Antiviral Therapy 10:343–348 In vitro combination studies of tenofovir and other analogues with ribavirin against HIV-1 Nicolas A Margot and Michael D Miller*

Gilead Sciences, Inc, Foster City, CA, USA

*Corresponding author: Tel: +1 650 522 5584; Fax: +1 650 522 5890; E-mail: [email protected]

In patients coinfected and treated for both HIV-1 and (ddI). In contrast, low-level anti-HIV antago- C (HCV), administration of ribavirin (RBV) nism was observed when RBV was combined with either may result in altered intracellular drug levels of nucleo- tenofovir or abacavir. Significantly stronger anti-HIV side reverse transcriptase inhibitors through inhibition of antagonism was observed when RBV was combined with 5′-monophosphate dehydrogenase. Drug interac- either , stavudine, emtricitabine or . tions between tenofovir and RBV were studied in vitro in Thus, although tenofovir and ddI are both order to provide insights into the safety of co- analogues, their in vitro interactions with RBV are administration of fumarate (DF) and markedly different. These results suggest a low potential RBV in HCV/HIV-1-coinfected patients. In accordance for increased toxicity upon co-administration of teno- with previous in vitro studies, strongly increased anti-HIV fovir DF with RBV in patients. activity was observed when RBV was combined with

Introduction

Coinfection with virus (HCV) in HIV-1- vitro [8,9]. Alterations in levels of intracellular active infected patients is common. One study found that 16% metabolites of anti-HIV drugs can also be measured of HIV-1-infected patients were coinfected with HCV, indirectly through altered anti-HIV activity in vitro. with as many as 72% of patients coinfected in cohorts The purpose of this study is to investigate in vitro anti- of haemophiliacs or intravenous drug users [1]. The HIV drug combinations of RBV with or primary treatment against HCV is a combination of nucleoside reverse transcriptase inhibitors (NRTIs) in pegylated -α with ribavirin (RBV) that order to address the potential for intracellular drug HCV/HIV-1-coinfected patients receive in addition to interactions of RBV with these agents when co-admin- their antiretroviral therapy for HIV-1. The exact mode istered in HCV/HIV-1-coinfected patients. of action of RBV is not fully understood but is thought to be due in part to inhibition of inosine 5′-monophos- Methods phate dehydrogenase (IMPDH), which is an enzyme involved in de novo nucleotide synthesis [2]. Drug combinations The US Food and Drug Administration (FDA) issued a Anti-HIV drug combination studies of RBV with warning that increased didanosine (ddI) toxicity has either tenofovir (TFV), emtricitabine (FTC), ddI, been observed in HCV/HIV-1-coinfected patients zidovudine (ZDV), lamivudine (3TC), stavudine receiving both ddI and RBV. Specifically, the US FDA (d4T), abacavir (ABC) or nelfinavir (NFV) were found a significant increase in the risk of developing performed. TFV and FTC were synthesized at Gilead mitochondrial toxicity in HCV/HIV-1-coinfected Sciences (Foster City, CA, USA); RBV, ZDV and ddI patients receiving RBV and ddI concomitantly [3]. This were purchased from Sigma (St Louis, MI, USA); 3TC finding was also documented by other groups [4–6], was obtained from Moravek Biochemicals (Brea, CA, leading to the conclusion that co-administration of RBV USA); ABC was supplied by GlaxoSmithKline and ddI should be avoided due to a high rate of clinically (Research Triangle Park, NC, USA); d4T was obtained significant toxicity [4]. These findings are in agreement from Bristol-Myers Squibb (New York, NY, USA); and with an earlier study demonstrating increased activity of NFV was obtained from Agouron Pharmaceuticals ddI in the presence of RBV [7] and with data showing (San Diego, CA, USA). The cytotoxic effect of RBV in that RBV increases the levels of the active metabolite MT-2 cells was measured using an XTT assay at RBV of ddI, dideoxyadenosine triphosphate (ddATP), in concentrations ranging from 500 nM to 3.2 mM. The

