TRANSACTIONS OF THE ROYAL SOCIETY OF TROPICAL MEDICINE AND HYGIENE (2003) 97, 573-576

Escherichia coli enterovirulent phenotypes in Zambians with AIDS-related diarrhoea

Paul KeUy 1,2, Susan Hicks% Joy Oloya 4, James Mwansa 3, Linda Sikakwa 3, Isaac Zulu 2 and Alan Phillips 4 1Department of Adult and Paediatric Gastroenterology, St Bartholomew's and the Royal London School of Medicine and Dentistry, London E1 2AD, UK; Departments of 2Medicine and SPathology and Microbiology, University Teaching Hospital, P.O. Box 50110, Lusaka, Zambia; 4Cemre for Paediatric Gastroenterology, Royal Free Hospital, Pond Street, London NW/3 2QG, U K

Abstract Persistent diarrhoea is a major cause of morbidity and mortality in AIDS patients, and consequently an important public health problem in sub-Saharan Africa. Although intestinal protozoa and bacteria are detected in many of these patients, a substantial proportion of disease remains unexplained even after intensive investigation. HEp-2 cell adherent Escherichia coli have been described in AIDS patients with persistent diarrhoea, but their contribution to the overall burden of disease is not yet defined. We studied HEp-2 cell adherence ofE. coli isolates from 116 adult Zambian AIDS patients and 153 healthy controls obtained in 1993 or 1998-99. Enteroaggregative, enteropathogenic, and diffusely adherent phenotypes were observed in E. coli isolates from both AIDS patients and controls, but cytotoxic phenotypes were only isolated from the AIDS patients. There was no evidence of seasonality in the frequency of isolation, and there was no evidence of long-term carriage. Light and electron microscopy of distal duodenal biopsies did not reveal any bacteria closely associated with the brush border. Isolates were less susceptible to amoxycillin, tetracycline, and sulfonamides than to newer antibiotics. Enterovirulent E. coli appear to contribute to intestinal disease in AIDS patients in Zambia but asymptomatic carriage is common. Antibiotic trials should be carried out.

Keywords: diarrhoea, AIDS, Escherichia coli, Hep-2 cell adherence, aetiology, Africa

Introduction Persistent diarrhoea due to HIV/AIDS is a major

cause of morbidity and mortality in adults and children in sub-Saharan Africa. Although intracellular protozoa are found in many cases (Kelly et al., 1996), bacterial enteropathogens such as Salmonella spp. and Shigella spp. contribute to morbidity to a variable degree in different populations (Nelson et al., 1992; Clerinx et al., 1995; Mwachari er al., 1998). The role of bacter- ial enteropathogens in the aetiology of persistent diar- rhoea in AIDS is as yet uncertain, as the clinical response to antibiotic therapy is variable.

Adherent bacteria were first observed in patients with AIDS-related persistent diarrhoea in colonic biopsies (Kotler et al., 1995; Mathewson et al., 1995), but their phenotypic identity has not yet been fully characterized. Escherichia coli which cause intestinal disease can be classified into 5 categories (Nataro and Kaper, 1998): enterotoxigenic (ETEC), enteroinvasive (EIEC), enterohaemorrhagic (EHEC), enteropathogenic (EPEC), and enteroaggregative (EAEC) E. coll. A sixth phenotypic category, diffusely adherent E. coli (DAEC), is of uncertain pathogenicity. ETEC produce secretory enterotoxins, but do not cause histological damage to the mucosa, whereas EIEC are related to Shigella spp., invade the mucosa, and are capable of causing dysentery, although precise pathogenetic me- chanisms are unclear. EPEC induce an attaching-effa- cing lesion in the host epithelial cell in response to intimin binding to the translocated intimin receptor (Kenny et al., 1997), leading to cytoskeletal rearrange- ment and the production of an enteropathy. EHEC are considered to cause a similar attaching-effacing lesion to EPEC and in addition produce Shiga-like toxin which is implicated in causing haemorrhagic colitis and the haemolytic uraemic syndrome. EAEC adhere to epithelial cells causing diarrhoea by an unknown me- chanism which may involve increased mucus secretion (Hicks et al., 1996; Nataro & Kaper, 1998), a cytotoxic

