Share this article on:

Phase I/II dose escalation and randomized withdrawal study with add-on azodicarbonamide in patients failing on current antiretroviral therapy

Goebel, Frank-D.a; Hemmer, Robertb; Schmit, Jean-Claudeb; Bogner, Johannes R.a; Clercq, Erik dec; Witvrouw, Myriamc; Pannecouque, Christophec; Valeyev, Rustemd; Vandevelde, Michele; Margery, Hélènee; Tassignon, Jean-Pierref

Clinical Science

Background Azodicarbonamide (ADA), a HIV-1 zinc finger inhibitor, targets a new step in viral replication and cell infectivity.

Objective A first phase I/II clinical study of ADA.

Methods ADA was administered at escalating doses concomitantly with current antiviral therapy during a 3-month open-label period in patients with advanced AIDS and documented virological failure. After 3 months, patients were randomized in a double-blind placebo-controlled withdrawal, ADA being given at the highest tolerated dosage.

Results Fifteen patients with advanced disease failing on combined antiretroviral therapy, 75% of them with proven phenotypic resistance, had a median baseline CD4 cell count of 85 × 106 cells/l, CD4/CD8 cell ratio of 0.09 and median plasma RNA viral load of 4.2 log10 copies/ml. Tolerance to ADA was dose dependent and some patients developed nephrolithiasis, glucose intolerance or showed an ADA-related cytotoxicity towards CD4 cells at higher dosages. No patient died during the study period. ADA increased CD4 cell percentage, increased the CD4/CD8 cell ratio and decreased plasma RNA viral load from baseline. At the end of the double-blind period, the ADA group, but not the placebo group, showed a significant response (P < 0.05). No phenotypic resistance to ADA was observed. Overall, 3/11 patients (27%) had consistent viral load reductions > 0.5 log10 copies/ml compared with baseline and 5/11 (45%) showed a CD4 cell recovery from baseline > 33%. In responders, ADA induced a median peak increase in CD4 cell percentage change from baseline of 65% (range 47–243%), and viral load decrease of 1.04 log10 copies/ml (range 0.52–1.23).

Conclusions The maximal tolerated dosage of ADA appears to be 2 g (three times daily). This study provides safety results that will allow larger clinical trials to confirm the preliminary efficacy data.

From the aMedizinische Poliklinik, Ludwig-Maximilians-Universität, Munich, Germany, the bInfectious Diseases Department, Centre, Hospitalier de Luxembourg, Luxembourg, the cAIDS Laboratory, Rega Institute of Medical Research, Katholieke Universiteit Leuven, Leuven, Belgium, the dData Management Department, PSI Pharma Support A.G., Zug, Switzerland, the eR&D Department, Hubriphar S.A., Brussels, Belgium and fClinical Operations Department, Tassignon & Partners S.A., Brussels, Belgium.

Received: 27 March 2000;

revised: 28 June 2000; accepted: 7 July 2000.

Requests for reprints to: Dr F.-D. Goebel, Med. Poliklinik, Pettenkoferstr. 8a,80336 München, Germany.

Sponsorship: This research was supported by a grant from Hubriphar S.A. The virology research and the development of an analytical method in blood and urine were supported by grants from the Government of the Region of Brussels.

Back to Top | Article Outline


Azodicarbonamide (ADA) is an anti-HIV compound [1,2] selectively targeting the retroviral NCp7 nucleocapsid zinc finger domains [3], without affecting mammalian nucleic and cytoplasmic zinc finger proteins [4]. The median effective concentration across many in vitro experiments is about 100 μmol/l and is similar for HIV isolates resistant to reverse transcriptase inhibitors (RTI) and/or protease inhibitors (PI) and for the wild type strains HIV-1IIIB and HIV-2ROD. Subsynergistic effects have been shown in vitro with didanosine and stavudine and additive effects with lamivudine and 9-(2-phosphomethoxyethyl) adenine. Escape mutations of the retroviral zinc finger domains are biologically unlikely [5]. No resistance has been observed so far in vitro with ADA after 30 subcultures.

However, ADA, is cytotoxic towards human peripheral blood monocytic cells (PBMC) in vitro at concentrations > 200 μmol/l [3]. ADA inhibits lymphocyte production of cytokines in vitro and in vivo, including tumour necrosis factor α, interferon γ and interleukins 2, 4 and 5, and it effectively delays graft rejection in murine models of major histocompatibility (MHC) class I and II incompatible skin allografts [6].

In the absence of a suitable and reproducible method for measuring ADA, pharmacokinetic parameters of ADA are not well determined for any route of administration. A radiolabelled study in rats showed that one third of the ADA dose administered orally was absorbed into the body and that distribution of radioactivity was rapid to all organs including the brain [7]. In vivo, ADA is almost completely converted to biurea, which has no anti-HIV properties [1–3]. ADA and biurea have been shown to induce kidney stones in rodents and in dogs when doses higher than 100 mg/kg body weight [8] were used in the most sensitive species. Dose-dependent methaemoglobin induction has been observed in one healthy volunteer after oral administration of ADA, with a peak at 1.5 h and a return to normal within 4 h after intake of ADA.

The objectives of the study described here were (i) to assess the safety and tolerability of ADA in patients with advanced AIDS at dosages up to 3 g three times daily; (ii) to describe the potential effects of ADA on critical clinical events (diarrhoea, opportunistic infections, survival and Karnofsky score) and on immunological (CD4 cell count in peripheral blood) and virological (RNA copies/ml) markers of disease progression, thereby estimating the therapeutic range; and (iii) to determine individual viral sensitivity to RTI, PI and ADA using in vitro cultures of HIV viruses before and at the end of the study and to correlate the results with the parameters of activity in vivo.