© 2005 International Medical Press 1359-6535 343 NA Margot & MD Miller

five concentrations of RBV tested in the combinations inhibit 50% of cell growth (CC50) was measured at ranged from 0.75–12 µM and were well below cyto- 70 µM (Figure 1) and was found to be approximately toxic levels. For the anti-HIV compounds, six sixfold above the highest concentration of RBV, concentrations of drug were tested, starting at a 12 µM, used in the combination experiments. The start maximum drug concentration of at least four times the of observed toxicity was found to be >12 µM in all of expected effective concentration to inhibit 50% of the CC50 experiments. viral replication (EC50) for each specific drug. In all The results for the combinations of RBV with the cases, the concentrations tested were well below cyto- eight HIV-1 inhibitors are shown in Figure 2, showing toxic levels. For each drug combination, the a representative experiment for each of the drug compounds were prepared separately by twofold combinations analysed. The values in Table 1 are the serial dilution and mixed in 96-well assay plates to average synergy and antagonism values from three to create a two-dimensional matrix of diluted drugs. five combination experiments and the standard devia- Each combination experiment was done in triplicate tions between experiments. RBV on its own did not and a minimum of three independent experiments exhibit any anti-HIV activity when tested at non- were conducted for each combination. A no-drug cytotoxic concentrations ranging from 0.75–12 µM in control was used for all the compounds tested. over 30 triplicate experiments. However, the presence of RBV had an effect on the anti-HIV activity of all the Antiviral assay and statistical analyses NRTIs analysed. The combination of RBV with ddI The antiviral effects of the drug combinations was showed strong enhancement of the anti-HIV activity of determined using an XTT assay in MT-2 cells as ddI, while the combinations with either ZDV, d4T, described earlier [10,11]. Briefly, 1.2 million MT-2 FTC or 3TC showed strong antagonism of the anti- cells were infected with wild-type HIV-1 (HXB2D) and HIV activity of these NRTIs. RBV combined with incubated in the assay plates for 5 days, starting at an either TFV or ABC resulted in moderate levels of anti- initial concentration of approximately 17 000 cells/well. HIV antagonism. The anti-HIV activity of the protease The antiviral effect of the combinations was measured inhibitor nelfinavir was not affected by RBV. by determining the HIV-1 cytopathic effect using the The effect of RBV on the anti-HIV EC50 values for vital dye XTT. The data were analysed with the soft- the NRTIs was also determined. The observed levels of ware MacSynergy II according to the method of RBV interaction in the drug combination studies corre-

Prichard et al. [12,13]. The software uses the indepen- lated closely with the antiretroviral EC50 observed for dent effects definition of additive interactions. In this the NRTIs. Fold-changes in the anti-HIV EC50 values model, the inhibition observed in the drug combination for the antiretroviral drugs in the presence of RBV are is compared with the predicted theoretical inhibition shown in Figure 3. Combinations of RBV with ddI obtained by simply adding the inhibitory effect of each drug alone. Any deviation from predicted is scored as synergy (positive deviation) or antagonism (negative Figure 1. Cytotoxicity of RBV in MT-2 cells deviation). The software calculates confidence inter- vals to assess the statistical significance of the deviations observed. In this study, data were analysed at the 95% confidence level. Synergy volumes were 100 CC = 70 µM defined by the program as follows: values <25 µM2% 50 reflect insignificant synergy (neither synergy nor antag- 80 onism); values ≥25 to <50 µM2% indicate minor synergy/antagonism; values ≥50 to <100 µM2% indi- 60 cate moderate synergy/antagonism; and values ≥ 2 40

100 µM % indicate strong synergy/antagonism. Cell death, %

Three-dimensional mesh plots and EC50 values were generated using SigmaPlot (SPSS, Inc, Chicago, IL, 20

USA). Statistical significance of EC50 fold-changes in the presence of RBV were calculated in Excel using 0 two-tailed paired t-tests. 0.0001 0.001 0.01 0.1 1 RBV concentration, mM Results The effect of ribavirin on cell viability was assayed using the vital dye XTT. The

CC50 value was obtained by averaging the results of four independent experi- The cytotoxicity of RBV in MT-2 cells was analysed in ments. The standard error bars are shown for each averaged value. RBV, four independent experiments. The concentration to ribavirin.

