This document provides information on the results of a clinical trial of trimethoprim/sulfamethoxazole (TMP/SMX) vs aerosolized pentamidine (AP) when given in conjunction with zidovudine for the prevention of recurrent Pneumocystis carinii pneumonia in persons with AIDS. This study was conducted by the AIDS Clinical Trials Group (ACTG) of the National Institute of Allergy and Infectious Diseases (NIAID). The information is provided to you, as a health care practitioner, to serve as preliminary information while a publication is being readied for submission to a peer-reviewed medical journal.
Introduction and Background
ACTG 021 is a Phase III comparison study of the toxicity and efficacy of zidovudine (ZDV, 100 mg every 4 hours) combined with either (1) trimethoprim/ sulfamethoxazole (TMP/SMX), one double-strength tablet daily, or (2) aerosolized pentamidine (AP), 300 mg every 4 weeks using the Marquest Respirgard II nebulizer, for preventing recurrence of Pneumocystis carinii pneumonia (PCP) and extending survival.
Interim data from this trial were recently reviewed by an independent Data and Safety Monitoring Board (DSMB) convened by NIAID. Members of the DSMB had no involvement in the conduct of this study. All are senior clinical trials researchers, including a statistician and physicians experienced in treating AIDS patients with opportunistic infections. The DSMB recommended early termination of the trial based on the results described below. A detailed description of these study results is attached.
Study Design and Results
This analysis is based on 310 participants accrued between July 1988 and November 1990 and followed through July 1, 1991. Eligible participants included those who had recovered from an initial episode of PCP and did not have known treatment-limiting adverse reactions to any of the trial therapies. The primary endpoint of the trial was the time to recurrent PCP. Protocol-directed management of adverse events included crossover to the other prophylactic medication. Twenty-seven percent of the participants randomized to TMP/SMX and 10 percent of the participants randomized to AP switched treatment during the trial. For all statistical analyses, participants were grouped according to the originally assigned treatment, regardless of whether or when they crossed over (i.e., an intent-to-treat analysis).
IMPORTANT THERAPEUTIC INFORMATION ON PREVENTION OF RECURRENT PNEUMOCYSTIS PNEUMONIA IN PERSONS WITH AIDS
The next portion of this document contains a more detailed summary of the data reviewed by the DSMB in an interim analysis. Due to the clinical importance of the results of ACTG 021, much of the interim data are being released in this document in order to assist health care providers and patients in their clinical decision-making concerning secondary PCP prophylaxis. All data are being reviewed for accuracy, and a manuscript is being prepared for submission to a peer-reviewed journal. The results that appear in the final manuscript may differ slightly from those that appear in this document: Updates to the data, however, will not change the substance of the conclusions.
This study was conducted at 23 sites of the AIDS Clinical Trials Group (ACTG), a network of AIDS clinical research centers across the United States that is funded by the National Institute of Allergy and Infectious Diseases (NIAID), part of the National Institutes of Health (NIH).
Protocol 021 was originally designed as a three-arm, randomized, open-label, comparison of AP, TMP/SMX, and Fansidar. The Fansidar arm was discontinued in March 1989 because of poor accrual; data from the 22 patients assigned to that treatment are not included in the analyses reported in this document. These analyses included only data from patients who were randomized between July 1988 and November 1990 to receive either AP or TMP/SMX.
Eligible patients included those who had recovered from an initial episode of morphologically confirmed PCP within 10 weeks prior to study enrollment. Treatment of this initial episode had to meet specified minimum criteria to be considered adequate. Patients were excluded if they had previously experienced an adverse reaction to one of the study therapies that required its discontinuation, had substantial compromise in hematologic, hepatic, or renal function, or had received more than 2 weeks of secondary prophylaxis following the PCP episode. Patients were randomized in a stratified manner by whether or not they had received conventional (at least 14 days' TMP/SMX or parenteral pentamidine) or unconventional therapy, and by whether they had less than or greater than 6 weeks' treatment with ZDV prior to the initial episode of PCP.
