In comparative 21-day trials, discontinuation of therapy because of adverse drug effects ranges from 50% to 57% for trimethoprim-sulfamethoxazole and from 14% to 55% for pentamidine [5-7]. During shorter treatment courses with drug-level monitoring and dose adjustments, more patients are able to complete therapy, but adverse reactions occur in most patients [8].

Atovaquone (formerly 566C80) is one of the oral treatment alternatives for P. carinii pneumonia. It is a hydroxynaphthoquinone whose mode of action against P. carinii is unknown. However, in plasmodia it acts by inhibiting de novo pyrimidine synthesis through reduction of dihydroorotate dehydrogenase activity [9]. Atovaquone has activity against P. carinii in experimental pneumonia and in human pneumonia [10-13]. In human studies, it has a favorable adverse effect profile [11-13]; the ability of mammalian cells to bypass de novo pyrimidine synthesis may explain the relative paucity of adverse effects.

Our multicenter randomized study compared oral atovaquone with intravenous pentamidine for treatment of P. carinii pneumonia in patients with AIDS. Results indicate that the efficacy of atovaquone and pentamidine for treatment of mild and moderate P. carinii pneumonia was similar, and there were fewer adverse events resulting in treatment discontinuation with atovaquone.

Methods

Top Methods Results Discussion Author & Article Info References

Experimental Design

This study was an open, randomized trial to compare the efficacy and safety of oral atovaquone (Mepron, Burroughs Wellcome Co., Research Triangle Park, North Carolina) with that of intravenous pentamidine (Pentam 300, Fujisawa Pharmaceutical Co., Deerfield, Illinois) for the treatment of mild and moderate P. carinii pneumonia in adults with AIDS who were intolerant of trimethoprim or sulfa antimicrobial agents. Mild P. carinii pneumonia was defined by an alveolar-to-arterial oxygen difference of less than 4.7 kPa (35 mm Hg), and moderate pneumonia was defined by an alveolar-to-arterial oxygen difference of 4.7 to 6.0 kPa (35 to 45 mm Hg). Patients with an alveolar-to-arterial oxygen difference of greater than 6.0 kPa (45 mm Hg) were excluded because oral therapy is often inappropriate for severe pneumonia. Patients with a history of intolerance to trimethoprim or sulfa (that is, they had stopped a previous therapeutic or prophylactic regimen because of an intolerance to these medicines) would receive study medication as primary therapy for this episode of P. carinii pneumonia. The sample size (156 patients with confirmed P. carinii pneumonia) was based on the intention to establish whether the difference in efficacy between treatment groups was fewer than 15 percentage points. Because of a lower-than-expected therapy success rate, statistical power for the efficacy component would not have been attained even with full enrollment. Therefore, enrollment was discontinued prematurely. Final enrollment was adequate for comparisons of safety with adequate statistical precision. The protocol was amended in July 1991 to remove the requirement for intolerance to trimethoprim-sulfamethoxazole after results of a trial comparing atovaquone and trimethoprim-sulfamethoxazole were known. At termination of the study, approximately 75% of the patients receiving primary therapy were intolerant of trimethoprim or sulfa.

Patients were to receive treatment with study medications for 21 days unless absence of a response or treatment-limiting toxicity was encountered. Patients were followed for 8 weeks after therapy was discontinued, and the status of their therapy was rated as one of the following: therapy success, therapy failure because of absence of response, therapy failure because of adverse events, or unevaluable.

This study was unmasked, that is, both patients and investigators were aware of which treatment the patient was receiving. Double-masking of this study would have required intravenous placebos for patients randomly assigned to receive atovaquone, and this option was rejected to protect patients from the risk of intravenous placebos. Single-masking, in which the investigator would not know which treatment the patient was receiving, was also rejected as an option. The logistics of single-masking were problematic, and it was thought probable that investigators would be able to distinguish which patients were receiving daily intravenous infusions. The disadvantage of an unmasked (or open) study is that investigators' previous perceptions may affect decisions to continue or discontinue study therapy for toxicities or absence of response. Although the protocol gave objective definitions of toxicities and criteria to determine response to therapy, the unmasked nature of the study may have resulted in some investigator classification bias.

