The Role of Microsporidia in the Pathogenesis of HIV-Related Chronic Diarrhea

Methods

Study Sample

In April 1991, we began a casecontrol study of HIV-infected patients who were enrolled if they had at least one recent (within 1 month) negative stool culture for bacterial pathogens and one negative stool examination for ova and parasites. Cases included patients with chronic diarrhea that was either continuous (defined by the Centers for Disease Control and Prevention as two or more loose stools per day for 30 days [12]) or intermittent (defined as episodes of two or more loose stools per day alternating with episodes of formed stools for 30 days). Patients without diarrhea served as controls. The patients were recruited from the Houston Veterans Affairs Medical Center Special Medicine Clinic, an outpatient facility that provides comprehensive care to approximately 500 veterans with HIV infection. The patients in our study were recruited directly by the investigators or were referred by the clinic staff.

We sought to enroll 100 patients, including an equal number of unmatched cases and controls. The study was designed with 50 patients in each group (1- = 80%, = 0.05) because we expected the proportion of controls with microsporidiosis to be low (that is, less than 20%). With 50 participants in each group, we expected to be able to detect clinically meaningful differences (30% or greater) in the proportions of cases and controls with microsporidiosis, allowing us to find statistical significance with an odds ratio of 4.0 or greater. The study was approved by the institutional review boards for human subjects at the Houston Veterans Affairs Medical Center and Baylor College of Medicine.

Clinical Data and CD4 Cell Count

Each patient in the study underwent a history and physical examination to determine the route of HIV exposure (subsequently referred to as transmission category) and to collect demographic, clinical (previous AIDS-defining diagnoses, stool frequency, stool consistency), and therapeutic data (treatment with zidovudine or trimethoprim-sulfamethoxazole). For all participants, a CD4 cell count was obtained within 1 month of recruitment. The information was recorded on a data collection form designed specifically for the project.

Microbiologic Studies

At the time of endoscopy, a repeated fresh stool specimen was obtained for bacterial culture and examination for ova and parasites. One mucosal biopsy specimen from each site (duodenum, sigmoid colon, rectum) was cultured for Mycobacterium species.

Endoscopy and Biopsy

Esophagogastroduodenoscopy and flexible sigmoidoscopy were done using Fujinon videoendoscopes (Fujinon Inc., Wayne, New Jersey). Biopsy specimens, taken with Jumbo forceps, were obtained at three sites: the distal duodenum, the sigmoid colon (25 to 60 cm from the anal verge), and the rectum (12 cm from the anal verge).

Light Microscopy

Three biopsy specimens from each site [duodenum, sigmoid colon, rectum] were fixed in 10% buffered formalin and embedded in paraffin. Serial 4-µmsections were cut, deposited on slides, and stained with 1) hematoxylin and eosin for the evaluation of the mucosal morphology and the detection of Microsporidia, Cryptosporidium species, and viral inclusions; 2) Ziehl-Neelsen for the detection of acid-fast bacteria such as Mycobacterium species; and 3) Giemsa and tissue Gram stain for the detection of Microsporidia. The histopathologist was initially blinded; after the electron microscopy results were available, the biopsy specimens were re-examined in an attempt to identify the organisms.

Electron Microscopy

One biopsy specimen from each site (duodenum, sigmoid colon, rectum) was fixed in phosphate-buffered 2% glutaraldehyde, postfixed in 1% phosphate-buffered osmium tetroxide, embedded in plastic resin, fine-sectioned on LKB 8800 or Reichert-Jung SUPER NOVA Ultramicrotomes, stained with uranyl acetate and lead citrate, and examined under the Philips 301 and JEOL 1200 electron microscopes. The pathologist-electron microscopist was blinded to the patient's clinical and laboratory data when the biopsy specimens were examined and the diagnosis was rendered.

Data Analysis

Data were entered into a computer database (Paradox, Borland International, Scotts Valley, California). Statistical analyses were done using STATA software (Computing Resource Center, Santa Monica, California). Differences in the mean age and CD4 cell count and in the distribution of variables (transmission category, intensity of infection) were assessed using the t-test and chi-square statistic, respectively. An overall odds ratio for having microsporidiosis was calculated; 95% CIs were calculated using the Woolf method [13]. Odds ratios adjusted for CD4 cell count and the presence of other enteric pathogens were calculated using the Mantel-Haenszel procedure and logistic regression.

