Such considerations have led to a 20-year search for an oral, well-tolerated therapeutic agent. The hypoxanthine analogue allopurinol, an inhibitor of uric acid formation in mammalian cells, was shown by LaFon and colleagues [3] to inhibit purine anabolism in Leishmania. In a randomized study done in Colombia, Martinez and Marr [4] found that a daily regimen of allopurinol (20 mg/kg per day) and antimony (20 mg/kg per day) for 15 days was much more effective than therapy with antimony alone (20 of 25 patients receiving allopurinol plus antimony [80%] recovered completely compared with 12 of 33 patients [36%] receiving antimony alone). However, the rate of cure for antimony alone in this study was surprisingly low [5]. Because of the controversy about the Colombian study and the lack of a well-controlled study on the efficacy of allopurinol in treating clinical leishmaniasis, we planned a randomized, blinded, controlled, large-scale clinical trial of the efficacy of allopurinol. To improve our ability to evaluate the efficacy of therapy, we chose the same endemic area (Colombia) and parasite (Leishmania panamensis) on which the major 1992 report focused.

Methods

Top Methods Results Discussion References

Study Design

In this randomized, controlled, partially double-blinded phase III study, we compared allopurinol with placebo and Glucantime (Rhone Poulenc, Paris, France). One hundred eighty-seven patients with cutaneous leishmaniasis were randomly assigned to one of three treatment groups. Patients in the first group received allopurinol, 300 mg (three 100-mg tablets) four times daily for 28 days, so that the dosage given was approximately 5 mg/kg four times daily or 20 mg/kg daily for 28 days. Patients in the second group received placebo, three tablets four times daily for 28 days. Patients in the third group received Glucantime, 20 mg of antimony/kg daily (no maximum daily dose) intramuscularly for 20 days. Patients in the allopurinol and placebo groups were assigned to treatment in a double-blinded manner.

Study Sample, Inclusion Criteria, and Exclusion Criteria

The study sample was composed of patients who were clinically suspected of having cutaneous leishmaniasis and who were from the following regions of Colombia: Arma, Dabeiba, Herveo, La Mesa, Marquetalia, Medellin, San Carlos, San Luis, Taraza, Valdivia, and Victoria. Patients were eligible for the study if they were 6 to 60 years of age, had cutaneous leishmaniasis as confirmed by the presence of parasites, had not received treatment for leishmaniasis with recognized agents during the previous 6 months, did not have lesions close to the eyes or on the mucosa, had body weight that was appropriate for height, and were amenable to prolonged follow-up. Exclusion criteria were the presence of concomitant diseases that required medical intervention, abnormalities in the complete blood count, abnormal glutamate oxaloacetate aminotransferase levels, abnormal creatinine levels, abnormal uric acid levels, and pregnancy. The ethical review committee of the Antioquia University School of Medicine and Hospital San Vicente de Paul, Medellin, Colombia, approved the study, and all patients gave written, informed consent.

Parasitologic Diagnosis

Leishmaniasis was diagnosed by visualizing amastigotes in lesion material or culturing promastigotes from lesion material. Two scrapings from one lesion on each patient (one scraping taken from the border of the lesion and one taken from the center) were examined with Giemsa stain for amastigotes; needle aspirates of the same lesion were added to culture medium to permit promastigotes to grow. Cultures were maintained at 26 °C for 4 weeks before they were reported as negative. In each isolate that was cultured, species were identified by the use of monoclonal antibodies [6].

Conduct of Clinical Trial

The first patient was entered into the study in April 1992; the last follow-up examination was completed in November 1995. Before the start of therapy, a complete history, a physical examination, and laboratory testing were done for each patient as described above. A photograph of each patient's lesions was also taken.

Patients were monitored every 10 days during the treatment phase of the study. Oral therapy was self-administered, and compliance was self-recorded. At each monitoring session, a 10-day supply of pills was dispensed and the patient's compliance record was checked and verified by counting the number of pills. Glucantime was administered by medical support personnel. In addition, the occurrence and severity of anticipated adverse effects were recorded at each monitoring session. Lesions were examined before the start of therapy; at the end of therapy; and 1.5, 3, 6, 9, and 12 months after the end of therapy. Attendance at monitoring sessions and follow-up appointments was aided by telephone calls and home visits by study staff.

At each evaluation, the induration (measured by using the ballpoint-pen technique) and the area of ulceration were both measured in two directions, which we designated R1 and R2. The areas of the indurated and ulcerated regions were calculated using the formula R1 x R2.

