Treatment of Shigellosis: V. Comparison of Azithromycin and Ciprofloxacin: A Double-Blind, Randomized, Controlled Trial

Methods

Patients

The study was conducted at the Dhaka Diarrhoeal Treatment Centre of the International Centre for Diarrhoeal Disease Research, Bangladesh. Study methods were similar to those of our previous studies of antimicrobial therapy for shigellosis [7-10]. Eligible patients were men 18 to 60 years of age who had grossly bloody-mucoid stool on examination, tenesmus, and illness that had lasted 72 hours or less. Patients were excluded if they had taken an effective antimicrobial agent for the current illness, had a coexisting illness requiring antimicrobial therapy, or had trophozoites of Entamoeba histolytica present on microscopic examination of stool. Signed informed consent was required before study entry.

Study Protocol

The study protocol was approved by the Ethical Review Committee. Patients were randomly assigned to receive either 1) one 500-mg dose of azithromycin on study day 1, followed by one daily 250-mg dose on the next 4 days [a total of 5 doses] or 2) 500 mg of ciprofloxacin every 12 hours for 5 days (a total of 10 doses). On admission, patients were given a consecutive study number to which treatment had been randomly preassigned by use of a random-number table; a block randomization method with a block size of six was used. The randomization list was developed and kept by a person not involved in the care or evaluation of the patients or in data analysis. Blinding was achieved by use of a “double-dummy” technique. Patients received an active form of the drug to which they were assigned and a placebo formulation of the other drug. Both active drug and placebo were given according to the routine schedule for the two drugs. Thus, patients assigned to the azithromycin group were given azithromycin tablets every 24 hours and received placebo tablets identical in appearance to the active ciprofloxacin tablets every 12 hours. Patients assigned to the ciprofloxacin group were given ciprofloxacin tablets every 12 hours and received placebo tablets identical in appearance to the active azithromycin tablets every 24 hours.

Study day 1 was considered to begin with the first administration of the study drug and to continue for the next 24 hours. Patients were asked to remain in the hospital for 6 full study days. On admission, a medical history was obtained and a physical examination was performed; each day thereafter, symptoms were reviewed and physical examination was done. Patients were asked to return for a follow-up visit 7 days after discharge; at this visit, an interval history was obtained and a physical examination was performed.

Laboratory Studies and Evaluation of Stool Samples

A stool sample was cultured for enteric bacterial pathogens on admission and on study day 3. Rectal swab samples for isolation of Shigella strains were obtained on admission, daily thereafter until discharge, and at the follow-up visit. On admission and on study days 3 and 5, microscopic examination of stool was done to enumerate leukocytes and erythrocytes and identify cysts and ova of parasites. Stools passed on each study day were enumerated; the stool was categorized as bloody-mucoid, watery, soft, or formed; and the macroscopic presence of blood or mucus was noted and recorded. Blood samples for determination of complete blood count, serum electrolyte levels, and total protein and creatinine levels were obtained on admission and thereafter if clinically indicated. All laboratory evaluations were done by using standard methods described elsewhere [7].

The minimum inhibitory concentration (MIC) of the study drugs for the Shigella isolates was determined by using the E-Test (AB Biodisk, Solna, Sweden). Serum and stool concentrations of the study drugs were determined by bioassay for the first 36 patients entered into the study before noon. Blood samples were taken 90 minutes, 3 hours, and 24 hours after the first dose of study drug; stool samples were obtained on study days 2 and 4.

Evaluation of Outcome

Treatment was considered to have failed clinically if, at the end of study day 3, a patient was having persistent dysentery or if, on study day 5, a patient was febrile (oral body temperature > 37.8 °C), had passed more than six stools, had any bloody-mucoid stools, or had more than one watery stool. Treatment was considered to have failed bacteriologically if Shigella strains could be isolated from a stool or rectal swab sample on study day 3 or thereafter. Patients in whom therapy failed and who continued to have dysentery were treated with pivamdinocillin, 400 mg every 6 hours for 5 days; treatment assignment was still unknown at the time pivamdinocillin was administered.

Determination of Sample Size

For determination of the sample size, the rate of clinical cure in the ciprofloxacin group was assumed to be 95% [8]. With a power of 80% and an α error of 0.05, each group had to have 33 patients in order to detect a difference of 30% in rates of clinical cure between ciprofloxacin and azithromycin treatment (that is, cure rate of 95% compared with cure rate of 65%). A larger sample size, which would have allowed greater power to detect smaller differences between the two groups, was precluded by limitations in funding.