Back to Top | Article Outline


Study design

Patients chosen for this study had to be suffering from advanced AIDS. Initially, the CD4 cell count had to be < 100 × 106 cells/l, but this criterion was extended to < 200 × 106 cells/l, in view of the acceptable overall tolerability and safety of ADA. A minimum baseline plasma viral RNA load of 10 000 copies/ml was required. Patients had to be clinically worsening according to their physicians. ADA was given as ‘add-on’ to the patient's background anti-HIV medications, which had to have been taken in a stable regimen for at least 2 months prior to study entry and then throughout the study. Biological markers were measured on at least two occasions during a 1-month run-in period. During the open-label period, ADA dosage [taken three times a day (t.i.d.) throughout] was escalated at monthly intervals over the range 1, 2 and 3 g t.i.d. The study drug was randomly continued under double-blind conditions for another 3 months at the highest tolerated dosage of ADA or an equivalent dosage of placebo. Randomization for placebo withdrawal occurred prospectively at baseline. The first visit was scheduled 1 week after the beginning of ADA and monthly thereafter. The safety parameters included physical examination, vital signs, extensive blood chemistry, including methaemoglobin, methaemoglobin reductase and glutathione, haematology, urine dipstick, urine microscopy if necessary and a 12-lead electrocardiograph. Lymphocyte subsets were counted at each visit. Viral load was measured after 1 month and at the end of the open-label and the double-blind periods. Viral sensitivity was tested at baseline and at the end of the study.

The protocol was approved by the institutional ethics committees and authorized by or notified to the competent regulatory authorities. The study was conducted in accordance with the Declaration of Helsinki and GCP guidelines. Patients gave their written informed consent prior to study entry. After three patients developed nephrotoxicity at the higher dosage, this dosage was cancelled in a protocol amendment. Likewise, potential CD4 cell toxicity of ADA in combination with zalcitabine in three patients led to a protocol amendment excluding patients receiving zalcitabine.

Back to Top | Article Outline

Marker determinations

Markers were measured at local hospital laboratories. A quantitative branched DNA method (Chiron Corporation, Emmeryville, California, USA) with sensitivity limit of 500 copies/ml was used for viral load and fluorescent antibody cell sorting (FACS) analysis for lymphocyte subsets.

Back to Top | Article Outline

Responder analysis using both markers

Evaluable patients

Evaluable patients are those who entered a given dosage level during the open-label period or who entered the double-blind period, respectively, took at least one dose of the scheduled study medication during the period of interest and had at least one marker measured at the end of the period of interest.

Back to Top | Article Outline

Immunological responders

Immunological responders are evaluable patients who showed an increase of absolute CD4 cell count > 33% compared with baseline and/or those who showed an increase in their ‘area under the curve’ for CD4 cell count, compared with that at baseline, together with a minimum of one CD4 cell count measurement that was > 33% greater than their baseline value.

Patients were not considered as immunological responders if their CD4 cell count showed a sharp fall of at least 33% under ADA at the end of a dosage period compared with the previous value, as this was thought to reflect cytotoxicity from ADA. This was classed as non-response even if the final value of CD4 cell counts at the end of the period was still 33% better than at baseline.

Patients with a positive area under the curve (minus baseline) for CD4 cells over the entire period of ADA treatment were also considered responders provided there was at least one measurement of CD4 cell count ≥ 33% of baseline during ADA therapy.

Back to Top | Article Outline

Virological responders

Virological responders are evaluable patients who show a fall in viral load > 0.5 log10 copies/ml at the end of the period of interest compared with their viral load at baseline.

Back to Top | Article Outline

Responder analysis using both markers

Non-responders do not meet the definitions of either immunological or virological responders, as defined above. Responders meet the definitions of either an immunological responder or a virological responder, or both.

Back to Top | Article Outline

Ex vivo HIV-1 cultures

Blood was sampled at the investigational sites in vacutainer CPT cell preparation tubes (Becton Dickinson, Franklin Lakes, New Jersey, USA) and centrifuged according to manufacturer's instructions before transfer to the analysis laboratory at the Rega Institute, Leuven. The supernatant was separated and centrifuged at 163 x g at 20°C; the supernatant from this step was stored in portions at − 80°C. The cells were resuspended in 15 ml peripheral blood lymphocyte (PBL) medium and centrifuged at 290 x g at 20°C; the supernatant was discarded and the cells resuspended in 15 ml fresh PBL medium. A small sample (100 μl) was taken for cell count and the tubes were again centrifuged at 290 x g at 20°C. The cells were suspended in freezing solution (RPMI 1640 medium, 10% dimethyl sulphoxide, 20% fetal calf serum and antibiotics) and stored in liquid nitrogen.

Because the number of cells in the treated samples was very low, virus was grown from the plasma and not by coculture. Approximately 10 × 106 PBMC isolated from the buffy coat of healthy donors and prestimulated for 3 days with phytohaemagglutinin (PHA, (Sigma Chemical Co., Bornem, Belgium) were cultured with 1 ml patient plasma and 9 ml PBL medium. Every seventh day, a supernatant sample was taken for p24 determination to evaluate the viral growth. When at least p24 levels of 50 000 pg/ml were reached, the cell culture was centrifuged and the supernatant was stored at −80oC in batches. These formed the virus stocks. If the threshold of 50 000 pg/ml p24 was not reached, half the culture (5 ml) was discarded and 10 × 106 fresh healthy donor PBMC suspended in 5 ml PBL medium was added and the culture was continued until the p24 content reached 50 000 pg/ml. If this was not achieved by 5 weeks of culture, the procedure was repeated two more times.

Back to Top | Article Outline

Phenotypic sensitivity determinations in vitro

PBMC that had been prestimulated for 3 days were separated by centrifugation at 180 x g at 20°C and resuspended in complete medium to give approximately 4 × 106 cells/ml. One quarter of this volume was taken and kept in a humidified CO2 incubator. It was replaced with the same volume of complete medium and the cells were harvested by centrifuged at 180 x g at 20°C. Virus stock was added to the pellet using the relation: volume virus stock (ml) = cells to infect/(1000 × TCID50/ml) (where TCID50 is the median tissue culture infective dose) and the cells were resuspended. Two series of wells were created using PBMC and virus-infected PBMC for toxicity and activity assays, respectively. Wells were always manipulated moving from the lowest to the highest drug concentration and care was taken not to cross-contaminate the activity and toxicity series. The anti-HIV drugs were prepared at various dilutions in complete medium supplemented with 10 U/ml interleukin-2 and 100 μl of each dilution was placed into wells (final concentrations for ADA 5, 10, 25, 50, 100 and 200 μmol/l). Wells containing drugs at increasing concentration were inoculated with 100 μl portions of resuspended PBMC or virus-infected PBMC. Plates were incubated in a humidified CO2 (4.5%) incubator for 4 days. On day 4, the cells were resuspended in the wells and 125 μl was removed for microscopic examination of the PBMC. The drug concentration in each well was readjusted for this removal by addition of drug solutions. On day 7, 100 μl supernatant was taken from each well and stored at −80°C until p24 antigen enzyme-linked immunosorbent assay (ELISA; NEN, Paris, France) was performed. The IC50 (concentration to give 50% inhibition) value was taken as the median of triplicate estimations. The percentage protection was calculated for each concentration of the compound as [1 − (p24 value of test concentration/p24 value of the untreated control)] × 100%. Toxicity was evaluated by the trypan blue exclusion method. The percentage toxicity was calculated for each concentration of compound.

