Optimal suppression of HIV replication by low-dose hydroxyurea through the combination of antiviral and cytostatic ('virostatic') mechanisms

Background: The hydroxyurea-didanosine combination has been shown to limit immune activation (a major pathogenic component of HIV/AIDS) and suppress viral load by both antiviral and cytostatic ('virostatic') activities. Virostatics action represent a novel approach to attack HIV/AIDS from multiple directions; however, the use of these drugs is limited by the lack of understanding of their dose-dependent mechanism of action and by fear of pancreatic toxicity, even though a large review of ACTG studies has shown that hydroxyurea does not increase the incidence of pancreatitis.

Methods: In vitro cytostatic and cytotoxic activity, inhibition of viral replication and immune activation by pharmacologically attainable plasma concentrations of hydroxyurea (10-100 μmol/l) and didanosine (1-5 μmol/l) were analyzed by cell proliferation, viability, apoptosis and infection assays using peripheral blood mononuclear cells. In vivo, 600, 900 and 1200 mg daily doses of hydroxyurea in combination with standard doses of didanosine and stavudine were studied in 115 randomized chronically infected patients.

Results: A cytostatic low (10 μmol/l) concentration of hydroxyurea inhibited cell proliferation and HIV replication in vitro. A gradual switch from cytostatic to cytotoxic effects was observed by increasing hydroxyurea concentration to 50-100 μmol/l, predicting that lower doses of hydroxyurea would be less toxic and more potent in vivo. The clinical results confirmed that 600 mg hydroxyurea was better tolerated, had fewer side effects and was more potent in suppressing HIV replication than the higher doses.

Conclusions: A bimodal, dose-dependent, cytostatic-cytotoxic switch is an immune-based mechanism explaining the apparent paradox that lowering the dose of hydroxyurea to 600 mg daily induces maximal antiviral suppression in HIV-infected patients.

Introduction

The mechanism of antiviral action of the hydroxyurea-didanosine regimen is based on the capacity of hydroxyurea to deplete intracellular components required for viral replication, namely deoxynucleotides ^[1,2], particularly deoxyadenosine trisphosphate (Didanosine competitor) ^[3], thus enabling it to act synergistically with didanosine. The hydroxyurea-didanosine combination offers a favorable resistance profile, as hydroxyurea targets a cellular enzyme that, unlike viral enzymes (e.g., reverse transcriptase), is not prone to mutation/resistance, and it has been shown to compensate for resistance to didanosine, explaining the long-term antiviral efficacy observed with this combination ^[4]. Both hydroxyurea and didanosine have a favorable antiviral profiles compared with other drugs in cells that are believed to represent the latent pools of replication-competent virus ^[5-7], such as resting/quiescent T cells, macrophages and dendritic cells ^[2,8].

The cytostatic and cytotoxic properties of hydroxyurea have been exploited for over 40 years for the treatment of myeloproliferative disorders ^[9,10], and might have implications in the HIV field. Treatment of CD4 T lymphocytes with hydroxyurea, in fact, renders them quiescent and refractory to productive HIV infection ^[1,2] (as HIV is mainly replicating in proliferating T lymphocytes ^[5-7]) and also increases beta-chemokines, inhibiting R5 HIV-1 ^[11]. Suppression of immune activation by cytostatic drugs such as hydroxyurea could complement antiretroviral treatment, because chronic immune activation plays a predominant role in the immune pathogenesis of AIDS ^[12,13].

Hydroxyurea blunts the CD4 cell count increase seen during antiretroviral therapy ^[14-17], and this has raised concerns regarding its potential use in a disease characterized by loss of CD4 T cells. Additional toxicity concerns were raised by the AIDS Clinical Trials Group (ACTG) 5025 study investigating a combination of didanosine, stavudine and indinavir with or without addition of hydroxyurea (1200 mg daily). The hydroxyurea-containing study arm showed a higher frequency of treatment failure and possibly pancreatic toxicity ^[18] and clinicians have become hesitant to use hydroxyurea for treatment of HIV-1 infection. Therefore, hydroxyurea remains an investigational new drug for HIV indication and it is not registered for clinical use in HIV-1 infection. However, a recent review on the risks of pancreatitis regarding 20 ACTG studies from 1989 to 1999, which involved over 8000 patients, concluded that although didanosine, stavudine, and indinavir combination bears the highest rate of incidence of pancreatitis, this is independent of the addition of hydroxyurea ^[19]. The study also concluded that hydroxyurea does not increase the risk of pancreatitis compared to didanosine alone ^[19].