Statistical Analysis

The significance of differences in proportions was tested by using the chi-square test; if the predicted size of any cell was five or less, the Fisher exact test was used. The significance of differences between continuous variables was assessed by using the Student t-test if the data were normally distributed or by using the Mann-Whitney U test if the data were not normally distributed. Differences in median values (along with 95% CIs for these differences) were calculated as the median of all possible differences. All tests of significance were two tailed. The initial data analysis was done without knowledge of treatment assignment. Data were entered into a computer by using StatPac Gold, version 3.0 (Walonick Associates, Minneapolis, Minnesota), and were analyzed by using the Statistical Package for the Social Sciences, version 6.0 for Windows (SPS, Inc., Chicago, Illinois), and EpiInfo version 6.0 (USD, Inc., Stone Mountain, Georgia). We used Confidence Interval Analysis, version 1.0 (British Medical Journal, London, United Kingdom), to calculate CIs.

Role of Industry Sponsor

The authors were solely responsible for designing the study; gathering, analyzing, and interpreting the data; writing the manuscript; reporting the results of the study; and determining whether, and where, the manuscript should be submitted for publication.

Results

Patients

Eighty-five patients were enrolled in the study between February 1995 and March 1996. No Shigella strains were isolated from an admission stool or rectal swab culture of 9 patients; these patients were not included in the analysis. An additional 6 patients (4 in the azithromycin group and 2 in the ciprofloxacin group) were excluded from the analysis because they withdrew before completion of the study (3 on study day 2, 2 on study day 3, and 1 on study day 4); thus, efficacy of therapy could not be determined.

Of the remaining 70 patients, 34 received azithromycin and 36 received ciprofloxacin. Shigella dysenteriae type 1 was isolated from half of the patients in each treatment group (Table 1). The treatment groups were well matched for clinical and laboratory features at admission; the only significant difference was in the proportion of patients who had more than 50 leukocytes per high-power field on microscopic examination of stool (Table 1). This difference was not statistically significant when the 6 patients who withdrew before completion of the study were included in the comparison.

Clinical Outcome

Treatment was clinically successful in 28 (82%) patients treated with azithromycin and in 32 (89%) patients treated with ciprofloxacin (P > 0.2) (Table 2). Treatment failed in 3 patients (2 in the azithromycin group and 1 in the ciprofloxacin group) at the end of study day 3 and in 7 patients (4 in the azithromycin group and 3 in the ciprofloxacin group) at the end of study day 5. Two of the 3 patients in whom therapy was determined to have failed on day 3 began receiving pivamdinocillin on study day 4.

The median number of stools was 29 in the azithromycin group and 24 in the ciprofloxacin group (P > 0.2) (Table 2). The two groups had similar numbers of stools on each of the 6 study days. With the exception of tenesmus lasting more than 72 hours (which occurred in 8 [24%] patients in the azithromycin group compared with 2 [6%] patients in the ciprofloxacin group), no significant differences were seen between the study groups in secondary outcome measures (duration of dysenteric or watery stools and duration of fever).

Bacteriologic Outcome

Rates of bacteriologic success of therapy were similar in the two groups (Table 2). Cultivatable Shigella strains were rapidly eradicated from the stool in both groups; at the end of study day 1, Shigella strains were isolated from rectal swab samples in only three (9%) patients in the azithromycin group and four (11%) patients in the ciprofloxacin group. In one patient in the ciprofloxacin group who was initially infected with S. dysenteriae type 1, S. boydii was isolated from stool on study day 6. At this time the patient was asymptomatic.

Clinical Outcome in Patients Infected with Shigella dysenteriae Type 1

Therapy clinically failed in 8 (23%) of the 35 patients infected with S. dysenteriae type 1 and in 2 (6%) of the 35 patients infected with other Shigella species (difference, 17% [95% CI, 1% to 33%]; P = 0.04). Both azithromycin and ciprofloxacin were less effective in patients infected with S. dysenteriae type 1 than in patients infected with other Shigella species. Azithromycin clinically failed in 5 (29%) of 17 patients infected with S. dysenteriae type 1 and in 1 (6%) of 17 patients infected with other Shigella species (P = 0.17); ciprofloxacin clinically failed in 3 (17%) of 18 patients infected with S. dysenteriae type 1 and in 1 (6%) of 18 patients infected with other Shigella species (P > 0.2).

The higher rate of treatment failure in patients infected with S. dysenteriae type 1 may in part reflect their initially more severe illness and the use of uniform outcome criteria on study day 5 regardless of Shigella species. When reduction in number of stools is used as an outcome measure, patients infected with S. dysenteriae type 1 had a greater reduction in the number of stools between study days 1 and 5 (median reduction, 82%) than did patients infected with other Shigella species (median reduction, 71%).

