| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
 |
ABSTRACT |
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
We postulated that tropical enteropathy is a small intestinal mucosal response to environmental influences in adult life. To test this hypothesis, we performed a longitudinal cohort study in healthy adults in a tropical population and related small intestinal structure and function to the frequency and type of asymptomatic intestinal infection. In this population, the seroprevalence of human immunodeficiency virus (HIV) is stable at approximately 20–30%,13 so the influence of HIV was also analyzed.
|
METHODS |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
During the setting up of the study, monosaccharide/ disaccharide absorption and permeability testing was conducted by members of the study team on themselves. These subjects, drawn from the same ethnic group but of higher socioeconomic status, were subject to the same exclusion criteria listed earlier and were used as a control group solely for comparison of sugar test results.
On the day of investigation for cohort participants, enteroscopy was performed with an Olympus SIF-10 enteroscope (KeyMed, Ltd., Southend on Sea, Essex, United Kingdom) under sedation. No significant adverse events were encountered. Three biopsies were collected under direct vision from proximal jejunum and placed in normal saline. One hundred milliliters of a solution of four test sugars was then instilled immediately into the jejunum down the biopsy channel and flushed through with another 100 mL of water. At the end of the procedure, biopsies were mounted on to cellulose acetate filter paper under a dissecting microscope and fixed in 10% formalin in saline ready for processing.
Morphometry.
Biopsies were all processed by a single operator (CN) and orientated into wax prior to cutting 3-µm sections for staining with hematoxylin and eosin. Sections were cut at three levels at 25-µm intervals. Slides were evaluated by a senior pathologist (RF) before morphometry was performed by a single observer (PK). Morphometry was performed using a Zeiss (Thornwood, NY) KS400 image analysis system and a macro (written by SG) available from http://www.smd.qmul.ac.uk/morbidanatomy/code/villi4.html. This macro was designed to allow morphometry of mucosa with varying degrees of enteropathy (Figure 1
). From each individual, one level of sectioning with the best orientation was chosen, and all mucosa were measured where the muscularis mucosae and epithelium were intact and where crypts could be seen to have been sectioned longitudinally. Where two villi were partly or fully fused, they were treated as one. In each field, a length of muscularis mucosae was identified underlying all mucosa where these criteria were satisfied and measured. The first two measurements were expressed in relation to a 100-µm length of muscularis mucosae15 at a magnification of 63x. The following variables were then measured: 1) cross-sectional area of the villous compartment, which is representative of villous compartment volume (VCV); 2) total length of epithelium, which is used as a measure of epithelial surface area (ESA); 3) villous height (VH), which is measured from villus tip to cryptvillus junction as judged by eye (care was taken to include every recognizable villous unit of well-orientated mucosa); 4) crypt depth (CD), which was measured in the same way, one for one with each villus; and 5) villous width (VW), which was measured automatically as the maximum width of the villus along a line perpendicular to VH in each unit of the villous compartment.
|
). To assess the variation between biopsies taken from the same individual, morphometric measurements were made on each biopsy individually in biopsies taken in triplicate from the same area of the jejunum. The variance ratio between biopsies from the same individual was 0.4 compared with 3.1 between individuals, indicating that only a small proportion (13%) of the variation seen between individuals could be accounted for by sampling variation between biopsies. Expressed another way, only 4 of 23 biopsies showed significant differences in VH between biopsies from the same individual (Figure 2). In biopsies taken in the second and third years, only VH and CD were measured.
|
|
Environment: measures of exposure. To assess the influence of the environment on mucosal structure and function, several indicator variables were used: seasonality, a household hygiene score, and direct measures of exposure to both pathogenic and non-pathogenic intestinal bacteria, protozoa, and helminths.
The household hygiene score used was constructed for the purpose of this study to allow comparison of the living conditions of each participant, and scores were given by the two nurses who visited the households on a daily basis. A scale of 1–10 was used, with up to two points being given for each of the following: overall cleanliness of the house, water storage facilities, food storage facilities, handwashing facilities and their use, and sanitation facilities.
Diarrhea incidence was measured by recall using a simple questionnaire administered every two weeks, and by active case-finding while offering investigation and treatment of each episode. Every month, participants submitted a stool sample that was analyzed for the presence of protozoa, bacteria, and helminths. Samples were plated onto MacConkey, xylose-lactose-sucrose, sorbitol-MacConkey, Yersinia selective, and thiosulfate-citrate-bile salt agar plates for culture of non-typhoid Salmonella spp., Shigella spp., Yersinia enterocolitica, Citrobacter rodentium, Aeromonas hydrophila, Plesiomonas shigelloides, and Vibrio cholerae. No isolates of Yersinia or Plesiomonas were obtained and they are not considered further. Colonies of non-lactose fermenting bacteria were subjected to biochemical characterization using triple-sugar-iron, lysine-iron, sulphide-indole-motility, urea, and Simmons citrate media, and interpreted using Cowan & Steel’s manual.17 Isolates picked at random were confirmed using the API 20E system (BioMerieux, Inc., Hazelwood, MO). Micro-aerophilic conditions for isolation of Campylobacter jejuni were difficult to maintain and results for this organism are omitted. Protozoa and helminths (Cryptosporidium parvum, Isospora belli, microsporidia, Giardia intestinalis, Entamoeba histolytica/dispar, Blastocystis hominis, Endolimax nana, Chilomastix mesnili, Iodamoeba butschlii, Entamoeba hartmanii, Hymenolepis nana, Strongyloides stercoralis, Ascaris lumbricoides, hookworm, and Taenia saginata) were identified using wet preparation smears and smears stained with modified Ziehl-Neelsen and modified trichrome stains (for microsporidia).