The primary objectives of this study were to compare the efficacy and toxicity of ZDV + TMP/SMX and ZDV + AP in (1) preventing the recurrence of PCP in and (2) extending survival of people with AIDS who had recovered from their initial episodes of PCP. Secondary objectives included an assessment of the impact of the two regimens on the occurrence of toxoplasmosis and serious bacterial infections.
The protocol specified that patients experiencing adverse events believed to be related to one study regimen were to be crossed over to the alternative regimen. Detailed algorithms were developed for the management of specific toxicities. It was anticipated that these algorithms and prevailing clinical practice would result in more crossovers among those assigned to TMP/SMX than among those assigned to AP. For example, patients receiving TMP/SMX and experiencing persistent grade 3 or worse hematologic toxicity (hemoglobin < 8.0 Gm/dl if due to hemolysis or platelets < 50,000/cm or granulocytes < 750/cm) were crossed over to AP, while patients receiving AP and experiencing the same level of hematologic toxicity were maintained on their assigned prophylactic therapy.
Several changes in the protocol were implemented during the course of the study. In March 1989, the Fansidar arm was eliminated to reduce the total sample size and to speed accrual. In March 1990, the dose of ZDV was changed from 1200 mg to 600 mg daily, in response to data from ACTG protocol 002 that demonstrated equivalent efficacy and reduced toxicity with the use of the lower dose. At the second DSMB review of ACTG protocol 021 in May 1990, the accrual goal for the trial was expanded to 300 patients (150/arm) and the follow-up was extended to 18 months because of the occurrence of fewer cases of recurrent PCP than had been expected and frequent treatment crossovers.
The primary analysis is based on intent to treat, that is, the comparison of groups defined by the treatment assignment at initial randomization.
The study was designed to have sufficient power to detect a difference of 15 percent versus 5 percent in a two-tailed test of the 1-year PCP recurrence rates. The accrual goal was increased to 150 participants per arm in May 1990 after an interim analysis revealed a low incidence rate. The revised sample size calculation was made assuming that as many as 15 percent of the patients would be lost to follow-up and that the crossover rate might be as high as 25 percent in the TMP/SMX group and 5 percent in the AP group. The study was designed to allow five interim reviews of the data by the DSMB prior to the final analysis.
Baseline characteristics and outcome variables were compared using the Wilcoxon and chi-square test, two-tailed Fisher's exact test or Student's t-test as appropriate. The distributions of times to PCP recurrence, first episode of severe toxicity, and first episode of ZDV discontinuation were compared using a stratified logrank test appropriate to the randomization schema. In these analyses, data from patients who died prior to experiencing a recurrence were considered as censored at the time of their death.
This report is based on the 310 patients who were randomized to AP or TMP/SMX between July 25, 1988 and November 27, 1990. Analyses are based on follow-up through July 1, 1991. As of that date, 189 participants were still being followed, 31 had been lost to follow-up, and 90 had died. Median follow-up was 17.4 months (18.1 months in the TMP/SMX arm and 17.0 months in the AP arm).
The median age of participants was 36 years. The majority were male (94.2 percent) and white (70.5 percent). Twenty- three percent were black. An additional stratification by reported ethnic background indicated that 17 percent of the participants were Hispanic, 60 percent were white, 22 percent were black, and 1 percent were of another ethnicity. As a group, these patients had the severe depletion of CD4 lymphocytes characteristic of advanced HIV infection: the median baseline CD4 count was 56 per mcL. Although 33 (11.2 percent) had CD4 cells greater than 200/mcL at study entry, the percent CD4 positive cells of the total lymphocyte count in these patients has not been assessed. The two treatment groups were not significantly different with respect to any baseline characteristics except hemoglobin (Table 1). The difference, 12.0 versus 11.5 Gm/dL, was not deemed clinically meaningful.