Moreover, none of the clinical investigations with hydroxyurea in over 500 HIV-infected patients preceding the ACTG 5025 study, namely the RIGHT 411 trial ^[14], the ACTG 307 study ^[15], the Swiss HIV Cohort Study ^[16] and the BMS 055 study ^[17], had reported cases of pancreatitis, but all showed superior efficacy of the hydroxyurea-containing arms. Since the 1200 mg daily dose of hydroxyurea used in the ACTG 5025 study was higher than in the preceding trials, it could be hypothesized that lowering the dose of hydroxyurea could decrease toxicity. The present study examines the in vitro effects of varying the dosage of hydroxyurea in terms of cytostatic and antiviral effects.

Methods

Mitochondrial toxicity assay

BxPC-3, a human pancreatic adenocarcinoma cell line (ATCC, Manassas, Virginia, USA), was cultured in RPMI-1640 supplemented with 10% heat-inactivated fetal calf serum, 2 mmol/l L-glutamine, 50 U/ml penicillin and 50 μg/ml streptomycin (all from Gibco Life Technologies, Glasgow, UK), referred to as complete media (CM), in 9.5 cm² of a 6-well plate (Costar, Cambridge, Massachusetts, USA) at a concentration of 2.5 × 10⁵ cells/ml, in the presence of hydroxyurea, didanosine and stavudine. Experiments were performed in triplicate. At day 7, cultured cells were split, counted and replated with drugs. At day 14, cells were harvested and stained with the lipophilic cationic probe JC-1 (Molecular probes, Eugene, Oregon, USA). Briefly, JC-1 is a lipophilic carbocyanine that exists in a monomeric form and is able to accumulate in mitochondria. In the presence of a high membrane potential, JC-1 can reversibly form aggregates that, after excitation at 488 nm, emit in the orange/red channel (namely, in the FL2 channel of commonly used flow cytometers). Monomers emit in the green channel (FL1). The collapse in membrane potential provokes a decrease in the number of JC-1 aggregates (revealed by a decrease in FL2) and a consequent increase of monomers (an increase in FL1). BxPC-3 cells were stained by adding 10 μmol/l JC-1 and were then incubated for 15 min in the dark. Within 30 min after staining, cells were analyzed using the Coulter Epics XL flow cytometer (Beckman-Coulter, Milan, Italy). Analysis of the data obtained from the flow cytometry was performed first by transforming the obtained results from logarithmic into linear scale and then by calculating the ratio of the median fluorescence intensity in FL2 divided by the median fluorescence in FL1. Cells obtained from the untreated sample were the negative control. Cells exposed for 5 min at room temperature to the mitochondrial toxicity-inducing agent valinomycin (500 nmol/l) represented the positive control. Maximal toxicity was measured in the positive control. The mitochondrial toxicity index (MTI) was calculated as the percentage change from 0 (negative control, untreated samples) to 100 (maximal toxicity, valinomicin treated samples). Values below 0 were considered to be equal to 0.

