacin, and sparfloxacin are potent inhibitors of P-gp-mediated efflux. We have also examined the CYP3A4-mediated drug–drug interactions by using an in vitro system. Carbamazepine is a substrate for CYP3A4. Both erythromycin and ketoconazole caused significant reduction of carbamazepine metabolism (Fig. 8).

Fig. 3
figure 3

Uptake of 3H ritonavir across MDCK-MDR1 cell monolayers in the presence of Ketoconazole (50 μM) and MK571 (25 μM). p < 0.05; n = 3–5; ±1 SD.

Full size image
Fig. 4
figure 4

Transport of 3H ritonavir across MDCK-MDR1 cell monolayers in Transwell® inserts in the presence of ketoconazole. 3H ritonavir permeability in apical to basolateral (A–B) direction was significantly enhanced (p < 0.05) in the presence of ketoconazole (50 μM). P app (A–B) cm/s × 10−6, control (3.0 ± 0.12), and ketoconazole (12.7 ± 1.31); n = 4–5; ±1 SD.

Full size image
Fig. 5
figure 5

Uptake of 3H erythromycin in the presence of saquinavir (75 μM) across MDCK-MDR1 cell monolayers. Erythromycin uptake was significantly elevated (p < 0.05) due to inhibition of efflux proteins (P-gp and MRP) by saquinavir. n = 4; ±1 SD.

Full size image
Fig. 6
figure 6

Uptake of 14C erythromycin across MDCK-MDR1 cell monolayers in the presence of various fluoroquinolones. Inhibition of P-gp/MRP-mediated efflux by grepafloxacin (1 mM), sparfloxacin (1 mM), and levofloxacin (1 mM) enhanced erythromycin absorption significantly. p < 0.05; n = 3–5; ±1 SD.

Full size image
Fig. 7
figure 7

Uptake of cyclosporin A in the presence of various fluoroquinolones across MDCK-MDR1 cells.

Full size image
Fig. 8
figure 8

Erythromycin and ketoconazole resulted in a decrease of carbamazepine metabolism due to inhibition of microsomal CYP3A activity. p < 0.05; n = 5; ±1 SD.

Full size image

Integrated functions of efflux and metabolism

P-gp is a robust efflux transporter and there are so far no structure delineations for its substrate specificity, but a number of drugs are found to interact with this transporter. CYP3A4 is the major phase 1 drug-metabolizing enzyme in human that metabolizes approximately 70% of xenobiotics. Therefore, when two drugs are taken orally, it is very likely that they will interact via metabolism. In human, CYP3A4 is the key enzyme responsible for intestinal and hepatic metabolism, and both P-gp and CYP3A4 have overlapping substrate specificity. Also, both proteins coexpress in the intestine. As a result, the concomitant administration of cyclosporine with rifampin, an inducer of both proteins, lowers the oral bioavailability (C max) of cyclosporine. Similarly, saquinavir concentration was reduced by 80% after coadministration with rifampin. Conversely, ketoconazole, an inhibitor of P-gp/CYP3A4, caused the elevation of cyclosporine levels (C max) in the blood resulting in higher toxicity (Wacher et al. 1998). However, the concomitant administration of ritonavir with saquinavir led to 5- to 6-fold increase of plasma concentration of saquinavir (C max) as a result of inhibition of both CYP3A4 and P-gp by ritonavir (Wacher et al. 1998). In addition to oxidative metabolism, conjugation reactions may play a crucial role in detoxification of xenobiotics from the small intestine. Many drug molecules are effluxed into intestinal lumen after being conjugated with a glucuronide or sulfate moiety. Transport mechanism for the cellular extrusion of the conjugated metabolites and organic anions has been described recently (Leslie et al. 2001). Several antiretroviral agents and other drugs coadministered in AIDS patients can pharmacokinetically act as inhibitors or inducers. However, coadministration of clarithromycin, an inhibitor of CYP3A4 with indinavir, did not alter the blood levels of indinavir (Boruchoff et al. 2000). Administration of antiretrovirals concurrently with other therapeutic agents may lead to increased absorption as a result of inhibition of P-gp-mediated efflux and CYP-mediated metabolism leading to potential toxicity. The chronic administration of certain PIs will enhance the production of MDR proteins (P-gp) and CYP enzymes, resulting in lower bioavailability and subtherapeutic plasma concentrations of coadministered agents. If the coadministered drug is a strong inhibitor or inducer of P-gp/CYP3A4, results will be variable. For example, the concomitant administration of PIs and clarithromycin resulted in higher AUC of both drugs because of the strong inhibition of CYP3A4 (Table 3). In contrast, opposite results were observed during the coadministration of PIs + rifampin due to heavy induction of CYP3A4 and P-gp (Table 3). Therefore, it is conclusively evident that integrated function of CYP3A4 and P-gp can drastically alter the oral bioavailability and may lead to the emergence of drug resistance viral strains. An understanding of the increased/decreased bioavailability of one drug in the presence of another may greatly aid in the design of appropriate drug regimen. Figure 9 depicts the coordination of efflux and metabolism in GI tract.