Address for correspondence: Dr Paul Kelly, Department of Adult and Paediatric Gastroenterology, Bart's & The London School of Medicine, Turner Street, London E1 2AD, UK; phone +44 (0)20 7295 7191, fax +44 (0)20 7295 7192, e-mail m.p.kelly@qmul.ac.uk

effect (Hicks et al., 1996; Nataro et al., 1996), and/or a secretory enterotoxin (Navarro-Garcia, 1999).

Enterovirulent E. coli can also be classified by their HEp-2 cell adherence pattern. EPEC are recognized by the localized adherence (LA) of small compact colonies of bacteria. They also exhibit characteristic fluorescent actin staining (Knutton et al., 1991) beneath the attach- ing bacteria, as actin condensation occurs at these sites. Classic EPEC possess both a pathogenicity island (termed the locus of enterocyte effacement) and a plasmid (EAF plasmid). The former includes genes necessary and sufficient to cause the attaching-effacing lesion, and the latter contains regulatory (Per) genes and a type IV bundle-forming pilus gene, which is involved in localized colony formation (Nataro & Kaper, 1998). EAEC show aligned aggregates (stacked brick pattern) and/or chains of cell-adherent bacteria. DAEC which show the scattered adhesion of individual bacteria have also been described, but this, effectively bacteriostatic, pattern of adhesion has not been corre- lated with genotype, and the pathogenetic potential of DAEC has yet to be established. The significance of these adherence phenotypes in vivo is as yet uncertain. In a previous study from Zambia, these were character- ized as EAEC (36%), DAEC (36%), and EPEC (26%) (Mathewson et al., 1995). In another study, isolates from AIDS patients showed the phenotypic adhesion pattern of EAEC (n = 5), DAEC (n = 1), and EPEC (n = 1) but did not hybridize with standard gene probes for the category in question (Polotsky et al., 1997). However, in a placebo-controlled crossover trial of HIV-infected aduhs with EAEC, ciprofloxacin reduced diarrhoea by 52% (Wanke et al., 1998), suggesting that EAEC do have a pathogenetic role in AIDS-related diarrhoea.

We analysed enterovirulent E. coli by HEp-2 cell adherence phenotype in Zambian AIDS patients with persistent diarrhoea1 disease and in local healthy adult controls. We also assessed whether carriage of these bacteria is seasonal, whether specific categories are carried over periods of several months, and we tested antibiotic resistance patterns.

Patients and Methods Isolates of E. coli were obtained during the course of

2 studies in 1993 or 1998-99. HIV-seropositive pa-

574 P. KELLY ETAL.

tients with diarrhoea for over 1 month were enrolled in to 2 studies at the University Teaching Hospital (UTH), Lusaka, Zambia, as previously described (Kelly et al., 1996; Zulu et al., 2002). Control faecal specimens were obtained over the same periods of time from 2 populations, first from healthy civil servants attending U T H to have a medical screening examina- tion at around the same time as the first U T H study, and second from a longitudinal population cohort study of healthy adults in 1 of the residential townships in Lusaka (Kelly et al., 2002) at around the time of the second study. To determine whether seasonal variation significantly affects isolation rates, paired isolates of E. coli from the same individuals in the community cohort were collected both during the rainy season and during the dry season. Both of these studies were approved by the Research Ethics Committee of the University of Zambia.

Faecal samples were plated out onto MacConkey agar and sorbitol-MacConkey agar (SMAC) according to standard methods (Barrow & Feltham, 1993). Colo- nies of presumptive E. coli were identified, confirmed as E. coli using the API20E system, and stored as stabs in agar in 1.5 mL microfuge tubes. Any colonies typical of E H E C on SMAC were serotyped for identification of the O157:H7 phenotype.

