Advances in HIV Pharmacology: Protein Binding, Pharmacogenomics, and Therapeutic Drug Monitoring

International AIDS Society–USA Topics in HIV Medicine

Perspective Advances in HIV Pharmacology: Protein Binding, Pharmacogenomics, and Therapeutic Drug Monitoring

Developing better antiretroviral drugs, nomics before this strategy finds a 50% inhibitory concentrations (IC50) of individualizing therapy through pati- place in clinical practice. Finally, studies PIs in the presence of AAG in vitro were ent genetic profiling, and maintaining are mixed on the benefits of TDM; correlated with reported binding affini- effective drug concentrations with although such monitoring may be ties: those drugs with higher AAG bind- therapeutic drug monitoring (TDM) appropriate in some settings, such as ing affinity showed less potent inhibi- represent 3 current areas of interest in pregnancy and pediatrics, data are cur- tion in the presence of that protein the field of HIV pharmacology. This rently lacking to support its routine use (Lazdins et al, J Infect Dis, 1997; Zhang article first examines antiretroviral in HIV care. Although data on these et al, J Infect Dis, 1999). Albumin also drug binding to plasma proteins, a fac- pharmacologic strategies do not cur- was found to increase PI IC values, but tor that affects the amount of free rently support their widespread clinical 90 drug available to enter cells. Protein application, ongoing research of such to a lesser degree than AAG (Molla, binding influences drug development, strategies offers hope for future Virology, 1998). raising questions about whether the improvement of the efficacy of anti- Given the inverse correlation be- drug levels required for appropriate retroviral therapy. This article summa- tween binding affinity and drug activity therapeutic effect can be achieved at rizes a presentation given by Charles in vitro, the question is whether protein tolerable doses. Second, individualized W. Flexner, MD, at the November 2002 binding affects in vivo performance of antiretroviral therapy has generated International AIDS Society–USA course highly protein-bound drugs. There are considerable interest, but much work in San Diego. some direct consequences of protein remains in the area of pharmacoge- binding in terms of clinical use. For example, cerebrospinal fluid concentra- Developing antiretroviral therapy that is protein binding assess binding to tions of many highly protein-bound drugs, including highly bound PIs, cor- more potent, safer, and better tolerated alpha1-acid glycoprotein (AAG) or albu- by patients requires consideration of min. AAG, which accounts for only relate better with plasma concentra- several factors, including drug binding to about 1% to 3% of plasma proteins, tions of free drug than with total drug plasma proteins, which affects the binds drug molecules with low capacity plasma concentrations. For a few highly amount of free drug available in the but high affinity, with the latter charac- protein-bound drugs, such as pheny- body. Individualizing drug therapy teristic making dissociation of the drug toin, procainamide, and lidocaine, free through patient genetic profiling and molecule from AAG more difficult than drug concentration correlates better therapeutic drug monitoring (TDM) are from albumin. Although albumin is a with activity than does total plasma also areas of interest in HIV pharmacol- major protein component of plasma, it concentration. However, for a number ogy. Each of these topics is discussed is a high-capacity but low-affinity of reasons, plasma protein binding does below. binder. Studies of protease inhibitors not generally need to be compensated (PIs) that test how much the inherent for in clinical use. These reasons include Plasma Protein Binding fluorescence of AAG is quenched by the fact that many drugs can be dosed high enough to achieve therapeutic lev- Antiretroviral drugs differ in the degree binding to drug molecules have shown a wide range of drug binding affinities for els of free drug even if they are highly to which they are bound to plasma pro- protein-bound, and that drug concentra- teins. Plasma protein binding is a con- AAG (Bakker et al, 12th World AIDS Conf, 1998). Of PIs tested in these stud- tions are significantly affected by other cern in drug development because, in aspects of drug pharmacokinetics. general, only free drug can penetrate ies, indinavir had the lowest affinity for cells or tissues and exert its therapeutic AAG, with an equilibrium association × 1 –1 Protein Binding In Vivo effect. In the case of antiretroviral drugs, constant of less than 1 10 M , fol- × 4 –1 any factor that reduces free-drug con- lowed by ritonavir at 1 10 M , nelfi- Drugs that bind to plasma proteins bind × 5 –1 × 5 centrations could in theory reduce drug navir at 2 10 M , saquinavir at 8 10 to and dissociate from those proteins at activity and thus promote HIV resis- M–1, and the investigational drug SC- particular rates, termed association and tance. 52151 at 2×106 M–1. SC-52151 thus had dissociation rates. At equilibrium, as For the most part, studies of plasma a binding affinity approximately 2 mil- much drug is associating with protein as lion times greater than indinavir and is dissociating at any given time, and Dr Flexner is Associate Professor of approximately 200 times greater than there is a constant concentration of free Medicine, Pharmacology, and Inter- ritonavir. The potential effect of greater drug. However, protein binding is not national Health at the Johns Hopkins binding affinity on activity against HIV the sole determinant of the amount of University in Baltimore, Md. is indicated by studies showing that the free drug that is available for therapeu-