Definitions of End Points

Survival was defined as being alive without ventilatory support 4 weeks after therapy was discontinued.

Successful therapy was defined as sustained clinical improvement 4 weeks after therapy was discontinued, with no additional or alternate antipneumocystis treatment during that time. Prophylaxis for P. carinii pneumonia was allowed after completion of acute therapy. Patients who received treatment for less than 21 days but did not require additional therapy for P. carinii pneumonia were considered therapeutic successes if they had sustained improvement for 4 weeks after treatment was discontinued.

Absence of response was defined by the following criteria: 1) mechanical ventilation required after the first 3 days of therapy; 2) any two of the following after 7 days: an increase in the alveolar-to-arterial oxygen difference of 2.7 kPa (20 mm Hg) or more over baseline, with an absolute value of more than 4.0 kPa [30 mm Hg]; worsening chest radiographs; or worsening clinical symptoms not due to another identifiable cause; or 3) no improvement after 10 days of any two of the second set of conditions. However, use of alternate antipneumocystis therapy within 4 weeks of discontinuation of therapy always defined a patient as a therapy failure.

Therapy failure because of treatment-limiting toxicity was defined as the discontinuation of therapy because of an adverse event and the requirement of additional treatment. Adverse events requiring discontinuation of therapy included the following: neutropenia (neutrophil count, 0.5 x 10⁹/L); thrombocytopenia (platelet count, 25.0 x 10⁹/L); hepatotoxicity (aspartate or alanine aminotransferase levels increased more than fivefold from baseline or >10 times the upper limit of normal; or bilirubin, >5 times the upper limit of normal); nephrotoxicity (creatinine level, >2.5 times the upper limit of normal; or calculated creatinine clearance, reduced >40% from baseline); anemia (hemoglobin concentration, <65.0 g/L despite transfusion); a prothrombin time of twice the upper limit of normal (or a prothrombin time >1.5 times the upper limit of normal and platelet counts of <40.0 x 10⁹/L); pancreatic toxicity (amylase level >5 times the upper limit of normal); vomiting for more than 2 consecutive days despite antiemetic therapy; the Stevens-Johnson syndrome, urticaria with systemic symptoms, or significant exfoliative dermatitis.

Unevaluable patients were those with documented P. carinii pneumonia who discontinued therapy for reasons other than treatment success, absence of response, or treatment-limiting toxicity (for example, noncompliance or lost to follow-up).

Patients

Eligible patients must have had HIV infection and clinical presentations consistent with P. carinii pneumonia. Histologic documentation of P. carinii pneumonia required within 7 days of study entry. Enrollment was limited to nonpregnant patients 13 years of age or older with an alveolar-to-arterial oxygen difference of 6.0 kPa ( 45 mm Hg) or less and a PaO₂ of 8.0 kPa ( 60 mm Hg) or greater while breathing room air; an absolute neutrophil count greater than 0.75 x 10⁹/L; a platelet count of 80 x 10⁹/L or greater; a hemoglobin concentration of 90 g/L or greater; a total bilirubin concentration of 43 µmol/L or less; an alanine aminotransferase level of 5 times the upper limit of normal or less; and a willingness and ability to give written informed consent. Exclusion criteria included the following: known intolerance to parenteral pentamidine or previous failure of pentamidine for treatment of P. carinii pneumonia, previous treatment for the current episode or another episode of P. carinii pneumonia within 4 weeks of entry, an investigator's opinion that the patient would require mechanical ventilator support in the near future, the presence of a concurrent pulmonary condition that would confound interpretation of drug effect, vomiting or malabsorption (including severe diarrhea), and known prolongation of the Q-T interval.

Clinical and Laboratory Assessments

Patients were monitored for therapeutic efficacy with chest radiographs, arterial blood gas measurements, and clinical symptom scores for dyspnea, chest pain or tightness, and cough (Table 1). Signs of drug toxicity were monitored through patient reports, clinical observation, electrocardiograms, and evaluation of laboratory results.