Isolates were recultured on MacConkey agar and 5 colonies of E. coli selected and subcultured overnight at 37 °C in brain heart infusion (BHI) broth. Cultures were stored in a microbank system (ProLab diagnostics Ltd, Neston, Cheshire, UK) at - 7 0 °C prior to use in the HEp-2 cell assay. For this, HEp-2 cells were grown on 13 mm diameter cover slips in 24-well tissue culture dishes in Dulbecco's modified Eagle's medium (DMEM) to 70-80% confluence. Cell cultures were washed in D M E M and 25 pL of an overnight bacterial culture in BHI was added to each well, which con- tained 1 mL fresh DMEM, and incubated for 3 h in a CO2 incubator at 37 °C, but occasional cultures were left to 6 h when 3 h patterns were equivocal. Cells were washed thoroughly with phosphate-buffered saline (PBS) to remove any non-adherent bacteria and fixed with 70% methanol for 30 rain. After fixation the cells were rinsed with PBS, stained with 10% Giemsa's stain, washed, air-dried, and mounted in DePeX prior to light microscopy (using a × 100 oil immersion objective lens) to observe patterns of bacterial adhe- sion. Bacterial strains were considered to be cytotoxic when the HEp-2 cells showed rounding up and detach- ment from the cover slip. A subset of isolates was tested for sensitivity to a range of antibiotics, using standard disc diffusion techniques.

In order to determine whether adherent bacteria were visible in the proximal small intestine, distal duodenal biopsies from the first group of AIDS patients which had been processed for light and electron micro- scopy as previously described (Kelly et al., 1997) were re-evaluated.

Fisher's exact test was used to compare the propor- tion of patients in the AIDS and control groups in which specific E. coli phenotypes were detected. Sea- sonal variation in paired samples from the same in- dividuals was evaluated using McNemar 's test to determine whether individuals retain the AA phenotype from one season to another. Fisher's exact test was also used to compare the proportion of AA phenotypes in the wet and the dry season.

Results Isolates of E. coli were obtained from 116 AIDS

patients attending U T H with persistent diarrhoea (48 from the first study, 68 from the second). Isolates from 28 civil servants obtained at around the time of the first study and 125 isolates from the community obtained at the time of the second study were also analysed. The prevalence of categories of pathogenic E. coli in patients

Table 1. HEp-2 cell adherence phenotypes of Escher i ch ia coli in AIDS patients and controls, Lusaka, Zambia

AIDS patients Controls

(n = 116) ( n = 153) Adherence pattern n (%) n (%) P~

None 48 (41) 63 (41) - Aggregative adherence 42 b (36) 69 (45) 0.17 Localized adherence 3 (3) 2 (1) 0.65 Poor localized adherence 0 2 (1) 0.51 Diffuse adherence 10 (9) 6 (4) 0.12 Cytotoxic 9 u (8) 0 <0.001 Indeterminate 7 (6) 11 (7) 0.81

~Statistical comparisons made using Fisher's exact test to com- pare each Escherichia coli category against non-adherent bac- teria. bThe numbers in the AIDS group total 103% as 3 isolates were both aggregarive and cytotoxic and are included in both groups.

with HIV-related persistent diarrhoea was similar over- all to that in the civil servants and in the community cohort (Table 1). No O157:H7 E. coli were detected. Light and electron microscopy of distal duodenal biop- sies from the first AIDS group did not reveal any bacteria adherent to the epithelial brush border.

To assess whether individuals are colonized with consistent categories, 27 paired isolates of E. coli ob- tained from the community during the rainy season were compared with isolates obtained during the dry season. In the wet season, 7 (26%) of 27 isolates showed aggregative adherence, and 11 (41%) of 27 isolates obtained from the same individuals in the dry season showed aggregative adherence (P=0 .39) . There was no relationship to suggest that categories were retained by the same individuals in different seasons (Table 2; P = 0.25). Most of the isolates were sensitive to antibiotics (Table 3).