40 Perspective - Advances in HIV Pharmacology Volume 11 Issue 2 March/April 2003

tic activity. For example, a highly pro- SC-52151, for example, was stopped the HLA-B27 marker and ankylosing tein-bound drug may be as potent in not because of the drug's high degree of spondylitis, which is one of the strong- vivo as a drug that is less protein- protein binding but because the drug est recognized genetic associations for a bound, since greater amounts of the could not be dosed to achieve an ade- common disease. However, another latter may be available for elimination quate anti-HIV effect in vivo. This was study that was performed in a larger or for entering therapeutically irrele- largely due to the drug's poor water sol- and more heterogeneous population, vant sites (eg, molecules, cells, or ubility, which required SC-52151 to be reported at about the same time, found organs other than the target sites). administered in an elixir that contained that only about 45% of patients with With regard to the impact of metab- large amounts of ethanol and thus limit- abacavir sensitivity had the HLA-B5701 olism and elimination on therapeutic ed dose (Fischl et al, J Acquir Immune marker (Hetherington et al, Lancet, activity as it relates to protein binding, Defic Syndr Hum Retrovirol, 1997). 2002). the average time to circulate plasma Other studies of genetic markers through the liver of an adult is 9 sec- Individualizing Treatment: have indicated weaker associations and onds. In an average individual, every Pharmacogenomics yielded findings that are more difficult molecule of a drug, even drugs that are to interpret than those in the Mallal highly protein-bound, is estimated to There is considerable enthusiasm about study, which may prove to be more typ- be free in the liver every few minutes, the prospect of individualizing antiretro- ical of data emerging in this field. One providing ample opportunity to clear viral therapy based on genetic profiling study examined the association of free drug for those agents that are of patients. However, much research mutations in the gene encoding the P- metabolized. In short, determining clin- glycoprotein drug transporter (the gene ical potency of a drug is much more associated with multidrug resistance in complicated than would be represented cancer chemotherapy) with outcomes by consideration of protein binding A problem in translating of antiretroviral treatment in HIV-infect- alone. ed patients (Fellay et al, Lancet, 2002). genetic findings into The investigators found that having the Modulating Protein Binding thymidine-thymidine (TT) genotype at Some drugs are known to affect AAG clinical practice is that position 3453 of the gene, rather than levels (eg, phenobarbital increases AAG cytidine-thymidine or cytidine-cytidine levels in cats), and it is possible that (CC), was associated with lower trough antiretroviral protein binding could be many associations concentrations of nelfinavir and modulated through use of drugs that efavirenz but higher CD4+ cell counts upregulate or downregulate AAG. do not pinpoint a after 6 months of treatment. The TT Studies to determine whether PIs that genotype is found in 25% of white are cytochrome P450 (CYP 450) induc- single gene responsible patients and in 13% of African- ers affect AAG levels found that, after 5 American patients, and a smaller study weeks of treatment (during which for a biologic effect (Wegner et al, 9th CROI, 2002) suggest- steady state was achieved in all ed that efavirenz, one of the drugs patients), neither nelfinavir nor riton- affected by the TT genotype, may be avir altered AAG levels in HIV-infected less effective in African-American patients (Flexner et al, 12th World AIDS remains to be done before this prospect patients than in white patients. African- Conf, 1998). There was some variabili- becomes reality. In a recent study per- American patients receiving efavirenz ty in AAG response among patients, formed by Mallal and colleagues had a significantly less durable plasma with levels increasing in some and (Lancet, 2002) in a relatively homoge- HIV-1 RNA response and a 2- to 3-fold decreasing in others; however, no sup- neous population of individuals of most- higher risk of relapse than did white raphysiologic AAG levels that might ly English or Irish descent in Western patients, with the time to treatment fail- have substantially reduced free drug Australia, 14 (78%) of 18 HIV-infected ure being approximately 400 days ver- were observed in any patients. Other patients with hypersensitivity to aba- sus 1400 days. The study concluded studies of the effects of drugs or HIV cavir had HLA type HLA-B5701, com- that these disparities were probably not disease on AAG levels similarly suggest pared with only 4 (2.4%) of 167 associated with differences in drug con- that in most cases the impact is not suf- patients with abacavir tolerance, yield- centrations or adherence, and no com- ficient to substantially alter free drug ing an odds ratio for sensitivity among parable racial differences were obser- concentrations long term. the former of 117. The combination of ved with nelfinavir or indinavir. For the most part, the problem of the 3 genetic markers HLA-B5701, HLA- Several factors may make it difficult protein binding is solved during clinical DR7, and HLA-DQ3 was present in 13 to assess the effect of the TT genotype drug development, by ascertaining (72%) of the abacavir-sensitive patients on antiretroviral treatment. This geno- whether therapeutically meaningful and in none of the abacavir-tolerant type is associated with lower efavirenz drug levels and good therapeutic effect patients, yielding an odds ratio for the concentrations but better CD4+ cell are achieved at tolerable doses. former group of 822. This association is count responses. It is linked with lower Development of the investigational PI similar in strength to the link between concentrations of some drugs (eg, fex-