Susceptibility of Shigella Isolates and Concentration of Study Drug in Serum and Stool

All Shigella isolates were susceptible to both azithromycin and ciprofloxacin when tested by the disc-diffusion method. The median MICs of azithromycin were 0.500 µg/mL for S. dysenteriae type 1 and for isolates of other Shigella serotypes; all isolates were inhibited by 1.5 µg of azithromycin per mL or less. For ciprofloxacin, the median MICs were 0.125 µg/mL for S. dysenteriae type 1 isolates and 0.016 µg/mL for isolates of other species; all isolates were inhibited by 0.380 µg of ciprofloxacin per mL or less. All 35 S. dysenteriae type 1 isolates were resistant to ampicillin and trimethoprim-sulfamethoxazole when tested by disc diffusion; of the Shigella isolates of other species, 46% were resistant to ampicillin and 60% were resistant to trimethoprim-sulfamethoxazole. Ninety-seven percent of the S. dysenteriae type 1 isolates and 6% of the isolates of other species were resistant to nalidixic acid; 3% of the S. dysenteriae type 1 isolates and none of the other isolates of Shigella serotypes were resistant to pivamdinocillin.

Both drugs achieved high stool concentrations relative to the MIC of the infecting strain of Shigella (Table 3). Serum concentrations of ciprofloxacin were a median of 28 times higher than the MIC, but serum concentrations of azithromycin were equivalent to the MIC (Table 3).

Follow-up

Twenty-two (65%) patients treated with azithromycin and 19 (53%) patients treated with ciprofloxacin returned for a follow-up evaluation. No Shigella strains were isolated from fecal samples of any patient at follow-up, and no patient had dysenteric symptoms. No serious adverse effects of study treatment occurred in any patient. Symptoms possibly related to drug therapy were reported in 3 patients treated with azithromycin (1 patient with vertigo and insomnia on day 2, 1 patient with chest-wall pain on day 1, and 1 patient with constipation on day 4) and in 4 patients in the ciprofloxacin group (1 patient with epigastric pain, 1 patient with vertigo on day 4, 1 patient with calf pain on day 2, and 1 patient with conjunctivitis on day 5).

Discussion

In our study, azithromycin was effective in the treatment of adult men with moderate to severe shigellosis. The clinical rate of cure (82% [CI, 70% to 95%]) and the bacteriologic rate of cure (94% [CI, 80% to 99%]) obtained with azithromycin in this study are similar to rates of cure we found with effective drugs in previous clinical studies conducted in the same patient population [7-10]. Azithromycin was more effective than the drugs evaluated in these studies, which had little or no efficacy in shigellosis. Cefixime was associated with a clinical rate of cure of only 53% (P = 0.07 compared with the clinical success rate of azithromycin in this study) and a bacteriologic rate of cure of 40% (P < 0.01 compared with azithromycin) [10]. In patients infected with ampicillin-resistant Shigella strains, ampicillin achieved a clinical rate of cure of 40% (P < 0.01 compared with azithromycin) and a bacteriologic rate of cure of 86% (P > 0.2 compared with azithromycin) [8].

Small but consistent differences in clinical outcome were seen between ciprofloxacin and azithromycin recipients in this study; ciprofloxacin was more effective than azithromycin in all outcome measures examined. With the exception of a secondary outcome measure (duration of tenesmus > 72 hours), none of these differences was statistically significant. It is possible that trials with a larger sample size would demonstrate statistically significant and clinically important differences between the efficacy of the two drugs, especially among patients infected with S. dysenteriae type 1. On the basis of standard outcome criteria, azithromycin failed clinically in 29% of patients infected with S. dysenteriae type 1 whereas ciprofloxacin failed in 17% of patients with such infection.

The failure rate of both drugs in patients infected with S. dysenteriae type 1 was substantially higher than the rate seen with other Shigella serotypes; both azithromycin and ciprofloxacin therapy failed in only 6% of patients infected with these serotypes. This difference in cure rates between species may in part reflect the outcome definitions used and the more severe illness caused by S. dysenteriae type 1; both study drugs effected a more marked reduction in stool frequency in patients infected with S. dysenteriae type 1 than in patients infected with other Shigella serotypes.

The rate of follow-up 1 week after discharge from hospital was 65% in the azithromycin group and 53% in the ciprofloxacin group. Relapse with shigellosis, however, is known to be rare; thus, in-hospital evaluation after 5 days of antimicrobial therapy (which was the case for all patients in our study) is likely to be a good measure of therapeutic efficacy.