PBMC were also isolated from healthy donors by density centrifugation centrifugation (Lymphoprep, Nycomed Pharma AS Diagnostics, Oslo, Norway) and stimulated with PHA for 3 days. The activated cells were washed with phosphate-buffered saline and infected with virus as descibed by the AIDS Clinical Trial Group protocols ([9]). Briefly , PBMC (2 x 105/200 μl) were plated in the presence of serial dilutions of the test compound (e.g. for ADA, final concentrations were 5, 10, 25, 50, 100 and 200 μmol/l) and were infected with HIV stocks at 1000 CC50/ml (cytotoxic concentration lethal to 50% of cells). At day 4 postinfection, 125 μl of the supernatant was removed from the infected cultures and replaced with 150 μl fresh medium containing the test compound at the appropriate concentration. At day 7 after plating the cells, p24 antigen was detected in the culture supernatant by ELISA.

Back to Top | Article Outline

Study populations

The safety open-label population (n = 15) included all patients who received at least one dose of ADA within the open-label period.

The efficacy open-label population (n = 11) included patients who received ADA during the open-label period and for whom a CD4 cell count was available at baseline, after week 1 or month 1 and after month 2.

The double-blind population (n = 11) included all patients who entered the double-blind period and received at least one dose of study drug. This population was considered both in the efficacy and safety analysis of the double-blind period.

Back to Top | Article Outline

Data validation and statistics

All data were recorded into case report forms frequently reviewed by independent monitors and verified with source documents at the hospital sites. All data were double-key entered into a dedicated SAS database (SAS Institute, Cary, North Carolina, USA). The statistical analysis plan was approved prior to unblinding the database. The blind was kept throughout the study, except for three patients, whose individual code envelopes were opened when nephrotoxicity was observed. Formal statistical analyses were performed using two-sided tests with the 5% level of significance. The use of one-sided tests on two occasions in the clinical responder analysis is justified by the preference to minimize the risk of type II error in this pilot study.

Back to Top | Article Outline

Study medications

ADA was purchased from Fluka AG, Switzerland, and purified by recrystallisation. Gelatin capsules were filled with 500 mg ADA powder. Placebo capsules were filled with lactose and dyed to match the yellow colour of ADA. Capsules were packed in bulk in opaque polyethylene bottles with a sealed screw cap. Capsule filling, labelling and packaging were performed according to GMP standards. The stability of the capsules was confirmed at the end of the study.

Back to Top | Article Outline


The structure of this phase I/II study is represented in Fig. 1 and the demographic characteristics of the population analysed for efficacy (n = 11) are presented in Table 1. This is a phase I/II trial focused on safety and tolerability. All but one of the patients were taking standard (at the time of the study) antiretrovirals, i.e. six patients were taking a triple combination of two dideoxynucleoside RTI and PI at study entry with the remainder taking a double combination. The drug regimens included five different dideoxynucleoside RTI (zidovudine, lamivudine, zalcitabine, didanosine and stavudine) and four PI (indinavir, nelfinavir, ritonavir and saquinavir), in various combinations. All patients were failing on their second drug regimen. Five patients (45%) were taking zidovudine plus lamivudine, the most frequently prescribed combination, with or without PI. A total of 75% of the patients with a positive viral culture at study entry showed resistance to at least one of the standard antiretrovirals. Many clinical laboratory parameters were abnormal in the patients prior to the study, in particular red blood cell counts and white blood cell counts were below normal in 87% and 67%, respectively, and mean erythrocyte corpuscular volume was above normal in 73%, which was essentially related to zidovudine therapy.

Fig. 1.

Fig. 1.

Table 1

Table 1

Of 15 randomized patients, three patients dropped out within the first month for reasons unrelated to ADA treatment and one patient had his medication bottles mixed up between the open-label and double-blind periods. During the double-blind period, four of five patients dropped out of the ADA group and one of six patients dropped out of the placebo group.

Back to Top | Article Outline

Safety results

The main reason for discontinuing ADA was nephrotoxicity: renal colic in one patient, nephrolithiasis with papillary necrosis in another and increase in serum creatinine in a patient treated with indinavir and with a history of renal insufficiency before this study. The common denominator is probably the precipitation of biurea and the formation of nephrolithiasis. Casts were found in urine at the end of the period on 2 g ADA t.i.d. in the first patient and the patient developed renal colic about 1 month later. Casts were also found in one patient after 1 month of ADA at 1 g t.i.d. and at the end of the open-label period with 2 g t.i.d. The patient's creatinine levels increased in parallel, probably in relation to ADA combined with indinavir, and the patient was eventually withdrawn from the study. Blood was found in the urine in a third patient at the 2 g t.i.d. ADA dosage and appeared again at the end of the study when they were taking placebo, but the latter was probably related to the intake of 3 g ADA t.i.d. the month before. In these three patients, the total daily dosage inducing nephrolithiasis (129−142 mg/kg) was in excess of the no-observed effect level (100 mg/kg) in the most sensitive animals. No renal effects have been observed during the first month at 1 g ADA t.i.d. No patient with normal renal function at study entry developed nephrolithiasis, papillary necrosis, renal colic, or increased serum creatinine at a dosage of 2 g t.i.d. or less.

Gastrointestinal intolerance occurred in isolated patients, such as abdominal pain, diarrhoea and flatulence. In one patient who had a history of abdominal lymphoma and gastrointestinal ulcerations related to exudative enteropathy, this led to his withdrawal from the study after 1 month at 1 g ADA t.i.d. Otherwise, these symptoms were mild to moderate and medication was continued.

In vitro, ADA is cytotoxic to human T lymphocytes. Two response patterns appear to underscore the risk of cytotoxicity towards CD4 helper cells in patients. The first is a continuous dose-related decline of > 33% in CD4 cells from baseline under ADA, followed by a rebound of > 33% after withdrawal of ADA. The second is a CD4 cell recovery of > 33% after one or more consecutive dosage levels of ADA, followed by a sharp fall (> 33%) at the next higher dosage, with or without rebound of > 33% after withdrawal of ADA.

