Holmberg SD, Moorman AC, Von Bargen JC, Palella FJ, Loveless MO, Ward DJ, Navin TR, for the HIV Outpatient Study (HOPS) Investigators . Possible Effectiveness of Clarithromycin and Rifabutin for Cryptosporidiosis Chemoprophylaxis in HIV Disease. JAMA. 1998;279(5):384-386. doi:10.1001/jama.279.5.384
From the Division of HIV/AIDS Prevention, National Center for HIV, STD, and TB Prevention (Dr Holmberg and Mss Moorman and Von Bargen), and the Division of Parasitic Diseases, National Center for Infectious Diseases (Dr Navin), Centers for Disease Control and Prevention, Atlanta, Ga; Northwestern University Medical School, Chicago, Ill (Dr Palella); and Oregon Health Sciences University, Portland (Dr Loveless). Dr Ward is in private practice in Washington, DC.
Context.— Cryptosporidium parvum infection, a common
cause of diarrhea in persons infected with the human immunodeficiency virus
(HIV), is difficult to treat or prevent.
Objective.— To evaluate relative rates of cryptosporidiosis in HIV-infected patients
who were either receiving or not receiving chemoprophylaxis or treatment for Mycobacterium avium complex.
Design.— Analysis of prospectively collected data from HIV-infected patients'
visits to their physicians since 1992.
Setting.— Ten (8 private, 2 publicly funded) HIV clinics in 9 US cities.
Patients.— A total of 1019 HIV-infected patients with CD4+ cell counts
less than 0.075×109/L.
Main Outcome Measures.— Incidence of clinical cryptosporidiosis during treatment with clarithromycin,
rifabutin, and azithromycin.
Results.— Five of the 312 patients reportedly taking clarithromycin developed
cryptosporidiosis vs 30 of the 707 patients not taking clarithromycin (relative
hazard [RH], 0.25 [95% confidence interval (CI), 0.10-0.67]; P =.004).Two of the 214 patients taking rifabutin developed cryptosporidiosis
vs 33 of the 805 not taking rifabutin (RH, 0.15 [95% CI, 0.04-0.62]; P=.01). Prophylactic efficacy of either drug was 75% or
greater. No protective effect was seen in the 54 patients reportedly taking
azithromycin (RH, 1.48 [95% CI, 0.44-5.04]; P=.46).
Conclusions.— Clarithromycin and rifabutin were highly protective against development
of cryptosporidiosis in immune-suppressed HIV-infected persons in this analysis;
further study is warranted.
CRYPTOSPORIDIUM PARVUM is a serious cause of
opportunistic infection in persons with the human immunodeficiency virus (HIV),
yet chemoprophylaxis or treatment for it with about 100 antimicrobial agents
has remained discouraging or inconclusive.1,2
For example, therapeutic efficacy of spiramycin or paromomycin has been difficult
to demonstrate in small clinical trials.1,3,4
Newer semisynthetic macrolide antibiotics, such as clarithromycin and azithromycin,
have shown promise in small therapeutic trials, but cryptosporidiosis has
been a life-threatening illness for an estimated 10% to 15% of patients with
the acquired immunodeficiency syndrome in the United States.1,2
Based on a report indicating that none of 63 HIV-infected persons reportedly
taking 500 mg of clarithromycin twice a day developed cryptosporidiosis vs
4 (5.5%) of 73 patients without the drug (not statistically significant),5 we were prompted to assess cryptosporidiosis risk
in those taking or not taking Mycobacterium avium
complex (MAC) chemoprophylaxis or treatment in the HIV Outpatient Study (HOPS).
HOPS is an actively recruiting cohort that now includes over 2800 HIV-infected
out-of-hospital patients seen in about 30000 symptom-driven visits since 1992.
Study sites include 10 (8 private and 2 public) clinics caring for HIV-infected
patients. HOPS physicians are almost all board certified in both internal
medicine and infectious disease and care for HIV-infected patients only.
Information in 5 categories—demographics and risk behaviors for
HIV infection; symptoms; diagnosed diseases ("definitive" and "presumptive");
medications taken (dose and duration); and laboratory test results—is
abstracted at each visit, electronically entered at each site using a common
form, collected centrally, and reviewed and corrected before entry in the
Centers for Disease Control and Prevention database. This analysis includes
patients seen up to April 1996. Much effort has been applied to the correctness
and completeness of HOPS data, including standardized training and extensive
verifying of data.