Discussion These findings suggest that adherent phenotypes of

E. coli are found frequently in AIDS patients in Zam- bia, although the background carriage rate in asympto- matic individuals was also high. Cytotoxic isolates were found only in the AIDS group; 3 of these also displayed the aggregative phenotype. Other phenotypes which would characterize EAEC, EPEC, or DAEC, were not significantly different in cases and controls. These results may indicate that clarification of the pathogenic role of these bacteria would require larger studies, or that further genotypic or phenotypic characterization would be required to identify subgroups associated with diarrhoea (see below). There was no evidence ofseaso-

Table 2. Paired isolates f r o m 27 healthy controls obtained in both the wet and dry seasons, Lusaka, Zambia

Dry season isolates

AA NA Total

Wet season AA 3 4 7 isolates NA 8 12 20

Total 11 16 27

NA, no adherence; AA, aggregative adherence. Altogether, 7/ 27 (26%) of wet season isolates were AA, and 11/27 (41%) of dry season isolates were AA (P = 0.39 by Fisher's exact test). McNemar's P - 0.25, making it unlikely that individuals carry isolates of the AA phenotype in both seasons.

ESCHERICHIA COLIIN AIDS PATIENTS IN ZAMBIA 575

Table 3. Ant ib io t ic sensi t ivi ty in re la t ion to HEp-2 cell phenotypes of E s c h e r i c h i a c o l i in isolates f rom heal thy controls , Lusaka, Z a m b i a

Drug

Aggregative adherence Diffuse Localized (n = 23) adherence adherence

n (%) (n = 3) (n = 3)

Ciprofioxacin 23 (100) 3 3 Gentamicin 23 (100) 3 3 Cefuroxime 23 (100) 3 3 Cefotaxime 23 (100) 3 3 Cefalexin 22 (96) 3 3 Azithromycin 22 (96) 3 3 Nalidixic acid 22 (96) 3 2 Amoxycillin-clavulanic acid 20 (87) 3 3 Chloramphenicol 18 (78) 2 2 Sulfamethoxazole 11 (48) 2 2 Tetracycline 10 (43) 2 2 Amoxycillin 9 (39) 1 1

nal variation in prevalence, nor that individuals in the community retain specific phenotypes over the months between repeated isolations. Enterovirulent E. eoli were susceptible to most antibiotics, but resistance to older cheaper antibiotics is emerging. Antibiotic resistance was not as common as in non-typhoid Salmonella and Shigella spp. isolates from the same community (Mwansa et al., 2002). No O157:H7 (EHEC) were confirmed and the LA phenotype (EPEC) was uncom- mon.

The dominance of the aggregative phenotype among our adherent isolates from AIDS patients differs from the results of a previous study in Zambia (Mathewson et al., 1995). In that study, which included patients with acute diarrhoea, the proportion of aggregative (36%), LA (26%), and DA (36%) phenotypes was more evenly balanced. EAEC have been implicated in persistent diarrhoea and growth failure in children in Central/South America, India, and Australasia (Nataro & Kaper, 1998) and Germany (Huppertz et al., 1997). Measures of inflammation are elevated in children with EAEC infection (Steiner et al., 1998), further support- ing a pathogenic role.