Mild glucose intolerance was associated with ADA in a dose-dependent manner. Fasting glucose (mean ± SD) increased from 96 ± 24 mg/dl at baseline to 115 ± 29 mg/dl at the end of the second month during open-label treatment. It appeared to stabilize thereafter. The differences between baseline and visit 5 (month 1) [P = 0.010, degrees of freedom (df) = 8, paired t-test] and baseline and visit 7 (month 3) (P = 0.028, df = 9) are statistically significant. In five out of six patients for whom data were available, including two with elevated glucose at baseline, glucose values decreased after withdrawal of ADA from a mean of 128 mg/dl under ADA to a mean of 102 mg/dl. The potential impact of concomitant use of PI is difficult to assess because of the sample size.

Methaemoglobin was slightly increased in two patients, confirming earlier observations in one healthy volunteer. The highest value observed was 3.5% at the dosage of 3 g ADA t.i.d., which is without clinical relevance.

Total glutathione was not affected by ADA. However, reduced glutathione increased from 175 ± 69 μmol/l at baseline to 191 ± 38 μmol/l after 1 week of therapy with ADA at 1 g t.i.d. (not statistically significantly different from baseline:P = 0.379, paired two-sided t-test; df = 10), and to 235 ± 33 μmol/l after 3 months of treatment (not statistically significantly different from baseline:P = 0.079; df = 7). Oxidized glutathione, on the contrary, decreased significantly from 501 ± 107 μmol/l at baseline to 443 ± 69 μmol/l after 1 week at 1 g ADA t.i.d. (P = 0.034; df = 10) and to 413 ± 115 μmol/l after 3 months of treatment (P = 0.022; df = 7), suggesting a biological effect of ADA but not of its metabolite biurea.

The Karnofsky index improved under ADA treatment as patients moved to higher dosages during the open-label period. The mean Karnofsky index at baseline for the whole population was 85 ± 12% and it increased progressively to 92 ± 8% towards the end of the third month. Compared with baseline, the mean indexes at visit 6 (month 2) and visit 7 (month 3) are statistically significant (month 2 P = 0.026; df = 11 and month 3 P = 0.053, df = 10; paired two-sided t-test). The paucity of data at the end of the double-blind study period did not allow any comparison with placebo.

Back to Top | Article Outline

Concomitant medications

Many concomitant medications were administered with the study medication, in particular antimicrobial medicines for treating/preventing bacterial, mycotic and viral opportunistic infections. Of key importance is the potential contribution of ADA to resistance development. From the viral sensitivity determinations, it appears that ADA did not influence sensitivity to RTI or PI utilized throughout the study and to which the patient's virus was either sensitive or resistant at baseline. No HIV strains resistant to ADA were isolated. Of interest is the decreased number of successful HIV cultures after ADA therapy: from 73% successful cultures at baseline, only 46% of the cultures were successful at the end of the study period with ADA therapy.

Back to Top | Article Outline

Safety parameters apparently not influenced by azodicarbonamide

ADA did not appear to have any effect on body weight, temperature, systolic or diastolic blood pressure or pulse. There was no apparent effect on electrocardiographic tracings, in particular there was no prolongation of QTc. No negative effects on haematology, coagulation, blood chemistry (including liver enzymes, bilirubin, total protein and protein electrophoresis, ions, blood lipids, urea, uric acid, thyroid hormones), urinalysis and urinary microscopy parameters could be attributed to ADA at any of the dosage levels.

Back to Top | Article Outline

Sensitivity to azodicarbonamide and standard antiretrovirals

Viral cultures were obtained at baseline from 11 of 15 patients (success rate of 73% for viral growth). However, two of the cultures did not yield sufficient virus for sensitivity testing.

The IC50 values for ADA ranged from 40 to 200 μmol/l with a CC50 of approximately 140 μmol/l (range 111−166). ADA seems to have a very small selectivity index, 1–3, in PBMC in vitro. There was no evidence of viral resistance to ADA.

In six out of eight patients (75%), viral resistance in vitro could be shown to one of the antiretroviral medicines taken by the patient at study entry. Few comparisons are possible with the double-blind end-of-study sample. Indeed, the success rate for growing the virus from the patient samples fell to 46%. It proved very difficult to grow the virus from samples received after add-on therapy with ADA, while the median plasma viral load did not change significantly during the treatment period In patients 6 and 7, sensitivity to the standard antiretrovirals taken throughout the study was determined. Patient 6 had a virus that was resistant to zidovudine and lamivudine at baseline and was still resistant to these at the end of the study. Patient 7 had a virus sensitive to zidovudine and zalcitabine at baseline and the virus was still sensitive at the end of the study. For these two patients, therefore, there was no evidence of a change in the sensitivity profile of the virus towards standard antiretrovirals following exposure to ADA.

Back to Top | Article Outline

Preliminary efficacy results

Figure 2 shows the evolution of surrogate markers during the open-label period. The CD4 cell change expressed as a percentage of the baseline value improved, reaching a peak of 32 ± 81% after the intermediate dosage of 2 g ADA t.i.d. However, the higher dosage of 2–3 g t.i.d. during month 3, although still showing positive recovery, was probably cytotoxic to CD4 lymphocytes in some patients. In the subgroup treated with zidovudine plus lamivudine, the CD4 cell percentage change was of the same pattern as for the open-label population, only with higher values, reaching a peak at 68 ± 110% after the intermediate dosage. During the open-label period, there was a slight but not significant reduction of viral load from that seen with 1 g ADA t.i.d. (month 1) to that seen with 2–3 g ADA t.i.d. (month 3), with a peak fall from baseline of 0.08 ± 0.33 log10 copies/ml after the middle dosage As with the CD4 cell count, there was a larger reduction of viral load in the subgroup receiving ADA as add-on to a regimen containing zidovudine plus lamivudine. The fall from baseline in this subgroup peaked at 0.17 ± 0.63 log10 copies/ml after the high dosage The CD4 cell recovery (coefficient of variation: approximately 150%) and viral load reduction (coefficient of variation approximately 400%) were not statistically significant (repeated measures ANOVA). However, in the subgroup of virological responders (n = 3; 27%), the peak fall in viral load was 0.75 log10 copies/ml at 12 weeks after the 2–3 g ADA t.i.d. dosage in this population of patients with advanced disease and no other new anti-HIV therapy.

Fig. 2.

Fig. 2.