Because we aimed to investigate those receiving clarithromycin and other
MAC prophylaxis (azithromycin or rifabutin) who were also at risk for clinical
cryptosporidiosis,6 we limited this analysis
to HOPS patients with CD4+ cell counts less than 0.075×109/L. We considered that a person had taken 1 of the 3 drugs if reportedly
taking it for at least 3 months continuously. The period of patient observation
was time between first clinic visit when CD4+ cell count was less
than 0.075×109/L until last clinic visit (before April 1996)
or development of cryptosporidiosis. Outcome measure was clinical cryptosporidiosis,
defined as diarrhea and other gastrointestinal symptoms, leading to a physician
visit, with a positive stool examination for C parvum
oocysts (acid-fast staining of unconcentrated fecal smears). We analyzed effects
of drugs of interest, demographic and socioeconomic variables, and other factors,
notably observation duration and most recent CD4+ cell count, on
Data were analyzed with Statistical Analysis Software (SAS Institute
Inc, Cary, NC), version 6.11. Incidence density of cryptosporidiosis, incidence
relative risk, confidence intervals, and P values
were calculated using maximum likelihood (likelihood ratio) tests. Potentially
confounding variables were analyzed by continuity-adjusted χ2,
Wilcoxon rank sum, and Fisher exact (2-tailed) tests.
A time-dependent accelerated failure time analysis,7
using SAS PROC LIFEREG, version 6.11, was done to assess relative contributions
of individual drugs and independent (demographic and immunologic) variables
to the likelihood of developing disease.
A change-in-estimate forward stepwise method was used to assess importance
and order of independent variables entered into the final model.8
A variable that changed the estimate of association between a MAC drug and
cryptosporidiosis by 5% or more was considered a potential confounder. The
variable that changed the association the most, and by at least 5%, was put
in the model, and remaining variables were added one at a time. When no new
variable changed the association estimate by more than 5%, the model was considered
We identified 535 HIV-infected patients who had taken clarithromycin,
azithromycin, or rifabutin for 5546 patient-months of observation and 484
patients not receiving these drugs and who were followed up for 4510 patient-months
(Table 1); all patients had CD4+ cell counts less than 0.075×109/L. Patients taking
these drugs had a mean of 13.6 (range, 2-66) visits and those not taking them
had a mean of 7.9 (range, 2-46) visits. Those taking any of the 3 drugs usually
did so for prophylaxis (77%) vs therapy for MAC (17%); 6% took any drug or
a combination for both prophylaxis and treatment of MAC.
The 535 patients reportedly taking MAC drugs and the 484 persons not
taking them were comparable in many ways (Table 1). Men who have sex with men were better educated, more likely
to have a private medical payment source, and more likely to receive drugs
(and earlier) than other groups. These variables, among others, were considered
potentially confounding and so included in the final statistical model.
We evaluated outcomes after and histories of cryptosporidiosis and diarrhea
before each patient observation. The 1019 patients had the following dispositions
over about 312 years of data collection: 458 (45%) were still actively being followed
by April 1, 1996; 353 (35%) died—without cryptosporidiosis diagnosis—while
under care; 76 (8%) were transferred; 74 (7.3%) were "inactive"; 38 (3.7%)
had no clinic visit for more than 6 months as of April 1, 1996; and 20 (2%)
could not be located. Prior to the censoring date of observation (first CD4+ cell count <0.075×109/L), 2 (0.4%) of 484 patients
with information of the 535 (later receiving MAC drugs) and 2 (0.5%) of 385
patients of the 484 (later not receiving them) were known to have had cryptosporidiosis;
61 (12.6%) of the 484 of those not taking the drugs and 55 (10.3%) of the
535 taking MAC drugs reported diarrhea. No difference was statistically significant.
There were no differences in most recent cell count between those taking
vs those not taking any drug (0.038 vs 0.039×109/L), nor
any drug individually: clarithromycin (0.040 vs 0.038×109/L);
rifabutin (both, 0.039×109/L); or azithromycin (0.038 vs
A statistically significant protective effect against cryptosporidiosis
in those taking clarithromycin or rifabutin was seen, and it remained significant
in a time-dependent accelerated failure regression model that controlled for
age, sex, race, transmission risk category, source of medical payment, CD4+ cell count, and use of the other 2 MAC drugs (Table 2). Prophylactic efficacy of either drug—calculated
as (1−relative hazard)—was 75% or greater. However, no observed
protective effect was seen for the few patients on azithromycin.