Time to recurrent PCP was the primary endpoint of thi study Fifty cases had occurred at the time of analysis:
14 (9.1 percent) among participants assigned to TMP/SMX and 36 (23.1 percent) among those assigned to AP (Table 2). The estimated 1-year rate of PCP recurrence was 4.5 percent in the TMP/SMX arm and 18.5 percent in the AP arm (Kaplan-Meier analysis). The estimated 18-month rates were 11.4 percent and 27.6 percent, respectively. Time to recurrence of PCP was significantly prolonged in the TMP/SMX group as compared with the AP group (stratified logrank test, p=.0002). In a proportional hazards model, with adjustment for baseline CD4 counts, the hazard of PCP recurrence for patients randomized to AP was 3.25 times that of patients randomized to TMP/SMX (p=.0005).
Thirty-nine of the 50 recurrences were confirmed by visualization of the pathogen. The other 11 diagnoses were made on a clinical basis only. When the analysis is limited to confirmed cases only, the statistical results are similar (p=.0018) to those obtained when both confirmed and clinically diagnosed (presumed) cases are considered. One case of extrapulmonary pneumocystosis was observed in a patient receiving AP.
Patients were stratified according to racial and ethnic background (white, black, and Hispanic) within each treatment group; no significant differences were found when comparing time-to-recurrence of PCP for each stratum in either the TMP/SMX or the AP treatment groups.
The results of analysis based on actual treatments being administered at the time of recurrence reinforces the conclusions of the intent-to-treat analysis. Early cases of PCP were distributed evenly in both groups. Three patients assigned to TMP/SMX and two patients assigned to AP developed recurrent PCP within 3 weeks of assignment. However, of the 14 PCP recurrences among those assigned to TMP/SMX, 7 events occurred after the patients had crossed over to AP in accord with protocol-mandated toxicity management. All seven had received AP for several months prior to the diagnosis of recurrent PCP. In contrast, all of the PCP recurrences among patients randomized to AP occurred while patients were either receiving AP or awaiting their initial treatment. Thus, only four patients who received more than 3 weeks of TMP/SMX and tolerated the continued use of the medication experienced recurrences.
There have been 90 deaths, 43 among those assigned to TMP/SMX and 47 among those assigned to AP (Table 3). The median survival was 25.8 months and 22.8 months, respectively (logrank test, p=.32). Three of the 14 PCP recurrences among participants assigned to TMP/SMX and 2 of the 36 recurrences among those assigned to AP resulted in death. This difference was not statistically significant (Fisher's test, p=0.22). Other deaths were attributed to a variety of conditions, including other pneumonias, opportunistic infections (primarily disseminated MAC and CMV) and malignancies, HIV encephalopathy, and sepsis.
There were 50 recurrences of PCP during the study: 14 among those assigned to TMP/SMX and 36 among those assigned to AP. Participants randomized to TMP/SMX experienced a significant delay in the time to recurrent PCP compared to those randomized to AP (stratified logrank, p=.0002). The 1-year PCP recurrence rate was estimated to be 4.5 percent for participants randomized to TMP/SMX, and 18.5 percent for those randomized to AP. The 18-month PCP recurrence rates were estimated to be 11.4 percent and 27.6 percent, respectively. The risk of developing recurrent PCP for participants randomized to AP was estimated to be 3.25 times that of participants randomized to TMP/SMX. Analysis of data based on the treatments actually received at the time of PCP recurrence yielded even more compelling results. No significant differences in time to PCP recurrence were observed among participants in each treatment group when they were stratified by race and ethnicity (white, black, Hispanic) and the experiences of these subgroups were compared.
There were no statistically significant differences between the treatment groups in either survival or the occurrence of hematologic or hepatic adverse events. Death due to PCP were few (five) and were distributed evenly between the two treatment arms. No significant differences in time to hematologic or hepatic abnormalities were observed among participants in each treatment group when they were stratified by race and ethnicity (white, black, Hispanic). Too few women were enrolled in the study to allow subgroup analyses based on gender.