41 International AIDS Society–USA Topics in HIV Medicine

ofenadine, as well as efavirenz) but leagues. The study found a fair amount they are highly metabolized by the CYP higher concentrations of others (eg, of overlap and lack of specificity 450 system, particularly CYP 3A4, with digoxin). Further, the TT/CC polymor- between these 2 patient groups in the some PIs being CYP 450 inducers, some phism is “silent” in that it does not HLA-B5701 locus. However, in another inhibitors, and some both. Levels of PIs affect the sequence or structure of the part of this immune response region of can be affected by the many other protein produced. Finally, the odds ratio the chromosome, encoding genes for drugs metabolized via the 3A4 enzyme for the impact of the TT mutation on the heat shock proteins, there was no over- system and by the inherent variability anti-HIV effect of efavirenz is weak, lap, suggesting that the gene responsi- of metabolism via this route. suggesting a weak association; indeed, ble for abacavir hypersensitivity actual- A number of studies have evaluated a number of studies that have yet to be ly resides in this region. Thus, HLA- use of TDM in patients receiving PI- published have not found an association B5701 is tightly linked to the trait for based antiretroviral therapy. In the between this genotype and antiretrovi- abacavir hypersensitivity but is not the PharmAdapt study, 256 treatment-expe- ral drug concentrations. gene responsible for this biologic effect. rienced patients were randomized in A major problem in translation of The specific causative gene remains to unblinded fashion to HIV genotyping or genetic findings into clinical practice is be identified. genotyping plus pharmacokinetic analy- that many associations do not pinpoint sis (Clevenbergh et al, 8th CROI, 2001; a single gene that is responsible for a Guaranteeing Success: TDM Clevenbergh et al, 41st ICAAC, 2001). biologic effect. Rather, they represent Genotyping and pharmacokinetic analy- an association between a previously There is considerable interest in moni- sis were performed at week 4, with identified marker and a biologic effect; toring antiretroviral drug levels in HIV- treatment modified on the basis of this it often remains to be determined infected patients to maintain concentra- information at week 8. Overall, there whether the marker is linked to another tions that provide maximal therapeutic was no difference between the 2 groups locus that is actually responsible for the effect with the least possible toxicity. with regard to virologic response at 12 biologic effect. Figure 1 shows a com- Particularly with regard to PIs, trough weeks, with plasma HIV-1 RNA levels parison of all marker variants in the serum concentrations are often predic- below assay detection limits in 43% of abacavir-sensitive and abacavir-tolerant tive of virologic outcome. There is a patients in the genotyping/TDM arm patients studied by Mallal and col- clear rationale for TDM for PIs, since and in 50% of patients in the genotyp-