The CD4/CD8 cell ratio (Fig. 3) showed a dose-dependent improvement at 1 g ADA t.i.d. (month 1) and 1-2 g ADA t.i.d. (month 2), and a consolidation at 2–3 g ADA t.i.d. (month 3).

Fig. 3.

Fig. 3.

During the double-blind period, the CD4 cell percentage fell in the placebo group from 59 × 106 cells/l at the beginning of the period to 41 × 106 cells/l at week 24. Patient attrition during the double-blind period in the ADA group was high because of two adverse dropouts related to high-dose ADA, one intercurrent infection and one administrative dropout. The evolution of CD4 cell count was not statistically significantly different between the two groups. In the placebo group, the CD4/CD8 cell ratio fell slightly from 0.10 at the beginning of the period to 0.09 at week 24, and viral load increased slightly from 4.43 to 4.6 log10 copies/ml.

Although the mean viral load reduction was small in the open-label population, there were three patients (7, 11 and 19) with clinically important responses with respect to both surrogate markers (Fig. 4). Consistent viral load reductions were observed in these patients over the entire treatment period with ADA (Fig. 2). Peak reduction at end of treatment (12 or 24 weeks depending on randomization) ranged from 0.52 to 1.2 log10 copies/ml. There was a continuous linear decline of plasma viral load over time. Evidence of efficacy in these three patients is based on a dose-dependent reduction of viral load > 0.5 log10 copies/ml and evidence of worsening viral load after withdrawal, if any, by at least 0.5 log10 copies/ml. One of the virological responders (patient 7) had a dose-dependent positive evolution of his CD4 cell count during the open-label period and a severe worsening after switching to placebo. Patient 11 also showed a sharp deterioration of CD4 cell count after withdrawal from the study. Patient 19 was resistant to the two PI he was receiving throughout the study (Fig. 4).

Fig. 4.

Fig. 4.

The efficacy analysis (Table 2) indicates a similar proportion of immunological responders at high and low dosage (5/11 and 6/15, respectively), but only the high dosage is associated with virological responders. There was a trend towards superiority of the high dosage with respect to the proportion of immunological and virological responders (P = 0.063, Fisher exact probability, 1-sided). During the double-blind period, the proportions of immunological, immunological and virological, and overall responders were greater in the ADA group. There was not a single immunological or virological responder in the placebo group. The overall response rate in the ADA group was statistically significantly greater than in the placebo group (P = 0.048, Fisher exact probability, 1-sided).

Table 2

Table 2

The improvement in CD4 cell count under ADA compares favourably with results with other antiretrovirals in the literature [10–30] when the median CD4 cell count at baseline is taken into account (Fig. 5). In Fig. 5, average CD4 cell changes at 12 weeks under various anti-HIV therapies (PI not included) have been plotted as a function of the median CD4 cell count at baseline. Except for one outlier value with little biological plausibility, the best results found in the literature can be linked by a straight line. The results observed in this study with ADA fall on this line. This would indicate that CD4 cell recovery with ADA was comparable in this study at 12 weeks with the best results reported so far in the literature, given the paucity of CD4 cells at baseline and given that the patient's viruses were resistant to PI.

Fig. 5.

Fig. 5.

There were no deaths during the study. Intercurrent infections occurred more frequently at the low dosage during month 1 and in the placebo group during the double-blind period (Table 3), but this requires careful interpretation because of the size of the sample and the short period of observation.

Table 3

Table 3

Back to Top | Article Outline


This is the first report about the safety, tolerability and preliminary efficacy of ADA in patients with advanced AIDS and the first report of a clinical investigation with a zinc finger inhibitor in HIV infection. In view of the small number of patients involved, the study should primarily be considered as a safety and tolerability study. Nevertheless, it is possible to look at the potential efficacy by assessment of surrogate markers.

The development of a suitable detection method for ADA in blood has not yet been possible. Consequently the pharmacokinetic parameters of ADA are unknown. This study provides some indications that ADA is absorbed via the gastrointestinal tract in its active form. Its metabolite biurea is inactive against HIV [1–3] and does not oxidize reduced glutathione. The observation of significant reductions in viral load in certain patients in this study, and the dose-dependent reduction of oxidized glutathione in serum, would suggest that ADA does get inside the cells. Indeed, the paradoxical increase of reduced glutathione in serum is an indication of oxidative stress, which could only result from systemic absorption of intact ADA, which has well-established cell toxicity. Interestingly, all three patients with serious nephrotoxicity related to high-dose ADA showed significant CD4 cell recovery during the open-label dose-escalation period, and one of these patients showed signs of CD4 cell toxicity at the 3 g ADA t.i.d. dosage. If nephrotoxicity is the consequence of large amounts of biurea being excreted via the kidneys, patients presenting such adverse events and showing evidence of antiretroviral activity testify that some, at least, of the parent compound ADA is absorbed intact via the gastrointestinal tract. Other evidence that ADA is absorbed with an intact double bond is the induction of methaemoglobin. This confirms the presence of ADA inside red blood cells, which appear to take up a large amount of the active form of ADA before a further release (on-going studies). This effect of ADA in this study does not appear to be related to the inhibition of methaemoglobin reductase It is also unlikely that biurea is responsible for the dose-related glucose intolerance observed. Rather, it is possible that ADA interferes with the formation of insulin from pro-insulin. Sulphydryl groups are involved in building the disulphide bridges in the functional insulin protein, and ADA, not biurea, is known to oxidize free sulphydryl groups.

ADA, but not biurea, induces cell death in human lymphocytes and PBMC in vitro at concentrations > 200 μmol/l [1–3]. Evidence in this study of competition between dose-related CD4 lymphocyte recovery and toxicity from ADA at high dosages is another argument supporting the hypothesis that ADA passes the gastrointestinal barrier in its active form.

Nephrolithiasis (including renal colic and renal insufficiency) is a known adverse reaction in rodents and dogs exposed for months to ADA or biurea, both insoluble in water, at dosage regimens in excess of 100 mg/kg per day [8]. The aggravation of renal failure in one patient taking indinavir with ADA indicates the need for a downward dose-adjustment study in patients with renal insufficiency and in patients receiving other potentially nephrotoxic drugs, such as acyclovir, foscarnet etc., in combination with ADA, as well as a careful monitoring of renal function. In humans, a total daily dose of 100 mg/kg perday should not be exceeded.