Cell proliferation assay

Peripheral blood mononuclear cells (PBMC) were obtained from healthy, normal donors by separation over Ficoll-Hypaque (Hypaque, Pharmacia Biotech, Upsala, Sweden). PBMC were immediately stained with 5 μmol/l carboxyfluorescein diacetate succinimidil ester (CFSE; Molecular Probes) according to the manufacturer's protocol. After staining, cells were resuspended in CM to a final concentration of 2.5 × 10⁶ cells/ml. Cells were then treated with three different concentrations of hydroxyurea, 10, 50 and 100 μmol/l, plus an untreated control sample, for 1 h at 37°C under 5% carbon dioxide. After incubation, cells were stimulated with 5 μg/ml phytohaemagglutinin (5 mg/ml stock stored at -20°C; Sigma, St Louis, Missouri, USA) and 20 IU/ml interleukin-2 (generous gift from Boehringer-Mannerheim, Mannerheim, Germany). At day 7, 1 × 10⁶ cells were washed and resuspended in 100 μl phosphate-buffered saline (Gibco Life Technologies). CFSE-labeled lymphocytes were stained with a dual-color analysis by FACS (Beckman-Coulter) with CD4-phycoerythrin or CD8-phycoerythrin-cyanin5 (Beckman-Coulter). For data analysis, the mitotic index (M) was used, calculating the sum of mitotic events from each proliferation cycle ^[20]. To extract a relative number, M was normalized to the total number of cells acquired.

Equation (Uncited)

where n is the number of cycles and X_n(T) is the number of acquired events per cycle.

Cell viability and apoptosis assay

PBMC were cultured for 7 days in 15 ml CM in a 25 cm² flask at a concentration of 2.5 × 10⁶ cells/ml and stimulated with 5 μg/ml phytohaemagglutinin plus 20 IU/ml interleukin-2. Before stimulation, cells were pretreated with hydroxyurea (10, 50, 100, 500 or 1000 μmol/l) for 1 h and untreated cells were prepared as a control. After culture, cells were counted by trypan blue exclusion (Eurobio, Les Ulis, France) to evaluate cell death. Cells were then resuspended in 500 μl binding buffer (Annexine V Apoptosis Detection Kit; Pharmingen, San Diego, California, USA) at 1 × 10⁶ cells/ml and 100 μl of this solution was incubated with annexin V and propidium iodide (Pharmingen) for 15 min, following manufacturer's instructions. After washing, the samples were analyzed by FACS. Data presented represent the average (±SD) of three separate experiments.

Antiviral activity in vitro

Phytohaemagglutinin-activated PBMC were infected with the HIV strain HIV-1_IIIB (Advanced Biotechnologies, Columbia, Maryland, USA) at 0.001 50% tissue culture infectious dose/cell. After incubation for 1 h at 37°C, cells were washed and plated in a 96-well round-bottom plate (Costar) at a concentration of 5 × 10⁵ cells/ml in CM supplemented with 20 IU/ml interleukin-2. Different concentrations of hydroxyurea were added 1 h after infection and replaced every 2 days when the medium was changed. After 7 days, the supernatants were assayed for HIV-1 p24 antigen by an enzyme-linked immunoadsorbent assay (Coulter, Miami, Florida, USA).

Clinical study

RIGHT 702 was a phase I/II, randomized open-label study of 115 HIV-infected patients (at least one third were naïve for antiretroviral drugs) at several clinical sites in the United States. It was designed to evaluate the safety and antiretroviral activity of three total daily doses of hydroxyurea (600, 800-900 and 1200 mg) in combination with stavudine (40 mg twice daily) and didanosine (initially administered as two 200 mg tablets daily and later as an enteric-coated 400 mg capsule daily) ^[21].

Results

To understand the cause of blunted CD4 cell increase and pancreatic toxicity observed in HIV-infected individuals treated with high doses of hydroxyurea and didanosine, in vitro assays on lymphocytes and pancreatic cells were performed using different concentrations of the drug. The concentrations chosen represented the trough, median and peak plasma concentrations (10, 50 and 100 μmol/l, respectively) plus two concentrations above the maximum concentration (500 and 1000 μmol/l) measured after administration of the most commonly used dosage (500 mg twice daily) of hydroxyurea ^[22].

The cell viability assay showed dose-dependent cytotoxicity of hydroxyurea. While a low concentration of hydroxyurea (10 μmol/l) did not significantly affect cell viability, at concentrations > 10 μmol/l, the number of dead cells progressively increased up to the point of inversion of the alive/dead cells ratio (Fig. 1a). Similarly, an increase of apoptotic cells and the expansion of the annexin^bright population within the pool of pre-apoptotic cells was observed at concentrations > 10 μmol/l (Fig. 1b and Table 1).