Chromosome 6p21.3 D6S1014*137 D6S273*135 MICA*194 MIB*344 TNF-238A DQ3/DR7NOTCH4C4A6 HSP MEGT1 B5701 C6 Centromere Telomere

Abacavir-Hypersensitive (n = 14)

Marker Present and Explained by 57.1 Haplotype

Marker Present but Explained by Abacavir-Tolerant non-57.1 Haplotype (n = 12)

Possible Susceptibility Region

Figure 1. Genetic mapping of the abacavir hypersensitivity region in abacavir-treated patients with polymorphisms as indicated. Adapted with permission from Mallal et al, Lancet, 2002.

42 Perspective - Advances in HIV Pharmacology Volume 11 Issue 2 March/April 2003

ing-only arm. However, a number of factors make these findings difficult to interpret. First, the target drug concentrations were relatively low, equivalent to the protein-adjusted IC50. In addition, about 60% of patients in both arms were receiving ritonavir, which acts pharmacokinetically to boost levels of other PIs. Finally, since 8 weeks of treatment elapsed prior to changes based on the pharmacokinetic analysis, modifica- tions based on this information may have been made too late to prevent development of viral resistance. It should also be noted that there was intrasubject variability with regard to drug levels, with some patients moving Figure 2. Association of decrease in plasma HIV-1 RNA levels with amprenavir normalized from “suboptimal” to “optimal” concen- inhibitory quotient (IQ) in patients receiving amprenavir. Adapted with permission from Piscitelli, ECCATH, 2001. trations between week 4 and week 8 with no change in drug dose. The GENOPHAR study had a design TDM at all were observed in patients noted above; in patients with renal or similar to PharmAdapt (eg, genotyping receiving indinavir/ritonavir. hepatic dysfunction; and in documenta- and pharmacokinetic analysis of treat- On balance, the available data sug- tion of adequate drug levels in the pres- ment-experienced patients at week 4 gest that a somewhat different ence of other drugs known to induce or and change in regimen at week 8), approach to TDM is needed in treat- inhibit the CYP 450 system. although it was conducted in blinded ment-experienced patients. Three fashion (Bossi et al, 9th CROI, 2002). recent studies reported a better correla- Presented by Dr Flexner in November 2002. First draft prepared from transcripts by Matthew This study also showed no difference in tion between drug concentrations and Stenger. Reviewed and updated by Dr Flexner in virologic outcome between genotyp- treatment outcome in treatment-experi- February 2003. ing/TDM and genotyping alone; as with enced patients if correction was made the PharmAdapt study, however, target for the level of drug resistance in the Financial Disclosure: Dr Flexner has received patient's viral population. Methods by grants and research support from Abbott, drug levels may have been too low and Agouron, and Merck. changes at week 8 based on pharma- which this correction can be achieved cokinetic analysis