Methaemoglobin formation was observed in a previous study in healthy volunteers and is confirmed here. Another azo compound, phenazopyridine [10], is known to induce methaemoglobin formation. However, the level of methaemoglobin is not clinically relevant. Careful monitoring should be considered for patients taking other compounds such as dapsone, which may be associated with increased methaemoglobin level.

The dose-related and statistically significant increase in fasting glucose in serum in this study was unexpected. Recently, ADA has been shown to have potent immunosuppressant properties in rodents [6]. Glucose intolerance is a well-known adverse reaction of immunosuppressant medicines such as cyclosporin and FK506 [11,12], mycophenolate mofetil [13] and glucocorticosteroids.

Also unexpected was the important toxicity towards CD4 lymphocytes of ADA in combination with zalcitabine. Zalcitabine is known to be a drug depressing intracellular dCTP levels [14] and to be cytotoxic towards lymphocyte subpopulations [15]. ADA is now also known to reduce the dCTP pool (unpublished results). Since human lymphocytes have a low dCTP pool and are devoid of any dCTP salvage pathway [16], any drug added to zalcitabine that further depresses the dCTP pool may precipitate cell death.

Because of a structural analogy between ADA and hydroxyurea, which is potentially active against HIV [17], it is perhaps useful to highlight that ADA has a mechanism of action that differs from hydroxyurea. Hydroxyurea inhibits ribonucleotide reductase [18,19] and depresses all dNTP inside the lymphocyte, whereas ADA has no significant activity on this enzyme and depresses the dCTP pool through a different mechanism (P. Robberecht, personal communication; M. Ussery, personal communication). Hydroxyurea does not affect retroviral zinc fingers [3] and has no relevant immunosuppressive properties, as demonstrated in graft rejection studies [20].

Several immunosuppressants have been shown to have potential benefits in the treatment of AIDS. Their effect is probably indirect through prevention of cell death induced by both viral replication and cytokine release. Some of these cytokines may trigger or sustain the viral replication. ADA is an effective inhibitor of lymphocyte proliferation and cytokine release in vivo in rodents after single doses and in vitro at concentrations similar to those inhibiting HIV replication [6]. In contrast to existing immunosuppressants, ADA has also a direct effect on retroviral zinc fingers and is apparently devoid of any effect on zinc fingers associated with mammalian cell proteins [4]. As a third mechanism of action, ADA could reduce the activated target cells, hence reducing their infectivity by circulating viral particles.

Our study shows that zinc finger inhibitors represent potentially an important contribution to combination therapy in AIDS because they are not likely to produce escape mutants [5]. There was no evidence of the emergence of resistance in the present study using quasi-monotherapy with ADA, since the patients’ viruses were mostly resistant to the combination antiretroviral therapy to which ADA was added. This must be confirmed in larger studies with long-term administration. New antiretroviral medications, including especially fusion inhibitors, should be evaluated for potential synergism and to reduce the risk of developing resistance against these molecules, as already shown for one such drug, T-20. Indeed, ADA could act both before and after fusion events and potentiate these new fusion drugs.

A clinically significant reduction of viral load was observed in this study only in three patients (27%). This was, however, a very unfavourable population for the initial evaluation of a new drug in AIDS: patients had advanced AIDS, the mean CD4 cell count at baseline was 77 × 106 cells/l, most patients had several years of prior therapy with a variety of RTI, 75% of patients had a virus phenotypically resistant to one or more drugs in the patient's antiviral therapy at baseline (indicated by sensitivity measurements at study entry), 20% of patients were taking a drug combination containing zalcitabine at entry, and 60% were taking one containing lamivudine. Lamivudine may, at certain doses, interact with ADA to induce cytotoxicity in CD4 lymphocytes and lamivudine and, particularly, zalcitabine depress dCTP in the cell [21]. The high heterogeneity in combination therapies, the common protocol violations and the high dropout rate (53% over 6 months), partly owing to the access to PI, all played a role in reducing the power of the study. These difficulties can certainly be overcome in future clinical trials. Despite such unfavourable characteristics, during the double-blind period, immunological and virological responders were found in the ADA group and not in the placebo group (P = 0.048, Fisher exact, 1-sided).

The improvement in CD4 cell count under ADA compares favourably with results in the literature for other antiretrovirals when the median CD4 cell count at baseline is taken into account (Fig. 5). The absolute increase in CD4 cells observed at 3 months with ADA in the open-label population (12 × 106 cells/l) is on the curve of the best results published so far in the literature at 12 weeks from initial therapy considering the median CD4 cell count at study entry (88 × 106 cells/l for ADA).

In contrast, the overall effect of ADA on viral load in this study looks rather modest, despite clinically significant reductions greater than 0.5 and 1.1 log10 copies/ml (in one and two patients, respectively). However, the viral load that is measured in peripheral blood with a branched DNA diagnostic kit (Chiron) does not discriminate between infectious and non-infectious particles. The results in patients who show no significant improvement in viral load must be interpreted in the light of the effect of ADA on zinc fingers. Without functional zinc fingers, viral particles are not infectious. If the concentration of the zinc finger inhibitor is insufficient inside infected cells to inhibit viral replication, it may nevertheless act on extracellular particles. A reduction in viral load would not then be observed because current test methods measure total genetic load and not the infectious load. However, there should be a CD4 cell count recovery if non-infected and newly formed CD4 cells are prevented from being infected. Patients 3, 4 and 6 clearly had a positive area under the curve of CD4 cell percentage change from baseline under ADA therapy even though their viral loads were apparently stable. It would be interesting in future studies, to test the infectivity of blood samples collected from patients treated with an optimum dosage of ADA.

Other groups have experienced weak effects of rescue therapy on viral load reduction in severely immunocompromised and heavily pretreated patients, despite the use of potent anti-HIV medicines. The addition of saquinavir to the antiretroviral regimen in patients with advanced HIV disease heavily treated with nucleoside drugs resulted in only transient decreases in HIV plasma RNA and this in only half the patients [22]. In a group of 18 patients who had failed on therapy with indinavir or ritonavir, only four patients (22%) had > 0.5 log10 copies/ml decrease in viral load after 24 weeks of therapy [23]. In 13 patients refractory to standard triple therapy, the combination of nelfinavir and saquinavir showed only short-term virological efficacy in a minority of patients (39% after 4 weeks, 23% after 8 weeks and 8% after 16 and 24 weeks) [24].