344 © 2005 International Medical Press Drug combinations of nucleoside analogues with RBV

Figure 2. Three-dimensional synergy plots

A 100 B 30 90 80 20 70 60 10 50 0 40 30 –10 Synergy, % Synergy, Synergy, % Synergy, 20 –20 10 0 –30 24 5380 -10 16 12 53 10 8 8 5310 8 6 5327 6 4 2 0 4 2 0 0 µ 0 I, µM RBV, µ RBV, M dd M NFV, nM

C D

0 0

53-10 53–10

53-20 53–20

53-30 53–30 gonism, % gonism, % a a 53-40 53–40 Ant Ant 53-50 53–50

53-60 16 53–60 1.8 10.7 5310 1.2 5310 8 6 5.3 8 6 4 0.6 4 2 0 0 µM 2 0 0 µM RBV, µM TFV, RBV, µM ABC,

E F

0 0

53–10 53–10

53–20 53–20

53–30 53–30 gonism, % gonism, % a a 53–40 53–40 Ant Ant 53–50 53–50

53–60 12 3 53–60 2 5310 8 5310 8 6 4 8 6 4 1 4 2 0 0 µM 2 0 0 µM RBV, µM 3TC, RBV, µM FTC,

G H

0 0

53–10 53–10

53–20 53–20

53–30 53–30 gonism, % gonism, % a a 53–40 53–40 Ant Ant 53–50 53–50

53–60 0.6 53–60 36 0.4 5310 24 5310 8 6 0.2 8 6 4 12 4 2 0 0 µM RBV, µ 2 0 0 µM RBV, µM ZDV, M d4T,

The level of synergy or antagonism found at the 95% confidence intervals for each of the combinations tested is shown. (A) ddI–RBV combination; (B) NFV–RBV combination; (C) TFV–RBV combination; (D) ABC–RBV combination; (E) 3TC–RBV combination; (F) FTC–RBV combination; (G) ZDV–RBV combination; (H) d4T–RBV combination. The colours indicate planes of 10% of volume of synergy or antagonism and change by increment of 10% with increasing synergy or antagonism. ABC, abacavir; ddI, didanosine; FTC, emtricitabine; 3TC, lamivudine; NFV, nelfinavir; RBV, ribavirin; d4T, stavudine; TFV, tenofovir; ZDV, zidovudine.

Antiviral Therapy 10:2 345 NA Margot & MD Miller

Table 1. Ribavirin–antiretroviral combination studies against HIV-1

Volume (µM2%)*

Drug combination Synergy Antagonism Combined effect

RBV–ddI 359 ±111 4.8 ±7.1 Strong potentiation RBV–NFV 6.9 ±2.9 9.0 ±8.2 No effect RBV–ABC 3.6 ±5.3 64.1 ±23.0 Moderate antagonism RBV–TFV 3.8 ±8.4 84.3 ±39.6 Moderate antagonism RBV–3TC 0 ±0 185 ±88.7 Strong antagonism RBV–FTC 2.0 ±3.5 246 ±184 Strong antagonism RBV–ZDV 0 ±0 339 ±119 Strong antagonism RBV–d4T 0.5 ±1.0 455 ±217 Strong antagonism

*Volume of synergy and antagonism were computed by the MacSynergy™ II program using a 95% confidence interval and are defined by the program as follows: values <25 µM2% indicate insignificant synergy (additive); values ≥25 to <50 µM2% indicate minor synergy/antagonism; values ≥50 to <100 µM2% indicate moderate synergy/antagonism; and values ≥100 µM2% indicate strong synergy/antagonism. ABC, abacavir; ddI, didanosine; FTC, emtricitabine; 3TC, lamivudine; NFV, nelfinavir; RBV, ribavirin; d4T, stavudine; TFV, tenofovir; ZDV, zidovudine.