Patients were evaluated at baseline; on days 2, 4, 7, 10, 14, 17, and 21 or the end of therapy; and 4 and 8 weeks after therapy. Chest radiographs were obtained at entry, on days 7 and 21, and 4 weeks after therapy; arterial blood gas evaluations were done at entry and on days 7 and 21; electrocardiograms were obtained at entry and on days 10, 14, and 17. Blood samples for determination of concentrations of atovaquone were collected at enrollment and on days 4, 7, 14, and 21 of therapy.

Medication, Dosage, and Duration

Patients were randomly assigned within each study center either to the group receiving 750 mg of atovaquone (three 250-mg tablets) three times daily with food or to the group receiving 3 to 4 mg/kg of intravenous pentamidine isethionate per day as a single infusion lasting at least 60 minutes. Both groups received therapy for 21 consecutive days. Patients with moderately severe P. carinii pneumonia also received oral prednisone [4].

When treatment was completed, P. carinii pneumonia prophylaxis was recommended. The choice of drug for prophylaxis was left to the discretion of the patient's physician.

Statistical Analysis

Our study was designed to test the hypothesis that the therapeutic success rate of oral atovaquone is not worse than that of intravenous pentamidine in the primary treatment of mild to moderate P. carinii pneumonia in patients with AIDS. In addition, the study was designed and powered to detect differences in the incidence of discontinuation because of toxicity. Study enrollment was terminated early when observed success rates for both treatment groups were considerably smaller than the expected success rates. The resulting sample size was still adequate to detect differences in the incidence of discontinuation because of toxicity.

The primary analyses of efficacy end points and variables were done using the intent-to-treat rule for patients with confirmed P. carinii pneumonia who received study medication. Unevaluable patients were considered treatment failures in the intent-to-treat analyses. An analysis of evaluable patients that excluded unevaluable patients was also done.

The primary analyses of safety variables included all patients who received study medication. Secondary analyses including only patients with histologically confirmed P. carinii pneumonia were also done to characterize the safety profiles of both therapies in the intended patient population.

The Fisher exact test was used to compare treatment group proportions in addition to an estimated 95% confidence interval of the treatment difference (atovaquone minus pentamidine) on the basis of the large-sample normal approximation.

Treatment medians were used to summarize measured values and changes from baseline. Differences between the treatment medians were examined using the Wilcoxon rank-sum test and the associated 95% confidence intervals. The analysis of time-to-event data was done using Kaplan-Meier survival analysis and the log-rank test. These time-to-event methods were applied to three different lengths of observation. Events during early therapy were analyzed during the first 7 days after randomization, later treatment-related events were analyzed for patients completing more than 7 days of therapy, and events over the full course of therapy were analyzed at the completion of therapy. A P value 0.05 was considered statistically significant.

Relationship to Sponsor

Burroughs Wellcome Company sponsored this work as one of the two studies required by the Food and Drug Administration as part of the new drug-approval process. All work was done under contracts between the Burroughs Wellcome Co. and the individual study centers. Data collected in case report forms by the individual investigators were collated and analyzed by Burroughs Wellcome employees. Raw data, analyzed data, and additional requested data were provided to the principal author during manuscript preparation. Authors from Burroughs Wellcome wrote the Methods section of the manuscript and reviewed the manuscript with all other authors.

Results

Top Methods Results Discussion Author & Article Info References

Patients Enrolled and Entry Characteristics

One hundred forty-four patients were enrolled into this trial; 73 patients were randomly assigned to receive atovaquone and 71 to receive intravenous pentamidine. Demographic characteristics of the randomly assigned patients are shown in Table 2.

Analyses of efficacy were limited to the 109 patients with confirmed P. carinii pneumonia: 56 in the atovaquone group and 53 in the pentamidine group. In the pentamidine group, 17 patients were initially treated with 3 mg/kg of pentamidine per day and 36 patients with 4 mg/kg per day.