Fig. 9
figure 9

Mechanism of drug interactions with efflux transporters and metabolizing enzyme CYP3A4 in enterocytes. Drug absorption is regulated by coordinated functions of both efflux transporters and CYP3A4. Because of efflux pump, limited amounts of drugs may enter the cell and then get metabolized by CYP3A4. Some metabolites can be pumped out of the cell into the lumen. Ultimately, very limited amounts drugs are absorbed and reach the blood for further hepatic metabolism. Therefore, any induction or inhibition of these two key factors results in altered bioavailability of drugs in a multidrug regimen.

Full size image

Activation of mammalian xenobiotic-sensing nuclear receptors PXR and CAR

P-gp/MDR1 is regarded as one of the major factors in the development of cellular drug resistance. Up-regulation of CYP enzymes in response to chronic administration of therapeutic agents may also play an important role in lowering their blood levels. A close chromosomal location of P-gp (MDR gene locus 7q21–7q21*7q21.1–7q21.1) and CYP3A4 (CYP3A4 gene locus 7q21.3–7q22.1*) genes, their expression in mature enterocytes, and similar substrate specificities indicate that the function of these two proteins are complementary in nature and may form a coordinated intestinal barrier (Katragadda et al. 2005; Patel et al. 2004).

Molecular mechanism of P-gp and CYP induction

Molecular mechanisms of P-gp and CYP induction are still being investigated by researchers. Figure 10 depicts a diagrammatic representation of molecular mechanism of such induction. The role of two members of the nuclear receptor superfamily of transcription factors, the pregnane X receptor (PXR) and the constitutive androstane receptor (CAR) have been recently reported. These receptors act as sensors for drugs including xenobiotics. The nuclear receptor superfamily consists of four modular domains: (1) a highly variable N-terminal region with an activation function (AF-1); (2) a DNA-binding domain with two zinc-finger motifs, a flexible domain, and a ligand binding domain (LBD) that also have an AF-2. After cellular entry, xenobiotics and other activators either trigger cytoplasmic-nuclear translocation of CAR by promoting the release of unknown proteins, or directly activate PXR in the nucleus. Subsequently, both PXR and CAR heterodimerize with retinoid X receptor (RXR) bind to their respective response elements on target gene and increase the transcription of those genes. In the flanking regions of several genes, response elements have been found that are activated by both PXR and CAR, thus allowing direct cross-communication between these two receptors. New insights into the transcriptional regulation of CYP3A4 revealed that vitamin D and its derivatives regulate the transcription by interacting with vitamin D response (VDR) element. This regulation of intestinal CYP3A4 by VDR is complementary to the apparent role of PXR in regulating hepatic CYP3A4 (Ambudkar et al. 2003; Handschin and Meyer 2003; Synold et al. 2001).

Fig. 10
figure 10

Activation of mammalian xenobiotic-sensing nuclear receptors PXR and CAR. Xenobiotic agent after entering the cell, triggers cytoplasmic CAR for nuclear translocation by an unknown mechanism, or directly activates PXR. Then, both CAR and PXR subsequently heterodimerize with RXR and bind to specific response elements leading to transcription.

Full size image

Conclusion

Drug–drug interactions may occur when more than one drug is indicated for concomitant therapy, particularly if overlapping mechanisms are involved in the absorption and/or metabolism of these agents. Drug efflux proteins and metabolizing enzymes are two central factors controlling the oral bioavailability of drugs and subsequently their therapeutic efficacy, and are therefore responsible for many drug–drug interactions. This review outlines such interactions that commonly occur in AIDS patients receiving concomitant multiple drugs in HAART and also other recreational drugs. In this review, we briefly discussed drug interactions of antiretrovirals with other agents causing alteration of oral bioavailability of each another. These compounds include drug classes comprising antimicobacterial, azole antifungals, statins, cardiovascular agents, immune modulators, and recreational drugs (benzodiazepines, cocaine, LSD, marijuana, Meth, MDMA, and opiates). A list of drugs that are substrates for P-gp is presented in Table 5. A number of these agents may interact with several PIs. Most clinical studies report primarily metabolism-mediated drug interactions. Interestingly, it appears that any agent that induces CYP may also simultaneously activate P-gp. So these two barriers act in harmony to eliminate xenobiotics and the magnitude of P-gp-mediated efflux on the coordination of drug metabolism seems to be dependent on the spatial relationship between P-gp and CYP3A4. Overlapping substrate specificities of these proteins result in complex and sometimes perplexing pharmacokinetic profiles of multidrug regimens. Therefore, a clear understanding of drug–drug interactions particularly with efflux transporters and metabolizing enzymes is pivotal to our understanding of such variable oral bioavailability and subsequent development of drug resistance as well as loss of anti-HIV activity from these antiretroviral drugs. Adverse drug interactions may be avoided simply by adjusting doses or substituting with drugs which are less or noninteracting.