resulted in an increase in ddI activity, while the combi- without RBV or in the presence of either 3 µM, 6 µM nations of RBV with either ZDV, d4T, 3TC, FTC, ABC or 12 µM RBV were 5.1, 8.0, 8.0 and 9.3 µM, respec- or TFV resulted in a decrease in anti-HIV activity. At tively. These concentrations of RBV represent an RBV concentration of 3 µM, the anti-HIV activity near-physiological plasma concentrations of RBV of ddI was increased 2.5-fold, while the anti-HIV based on pharmacokinetic studies in patients where a activity of ABC, TFV, 3TC, FTC, ZDV and d4T was Cmax of 3.9 µM has been reported [14]. The anti-HIV decreased 1.4-, 1.6-, 3.2-, 3.2-, 3.5- and 3.4-fold, activity of NFV was not modified in the presence respectively, when compared with the no-RBV control. of RBV. These changes in anti-HIV effect were further enhanced in the presence of 6 µM RBV and 12 µM Discussion RBV (Figure 3). For TFV, the EC50 values obtained Our findings with regards to cytotoxicity and lack of anti-HIV activity of RBV are in agreement with Figure 3. Changes in anti-HIV activity in the presence of previous experiments in MT-4 cells [7,15]. In contrast ribavirin to these findings, Klein et al. [16] found that RBV had some low-level anti-HIV activity described as >50 µM in an assay system using human cord blood mononu- 20 clear cells. Thus, the level of activity of RBV against 18 HIV-1 may depend on the cell type used in the exper- 16 iments and a clear delineation between antiviral effect 50 14 and cytotoxicity must be made. In our experiments, 3 µM RBV 12 6 µM RBV cytotoxicity could result in the appearance of anti- e in EC g 10 12 µM RBV HIV antagonism. However, the concentrations of RBV No RBV 8 employed were below the levels of cytotoxicity. In this study, the reported increases in ddI-associ- 6 Fold-chan ated toxicities with RBV co-administration were 4 correlated with a strong enhancement of the anti-HIV 2 activity of ddI when combined with RBV in vitro, in 0 agreement with earlier studies [7,17]. In contrast, TFV ddI NFV TFV ABC FTC d4T 3TC ZDV anti-HIV activity was not enhanced with RBV, but low-level antagonism was observed. Although TFV The effects of the presence of ribavirin at 3 µM, 6 µM and 12 µM on the anti- HIV activity of each of the compounds tested in combination with RBV are and ddI are both adenosine analogues, their in vitro shown as the average of the fold-change in EC50 for each experiment in the interactions with RBV are markedly different. It has presence of RBV compared with no-RBV controls. Each average was obtained from at least three independent experiments. At least one of the observed been suggested that increased levels of the active fold-changes for all compounds was statistically significant in paired t-tests metabolite of ddI, ddATP, in the presence of RBV is (P<0.05) except for NFV. Error bars show the standard deviation. ABC, abacavir; ddI, didanosine; FTC, emtricitabine; 3TC, lamivudine; NFV, nelfinavir; RBV, due to the increased level of inosine monophosphate ribavirin; d4T, stavudine; TFV, tenofovir; ZDV, zidovudine. (IMP) as a consequence of inhibition of IMP