Initial clinical symptom scores, duration of symptoms before enrollment, measures of initial disease severity, P. carinii pneumonia prophylaxis, previous episodes of P. carinii pneumonia, and the use of anitretroviral therapy were tabulated (Table 3). Initial chest radiographs were abnormal in more than 90% of patients.

Therapy Outcome

Thirty-two patients (57%) who received atovaquone were successfully treated compared with 21 (40%) who received pentamidine, yielding a difference of 17% (95% CI for the difference, –3% to 38%; P = 0.085) (Table 4). Exclusion of unevaluable patients from each of the therapy groups does not substantially change the results for successful therapy (64% for patients treated with atovaquone compared with 43% for those treated with pentamidine; CI, 0% to 42%; P = 0.045). The incidence of treatment-limiting events accounted for most of the difference in successful therapy rates between the drugs. More patients treated with atovaquone (29%) than those treated with pentamidine (17%) were considered treatment failures because of the absence of response (CI for the difference, –6% to 29%; P = 0.18); however, more patients treated with pentamidine (36%) than with atovaquone (4%) were considered treatment failures because of treatment-limiting toxicity (CI for the difference, –48% to –17%;P < 0.001).

If patients who could not tolerate initial therapy are excluded from the analysis, 59% (32 of 54) of the patients receiving atovaquone and 62% (21 of 34) of those receiving pentamidine were successfully treated (CI for the difference, –26% to 21%; P > 0.2). In this group, absence of response was observed in 16 (30%) patients receiving atovaquone and in 9 (27%) patients receiving pentamidine. These results imply that the two drugs have similar response rates in patients who can tolerate them.

Survival was similar in both treatment groups (log-rank; P = 0.65). Death was attributed to P. carinii pneumonia in four patients treated with atovaquone and in three treated with pentamidine. The other deaths were attributed to causes other than P. carinii pneumonia (two patients treated with atovaquone and one with pentamidine); these deaths were specifically attributed to infections with other organisms. Four additional deaths (two patients receiving atovaquone and two receiving pentamidine) were likely caused by infections other than that with P. carinii pneumonia.

During the study, no obvious differences in the resolution of clinical symptoms between the treatment groups were found (data not shown). An increase in cough was reported by some patients receiving atovaquone. Oxygenation improved equally in patients in both treatment groups. Serum lactate dehydrogenase levels also returned to normal equally in patients in both treatment groups. No differences were found in therapy outcome or survival related to diarrhea the week before initiation of treatment.

All patients for whom therapy was discontinued prematurely were given the option to continue therapy with the other study drug. Of the seven patients who crossed over to pentamidine, three finished therapy successfully, three had therapy discontinued because of adverse events, and one was lost to follow-up. Seventeen of the 18 patients who crossed over to atovaquone successfully completed therapy, including 4 of the 5 patients who had failed to respond to therapy with pentamidine.

Mean steady state atovaquone levels were available for 28 of 32 successfully treated patients. The median value for these patients was 14.3 µg/mL (range, 1.66 to 35.58 µg/mL). In 13 patients receiving primary therapy who failed to respond to atovaquone and for whom mean steady state levels were available, the median value was 10.7 µg/mL (range, 1.29 to 19.41 µg/mL). In general, a higher mean steady state level of atovaquone was associated with a better therapeutic outcome and a higher likelihood that the alveolar-to-arterial oxygen difference and dyspnea score would return to normal, but large overlaps existed between the groups. Among successfully treated patients, eight (29%) had mean steady state levels less than the median value of the group that failed to respond to therapy; four (31%) in the group that failed to respond to therapy had mean steady state levels greater than the median value of those successfully treated. Only two patients failed atovaquone therapy because of adverse events; their mean steady state atovaquone levels were 9.85 and 17.76 µg/mL, respectively.

Survival was strongly correlated with mean steady state atovaquone levels (14.1 µg/mL in survivors compared with 6.0 µg/mL in nonsurvivors; P = 0.006). Patients with diarrhea had higher mean steady state levels compared with patients without diarrhea (19.69 µg/mL compared with 11.43 µg/mL; P = 0.046).