Definition of Responses

Response of lesions to therapy was determined clinically. Lesions treated with standard Glucantime regimens may increase in size or may not completely heal by the end of therapy; however, they generally heal by 1.5 months after therapy. We therefore used the response pattern of lesions treated with Glucantime as our standard in developing our definitions. For each lesion, we used the following definitions and dispositions:

Complete clinical response: Complete reepithelialization of the ulcer and disappearance of all induration. Lesions that showed a complete clinical response were followed for as long as 12 months to verify lack of relapse.

Clinical improvement: Fifty percent to 99% reepithelialization of the ulcer area and diminution of induration relative to the previous examination. Lesions that showed clinical improvement at the end of therapy or 1.5 months after the end of therapy were followed until either a complete clinical response or no clinical response was seen at subsequent follow-up sessions.

No clinical response: Less than 50% enlargement or diminution of the ulcer area and of induration. If no clinical response was seen at the end of therapy, the lesion was monitored further. If no clinical response was seen at a later follow-up, therapy was considered to have failed.

Failure to respond: 1) Greater than 50% enlargement of lesion size at the end of therapy or at subsequent follow-up or 2) no clinical response at an examination done 1.5 months or more after the end of therapy.

Relapse: The reappearance of the lesion at the original site after a complete clinical response or the appearance of lesions involving the mucosa.

A patient was considered cured if all of the patient's lesions had a complete clinical response by the third month of therapy and no relapse had occurred by the 12-month follow-up appointment. Therapy was considered to have failed if any of the patient's lesions did not respond to therapy or relapsed.

Evaluation of Responses

Patients, study investigators, and monitors were blinded to therapy with allopurinol compared with placebo. After the end of follow-up for the last patient, three independent, blinded evaluators determined efficacy and reached a consensus for each patient. The randomization code was then broken. Toxicity was determined by one evaluator before the code was broken.

Results

Top Methods Results Discussion References

Patient Characteristics

Five of the original 187 randomly assigned patients were excluded from the study: Two patients violated the study protocol, 1 had an uncertain parasitologic diagnosis, 1 had a clinical course that could not be interpreted, and 1 had co-infection with Sporothrix schenckii.

The characteristics of the 182 patients studied are shown in Table 1. The groups were well matched for age, sex, number of lesions per patient, location and characteristics of lesions, and duration of the presence of lesions before treatment.

Study Site Characteristics and Parasitology

Patients from Arma and La Mesa had disease caused by L. braziliensis. Patients from the other nine regions had disease caused by L. panamensis. Ninety-seven percent to 100% of lesions were smear positive and 61% to 75% were culture positive for Leishmania organisms. Because our study was designed to investigate the efficacy of treatment for infection with L. panamensis, we stopped entering patients into the study at Arma and La Mesa as soon as the strain present at those sites was recognized. For the 182 analyzable patients, 153 patients (84%) had documented infection with L. panamensis or were from regions in which L. panamensis was endemic; 29 patients (16%) had infection with L. braziliensis or were from regions in which L. braziliensis was endemic. Most patients with L. panamensis infection were from or were seen in Medellin (56 patients), Dabeiba (21 patients), San Carlos (17 patients), San Luis (15 patients), and Valdivia (14 patients).

Efficacy of Treatment

Among the 182 patients studied, treatment could be evaluated in 157 (86%). None of the 25 excluded patients returned for follow-up.

Therapy failed in approximately two thirds of evaluable patients in the allopurinol and placebo groups (Table 2). The failure rate for allopurinol in patients from regions in which L. panamensis was endemic (66%) did not differ significantly from the failure rate for allopurinol in patients from regions in which L. braziliensis was endemic (73%; P = 1.0; Fisher exact test); the failure rate for placebo did not differ between the two regions (66% compared with 40%; P = 0.30; Fisher exact test) We therefore pooled the data for the patients with L. panamensis infection and those with L. braziliensis infection for further analysis.

Failure of therapy was determined at the discrete time points at which lesions were measured. Almost all failures were recognized at the end of therapy (lesions enlarged >50% constituted failure) or 1.5 months after the end of therapy (lesions decreased <50% constituted failure). The time to failure did not statistically significantly differ between the placebo group and the allopurinol group (P > 0.2; chi-square test). Although therapy was considered to have failed if only one lesion had not healed, disparities between the clinical course for one lesion and that for other lesions on the same patient were unusual. Among the 35 patients in whom therapy failed and who had more than one lesion, therapy failed to heal more than 50% of the lesions in 28 (80%) (data not shown).