346 © 2005 International Medical Press Drug combinations of nucleoside analogues with RBV

dehydrogenase by RBV [8]. As IMP is the major phos- Acknowledgements phate donor for the initial phosphorylation step in the pathway from ddI to ddATP, this would result in The authors wish to thank Adrian Ray for his editorial increased ddATP levels that can be observed in vitro review and Margaret Benton and Susan Edl for tech- through increased anti-HIV activity and in vivo nical assistance in the writing of this manuscript. through increased mitochondrial toxicity. TFV, on the other hand, is a monophosphate molecule and there- References fore does not require initial phosphorylation in the 1. Sherman KE, Rouster SD, Chung RT & Rajicic N. manner of ddI. The results presented here are in prevelance among patients infected with support of the differential metabolic pathways for human immunodeficiency virus: a cross-sectional analysis of the US adult AIDS Clinical Trials Group. Clinical TFV and ddI based on their different initial phospho- Infectious Diseases 2002; 34:831–837. rylation requirements. 2. McHutchison JG & Patel K. Future therapy of hepatitis C. Thus, unlike the case of ddI, the results of this study Hepatology 2002; 36:S245–S252. suggest a very low potential for increased toxicity 3. Fleischer R, Boxwell D & Sherman KE. Nucleoside analogues and mitochondrial toxicity. Clinical Infectious upon co-administration of TFV with RBV in patients. Diseases 2004; 38:E79–E80. The result of low-level antagonism of TFV anti-HIV 4. Moreno A, Quereda C, Moreno L, Perez-Elias MJ, Muriel activity with RBV is of interest. Similar low levels of A, Casado JL, Antela A, Dronda F, Navas E, Barcena R & Moreno S. High rate of didanosine-related mitochondrial RBV-mediated antagonism were observed for ABC toxicity in HIV/HCV-coinfected patients receiving while much higher levels of antagonism were observed ribavirin. Antiviral Therapy 2004; 9:133–138. 5. Kakuda TN & Brinkman K. Mitochondrial toxic effects for ZDV, d4T, FTC and 3TC. The strong antagonistic and ribavirin. Lancet 2001; 357:1802–1803. effects seen for those four drugs in the presence of 6. Salmon-Ceron D, Chauvelot-Moachon L, Abad S, RBV correlates with previous reports that have shown Silbermann B & Sogni P. Mitochondrial toxic effects and that RBV could inhibit the phosphorylation of pyrim- ribavirin. Lancet 2001; 357:1803–1804. ′ ′ 7. Balzarini J, Naesens L, Robins MJ & De Clercq E. idine 2 ,3 -deoxynucleotides such as ZDV, d4T, FTC Potentiating effect of ribavirin on the in vitro and in vivo and 3TC in vitro [15,17–19]. Alternatively, the levels antiretrovirus activities of 2′,3′-dideoxyinosine and 2′,3′- dideoxy-2,6-diaminopurine riboside. Journal of Acquired of antagonism observed for the various combinations Immune Deficiency Syndromes & Human Retrovirology could be due to differential toxicity between the drugs 1990; 3:1140–1147. on their own and when they are in combination with 8. Hartman NR, Ahluwalia GS, Cooney DA, Mitsuya H, Kageyama S, Fridland A, Broder S & Johns DG. Inhibitors RBV. This hypothesis appears unlikely based on the of IMP dehydrogenase stimulate the phosphorylation of the high level of antagonism observed for otherwise rela- anti-human immunodeficiency virus 2′,3′- dideoxyadenosine and 2′,3′-dideoxyinosine. Molecular tively non-toxic drugs such as 3TC and FTC. The Pharmacology 1991; 40:118–124. mechanistic basis for the observation of low-level 9. Balzarini J, Lee C-K, Herdewijn P & De Clercq E. antagonism with tenofovir and abacavir is unknown, Mechanism of the potentiating effect of ribavirin on the activity of 2′,3′-dideoxyinosine against human but may relate to a general perturbation of the intra- immunodeficiency virus. Journal of Biological Chemistry cellular phosphorylation pathways as a result of 1991; 266:21509–21514. 10. Gu Z, Salomon H, Cherrington JM, Mulato AS, Chen MS, RBV activity. Yarchoan R, Foli A, Sogocio KM & Wainberg MA. K65R To our knowledge, there has been no report of a mutation of human immunodeficiency virus type 1 reverse transcriptase encodes cross-resistance to 9-(2-phosphonyl- negative clinical consequence of the antagonistic anti- methoxyethyl). Antimicrobial Agents & HIV effect of RBV in combination with ZDV, d4T, Chemotherapy 1995; 39:1888–1891. FTC or 3TC. Two studies have shown that anti-HIV 11. Weislow OS, Kiser R, Fine DL, Bader J, Shoemaker RH & Boyd MR. New soluble-formazan assay for HIV-1 cyto- treatment that included the combination of d4T and pathic effects: application to high-flux screening of 3TC or ZDV and 3TC was not adversely affected in synthetic and natural products for AIDS-antiviral activity. Journal of the National Cancer Institute 1989; 81:577–586. patients receiving RBV as a part of their anti-HCV 12. Prichard MN, Aseltine KR & Shipman C Jr. MacSynergy™ treatment [20,21]. The concomitant use of interferon II, Version 1.0. 1993. Ann Arbor, MI: University of with RBV to treat the HCV may also be playing a role Michigan. in maintaining virological control of HIV. Overall, 13. Prichard MN & Shipman C Jr. A three-dimensional model to analyze drug-drug interactions. Antiviral Research 1990; these clinical findings suggest that the in vitro anti- 14:181–205. retroviral drug antagonism observed in the 14. Glue P, Schenker S, Gupta S, Clement RP, Zambas D & combinations with RBV are not clinically significant Salfi M. The single dose of ribavirin in subjects with chronic . British Journal of for the combined treatment of HCV and HIV-1 in Clinical Pharmacology 2000; 49:417–421. HCV/HIV-1-coinfected patients. Moreover, with the 15. Baba M, Pauwels R, Balzarini J, Herdewijn P, De Clercq E & Desmyter J. Ribavirin antagonizes inhibitory effects of exception of ddI, none of the licensed NRTIs tested pyrimidine 2′,3′-dideoxynucleosides but enhances showed any synergistic activity with RBV that could be inhibitory effects of 2′,3′-dideoxynucleosides of replication of human immunodeficiency virus in vitro. associated with increased toxicity in HCV/HIV-1- Antimicrobial Agents & Chemotherapy 1987; coinfected patients. 31:1613–1617.

