Placebo-controlled trial of prednisone in advanced HIV-1 infection

Objective: To examine the safety and the immunologic and virologic consequences of corticosteroid use in HIV-1 infection.

Cited Here...: A randomized, double-blinded, placebo-controlled trial of corticosteroid administration in 41 patients with advanced HIV-1 infection. Patients had a baseline median CD4 cell count of 131 × 10⁶ cells/l at enrollment and 85% had a history of opportunistic infection. All but one of the patients had been taking stable antiretroviral regimen, including a protease inhibitor in 36, for a median duration of 158 days. Patients were randomized to 8 weeks of prednisone 0.5 mg/kg daily or placebo.

Cited Here...: No AIDS-defining events occurred; two patients in each group developed oral candidiasis, and two patients on prednisone developed mild herpes simplex flares. None who developed oral candidiasis or herpes simplex was receiving prophylaxis and each responded promptly to therapy. In the prednisone group, two patients developed hyperglycemia and one diabetic increased insulin requirements. CD4 cell counts and plasma HIV-1 RNA levels did not change, but plasma tumor necrosis factor α levels and CD38+CD8+ cells decreased significantly in those taking prednisone.

Conclusion: Short-term prednisone administration is well tolerated and reasonably safe in advanced HIV-1 disease and decreases immune activation without effects on HIV-1 RNA levels or CD4 cell counts. These results suggest that, in stable HIV-1 disease, these immune activation markers are more likely consequences of but not inducers of HIV-1 replication.

Introduction

HIV-1 replication and disease pathogenesis may be linked to the degree of immune activation. Biologic markers of increased immune activation are predictive of CD4 T cell decline and of clinical deterioration ^[1-3]. The effect of immune activation on HIV-1 expression in vivo is supported by studies that show increased plasma HIV-1 RNA levels in patients with concurrent infection and by reports that administration of proinflammatory cytokine inhibitors may decrease HIV-1 replication in vivo^[4]. Consequently, interventions that suppress cellular activation may result in decreased HIV-1 expression, with a favorable effect on the progression of HIV-1 disease. Uncontrolled clinical trials of corticosteroids in HIV-1 infection have suggested a beneficial effect on serum p24 antigen levels and on CD4 cell counts ^[5-7]. We report here the results of the first randomized, double-blind, placebo-controlled trial of corticosteroid administration in patients with advanced HIV-1 infection.

Methods

Patients

Subjects were enrolled in the study if they fulfilled the following criteria: HIV-1 infection as confirmed by enzyme-linked immunosorbant assays (ELISA) and Western blot, age 18 years or over, able to give written informed consent, no evidence for acute systemic fungal or mycobacterial infection, blood CD4 cell count < 200 × 10⁶ cells/l and receiving a stable antiretroviral regimen for a minimum of 8 weeks prior to study entry. This antiretroviral regimen was maintained throughout the course of this study. Concomitant use of other immunomodulators (e.g. thalidomide, interleukin 2, pentoxyfyline (oxpentifyline), non-steroidal anti-inflammatory drugs) was not allowed during the study period.

Trial design and treatment

Patients were randomized with equal probability and in a double-blinded fashion to receive prednisone at a daily dose of 0.5 mg/kg or placebo. The prednisone (Roxane, Columbus, Ohio, USA) and the placebo (prepared in-house) were provided in suspensions with similar taste, smell and appearance. After 8 weeks of treatment, patients were provided with 2-week tapering doses of the study drug. All patients received prophylaxis for Pneumocystis carinii pneumonia with trimethoprim-sulfamethoxazole, dapsone or aerosolized pentamidine. Patients with recurring herpes simplex virus infection or oral candidiasis (thrush) were placed or maintained on suppressive doses of acyclovir (aciclovir) or an antifungal drug, respectively. Patients with a previous history of a positive tuberculin skin test must have completed at least 6 months of preventive therapy for tuberculosis. Patients with Mycobacterium avium complex (MAC) infection must have had no acute symptoms of infection and must have been receiving a stable approved antimycobacterial regimen.