We could not confirm drug adherence in the 9 persons receiving MAC chemoprophylaxis
who developed cryptosporidiosis, since 7 had died; however, 1 patient who
died had a history of nonadherence with other drug regimens. Prescribed drug
dosage was generally higher (except 1 case) in those not developing cryptosporidiosis
than in those developing it, although the numbers are too small for meaningful
statistical analysis (data not shown).
This analysis indicates that those reportedly on MAC prophylaxis with
clarithromycin or rifabutin also had a several-fold decreased risk of developing
cryptosporidiosis, a statistically robust effect in regression analyses controlling
for observation duration and many demographic (age, sex, race, transmission
risk group, main source of medical payment) and immunologic (CD4+
cell count) variables. We did not find statistically significant differences
in several possible confounders, such as geographic site (municipal water
source9) and drugs considered unrelated to
development of cryptosporidiosis (acyclovir and trimethoprim-sulfamethoxazole),
that might be indexes of access to medical care (data not shown).
There are unavoidable limitations; these data were derived from an observed
cohort, not a controlled clinical trial, and unmeasured factors may confound
associations. The 3 drugs, recommended in HIV patients only for prevention
and treatment of MAC,10 may have been differentially
prescribed and used by HIV patients in this study. However, patients who did
and did not receive MAC drugs did not differ markedly in demographic or immunologic
profiles (Table 1), and our analyses
controlled for these variables and others. Because we lacked an absolute standard
or another commonly used test for Cryptosporidium
infection, we do not know if some patients taking macrolides had suppressed
and undetectable stool organisms, yet suffered from cryptosporidiosis. Still,
only a high rate of "false-negative" stool examination results—and that
occurring in persons suffering from disease despite undetectable stool C parvum oocytes—would negate the large and robust
prophylactic efficacy of clarithromycin and rifabutin seen.
Of factors possibly confounding the observed protective effect, host
immunologic function is putatively most important. Cryptosporidial diarrhea
is less likely in those with high or increasing CD4+ cell counts.11,12 Besides limiting our analyses to
those with a CD4+ cell count of less than 0.075×109/L, the time-dependent regression model controlled for this factor.
Also, means of most recent CD4+ cell counts of those with or without
MAC prophylaxis or treatment were both about 0.040×109/L.
Thus, no confounding variable—such as combination antiretroviral or
protease inhibitor use (not approved for use or available generally until
after periods of observation in this analysis)—was detected that would
enhance immunologic protection.
Data regarding biologic mechanisms and efficacy of these drugs are relatively
scant and inconclusive. Some semisynthetic macrolides may reduce cryptosporidial
oocyst load in stools of infected persons13,14:
clarithromycin and azithromycin activity against Cryptosporidium has been seen in vitro and in experimental models.2
Lower activity of clarithromycin vs azithromycin in animal models15,16 may result from animal inability
to produce the active 14-hydroxy-metabolite seen in humans receiving clarithromycin.
A few controlled trials of clarithromycin and azithromycin in treating cryptosporidiosis
in about 90 patients have shown decreased stool oocytes and some clinical
benefit.17 anecdotal reports of success with
spiramycin, another macrolide antibiotic, indicate possible benefit in placebo-controlled
trials in 54 patients with acquired immunodeficiency syndrome.3,18
We are unaware of animal models or human trials of rifabutin anticryptosporidial
activity, a semisynthetic ansamycin antibiotic with a mechanism of action
distinct from that of macrolides. Although there may be few or no published
studies of rifabutin in treating cryptosporidiosis, a possible protective
effect should be considered.
The lack of protective effect in those taking azithromycin is confusing,
since it is similar in structure and action to clarithromycin and spiramycin,
which seem to have some therapeutic efficacy in animals and humans.13 Therapeutic efficacy of a 500-mg daily dose of azithromycin,
however, was not seen in a recent study.19
The lack of prophylactic efficacy seen in our data may partly be attributable
to small numbers of patients receiving azithromycin (54 persons), of whom
only 2 developed cryptosporidiosis and who may not have been adherent. The
small numbers preclude a definitive conclusion about its potential utility.
Randomized clinical trials of this issue will be difficult, as study
participants with low CD4+ cell counts cannot ethically be randomized
for MAC chemoprophylaxis, and newer antiretroviral therapy will confound the
We found robust and statistically significant protective effects of
clarithromycin and rifabutin, used for MAC prophylaxis and treatment, in preventing
cryptosporidiosis. These data, primarily involving chemoprophylaxis, cannot
necessarily be extended to treatment of HIV-infected persons with cryptosporidiosis.
We conclude that similar analyses of other data sets and expanded trials of
new macrolide and other agents for preventing cryptosporidiosis in HIV-infected
persons are warranted.