EAEC is currently the focus of considerable interest, but as yet the relationship between genotype and phe- notype is inadequately defined (Bernier et al., 2002). Some EAEC have been shown to express Stx2 which normally characterizes EHEC (Mossoro et al., 2002). In challenge studies with human volunteers, only 5/16 developed diarrhoea (Gomez et al., 1995), and it is possible that only a subset of enteroaggregative strains are responsible for diarrhoea. The EAEC heat stable enterotoxin-1 (EAST-l) gene was detected in only 53% ofEAEC (Yamamoto et al., 1997), and it has been found in EPEC, ETEC, and EHEC (Savarino et al., 1996). Other cytotoxic effects have been noted in vitro (Hicks et al., 1996). It has been suggested that the cytotoxicity of EAEC is related to a plasmid-borne genetic determinant which may confer the cytotoxic properties of some isolates (3 of 9 in our study) and which might explain the heterogeneity of clinical dis- ease expression (Nataro & Kaper, 1998). Newly de- scribed genes may play an as yet undetermined role in defining pathogenic subtypes. These may include aap, which encodes a protein termed dispersin which pro- motes bacterial dispersion through mucus and along cell monolayers, and aggR which encodes a regulatory protein which controls transcription of aap and other adherence fimbrial factor genes (Sheikh et al., 2002).

DAEC have not been associated with diarrhoea in many of the studies which have been conducted in children, but it has been suggested that there may be an

effect in children aged > 1 year (Nataro & Kaper, 1998). A recent study from New Caledonia found that DAEC were only associated with diarrhoea if they carried the afa/daa gene sequences, genetic markers found in uropathogenic E. coli and in a proportion of enteric isolates of DAEC (Germani et al., 1996). DAEC have only once to our knowledge been impli- cated in HIV-related enteric disease (Mathewson et al., 1995). As with EAEC, DAEC cannot yet reliably be identified using genetic markers. Recently, representa- tional difference analysis has been employed to find differences between a pathogenic and a non-pathogenic isolate. No genetic evidence was obtained of type III secretion system or intimin genes typical of EPEC, but the pathogenic strain does possess siderophore and daa genes and other sequences of unknown significance (Blanc-Potard et al., 2002).

We conclude that enterovirulent E. coli are associated with persistent diarrhoea in Zambian AIDS patients, but the background rate of carriage in healthy adults was high. Further work is needed to determine whether there are genetic subsets of EAEC and DAEC which are pathogenic, for example genotypes associated with cytotoxicity, or whether disease expression is a conse- quence of host factors. The antibiotic resistance profile was similar to that found in Thailand, in that the least useful antibiotics were amoxycillin, tetracycline, and sulfonamides (Suthienkul et al., 2001). It should now be possible to design further clinical trials using appro- priate antibiotics.

Acknowledgements We are grateful to Saghar Navabpour for help with the

HEp-2 assay. This work was supported by the Wellcome Trust, London, UK.

References Barrow, G. I. & Feltham, R. K. A. (1993). Cowan & Steel's

Manual for the Identification of Medical Bacteria, 3rd edition. Cambridge: Cambridge University Press.

Bernier, C., Gounon, P. & LeBouguenec, C. (2002). Identifi- cation of an aggregative adhesion fimbria (AAF) type III- encoding operon in enteroaggregative Escherichia coli as a sensitive probe for detecting the AAF-encoding operon family. Infection and Immunity, 70, 4302-4311.

Blanc-Potard, A.-B., Tinsley, C., Scaletsky, I., Le Bouguenec, C., Guignot, J., Servin, A. L., Nassif, X. & Bernet-Camard, M.-F. (2002). Representational difference analysis between Afa/Dr diffusely adhering Escherichia coli and non-patho- genic E. coli K12. Infection and lmmunity, 70, 5503-5511.

Clerinx, J., Bogaert, J., Taelman, H., Habyarimana, J. B., Nyirabareja, A., Ngendahayo, P. & Van de Perre, P. (1995). Chronic diarrhoea among adults in Kigali, Rwanda: associa- tion with bacterial enteropathogens, rectocolonic inflamma-

576 P. KELLY ETAL.

tion, and HIV infection. Clinical Infectious Diseases, 21, 1282-1284.