may have been made include measurement of the inhibitory too late. A third study performed in quotient (IQ), which is the trough con- treatment-experienced patients, the centration divided by the IC50 for the Suggested Reading GART study, also showed no benefit of drug; the virtual IQ, which is the trough treatment based on optimal PI levels. concentration divided by the virtual IC50 Back D, Gatti G, Fletcher C, et al. Therapeutic (Baxter et al, AIDS, 2000) derived from a virtual phenotype drug monitoring in HIV infection: current status The ATHENA study has provided database; and the normalized IQ, which and future directions. AIDS. 2002;16:S5-S37. some evidence of benefit of TDM in is the patient’s virtual IQ divided by a Bakker J, Tazartes D, Flexner C. A fluorescence treatment-naive patients. In this “blind- population mean virtual IQ for a patient quenching assay for determining the binding ed” study, all patients underwent TDM, with drug-sensitive virus. Figure 2 affinity (Ka) of HIV protease inhibitors to alpha- with their physicians given either dos- shows the correlation between ampre- 1 acid glycoprotein. [Abstract 42268.] 12th ing advice based on monitoring or no navir normalized IQ and decrease in World AIDS Conference. June 28-July 3, 1998; Geneva, Switzerland. advice (Burger et al, 1st IAS Conf, plasma HIV-1 RNA level found by 2001). Advice resulted in significant Piscitelli and colleagues (ECCATH, 2001) Baxter JD, Mayers DL, Wentworth DN, et al. A decreases in discontinuation rates at 1 indicating greater decreases in viral load randomized study of antiretroviral manage- year among patients receiving nelfi- with higher IQ values. ment based on plasma genotypic antiretroviral navir (2.4% vs 17.6% with no advice) or Much work remains to be done resistance testing in patients failing therapy. before clinical guidelines for TDM can CPCRA 046 Study Team for the Terry Beirn indinavir (9.5% vs 40.0%). However, Community Programs for Clinical Research on results of this study are also difficult to be developed. Until then, such monitor- AIDS. AIDS. 2000;14:F83-F93. interpret, since many practitioners who ing may be of benefit in some clinical received advice based on TDM did not situations. These include the settings of Bossi P, Peytavin G, Delaugerre C, et al. institute it. Further, while virologic pregnancy and pediatrics, in which drug GENOPHAR: A randomized study of plasmatic response improved among patients concentrations can change rapidly and drug measurements associated with genotypic resistance testing in patients failing antiretrovi- receiving nelfinavir, no virologic benefit are difficult to predict; use with pheno- ral therapy. [Abstract 585-T.] 9th Conference on of TDM was observed among patients typing to determine optimal drug con- Retroviruses and Opportunistic Infections. receiving indinavir, and no benefits of centrations in salvage treatment, as February 24-28, 2002; Seattle, Wash.