While both regimens are effective, the results of this trial provide conclusive evidence of the superior efficacy of TMP/SMX in preventing or delaying recurrent PCP in participants who have had an initial episode of HIV-related PCP. The results may not apply to patients with less severe immunosuppression, intolerance to one or more of the study medicines, or no prior episode of PCP. Although the number of women enrolled in this study was too low to permit a separate analysis, we have no reason to believe that the efficacy and toxicity findings are not applicable to women.
There were four cases of toxoplasmosis in the TMP/SMX arm and six in the AP arm. Three of the patients assigned to TMP/SMX had been receiving AP for at least 3 months prior to onset of clinical toxoplasmosis. All of the patients assigned to AP were receiving AP at the time of development of toxoplasmosis. Serious bacterial infections were infrequent and further analysis is required before conclusions can be drawn.
Toxicity. In contrast to what many clinicians might have predicted, the treatment groups were similar with respect to hematologic and hepatic complications of therapy (Table 4). Forty-seven percent of those assigned to TMP/SMX and 44 percent of patients assigned to AP experienced at least one episode of grade 3 or 4 hematologic toxicity (hemoglobin < 8.0 Gm/dL or platelets < 50,000/cm or granulocytes <750/cm). The time to the first episode of severe hematologic toxicity was similar for both study arms (stratified logrank test p=.98), as was the time to the first interruption of ZDV (p=.72). Since these numbers are based on an intent-to-treat analysis, they reflect the toxicity experienced in the two groups and not the relative toxicities of TMP/SMX and AP.
Patients were stratified according to racial and ethnic background (white, black, Hispanic) within each treatment group: no significant differences were found when comparing time to hematologic adverse events or time to hepatic adverse events for each stratum in either the TMP/SMX or the AP treatment groups.
With regard to severe symptoms, five participants (three assigned to TMP/SMX and two assigned to AP) were diagnosed with pancreatitis. Patients assigned to TMP/SMX had a marginally greater frequency of severe rash than those assigned to AP (p=.08). The only statistically significant difference in toxicity were severe asthenia (more among those assigned to AP, p=.03) and severe abdominal pain (more among those assigned to TMP/SMX, p<.001). Other gastrointestinal symptoms were distributed similarly among both groups.
Twenty-seven percent of the patients assigned to TMP/SMX and 4 percent of those assigned to AP switched treatment in accord with the protocol-directed management of adverse events. An additional 6 percent of the patients assigned to AP were crossed over to TMP/SMX following recovery from recurrent PCP because they or their physicians were unwilling to return to the originally assigned medication.
ACTG protocol 021 compared two different strategies for the prophylaxis of recurrent PCP in patients who had recovered from an initial episode of PCP and were not known to be intolerant of ZDV, TMP/SMX, or AP. In this patient population, the routine administration of TMP/SMX as prophylaxis, in conjunction with ZDV, resulted in fewer recurrences of PCP than did the use of AP plus ZDV. No differences in time to recurrent PCP or major toxicities were noted when the experiences of patients with different racial and ethnic backgrounds were compared.
Although TMP-SMX was clearly the more effective first line prophylactic agent for patients who had experienced an episode of PCP, similar survival may be expected whether AP or TMP/SMX is chosen. Deaths due to PCP were few: Other causes of death appear to play a far more important role in determining survival during administration of an effective prophylactic regimen.
Finally, the combination of TMP/SMX plus ZDV was not associated with an earlier onset of hematologic or hepatic toxicity when compared to AP plus ZDV in patients who had recovered from an initial episode of PCP.
1 Kaplan-Meier estimate of the time at which 50% of the group would have experienced their first episode of toxicity. Grade 3 or 4 hematologic toxicity defined as hemoglobin < 8.0 gm/dl or platelets < 50,000/cm or granulocytes < 750/cm.
2 Kaplan-Meier estimate of the time at which 50% of the group would have experienced their first interruption of ZDV treatment.