Clinical endpoints and laboratory measurements

Evaluations were performed at baseline and at weeks 4 and 8. At each evaluation, patients underwent a complete medical history, quality of life questionnaire, Karnofsky Performance Score determination and a physical examination that included weight and blood pressure measurements. Laboratory evaluations included serum glucose, blood chemistries, complete blood counts, and flow cytometric measurement of CD4, CD8 and CD38+CD8+ lymphocytes. Plasma tumor necrosis factor α (TNF-α) and interleukin 6 (IL-6) were measured in batch by ELISA (Medgenix, INCSTAR, Stillwater, Minnesota, USA and R&D Systems, Minneapolis, Minnesota, USA, respectively) on frozen samples. Plasma HIV-1 RNA was measured using the Roche HIV-1 Monitor Test (linear range 400-750 000 copies/ml; Roche Diagnostic System, Brachburg, New Jersey, USA). If the prestudy RNA was < 5000 copies/ml, then an ultrasensitive assay was used (linear range 50-75 000 copies/ml). Lymphocyte apoptosis was measured using propidium iodide staining as described previously ^[8,9].

Statistical methods

Measures of central tendency were estimated for clinical, immunologic and virologic characteristics according to treatment group at baseline, 4 and 8 weeks. Mean values at baseline were compared between the prednisone and placebo group with Student's t-test. Correlations between measurements were estimated using the Pearson correlation coefficient. A series of linear regression models for repeated measures compared the rate of change between the two groups for each outcome measure ^[10].

Results

Patient characteristics

Of the 41 enrolled patients, 39 were male and two were female. Karnofsky Performance Score was ≥ 90 in all patients. Most patients had a history of prior opportunistic infection, as shown in Table 1. All but one of the patients, were receiving antiretroviral therapies; 36 patients were receiving a protease inhibitor. The median duration of stable antiretroviral therapy was 187 days (range 60-704). The immunologic and virologic characteristics at study entry are detailed in Table 1. The baseline mean CD4 cell count was 129 × 10⁶ cells/l (range 0-230). The baseline median plasma HIV-1 RNA level was 4253 copies/ml (range < 50 to > 750 000). Baseline characteristics were well matched except for higher (0.5 log₁₀,NS) plasma HIV-RNA in the prednisone group.

Table 1

Clinical evaluations

Three patients did not complete the study; two self-withdrew, the first (in the placebo group) at week 7 to change antiretroviral therapy and the second (in the prednisone group) at week 5 after surgery for gallstone cholecystitis. One patient in the prednisone group was taken off study because of adverse effects (hyperglycemia and oral candidiasis). No other major toxicities or AIDS-defining events occurred. In the prednisone group, mean weight (± SD) increased significantly [from 170.8 ± 8.4 lb (77.6 ± 3.8 kg) at baseline to 174.6 ± 8.7 lb (79 ± 4 kg) at week 4 (P ≤ 0.01) and 176.8 ± 9.1 lb (80.4 ± 4.1 kg) at week 8 (P ≤ 0.05)] compared with baseline. Two patients in each group developed oral candidiasis, which responded promptly to antifungal therapies. Two patients in the prednisone group developed mucocutaneous herpes simplex virus recurrence, which resolved promptly with therapy. The patients who developed oral candidiasis or herpes simplex virus recurrence were not receiving recommended prophylaxis. Overall, nine patients in the prednisone group had a history of previous herpes simplex virus infection; two were not on prophylaxis and had a herpes flare, and seven were on prophylaxis and did not experience a flare. Two patients in the prednisone group developed hyperglycemia, mild in the first but severe (serum glucose of 550 g/dl) in the second. Hyperglycemia in the more severe case resolved promptly after discontinuing the prednisone, while the patient with mild hyperglycemia remained on oral hypoglycemics after the study. Of note, central obesity was noted at the same time, which suggest a possible relationship between the hyperglycemia and protease inhibitor therapy. In addition, one diabetic patient in the prednisone group developed increased insulin requirements during the study period.

Laboratory evaluations

During the course of the trial, CD4 cell counts (mean ± SD) remained stable in the placebo group, but there was a tendency toward a decrease in CD4 cell count at week 4 in the prednisone group from 128.5 (± 15) × 10⁶ cells/l at baseline to 110 (± 23) × 10⁶ cells/l at week 4 and 126 (± 23) × 10⁶ cells/l at week 8 (P > 0.05). The changes seen for CD4 percentage (mean ± SD) paralleled those seen for absolute CD4 cell count: from 8.5 ± 1% at week 0 to 7.9 ± 1% at week 4 and 8 ± 1.25% at week 8 (P > 0.05). By contrast, the decrease in CD8 cell count (mean ± SD) was significant in the prednisone group, from 1024 (± 483) × 10⁶ cells/l at baseline to 718 (± 476) × 10⁶ cells/l at week 4 (P > 0.05). The percentage of CD8 cells displaying the surface activation marker CD38 (CD38+CD8+) decreased significantly at both week 4 and week 8 in the prednisone group from 33.75 ± 18.02% at baseline to 26.21 ± 18.62% at week 4 (P < 0.01) and to 28.47 ± 16.51% at week 8 (P < 0.05) (Fig. 1).