43 International AIDS Society–USA Topics in HIV Medicine

Burger D, Hugen P, Droste J, Huitema A. and clinical benefits. Annu Rev Pharmacol Mallal S, Nolan D, Witt C, et al. Association Therapeutic drug monitoring (TDM) of nelfi- Toxicol. 2000;40:649-674. between presence of HLA-B*5701, HLA-DR7, navir (NFV) and indinavir (IDV) in treatment- and HLA-DQ3 and hypersensitivity to HIV-1 naive patients improves therapeutic outcome Flexner C, Hsu A, Kerr B, Gallant J, Heath- reverse-transcriptase inhibitor abacavir. Lancet. after 1 year: results from ATHENA. [Abstract Chiozzi M, Anderson R. Steady-state pharma- 2002;359:727-732. 30.] 1st International AIDS Society Conference cokinetic interactions between ritonavir (RTV), on HIV Pathogenesis and Treatment. July 8-11, nelfinavir (NFV), and the nelfinavir active Molla A, Vasavanonda S, Kumar G, et al. Human 2001; Buenos Aires, Argentina. metabolite M8 (AG1402). [Abstract 42265.] serum attenuates the activity of protease 12th World AIDS Conference. June 28-July 3, inhibitors toward wild-type and mutant human Clevenbergh P, Durant J, Garraffo R, et al. 1998; Geneva, Switzerland. immunodeficiency virus. Virology. 1998;250: Usefulness of protease inhibitor therapeutic 255-262. drug monitoring? PharmAdapt: a prospective multicentric randomized controlled trial: 12 Flexner CW, Piscitelli SC. Concentration-target- weeks results. [Abstract 260B.] 8th Conference ed therapy and the future of HIV management. Piscitelli SC, Gallicano KD. Interactions among on Retroviruses and Opportunistic Infections. AIDS. 2002;16:S1-S3. drugs for HIV and opportunistic infections. N February 4-8, 2001; Chicago, Ill. Engl J Med. 2001;34:984-996. Flexner C, Piscitelli SC. Drug administration Clevenbergh P, Garraffo R, Durant J, et al. Long and interactions. In: Dolin R, Masur H, Saag Piscitelli SC, Metcalf J, Hoetelmans R, Falloon J. term efficacy of protease inhibitor plasma drug MS, eds. AIDS Therapy. New York, NY: Churchill The relative inhibitory quotient (RIQ): a method levels monitoring (PI TDM): PharmAdapt W 32 Livingstone; 2003:923-939. for predicting response to protease inhibitors. results. [Abstract 1730.] 41st Interscience [Abstract 164.] 8th European Congress on Conference on Antimicrobial Agents and Flexner C, Piscitelli SC. Drug-drug interactions Clinical Aspects and Treatment of HIV Infection. Chemotherapy. December 16-19, 2001; in human immunodeficiency virus infection. October 28-31, 2001; Athens, Greece. Chicago, Ill. In: DeClercq E, ed. Antiretroviral Therapy. Washington, DC: ASM Publications; 2001:339- Wegner S, Vahey M, Dolan M, et al. Racial dif- Fellay J, Marzolini C, Meaden ER, et al. 350. ferences in clinical efficacy of efavirenz-based Response to antiretroviral treatment in HIV-1- antiretroviral therapy. [Abstract 428-W.] 9th infected individuals with allelic variants of the Conference on Retroviruses and Opportunistic Hetherington S, Hughes AR, Mosteller M, et al. multidrug resistance transporter 1: a pharma- Infections. February 24-28, 2002; Seattle, Wash. cogenetics study. Lancet. 2002;359:30-36. Genetic variations in HLA-B region and hypersensitivity reactions to abacavir. Lancet. 2002;359:1121-1122. Zhang XQ, Schooley RT, Gerber JG. The effect of Fischl MA, Richman DD, Flexner C, et al. Phase increasing alpha1-acid glycoprotein concentra- I/II study of the toxicity, pharmacokinetics, and tion on the antiviral efficacy of human im- activity of the HIV protease inhibitor SC-52151. Lazdins JK, Mestan J, Goutte G, et al. In vitro munodeficiency virus protease inhibitors. J Acquir Immune Defic Syndr Hum Retrovirol. effect of alpha1-acid glycoprotein on the anti- J Infect Dis. 1999;180:1833-1837. 1997;15:28-34. human immunodeficiency virus (HIV) activity of the protease inhibitor CGP 61755: a compar- Flexner C. Dual protease inhibitor therapy in ative study with other relevant HIV protease Top HIV Med. 2003;11(2):40-44. HIV-infected patients: pharmacologic rationale inhibitors. J Infect Dis. 1997;175:1063-1070. Copyright © 2003 International AIDS Society–USA

Courtesy Announcement. The International AIDS Society–USA is not affiliated with Warmline or PEPline.