Germani, Y., Begaud, E., Duval, P. & Le Bouguenec, C. (1996). Prevalence of enteropathogenic, enteroaggregative and diffusely adherent Escherichia coli among isolates from children with diarrhoea in New Caledonia. Journal of Infectious Diseases, 174, 1124-1126.

Gomez, H. F., Mathewson, J. J., Johnson, P. C. & DuPont, H. L. (1995). Intestinal immune response of volunteers ingest- ing a strain of enteroadherent (HEp-2 cell adherent) Escher- ichia coli. Clinical and Diagnostic Laboratory Immunology, 2, 10-13.

Hicks, S., Candy, D. C. A. & Phillips, A. D. (1996). Adhesion of enteroaggregative Escherichia coli to paediatric intestinal mucosa in vitro. Infection and Immuni~y, 64, 4751-4760.

Huppertz, H. I., Rutkowski, S., Aleksic, S. & Karch, H. (1997). Acute and chronic diarrhoea and abdominal colic associated with enteroaggregative Escherichia coli in young children living in western Europe. Lancet, 349, 1660-1662.

Kelly, P., Baboo, K. S., Woolf, M., Ngwenya, B., Luo, N. & Farthing, M. J. G. (1996). Prevalence and aetiology of persistent diarrhoea in adults in urban Zambia. Acta Tropica, 61, 183-190.

Kelly, P., Davies, S. E., Mandanda, B., Veitch, A., McPhail, G., Zulu, I., Drobniewski, F., Fuchs, D., Summerbell, C., Luo, N. P., Pobee, J. O. M. & Farthing, M. J. G. (1997). Enteropathy in Zambians with HIV related diarrhoea: re- gression modelling of potential determinants of mucosal damage. Gut, 41, 811-816.

Kelly, P., Zulu, I., Amadi, B., Munkanta, M., Banda, J., Rodrigues, L. C., Mabey, D., Feldman, R. & Farthing, M. J. G. (2002). Morbidity and nutritional impairment in rela- tion to CD4 count in a Zambian population with high HIV prevalence. Acta Tropica, 83, 151-158.

Kermy, B., DeVinney, R., Stein, M., Reinscheid, D. J., Frey, E. A. & Finlay, B. B. (1997). Enteropathogenic E. coli (EPEC) transfers its receptor for intimate adherence into mammalian cells. Cell, 91, 511-520.

Knutton, S., Phillips, A. D., Smith, H. R., Gross, R. J., Shaw, R., Watson, P. & Price, E. (1991). Screening of entero- pathogenic Escherichia coli in infants with diarrhoea by the fluorescent-actin staining test. Infection and Immunity, 59, 365-371.

Kotler, D. P., Giang, T. T., Thiim, M., Nataro, J. P., Sordillo, E. M. & Orenstein, J. M. (1995). Chronic bacterial entero- pathy in patients with AIDS. Journal of Infectious Diseases, 171, 552-558.

Mathewson, J. J., Jiang, Z. D., Zumla, A., Chintu, C. C., Luo, N., Calamari, S. R., Genta, R. M., Steephen, A., Schwartz, P. & DuPont, H. L. (1995). HEp-2 cell adherent Escherichia coli in patients with HIV-associated diarrhoea. Journal of Infectious Diseases, 171, 1636-1639.

Mossoro, C., Glaziou, P., Yassibanda, S., I_an N. T., Bekondi, C., Minssart, P., Bernier, C., LeBouguennec, C. & Germs- hi, Y. (2002). Chronic diarrhoea, haemorrhagic colitis and haemolytic-uraemic syndrome associated with HEp-2 cell adherent Escherichia coli in adults infected with HIV in Bangui, Central African Republic. Journal of Clinical Micro- biology, 40, 3086-3088.