The present study shows that it is possible to combine ADA with other antiretrovirals belonging to the two main classes: dideoxynucleoside RTI and PI. In combination with RTI that depress the intracellular dCTP pool, such as zalcitabine and lamivudine, ADA's cytotoxicity towards CD4 lymphocytes may be enhanced and dosage should be modified appropriately. The somewhat complex toxicity of the drug is of concern, and patients in future trials should be carefully monitored, in particular for renal function and the metabolic adverse events that have already been shown to be a problem with some of the antiretroviral drugs. Certain PI may be nephrotoxic (e.g., indinavir) and care should be taken when co-administering ADA.

In order to reduce the total body burden of biurea and/or improve the bioavailability of ADA, which is known to be only one third absorbed after oral administration in rats [7], and hence increase the intracellular levels of ADA, new pharmaceutical developments would be needed. Indeed the median particle size of the current formulation is greater than 30 μm.

The current formulation appears to show some activity on surrogate marker endpoints in this study. Larger clinical trials should be undertaken in order to confirm these very preliminary, yet encouraging, results and further approaches should also focus on the pharmacokinetics of this first candidate of the family of zinc finger inhibitors.

The optimum dosage recommended for further clinical development of this current pharmaceutical formulation appears to be 2 g t.i.d., because this dosage showed the highest mean CD4 cell percentage change from baseline, the highest mean absolute change in CD4/CD8 cell ratio from baseline, significant viral load reduction to < 1.0 log10 copies/ml after 6 months of treatment, a statistically significant improvement of the Karnofsky performance index and fewer intercurrent infections. In addition, cytotoxicity towards CD4 lymphocytes, nephrotoxicity and glucose intolerance were less frequent than at a dosage of 3 g t.i.d.

Back to Top | Article Outline


We thank Professor L. Thunus, Institute of Pharmacy, University of Liège, Belgium, for purifying azodicarbonamide to clinical grade and for testing the purity and stability of the supplies,. SANICO S.A., Turnhout, Belgium for the manufacture of study supplies to GMP standards following the randomization code kindly provided by PSI Pharma Support Inc., Switzerland and Professor P. Hermans for help in writing.