Fig. 1

Table 1

Mitochondrial toxicity ^[23-27] was evaluated in the pancreatic cell line BxPC3 after 14 days of continuous exposure to stavudine (1, 5 and 10 μmol/l), didanosine (1, 5 and 10 μmol/l), or hydroxyurea (10, 50 and 100 μmol/l). For both didanosine and stavudine, the concentrations chosen were one concentration below and two concentrations above the maximum concentration measured in vivo after the administration of clinically recommended doses of each drug ^[28-30]. Exposure to stavudine alone induced a dose-dependent increase in the MTI. The MTI increased from 2.9 ± 2.5 to 56.2 ± 30.3 with stavudine 1 and 10 μmol/l, respectively (Fig. 2a). In accordance with previously published data ^[31], didanosine induced moderate mitochondrial toxicity [MTI, 16.6 ± 13.8 at the maximum dose tested (10 μmol/l)]. The effects of hydroxyurea alone on mitochondrial functionality were negligible at any of the concentrations tested.

Fig. 2

Mitochondrial toxicity in pancreatic cells was increased by escalating concentrations of didanosine in combination with stavudine, even in the absence of hydroxyurea, as expected from clinical experience ^[32,33] (Fig. 2b). Adding a high concentration of hydroxyurea increased mitochondrial toxicity. In fact, 100 μmol/l hydroxyurea in combination with 1 μmol/l didanosine and 1 μmol/l stavudine resulted in a marked elevation of the MTI to 85 ± 7, reflecting significant impairment of mitochondrial functions. Reducing the hydroxyurea concentration lowered the MTI, particularly, as previously described, at low didanosine concentration ^[31]. For example, decreasing the hydroxyurea concentration to 50 and 10 μmol/l at 1 μmol/l didanosine and stavudine lowered the MTI to 69 ± 21 and 24 ± 30, respectively. A less-evident but still linear trend for decreased mitochondrial toxicity of didanosine and stavudine with decreasing doses of hydroxyurea was shown for two further concentrations (5 and 10 μmol/l) of didanosine.

The effect of varying the concentration of hydroxyurea on its cytostatic and antiviral properties was then examined. The containment of cell proliferation paralleled the suppression of viral replication when the drug concentration was increased from 0 to 100 μmol/l, suggesting that the cytostatic effects of hydroxyurea correlated with its antiviral activity (Fig. 3a). The antiviral curve had, however, a biphasic mode. Most of the decrease was observed when the hydroxyurea concentration varied from 0 to 10 μmol/l, which was the low concentration of the drug. Subsequently, the curve tended to plateau. Moreover, when the p24 levels were normalized for the number of cells, the low hydroxyurea concentration (10 μmol/l) inhibited viral replication as efficiently as the high concentration (100 μmol/l) (Fig. 3b). These results indicated that HIV-1 replication was maximally inhibited by low concentration of hydroxyurea. Higher doses of hydroxyurea did not improve the antiretroviral activity whereas, as shown before, they did increase cytotoxicity. Based on the in vitro results, it was predicted that lowering the dose of hydroxyurea would preserve its antiviral effects while minimizing side effects, including cytotoxic effects on CD4 T cells and pancreatic toxicity.