Results

Characteristics of the Study Sample

Between 1 April 1991 and 30 November 1992, we enrolled 106 HIV-infected men, 55 with chronic diarrhea and 51 without diarrhea. The clinical characteristics of the patients are summarized in Table 1. The mean age of all participants was 43 years, and the mean age of patients in the two groups was similar. The transmission categories were as follows: homosexual or bisexual men with or without a history of intravenous drug use (95 of 106 [90%]); and heterosexual men with or without a history of intravenous drug use (11 of 106 [10%]). The proportion of homosexual or bisexual men among the cases and controls was similar (Table 1). The mean CD4 cell count of all study participants was 151 cells/mm³ (range, 1 to 900 cells/mm³); the mean CD4 cell counts were significantly different (P = 0.02) in cases (113 cells/mm³) and controls (193 cells/mm³) (normal range, 517 to 1677 cells/mm³).

Presence of Other Enteric Pathogens

In the overall study sample, enteric pathogens other than Microsporidia were identified in 24 of 106 men (23%). The proportions of cases and controls with other enteric pathogens were 29% (16 of 55) and 16% (8 of 51), respectively (P = 0.10) (Table 1). These organisms were detected in stool specimens in 9 men (Giardia lamblia in 4, E. histolytica in 3, Shigella and Campylobacter species in 1 each); light microscopy showed organisms in an additional 10 men (cytomegalovirus in 5, acid-fast bacilli in 4, and both organisms in 1); electron microscopy showed cytomegalovirus in an additional man; and mycobacterial culture of biopsy tissue showed Mycobacterium avium complex in an additional four men.

Microsporidia Infection

Prevalence and Sites of Infection

Electron microscopic examination of duodenal biopsy specimens showed E. bieneusi infection in 31 of 106 men (29%). Organisms were detected in specimens from the sigmoid colon in 3 of these 31 men, and 1 of these 3 men also had organisms in a biopsy specimen from the rectum. The electron micrographs in Figures 1 and 2 show the characteristic features [14] of E. bieneusi observed in the duodenal biopsy specimens.

Light Microscopy Compared with Electron Microscopy for Detecting Microsporidia

At the initial light microscopic examination, Microsporidia were not detected. However, after the electron microscopy results were available, the biopsy specimens of the 31 men with microsporidiosis were re-examined. Sections stained with hematoxylin and eosin revealed lightly stained oval bodies in the villous tips [3] in 11 of the 31 men. In 2 men, rare aggregates of small refractile supranuclear bodies, presumably spores, were identified. No Microsporidia were detected in sections stained with Giemsa or Churukian tissue Gram stain [15]. No Microsporidia were detected in sections from the sigmoid colon and rectum. The proportions of cases and controls in whom Microsporidia were observed using light microscopic examination were the same (6 of 55 [11%] and 5 of 51 [11%], respectively).

Clinical Features

All 31 men with microsporidiosis were homosexual except 1 who was heterosexual and had a history of previous intravenous drug use. Ten men had AIDS [12], and all men except 1 were receiving antiretroviral therapy or antimicrobial therapy. The CD4 cell counts ranged from 10 to 660 cells/mm³; 3 men had normal counts. Of the 18 men with diarrhea, 11 had chronic continuous diarrhea and 7 had intermittent diarrhea. Of the 13 men without diarrhea, 7 had no gastrointestinal symptoms and 6 had symptoms (rectal discharge in 2 men, abdominal pain in 1 man, dysphagia or odynophagia in 2 men, and nausea in 1 man).

Microsporidia Infection and Chronic Diarrhea

No statistical difference was observed in the prevalence of microsporidiosis in participants with (18 of 55 [33%]) and without (13 of 51 [25%]) chronic diarrhea (odds ratio, 1.42; 95% CI, 0.61 to 3.31) in the full study sample (Table 2). The Mantel-Haenszel procedure was used to adjust the odds ratio for the presence of other enteric pathogens, and, again, no clear relation between microsporidiosis and diarrhea was observed (adjusted odds ratio, 1.53; 95% CI, 0.65 to 3.60).

Having observed no statistically significant difference in the crude odds ratio in the overall study sample, or after adjustments were made for other enteric pathogens, we next sought to determine whether an association between microsporidiosis and diarrhea existed in the subgroup of men with severe immunodeficiency. Mantel-Haenszel odds ratios, adjusted for CD4 count ( 100 or >100 cells/mm³), were calculated, and no clear relation between microsporidiosis and diarrhea was observed in either the overall study sample (adjusted odds ratio, 1.57; 95% CI, 0.66 to 3.75) or patients (n = 82) without other enteric pathogens (odds ratio, 1.95; 95% CI, 0.74 to 5.16) (data not shown in Table 2). Using logistic regression to control for CD4 cell count and other enteric pathogens, we found, again, no relation between microsporidiosis and diarrhea [adjusted odds ratio, 1.66; 95% CI, 0.68 to 4.06] (data not shown in Table 2).