Back to Top | Article Outline


1. Vandevelde M, Witvrouw M, Sprecher S, de Clercq E, Tassignon J-P. ADA, a potential anti-HIV drug. AIDS Res Hum Retroviruses 1996, 12: 567 –568.
2. Witvrouw M, Vandevelde M, Tassignon J-P, Schmit J-C, Desmyter J, de Clercq E . ADA, a potential anti-HIV drug. 9th International Conference on Antiviral Research. Urabandai, Fukushima, Japan, May 1996 [abstract].
3. Rice WG, Turpin JA, Huang M. et al. Azodicarbonamide inhibits HIV-1 replication by targeting the nucleocapsid protein. Nat Med 1997, 3: 341 –345.
4. Huang M, Maynard A, Turpin JA. et al. Anti-HIV agents that selectively target retroviral nucleocapsid protein zinc fingers without affecting cellular zinc finger proteins. J Med Chem 1998, 41: 1371 –1381.
5. Rice WG, Turpin JA. Virus-encoded zinc fingers as targets for antiviral chemotherapy. Med Virol 1996, 6: 187 –199.
6. Tassignon J, Ismaili J, Lemoine A. et al. Azodicarbonamide inhibits T cell responses in vitro and in vivo . Nat Med 1999, 5: 947 –950.
7. Mewhinney JA, Ayres PH, Bechtold WE. et al. The fate of inhaled azodicarbonamide in rats. Fund Appl Toxicol 1987, 8: 372 –381.
8. BG Chemie. Azodicarbonamide. In Toxicological Evaluations 6. Potential Health Hazards of Existing Chemicals. Berlin: Springer-Verlag; 1992: 133 –155.
9. Japour AJ, Mayers DL, Johnson VA. et al. Standardized peripheral blood mononuclear cell culture assay for determination of drug susceptibilities of clinical human immunodeficiency virus type 1 isolates. Antimicrob Agents Chemother 1993, 37: 1095 –1101.
10. Reynolds JEF (ed.). Phenazopyridine hydrochloride. Martindale: The Extra Pharmacopoeia, 29th edn. London: The Pharmaceutical Press; 1989: 35 35.
11. Hirano Y, Mitamura T, Tamura T, Ohara K, Mine Y, Noguchi H. Mechanism of FK506-induced glucose intolerance in rats. J Toxicol Sci 1994, 19: 61 –65.
12. Ishizuka J, Gugliuzza KK, Wassmuth Z. et al. Effects of FK506 and cyclosporine on dynamic insulin secretion from isolated dog pancreatic islets. Transplantation 1993, 56: 1486 –1490.
13. EMEA (European Medicines Evaluation Agency). European Public Assessment Report for Cellcept, CPMP/767/95. London: European Medicines Evaluation Agency; February 1996.
14. Gao W-Y, Shirasaka T, Johns DG, Broder S, Mitsuya H. Differential phosphorylation of azothymidine, dideoxycytidine, and dideoxyinosine in resting and activated peripheral blood mononuclear cells. J Clin Invest 1993, 91: 2326 –2333.
15. Taylor LD, Binienda Z, Schmued L, Slikker W Jr. The effect of dideoxycytidine on lymphocyte subpopulations in nonhuman primates. Fund Appl Toxicol 1994, 23: 434 –438.
16. Bofill M, Fairbanks LD, Ruckemann K, Lipman M, Simmonds HA. T-lymphocytes from AIDS patients are unable to synthesize ribonucleotides de novo in response to mitogenic stimulation. J Biol Chem 1995, 270: 2969 –2967.
17. Malley SD, Grange JM, Hamedi-Sangsari F, Vila JR. Synergistic anti-human immunodeficiency virus type 1 effect of hydroxamate compounds with 2',3'-dideoxyinosine in infected resting human lymphocytes. Proc Natl Acad Sci USA 1994, 91: 11017 –11021.
18. Tai AW, Lien EJ, Moore EC, Chun Y, Roberts JD. Studies of N -hydroxy-N'-aminoguanidine derivatives by nitrogen 15 nuclear magnetic resonance spectroscopy and as ribonucleotide reductase inhibitors. J Med Chem 1983, 26: 1326 –1329.
19. Weckbecker G, Lien EJ, Cory JG. Properties of N -hydroxy-N'-aminoguanidine derivatives as inhibitors of mammalian ribonucleotide reductase. Biochem Pharmacol 1988, 37: 529 –534.
20. Allegra CJ, Grem JL. Hydroxyurea. Antimetabolites in Cancer: Principles and Practice of Oncology, 5th edn, Ch. 19.6. Edited by DeVita VT, Hellman S, Rosenberg SA. Philadelphia: Lipincott-Raven; 1997: 448 448.
21. Gao W-Y, Agbaria R, Driscoll JS, Mitsuya H. Divergent anti-human immunodeficiency virus activity and anabolic phosphorylation of 2',3'-dideoxynucleoside analogs in resting and activated human cells. J Biol Chem 1994, 269: 12633 –12638.
22. Walmsley S, Walach C, Fletcher D et al. Efficacy of saquinavir in patients with advanced HIV infection.. Fourth Conference on Retroviruses and Opportunistic Infections. Washington DC, January 1997 [abstract 192].
23. Deeks SG, Grant RM, Beatty GW, Horton C, Detmer J, Eastman S. Activity of a ritonavir plus saquinavir-containing regimen in patients with virologic evidence of indinavir or ritonavir failure. AIDS 1998, 12: F97 –F102.
24. Reiser M, Salzberger B, Stiepel A, Ivette A, Hoetelmans R, Fatkenheuer G. Virological efficacy and drug levels of nelfinavir plus saquinavir as salvage therapy in HIV-infected patients refractory to standard triple therapy XII International Conference on AIDS. Geneva, June1998 [abstract 22338].
25. Abrams DI, Goldman AI, Launer C. et al. A comparative trial of didanosine or zalcitabine after treatment with zidovudine in patients with human immunodeficiency virus infection. :The Terry Beirn Community Programs for Clinical Research on AIDS [see comments]. N Engl J Med 1994, 330: 657 –662.
26. Bartlett JA, Benoit SL, Johnson VA. et al. Lamivudine plus zidovudine compared with zalcitabine plus zidovudine in patients with HIV infection. :A randomized, double-blind, placebo-controlled trial. North American HIV Working Party. Ann Intern Med 1996, 125: 161 –172.
27. Biron F, Lucht F, Peyramond D. et al. Anti-HIV activity of the combination of didanosine and hydroxyurea in HIV-1-infected individuals. J Acquir Immune Defic Syndr Hum Retrovirol 1995, 10: 36 –40.
28. Carr A, Vella S, de Jong MD, Sorice F. et al. A controlled trial of nevirapine plus zidovudine versus zidovudine alone in p24 antigenaemic HIV-infected patients. :The Dutch–Italian–Australian Nevirapine Study Group. AIDS 1996, 10: 635 –641.
29. Cheeseman SH, Havlir D, McLaughlin MM. et al. Phase I/II evaluation of nevirapine alone and in combination with zidovudine for infection with human immunodeficiency virus. J Acquir Immune Defic Syndr Hum Retrovirol 1995, 8: 141 –151.
30. Danner SA, Carr A, Leonard JM. et al. A short-term study of the safety, pharmacokinetics, and efficacy of ritonavir, an inhibitor of HIV-1 protease. :European–Australian Collaborative Ritonavir Study Group. N Engl J Med 1995, 333: 1528 –1533.
31. Fischl MA, Richman DD, Hansen N. et al. The safety and efficacy of zidovudine (AZT) in the treatment of subjects with mildly symptomatic human immunodeficiency virus type 1 (HIV) infection. :A double-blind, placebo-controlled trial. The AIDS Clinical Trials Group [see comments]. Ann Intern Med 1990, 112: 727 –737.
    32. Kitchen VS, Skinner C, Ariyoshi K. et al. Safety and activity of saquinavir in HIV infection [see comments]. Lancet 1995, 345: 952 –955.
      33. Markowitz M, Saag M, Powderly WG. et al. A preliminary study of ritonavir, an inhibitor of HIV-1 protease, to treat HIV-1 infection. N Engl J Med 1995, 333: 1534 –1539.
        34. Meng TC, Fischl MA, Boota AM. et al. Combination therapy with zidovudine and dideoxycytidine in patients with advanced human immunodeficiency virus infection. :A phase I/II study [see comments]. Ann Intern Med 1992, 116: 13 –20.
          35. Pluda JM, Cooley TP, Montaner JS. et al. A phase I/II study of 2′-deoxy-3′-thiacytidine (lamivudine) in patients with advanced human immunodeficiency virus infection. J Infect Dis 1995, 171: 1438 –1447.
            36. Rutschmann OT, Kaiser L, Perrin L, Fathi M, Hirschel BJ. Adding saquinavir to stavudine in patients with advanced HIV-1 infection. AIDS 1997, 11 (4): 548. 548.
              37. Skowron G, Bozzette SA, Lim L. et al. Alternating and intermittent regimens of zidovudine and dideoxycytidine in patients with AIDS or AIDS-related complex. Ann Intern Med 1993, 118: 321 –330.
                38. Staszewski S, Miller V, Rehmet S. et al. Virological and immunological analysis of a triple combination pilot study with loviride, lamivudine and zidovudine in HIV-1-infected patients. AIDS 1996, 10: F1 –F7.
                  39. Yarchoan R, Perno CF, Thomas RV. et al. Phase I studies of 2',3'-dideoxycytidine in severe human immunodeficiency virus infection as a single agent and alternating with zidovudine (AZT). Lancet 1988, i: 76 –81.
                    40. Jablonowski H. Studies of zidovudine in combination with didanosine and zalcitabine. J Acquir Immune Defic Syndr Hum Retrovirol 1995, 10: S52 –S56.
                      41. Spruance SL, Pavia AT, Peterson D. et al. Didanosine compared with continuation of zidovudine in HIV-infected patients with signs of clinical deterioration while receiving zidovudine. :A randomized, double-blind clinical trial. The Bristol–Myers Squibb AI454-010 Study Group [see comments]. Ann Intern Med 1994, 120: 360 –368.
                        42. Giacca M, Zanussi S, Comar M. et al. Treatment of human immunodeficiency virus infection with hydroxyurea: virologic and clinical evaluation. J Infect Dis 1996, 174: 204 –209.
                          43. Hubriphar study HU/ADA-03 EU.
                            44. Greka P, Michaliou S, Kavatha D. et al. Treatment with ziduvodine and lamivudine combination in HIV(+) patients: a preliminary report. AIDS 1996, 10 (suppl 2): P42. P42.
                              © 2001 Lippincott Williams & Wilkins, Inc.