Neither the choice of prophylactic agent nor the timing of prophylaxis initiation after treatment termination had any discernible effect on therapy success rates.

Adverse Events

Differences were found in the adverse events associated with atovaquone and pentamidine. Among all patients Table 5, treatment-limiting adverse events occurred more frequently in patients in the pentamidine group. Forty-one percent of all patients treated with pentamidine prematurely discontinued therapy because of adverse events compared with 7% of patients treated with atovaquone, a difference of –34%(CI, –47% to –21%;P < 0.001). The same difference was seen in the subset of patients with histologically confirmed P. carinii pneumonia, in which 53% of patients treated with pentamidine and 9% of those treated with atovaquone discontinued therapy because of an adverse event (CI for the difference, –61% to –27%;P < 0.001). Five events were cited in more than 5% of the patients receiving pentamidine as reasons for discontinuation of therapy: hypoglycemia (11%); vomiting (8%); nausea (7%); elevated creatinine level (6%); and rash (6%). The only adverse event causing discontinuation of atovaquone therapy in more than one patient was rash (two patients [4%]).

In addition to the adverse events causing an interruption in therapy, four others occurred at different rates. Increased cough was reported more frequently in the atovaquone group (14%) than in the pentamidine group (1%) (P = 0.009). In the pentamidine group, the following toxicities occurred more frequently: altered taste (3% in the atovaquone group compared with 13% in the pentamidine group; P = 0.03), hypoglycemia (1% compared with 15%, respectively; P = 0.002), and hypotension (1% compared with 10%, respectively; P = 0.032).

No clear association was found between adverse events and mean steady state atovaquone levels. When patients with the indicated symptom were compared with patients without the indicated symptom, higher mean steady state levels were observed in patients with rash (15.5 µg/mL in patients with rash compared with 11.4 µg/mL in patients without rash; P = 0.18) and fever (17.4 µg/mL in patients with fever compared with 12.5 µg/mL in patients without fever; P = 0.39), and lower mean steady state levels were observed in patients with gastrointestinal disturbances (11.1 µg/mL in patients with disturbances compared with 14.4 µg/mL in those without disturbances; P = 0.26). No consistent correlations were found between steady state plasma levels of atovaquone and laboratory abnormalities that developed during therapy.

Discussion

Top Methods Results Discussion Author & Article Info References

In our study, pentamidine was more toxic than atovaquone when given to patients with AIDS for mild and moderate P. carinii pneumonia. This finding is not surprising, considering the known adverse effects accompanying pentamidine therapy and the relatively good tolerance of patients for atovaquone. In previous clinical studies for treatment of P. carinii pneumonia, atovaquone has been well tolerated with few serious adverse events [11, 13], even when continuously administered for as long as 42 days [12].

Atovaquone and pentamidine had similar overall successful therapy rates in our study. More patients receiving atovaquone failed to respond to therapy. There was a trend toward more successful therapy with atovaquone, which was more pronounced in the group of evaluable patients with histologically proven P. carinii pneumonia. The overall response rates and rates of successful therapy were lower in both the atovaquone and pentamidine groups than had been anticipated but were within the ranges reported for pentamidine [5, 8, 14]. The stringent definition of successful therapy that required sustained improvement for 4 weeks, no additional antipneumocystis therapy, and inclusion of all unevaluable patients as therapy failures probably accounts for the relatively low rates of successful therapy.

Our study was originally designed and powered to test the hypothesis that atovaquone was not worse than intravenous pentamidine as initial treatment for P. carinii pneumonia in this patient population and to detect differences in toxicity rates. It was not originally powered to show differences in rates of successful therapy or of absence of response. Thus, the power of the study to detect differences in successful therapy rates or absence of response rates is limited [15]. A significant difference in these rates may exist that this study would be unlikely to detect [16].

Factors that might effect the outcome of an episode of acute P. carinii pneumonia include the severity of the pneumonia, previous prophylactic therapy, and previous episodes of P. carinii pneumonia [14]. Patients in this study were stratified according to whether they had mild or moderate disease, and no statistically significant differences in disease severity were found between the treatment groups. The use of adjunctive corticosteroid therapy was nearly universal for patients with moderately severe disease; only one patient in each treatment group did not receive corticosteroids. The proportions of patients who had received prophylaxis for P. carinii pneumonia or who had had previous episodes of the disease were evenly distributed between the treatment groups.