Azithromycin was effective despite serum concentrations that were, on average, only equal to the MIC of azithromycin for the infecting Shigella strain. Azithromycin serum concentrations in relation to the MIC were lower than the relative concentrations previously obtained with cefixime, which was ineffective in the treatment of shigellosis [10]. The efficacy of azithromycin in treating shigellosis despite low serum concentrations probably results from the fact that, unlike cefixime, azithromycin attains a high intracellular concentration and the fact that Shigella organisms invade colonic epithelial cells and multiply intracellularly [2]. Although we did not measure azithromycin concentrations within colonic epithelium, intracellular concentrations of azithromycin are known to exceed serum concentrations by 100-fold or more [24-26]. High drug concentrations are also attained within polymorphonuclear leukocytes [25]. Given the marked local inflammatory reaction that occurs with shigellosis [2], carriage of azithromycin within polymorphonuclear leukocytes may have further increased the concentration of drug at the site of infection and thereby enhanced the drug's in vivo efficacy [25, 28]. It is not likely that the high stool concentrations of azithromycin accounted for the efficacy of the drug; nonabsorbable antimicrobial agents with good in vitro activity against Shigella have repeatedly been shown to be ineffective in the treatment of shigellosis [6, 29, 30].

Cost of therapy remains an impediment to treatment of multidrug-resistant shigellosis with azithromycin or other new drugs. In Bangladesh (where drug prices are regulated by the government and drug patents are not honored), the course of azithromycin used in our study costs $4.30, the course of ciprofloxacin used costs $3.90, and a 5-day course of pivamdinocillin costs $7.70. Although these prices are relatively inexpensive by the standards of rich countries, they remain prohibitively expensive for many persons in Bangladesh and other poor countries, where the per-capita income is often $1 per day or less [31].

The cost of therapy could be reduced and the likelihood of compliance could be increased [32] if shorter courses of azithromycin are shown to be effective in the treatment of shigellosis. Short courses of azithromycin have been effective in the treatment of urogenital chlamydial infections and in trachoma [33, 34] and in the treatment of chancroid [35]. A 3-day treatment course has been found to be effective in eradicating bacteria in Campylobacter jejuni enteritis [36]. Because intracellular concentrations of azithromycin can remain high for days after a single dose of the drug [26], there is a rational pharmacologic basis for studying shorter courses of therapy. Studies that examine this issue are needed. In addition, single doses of ampicillin and ciprofloxacin, which have shorter half-lives than azithromycin, have been found to be effective in treating Shigella infections in adults [9, 37], although single-dose ciprofloxacin was less effective in S. dysenteriae type 1 infections than 5-day therapy [9].

Because of the increasing prevalence of multidrug-resistant strains of Shigella, it is likely that azithromycin will have an important role in the treatment of infection with these organisms. Shigella strains resistant to azithromycin have not yet been identified in Bangladesh. Azithromycin is associated with a low incidence of adverse effects [26, 27] and is safe for use in children, among whom most Shigella infections occur [28]. Because azithromycin belongs to a different class of antimicrobial agents than the drugs currently used to treat shigellosis, development of resistance to agents currently in use should not result in resistance to azithromycin.

Decisions on whether to use azithromycin as a first-line agent for the treatment of shigellosis will depend on local patterns of resistance, the spectrum of enteric infections for which antimicrobial agents are routinely prescribed, and the other antimicrobial agents available for use. In Bangladesh, empirical antimicrobial therapy for enteric infections is recommended for only three clinically distinct syndromes: cholera caused by infection with Vibrio cholerae O1 or O139; enteric fever caused by Salmonella typhi or Salmonella paratyphi; and dysentery, caused primarily by infection with Shigella strains. Campylobacter jejuni infection, for which use of azithromycin has been evaluated in other countries [36], is not a major cause of the dysentery syndrome in Bangladesh [38]. When compared with the two currently available options for the treatment of the dysentery syndrome-the fluoroquinolones and pivamdinocillin-azithromycin has two advantages: It is available in a liquid preparation (thereby facilitating treatment of children), and it is more readily available than pivamdinocillin. Should additional studies find that azithromycin is effective in the treatment of shigellosis in children and should shorter courses of therapy be proven effective (thereby reducing cost), it is likely that azithromycin will become one of the mainstays of dysentery treatment in Bangladesh.

If azithromycin is to be used for empirical treatment of shigellosis, efforts to control the development of resistance to this agent are required. Without such efforts, it is likely that the organisms will develop resistance to azithromycin, just as occurred with so many other drugs once useful for the treatment of shigellosis and occurred when azithromycin was used to treat C. jejuni infections [39]. Delaying the onset of resistance requires that the drug not be used for conditions for which antimicrobial therapy is not required or for conditions in which an older, less expensive, and narrower-spectrum agent might be used. Limiting the use of azithromycin requires the combined efforts of pharmaceutical companies that market the drug, practitioners who prescribe drugs, and governmental and nongovernmental organizations that play a major role in health care delivery in the poor countries where shigellosis is a major public health problem.

Dr. Seas: Unversidad Peruana Cayetano Heredia, Instituto de Medicina Tropical Alexander von Humboldt, A.P. 4314, Lima 100, Peru.

Dr. Bennish: New England Medical Center, 750 Washington Street, Box 041, Boston, MA 02111