Fig. 3

A clinical trial studying different doses of hydroxyurea (600, 800-900 and 1200 mg total daily dose) in combination with didanosine and stavudine confirmed this hypothesis. Patients were randomized to receive low (n = 39), medium (n = 38) or high (n = 38) hydroxyurea dosage. The low daily dosage of 600 mg hydroxyurea was significantly more efficacious than higher doses throughout the different analyses ^[21]. Figure 4 shows the median log₁₀ transformed HIV RNA for the three groups and the change from baseline to week 24, with the greatest decrease (1.95 log₁₀ copies/ml; P = 0.02) in HIV RNA together with the highest increase in CD4 cell count (65 × 10⁶ cells/l; P = 0.0019) seen in the 600 mg daily dosage group. The absolute frequency of adverse events was decreased by lowering the daily hydroxyurea dosage, as there was a total of 40, 47 and 62 events in the 600, 800-900 and 1200 mg groups, respectively. The most frequently reported adverse events related to treatment in all groups were nausea, peripheral neuropathy and diarrhea. One patient, who was randomized to the treatment arm with the highest daily dose of hydroxyurea (1200 mg), died of necrotizing pancreatitis, confirming the risk of using this hydroxyurea dosage ^[18] that is twice the dosage found to be optimal in the present study. All data presented here refer to the intent-to-treat analysis. A more extensive analysis of the RIGHT 702 trial can be found in a related paper entirely dedicated to the clinical outcome of the trial ^[21].

Fig. 4

Discussion

The results described here reveal the antiretroviral mechanism of a 'virostatic' (antiviral and cytostatic) drug, based on a dose-dependent, cytostatic/cytotoxic bimodal effect. They also clarify why reducing the hydroxyurea dose from 1200 to 600 mg decreased toxicity without loss of antiviral efficacy, measured as viral load reduction and CD4 T cell count increase, in HIV-1-infected patients. Lowering the dose of hydroxyurea decreased the cytotoxic effects of the drug in vitro, explaining the increase of CD4 cell counts observed in vivo with the low (600 mg) total daily dose of the drug. The maintenance of antiviral activity at the 600 mg total daily dose, compared with the higher dosage, is consistent with the in vitro data showing that the antiviral effects of the drug were already optimal at the lowest concentration. However, further research is required to explain why lowering the dose of hydroxyurea even increased the antiviral efficacy in vivo. We are currently exploring whether other immunomodulating properties of the drug might be, in part, responsible for this phenomenon.

In particular, the results provide an explanation for the increased occurrence of fatal pancreatitis when hydroxyurea 1200 mg daily dose is administered with didanosine and stavudine, which is twice the optimal dose established by the RIGHT 702 study. After re-examining pancreatitis-related mortality in the previous randomized, controlled hydroxyurea studies ^[14-18,34] (all together accounting for approximately 1000 patients, 200 of whom had received 1200 mg total daily dose hydroxyurea), we discovered that only the dose of 1200 mg daily hydroxyurea (used in the RIGHT 702 study described here and in the ACTG 5025 study ^[18]) was associated with fatal pancreatitis. Another common feature of the RIGHT 702 and ACTG 5025 studies is the particular formulation of didanosine. Before the enteric-coated capsule became available, didanosine was administered as a buffered formulation, 200 mg twice daily. Later, the once-daily 400 mg dose of the buffered formulation was approved, as administered in the ACTG 5025 study. In the RIGHT 702 study, patients initially received 400 mg daily of the buffered formulation, which was later changed to 400 mg daily of the enteric-coated formulation, as it became available during the course of the study. The peak plasma concentration of the enteric-coated formulation is reduced approximately 40% compared with the buffered formulation ^[30,31,35], which could also influence toxicity. This might have contributed to the differences seen between the studies, as patients who died of pancreatitis in the ACTG 5025 study and in the RIGHT 702 study had received the old 400 mg daily buffered didanosine formulation. Our study is limited by the fact that it has been performed in vitro, and the increased toxicity that we observed by adding hydroxyurea to didanosine might not have a clinical significance. In fact, a recent review of ACTG studies has concluded that hydroxyurea/didanosine does not increase the rate of pancreatitis compared to didanosine alone ^[19].