Antiviral Therapy 10:2 347 NA Margot & MD Miller

16. Klein MB, Campeol N, Lalonde RG, Brenner B & 19. Hoggard PG, Kewn S, Barry MG, Khoo SH & Back DJ. Wainberg MA. Didanosine, interferon-alfa and ribavirin: a Effects of drugs on 2′,3′-dideoxy-2′,3′-didehydrothymidine highly synergistic combination with potential activity phosphorylation in vitro. Antimicrobial Agents & against HIV-1 and hepatitis C virus. AIDS 2003; Chemotherapy 1997; 41:1231–1236. 17:1001–1008. 20. Morsica G, De Bona A, Foppa CU, Sitia G, Finazzi R & 17. Vogt MW, Hartshorn KL, Furman PA, Chou TC, Fyfe JA, Lazzarin A. Ribavirin therapy for chronic hepatitis C does Coleman LA, Crumpacker C, Schooley RT & Hirsch MS. not modify HIV viral load in HIV-1 positive patients under Ribavirin antagonizes the effect of azidothymidine on HIV antiretroviral treatment. AIDS 2000; 14:1656–1658. replication. Science 1987; 235:1376–1379. 21. Landau A, Batisse D, Piketty C, Jian R & Kazatchkine 18. Hoggard PG, Veal GJ, Wild MJ, Barry MG & Back DJ. MD. Lack of interference between ribavirin and nucleoside Drug interactions with zidovudine phosphorylation in analogues in HIV/HCV co-infected individuals undergoing vitro. Antimicrobial Agents & Chemotherapy 1995; concomitant antiretroviral and anti-HCV combination 39:1376–1378. therapy. AIDS 2000; 14:1857–1858.

Received 20 October 2004, accepted 17 February 2005

348 © 2005 International Medical Press