Fig. 1

Plasma TNF-α levels (mean ± SD) fell from a mean of 31.20 ± 13.45 pg/ml at baseline in the prednisone group to 18.65 ± 9.26 pg/ml at 4 weeks and 21.58 ± 8.36 pg/ml at 8 weeks (P < 0.01 at both time points) (Fig. 2). Similarly, plasma IL-6 levels decreased significantly from 2.90 ± 3.78 pg/ml at baseline to 2.28 ± 3.58 pg/ml at 4 weeks (P < 0.05) and to 2.32 ± 3.22 pg/ml at 8 weeks (P > 0.05) (Fig. 2). We also observed a transient decrease in the percentage of apoptotic lymphocytes at week 4 in the prednisone arm, but this change was not significant (Fig. 3).

Fig. 2

Fig. 3

There was no significant change in plasma HIV-1 RNA levels in either study arm (Fig. 3). Plasma HIV-1 RNA levels were < 400 copies/ml at study entry in 14 patients (six in the prednisone arm); four of these patients (one in the prednisone group and three receiving placebo) had plasma HIV-1 RNA levels of > 400 copies/ml during the study period.

Discussion

Although prednisone is used to treat a variety of complications of HIV-1 infection, little is known about the virologic and immunologic consequences of this therapy. The results of this study demonstrate that a short, 8-week course of prednisone is apparently reasonably safe, even in the setting of advanced HIV-1 infection and multiple prior opportunistic infections. Apart from hyperglycemia, a recognized complication of corticosteroid administration, no major toxicities were observed and no AIDS-defining events occurred. Indices of immune activation decreased and CD4 cell counts and plasma HIV-1 RNA levels remained stable. Antiviral prophylaxis should be provided to persons with HIV-1 infection and prior herpes simplex viral infection who are treated with corticosteroids. Likewise, antifungal prophylaxis may prevent recurrence of mucosal candidiasis in HIV-1-infected persons receiving prednisone.

Earlier studies had reported corticosteroid-associated accelerated clinical progression of Kaposi's sarcoma in the pre-AIDS era ^[11-13], as well as in AIDS patients who received corticosteroids for other complications of AIDS including Pneumocystis carinii pneumonia, toxoplasmic encephalitis, and non-Hodgkin's lymphoma ^[12-17]. In one of these studies ^[17], the median dosage of glucocorticosteroids (prednisone) was 180 mg/day (range 50-400) and median duration was 20 days (range 14-60). Others described the development of cytomegalovirus and Mycobacterium avium complex infections in HIV-1-infected patients who received relatively lower doses of prednisone (60 mg/day) for 2-11 weeks (median 5 weeks), but in the setting of more advanced disease (median CD4 cell count 48 × 10⁶ cells/l) and HIV nephropathy ^[14,15]. Our 8-week study found that 0.5 mg/kg body weight prednisone daily [a median daily dose of 36 mg (range 27-65)] was relatively safe in patients with less advanced HIV disease (median CD4 cell count 130 × 10⁶ cells/l). This is consistent with the observations of Andrieu et al., who showed that individuals with higher CD4 cell counts (> 200 × 10⁶ cells/l) tolerated 0.5 mg/kg of prednisolone daily very well, except for a few episodes of oral candidiasis and herpes zoster ^[5]. Therefore, the risk of glucocorticoids is greatest when higher doses (> 60 mg prednisone daily) are used, or when lower doses are used in the setting of very low CD4 cell numbers or of another immunosuppressive condition, such as renal failure.

We found that administration of corticosteroids was associated with a decrease in immune activation, as reflected in decreases in plasma TNF-α levels, a more modest and transient decrease in plasma IL-6 levels and decreases in the frequency of CD8 T cells that express the activation marker CD38. TNF-α enhances the in vitro expression of HIV-1 by activating the cellular transcription factor NF-κB ^[18-20]; although corticosteroids also can block NF-κB activity directly through enhancing iκB activity ^[21], we could demonstrate no effect of prednisone administration on plasma HIV-1 RNA levels. As a result, it is likely that the elevated levels of TNF-α that are seen in untreated, uncomplicated HIV-1 infection and that are decreased by the administration of antiretroviral therapies ^[22] are primarily a consequence of HIV-1 replication and may contribute little to enhancing HIV-1 replication directly. Alternatively, the modest inhibition of TNF-α production by prednisone may have been insufficient to affect HIV-1 replication. A different relationship between TNF-α levels and HIV-1 replication may be operative in HIV-1 infection complicated by opportunistic infections such as tuberculosis ^[23], and the role of immunosuppression in these settings should be explored cautiously.