In a randomized, placebo-controlled study comparing atovaquone with trimethoprim-sulfamethoxazole, preexisting diarrhea was associated with lower plasma drug levels and death in patients receiving atovaquone [13]. However, in our study, the same general degree of diarrhea was not associated with either death or low plasma levels of atovaquone; the steady state plasma levels of atovaquone were higher in patients with initial or preexisting diarrhea than in those without diarrhea. In both studies, patients with severe diarrhea were excluded. Survival was correlated with mean steady state plasma levels of atovaquone, indicating that poor drug absorption could limit the effectiveness of atovaquone. Absorption of atovaquone is enhanced by ingestion with foods. The absorption of the current formulation of atovaquone limits its usefulness to patients who can take oral medications and eat dependably. Therapeutic drug monitoring and parenteral or more bioavailable oral formulations could enhance atovaquone efficacy.

The Food and Drug Administration licenses four drug therapies for P. carinii pneumonia. Trimethoprim-sulfamethoxazole is considered the best choice because of its combination of efficacy, cost, and availability for oral or intravenous administration [3]. Intravenous pentamidine is also licensed without restriction for treatment of P. carinii pneumonia. Intravenous trimetrexate is licensed for moderate and severe pneumonia in patients intolerant of trimethoprim-sulfamethoxazole. Atovaquone is licensed for treatment of mild and moderate cases in patients intolerant of trimethoprim-sulfamethoxazole.

Other oral therapies for P. carinii pneumonia include the combinations of trimethoprim with dapsone and of clindamycin with primaquine. Both these combinations are efficacious, and trimethoprim with dapsone is reported to have fewer adverse effects than trimethoprim-sulfamethoxazole [7, 17-23]. Results of any large randomized series have not been published for either combination, although these combination therapies are being studied in a trial of the AIDS Clinical Trials Group.

Among the choices for treatment of mild and moderate P. carinii pneumonia, both atovaquone and intravenous pentamidine are available to patients who are intolerant of trimethoprim-sulfamethoxazole. Atovaquone may represent the better choice for intolerant patients who can be treated with oral therapy taken with meals. Results from our study indicate that atovaquone and pentamidine have similar rates of successful therapy in the treatment of mild and moderate P. carinii pneumonia and that atovaquone is better tolerated. Patients receiving atovaquone more frequently fail to respond to therapy, and patients receiving pentamidine have more treatment-limiting adverse drug toxicities; patients treated with either of these agents should be carefully monitored for these occurrences.

Appendix

The following are members of the Atovaquone Study Group: Peter T. Frame, MD, Protocol Chair (University of Cincinnati, Cincinnati, Ohio); Zab Mohsenifar, MD (Cedars-Sinai Medical Center, Los Angeles, California); R. Michael Buckley, MD (Pennsylvania Hospital, Philadelphia, Pennsylvania); Tony Cheung, MD (Mount Sinai Medical Center, New York, New York); Robert Hyland, MD, and Charles Chan, MD (The Wellesley Hospital, Toronto, Ontario, Canada); William Lang, MD (ViRx Inc., San Francisco, California); Donna Mildvan, MD (Beth Israel Medical Center, New York, New York); Stephen B. Greenberg, MD (Baylor College of Medicine, Houston, Texas); Donald Craven, MD (Boston City Hospital, Boston, Massachusetts); Martin Hirsch, MD (Massachusetts General Hospital, Boston, Massachusetts); Wafaa El-Sadr, MD (Harlem Hospital Center, New York, New York); Patrick Joseph, MD (Summit Medical Center, Oakland, California); David Hardy, MD (University of California Los Angeles, Los Angeles, California); Nathaniel Brown, MD, and Michael Rogers, PhD (Burroughs Wellcome Co., Research Triangle Park, North Carolina).