We then examined the effect of intensity of microsporidia infection. We graded the intensity of infection on the basis of the electron microscopic findings, creating the following three categories: few (organisms detected only after an extensive search), moderate (detected without extensive search in fewer than one half the enterocytes), and abundant organisms (readily detected in more than one half the enterocytes). Among patients with microsporidiosis, no difference was observed in the intensity of infection among the cases and controls (P > 0.2) (data not shown in Table 2).

Discussion

Progress in understanding the role of Microsporidia infection in humans has been hampered by the lack of a simple, noninvasive diagnostic test. Light microscopic detection of Microsporidia in paraffin sections of mucosal biopsy tissue is difficult because the organisms are small and stain poorly with routinely used histologic stains, such as hematoxylin and eosin [6, 16]. Our results confirm these previous observations. We identified Microsporidia by light microscopy in only 11 of 31 men and only after the diagnosis had been established by electron microscopy. We agree with others that sections stained with hematoxylin and eosin are superior to Giemsa-stained or Gram-stained sections for detecting the organisms [3]. Although some investigators have advocated Giemsa-stained smears or touch preparations of biopsy specimens [17], we did not use these methods. For light microscopic diagnosis, the use of semi-thin (1- microns) plastic sections of mucosal biopsy tissue is preferable [3]; however, plastic embedding for histopathologic examination is available in only a few institutions. Detection of the parasite has largely depended on ultrastructural features [14], although recently a new modified trichrome technique for identifying Microsporidia spores in stool samples has been reported [18].

In our study sample, we identified E. bieneusi in asymptomatic men, contrary to published reports in which microsporidiosis has been detected only in persons with chronic diarrhea. Several explanations can be proposed for our findings. First, even though the organisms may be present in enterocytes, E. bieneusi may not lead to chronic diarrhea; infection may be coincidental. Indeed, the relation we observed between E. bieneusi infection and chronic diarrhea was not statistically significant, even after adjustments were made for the CD4 cell count and the presence of other enteric pathogens. However, given our unexpected finding of microsporidiosis in 25% of patients without diarrhea, our sample size did not allow us to exclude odds ratios within the CIs reported above.

Careful review of previous reports shows that of the four earlier studies that found an association between microsporidiosis and chronic diarrhea, one had no control group [3] and another had a control group consisting of normal heterosexual [19] rather than HIV-infected persons without diarrhea; thus, the research question we studied could not be addressed. Two studies did have a control group of patients with HIV-related chronic diarrhea. However, CD4 cell counts were not reported in one study [4], so it is difficult to determine whether the patients were similar to those we studied. In the remaining study [11], light microscopic examination of Giemsa-stained smears was used to detect Microsporidia, whereas we relied on electron microscopy, which is probably more sensitive.

A second explanation for our findings is that the Microsporidia infection in our patients was not sufficiently intense to cause diarrhea. A valid, reliable method for evaluating the intensity of E. bieneusi infection is not available, although a technique for quantitating fecal oocyst excretion in Cryptosporidium infection has been reported [20]. Lacking such a method for Microsporidia, we graded the intensity of infection based on electron microscopic examination of the duodenal biopsy specimens. Although our results do not suggest a relation between intensity of infection and symptoms, an important limitation of our approach is that the number of organisms detected in a single biopsy specimen may not reflect the overall intensity of infection in a patient with microsporidiosis.

We observed E. bieneusi in men with CD4 cell counts above the severely immunodeficient range of 100 cells/mm³ or fewer previously reported [11]. How commonly this occurs is unknown. Because of the lack (until recently) of a simple, noninvasive diagnostic test, microsporidiosis has been diagnosed largely as a result of intensive workup of subsets of severely immunodeficient HIV-infected patients referred for investigation of unexplained chronic diarrhea. The prevalence of microsporidiosis in patients representative of the full clinical and immunologic spectrum of HIV infection has not been studied adequately.

We found that E. bieneusi was present in the sigmoid colons of three men, one of whom also had the organism present in the rectum. We are not aware of any previous reports documenting human colonic infection with E. bieneusi, although the organism was recently detected in the duodenal tissue and bile of HIV-infected patients with cholangitis [21].

The study results highlight major gaps in our understanding of the epidemiology of E. bieneusi infection and the role of this protozoan in the pathogenesis of HIV-related chronic diarrhea. The relation we observed between E. bieneusi infection and HIV-related diarrhea was not statistically significant, suggesting that if microsporidiosis is associated with diarrhea, the association may be weak. Clearly, further study is needed to determine the prevalence of microsporidiosis in HIV-infected patients and to evaluate the relation between infection and disease.