The modest decrease in CD38 expression as a result of prednisone administration may provide some insight as to the regulation of expression of this molecule in HIV-1 infection. Increased expression of CD38 on CD4 and CD8 T cells is seen in HIV-1 infection ^[3,22] and although CD38 levels and plasma HIV-1 RNA levels are tightly correlated, CD38 expression may be a better predictor of immune deterioration than is plasma HIV-1 RNA ^[24]. Since we have observed a decrease in CD38 expression without a significant change in HIV-1 replication, this would suggest that the effect of HIV-1 replication on CD38 expression is mediated indirectly through a corticosteroid-sensitive pathway. This observation is consistent with a model wherein HIV-1 replication drives TNF-α expression, which induces CD38 expression and this is a marker for (rather than a mediator of) CD4 cell losses ^[25]. Whether decreasing CD38 expression will have a beneficial effect on HIV-1 disease outcome cannot be predicted. However, in this short-term study, there was no effect of corticosteroid administration on CD4 T cell numbers. In this regard, an earlier open label study performed among HIV-1 infected patients with a median baseline CD4 T cell count of 438 × 10⁶ cells/l demonstrated a rapid and durable increase in circulating CD4 T cells that averaged 119 × 10⁶ cells/l after 1 year of treatment with daily prednisolone 0.3-0.5 mg/kg ^[5]. The patients with more advanced disease studied in the present trial had no demonstrable increase in circulating CD4 T cells. In the earlier study ^[5], circulating CD8 T cell numbers did not change, but in our patients with more advanced disease there was a small decrease in circulating CD8 T cells in those treated with corticosteroids. This decrease was primarily among CD8 cells that coexpressed CD38.

In our study, there was no significant effect of prednisone therapy on plasma HIV-1 RNA levels. This is in agreement with earlier findings in a similar population of patients with advanced HIV infection ^[26] but contrasts with other studies where significant, albeit transient, declines in p24 antigen and plasma HIV-1 RNA levels were observed ^[5-7]. Since increases in immune activation markers may precede changes in HIV-1 RNA levels and CD4 cell counts by many months ^[27] it is conceivable that the brief duration of our study may have missed a later fall in plasma HIV-1 RNA levels caused by decreases in immune activation. We suspect it is more likely, however, that in this stable yet advanced population, immune activation is a consequence rather than a cause of HIV-1 replication.

Progressive HIV-1 infection is characterized by an inexorable loss of circulating T lymphocytes. One postulated mechanism of cell loss is an accelerated tendency to programmed cell death or apoptosis ^[28]. Enhanced spontaneous and inducible lymphocyte apoptosis has been demonstrated among patients infected with HIV and this may be related to heightened immune activation ^[29,30]. The effect of corticosteroids on apoptosis is complex; T lymphocyte apoptosis can be induced by a variety of stimuli including exposure to corticosteroids. However, HIV-1-infected cells appear to be resistant to glucocorticoid-induced cell death; this may be related to alterations in glucocorticoid receptor expression ^[31]. We found a trend to a decrease in the frequency of spontaneous apoptosis in circulating lymphocytes of HIV-1-infected patients who were treated with corticosteroids and Andrieu et al. found that prednisolone therapy decreased activation-induced apoptosis in HIV-1-infected persons ^[5]. These observations suggest that, even though the inducers of these cell death mechanisms may differ in HIV-1 disease, a more proximate mediator of apoptosis in HIV-1 disease may be glucocorticoid sensitive, for example caspase 3 activity ^[32].

A short-term, 8-week course of glucocorticoids was well tolerated and reasonably safe in a small group of HIV-1-infected patients with advanced disease and multiple prior opportunistic infections. HIV-1-infected patients receiving corticosteroids should receive prophylaxis if they have a history of recurrent herpes simplex infections. CD4 cell counts and plasma HIV-1 RNA levels did not change during the study period, despite a significant decrease in plasma TNF-α levels and in the numbers of CD38+CD8+ T cells. These findings indicate that immune activation in uncomplicated HIV-1 infection is likely to be a consequence of HIV-1 replication. As corticosteroids may be used to treat complications of HIV-1 infection, such as HIV-associated nephropathy and other rheumatologic disorders, the long-term safety and activities of longer durations of therapy should be evaluated.