Therapy was considered to have failed in six patients because of relapse after initial healing. Lesions in three patients recurred at the original site of infection and were first seen 1.5 and 3 months after healing (Glucantime group) and 5 months after healing (allopurinol group). Lesions in three patients occurred at the nose as a manifestation of mucosal disease; these lesions were seen at the end of treatment and 1.5 months later (allopurinol group) and 12 months after healing (placebo group).

The study protocol required that patients whose lesions initially reepithelialized be followed for 9 to 12 months to verify recovery. Such follow-up was done in 81 of 88 patients (92%) who were initially cured. The other 7 patients were followed for 1.5 to 6 months. The rates of cure for evaluable patients were 33% in the allopurinol group, 37% in the placebo group, and 93% in the Glucantime group. The difference in the rates of cure for evaluable patients in the placebo and allopurinol groups was not significant (difference, 4% [95% CI, –14% to 22%]; P = 0.68). Even if it is assumed that the five patients in the allopurinol group who could not be evaluated would be cured and that therapy would fail in the 10 similar patients in the placebo group, the rates of complete response for allopurinol (38%) and for placebo (30%) would still not statistically significantly differ (P = 0.44). The difference in rates of cure for evaluable patients between the groups that received oral therapy and the Glucantime group was highly significant (P < 0.001).

Toxicity

Because the trial was double-blinded between the allopurinol and placebo groups, side effects caused by allopurinol could be quantified. Fifteen patients in the allopurinol group and 6 patients in the placebo group had moderate to severe side effects (P = 0.08). The only severe side effects attributable to allopurinol were headache and epigastric pain. Fifty-three of the 67 patients in the unblinded glucan-time group had moderate side effects, and 35 had severe adverse effects. As reported elsewhere [1, 2], myalgias, arthralgias, anorexia, nausea, and headache were common adverse effects.

Discussion

Top Methods Results Discussion References

When used to treat Colombian cutaneous leishmaniasis predominantly caused by L. panamensis, allopurinol had the same efficacy as placebo. Cumulative rates of cure were 33% for patients in the allopurinol group and 37% for patients in the placebo group. Almost all patients who did not to respond to therapy had treatment failure according to the criterion of greater than 50% enlargement of the lesion at the end of therapy or less than 50% diminution of the lesion at the first follow-up examination 1.5 months later. Two of the three patients in whom treatment was considered to have failed because of relapse in the nasal mucosa were in the allopurinol group. In comparison, standard therapy with pentavalent antimony in the form of Glucan-time cured 93% of patients.

Our trial had several strengths. First, a large number of patients were randomly assigned to treatment. Patient characteristics were similar among the three treatment groups, and the percentage of patients who could not be evaluated after 9 to 12 months of follow-up (14%) was modest for a field trial that had a long follow-up period. The trial was double-blinded between the allopurinol and placebo groups, and the rate of complete response with standard Glucantime therapy was the same as that in another Colombian study (91%) [7].

Our study shows that allopurinol was not effective as monotherapy for Colombian cutaneous leishmaniasis caused by L. panamensis. Although the efficacy of an antileishmanial drug might vary with the infecting species or the patient population, no published data suggest that allopurinol monotherapy is effective against any other form of cutaneous leishmaniasis. Whether the combination of allopurinol and an agent with established antileishmanial efficacy is more effective than the latter agent alone must be studied in carefully conducted clinical trials. For visceral leishmaniasis, 5 of 5 patients who did not respond to therapy with antimony were cured by a combination of antimony plus allopurinol [8], and 24 of 25 patients with chronic Iranian cutaneous leishmaniasis who did not respond to conventional therapy were successfully treated with a combination of allopurinol (20 mg/kg daily for 30 days) and antimony [9]. The presumption in these uncontrolled studies was that a repeated course of antimony alone would not have been markedly effective.

In leishmaniasis, as in other diseases, considerable hope will be placed in agents that have some preclinical or clinical rationale supporting their efficacy and that are easily administered, well tolerated, and inexpensive. Although the desire for such agents may be particularly strong in developing nations, these agents will also benefit patients and providers in the developed world. Our study shows that despite the reasonableness of this desire, well-designed clinical trials are needed to evaluate the efficacy of treatments.

From the Universidad de Antioquia, Medellin, Colombia; and the World Health Organization, Geneva, Switzerland.

Dr. Grogl: Avenida Guedes da Fontoura 102, Barra da Tijuca, RJ Brazil.

Dr. Modabber: Tropical Disease Research, World Health Organization, Geneva, Switzerland.

Dr. Berman: 9804 Culver Court, Kensington, MD 20895.