The cytostatic activity of low doses of hydroxyurea is of particular interest because immune activation plays a fundamental role in HIV pathogenesis, possibly through multiple mechanisms, such as deleting both reactive T cells ^[36,37] and bystander cells that are not infected by the virus ^[38], increasing susceptibility to apoptosis ^[39], activating 'auto-immunity' ^[40], and reducing the generation of long-lived, potential progenitor T cells ^[13]. T cell activation and CD4 T cell depletion are more closely associated with disease progression than plasma virus burden during both HIV-1 and HIV-2 infection ^[41,42]; in the SIV model, increased turnover in the lymph node compartments is driven by a generalized immune activation ^[43], and SIV infection of sooty mangabeys becomes non-pathogenic when immune activation and bystander immunopathology are limited, despite high viral load ^[12,44], consistent with an immune system overactivation-pathogenesis relationship. In addition, immune activation has been unexpectedly shown as an independent factor correlated with the onset of metabolic alterations and lipoatrophy during HIV infection ^[45]. Cytostatic drugs might correct some of these pathogenetic alterations. Recent evidence is emerging in support of this hypothesis, as the use of hydroxyurea has been associated with the repopulation of the long-lived T cell compartment ^[46] and the reduction in incidence of hypercholesterolemia, hypertriglyceridemia, lipohypertrophy and lipoatrophy ^[47].

The characteristics of the combination of didanosine plus hydroxyurea may represent a valid option for the treatment of HIV infection in resource-poor settings. However, the risk of development of didanosine-resistant HIV strains should be taken into consideration. The resistance profile of the combination of hydroxyurea plus didanosine has been tested both in vitro and in vivo. During 24 weeks of therapy, patients receiving didanosine plus hydroxyurea combination therapy developed didanosine-resistant mutations more rapidly than patients treated with didanosine monotherapy ^[48]. Despite the emergence of didanosine-resistant mutations, no viral rebound was observed in the patients receiving didanosine plus hydroxyurea ^[14,48]. In vitro experiments showed that hydroxyurea compensated for didanosine resistance, thus explaining the clinical findings ^[48,49]. Moreover, hydroxyurea has not been shown to induce resistance during > 40 years of clinical experience ^[9]. This evidence indicates that didanosine in combination with hydroxyurea retains its antiviral potency in the presence of didanosine-resistant mutations, suggesting that this combination would be suitable for long-term control of HIV.

Zidovudine, the first drug approved for treatment of AIDS, was initially approved at a daily dose of 1200 to 1500 mg. Since this dose of zidovudine was associated with multiple side effects, particularly myelotoxicity, it was subsequently tested at a lower daily dose of 600 mg or less, with similar suppression of viremia while having better tolerability ^[50]. Didanosine was originally investigated at dose levels as high as 750 mg daily ^[51-53], which also proved to be too toxic. Toxicity of didanosine is dose related; the risk for pancreatitis increases further at doses above currently recommended levels, and lower doses have not been associated with a reduction in efficacy. Didanosine is currently approved for treatment at doses of 400 mg daily or less. These examples illustrate how drugs that were initially used at excessively high (toxic) doses eventually became invaluable tools against HIV at lower, more appropriate doses, particularly in view of the recent demonstration that hydroxyurea does not increase the risk of didanosine associated pancreatitis ^[19].

The identification of the optimal daily dose of 600 mg hydroxyurea should mitigate the concerns over mitochondrial toxicity and blunted CD4 cell increase. As has happened with zidovudine and didanosine, hydroxyurea deserves re-evaluation at a low dosage in HIV/AIDS, in combination with didanosine, as this synergistic virostatic combination remains a unique therapeutic approach for the treatment of HIV-1 infection and could became a valuable therapeutic option for the long-term management of HIV/AIDS worldwide.

Acknowledgements

The authors would like to acknowledge all the RIGHT 702 team members: G. Blick, P. Shalit, D. Peterson, A. Tennenberg, S. Schrader, B. Rashbaum, C. Farthing, D. Herman, D. Norris, P. Greiger, I. Frank, D. Asmuth, T. West, H. Lampiris, T. Babinchak, D. W. Seekins and P. Skolnik. The authors are thankful to J. Trocio, J. Xu and T. Battle for technical assistance, and to S. Petrocchi for editorial assistance. The work was, in part, supported by a grant from Fondazione Cariparma, Parma Italy.