Mwachari, C., Batchelor, B. I. F., Paul, J., Waiyaki, P. G. & Gilks, C. F. (1998). Chronic diarrhoea among HIV-infected adult patients in Nairobi, Kenya. Journal of Infection 37,

48-53. Mwansa, J. C. L., Mutela, K., Zulu, I., Amadi, B. C. & Kelly,

P. (2002). Antimicrobial resistance in enterobacteria from African AIDS patients. Emerging Infectious Diseases, 8, 92-93.

Nataro, J. P. & Kaper, J. B. (1998). Diarrheagenic Escherichia coli. Clinical Microbiology Reviews, 11, 142-201.

Nataro, J. P., Hicks, S., Phillips, A. D., Vial, P. A. & Sears, C. L. (1996). T84 cells in culture as a model for enteroaggre- gative Escherichia coli pathogenesis. Infection and Immunity, 64, 4761-4768.

Navarro-Garcia, F., Sears, C., Eslava, C., Cravioto, A. & Nataro, J. P. (1999). Cytoskeletal effects induced by Pet, the serine protease enterotoxin of enteroaggregative E. coll. Infection and Immunity, 67, 2184-2192.

Nelson, M. R., Shanson, D. C., Hawkins, D. A. & Gazzard B. G. (1992). Salmonella, Campylobacter and Shigella in HIV- seropositive patients. AIDS, 6, 1495-1498.

Polotsky, Y., Nataro, J. P., Kotler, D. P., Barrett, T. J. & Orenstein, J. M. (1997). HEp-2 cell adherence patterns, serotyping and DNA analysis of Escherichia coli isolates from eight patients with AIDS and chronic diarrhoea. Journal of Clinical Microbiology, 35~ 1952-1958.

Savarino, S. J., McVeigh, A., Watson, J., Cravioto, A., Molina, J., Echeverria, P., Bhan, M. K., Levine, M. M. & Fasano, A. (1996). Enteroaggregative Escherichia coli heat-stable en- terotoxin is not restricted to enteroaggregative Escherichia coll. Journal of Infectious Diseases, 173, 1019 - 1022.

Sheikh, J., Czeczulin, J. R., Harrington, S., Hicks, S., Hender- son, I. R., LeBouguenec, C. L., Gounon, P., Phillips, A. & Nataro, J. P. (2002). A novel dispersin protein in entero- aggregative Escherichia coli. Journal of Clinical Investigation, 110, 1329-1337.

Steiner, T. S., Lima, A. A., Nataro, J. P. & Guerrant, R. L. (1998). Enteroaggregative Escherichia coli produce intestinal inflammation and growth impairment and cause inter- leukin-8 release from intestinal epithelial cells. Journal of Infectious Diseases, 177, 88-96.

Suthienkul, O., Aiumlaor, P., Siripanichgon, K., Eampokalap, B., Likhanonsa!ml, S., Utrarachkij, F. & Rakue, Y. (2001). Bacterial causes of AIDS-associated diarrhoea in Thailand. Southeast Asian Journal of Tropical Medicine and Public Health, 32, 158-170.

Wanke, C. A., Gerrior, J., Blais, V., Mayer, H. & Acheson, D. (1998). Successful treatment of diarrheal disease associated with enteroaggregative E. coil in adults infected with HIV. Journal of Infectious Diseases, 178, 1369-1372.

Yamamoto, T., Wakisaka, N., Sato, F. & Kato, A. (1997). Comparison of the nucleotide sequence of enteroaggregative Escherichia coli heat-stable enterotoxin 1 genes among diar- rhoea-associated Escherichia coll. FEMS Microbiology Letters, 147, 89-95.

Zulu, I., Veitch, A., Sianongo, S., McPhail, G., Feakins, R., Farthing, M. J. G. & Kelly, P. (2002). Albendazole chemo- therapy for AIDS-related diarrhoea in Zambia-clinical, parasitological and mucosal responses. Alimentary Pharma- cology and Therapeutics, 16, 595-601.

Received 26 November 2002; revised 7 March 2003; accepted for publication 10 March 2003