Data analysis. Analysis was carried out using STATA version 7.0 (Stata Corp., College Station, TX). Morphometric variables and three of the sugar recoveries were normally distributed, but lactulose recovery was markedly skewed. Therefore, for most variables, parametric tests (t-test or analysis of variance) and linear correlations were used, but for comparisons involving lactulose recovery, a non-parametric test (the Kruskal-Wallis test) and Spearman’s rank correlation coefficient were used. Correlations are only presented when statistically significant (P < 0.05).
Longitudinal changes were analyzed as continuous variables and by identification of individuals in whom changes exceeded pre-specified confidence limits. These were 25 µm or 9 µm for VH and CD, respectively, being the 95% confidence limits of reproducibility (Table 1
), and 0.018 and 0.040 for L:R ratio and R:G ratio, respectively, being the 95% confidence limits around measurements in healthy residents of the United Kingdom.16 To detect seasonal variation, measurements were made at different times of the year and analyzed in two-month periods that were related to rainfall and temperature (National Meteorological Office, Lusaka, Zambia). Matching between seasonal patterns was assessed by using the putative independent variable (e.g., incidence of diarrhea) to rank the two-month periods in ascending order and using a trend test to assess the effect of the ranked seasons on the dependent variable (e.g., VH).
Logistic regression models were constructed to evaluate the independent contributions of demographic, nutritional, and environmental variables. Morphometric or functional variables (VH, CD, xylose recovery, L:R ratio, and R:G ratio) were dichotomized around the median for use as outcome variables.
|
RESULTS |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
200 cells/µL (n = 50), and HIV seropositive with CD4 cell counts < 200 cells/µL (n = 25).
Morphometry: cross-sectional assessment at entry.
Stereo-microscopy was performed after 164 enteroscopies to orientate the biopsies. None of these jejunal biopsies showed finger-like villi as the predominant form, but the severity of enteropathy varied considerably (Figure 1). Morphometric measures did not differ between men and women, nor were they related to age. The CD and VW were significantly greater in HIV-infected individuals (Table 2). In HIV-seropositive adults, VH was positively correlated with CD4 cell count (r = 0.33, P = 0.03).
|
Within the study population, recoveries of rhamnose, D-xylose, and 3-O-methyl-D-glucose, and the R:G ratio were normally distributed and there was no difference between men and women. Lactulose recovery and the L:R ratio were higher in women than men (P = 0.0001). The L:R ratio, but not the R:G ratio, was higher in HIV-seropositive adults, more so in those with low CD4 cell counts (Table 2). The higher permeability in young women was still statistically significant after stratification by HIV status. In HIV-seropositive individuals only, xylose recovery was correlated with CD4 cell counts (r = 0.53, P < 0.001).
Structure: function correlation at baseline.
There were few significant correlations between structure and function, and these correlations were modest: xylose recovery and VH (r = 0.22, P < 0.01), rhamnose recovery and ESA (r = 0.19, P < 0.05), L:R ratio and VH (
= –0.26, P = 0.01), and L:R ratio and ESA ( = –0.30, P = 0.004). Other possible pairwise correlations were not significant.
Longitudinal changes in architectural and functional parameters.
Structure and function varied substantially within individuals over time (Figure 3). Over the two-year period between initial and final tests, 14 (24%) of 59 individuals with satisfactory biopsies at both time points showed an increase in VH over 25 µm, and 28 (47%) showed a decrease of greater than 25 µm. The CD varied in 69% of the participants over this time. Over two years, permeability (L:R ratio) increased in 21 (38%) of 55 adults in whom measurements were available throughout and decreased in 16 (29%). The R:G ratio showed less change, with 58% remaining within the limits described earlier. The HIV-seropositive participants were more likely to demonstrate a decrease in VH over one (risk ratio [RR] = 2.5, 95% confidence interval [CI] = 1.03–6.2, P = 0.03) or two (RR = 3.0, 95% CI = 1.3–6.9, P = 0.006) years. Permeability changes did not differ significantly in HIV-seropositive and HIV-seronegative participants. Changes over one year in VH and CD were positively correlated (r = 0.26, P = 0.008) and changes in VH and permeability were negatively correlated ( = –0.32; P = 0.01).
|
). Diarrhea incidence was 0.52 episodes per year in HIV-negative participants and 1.33 episodes per year in HIV-seropositive participants. Diarrhea incidence showed a seasonal pattern with a peak in December-January (peak rainfall months) and a nadir in June-July (cold, dry season), which did not match the seasonality of VH or permeability. Only in HIV-seropositive adults was there an association between diarrhea incidence and mucosal parameters: the L:R ratio and R:G ratio were correlated with the rate of diarrhea two months prior to the date of investigation ( = 0.54 and r = –0.48, respectively, P < 0.001 for both).
|
). Isolation rates of several organisms were also seasonal, i.e., C. parvum, microsporidia, S. stercoralis, E. nana, H. nana, I. butschlii, E. hartmanii, Shigella spp., A. hydrophila, and C. rodentium. The seasonal variation in VH matched that seen in C. rodentium (P < 0.0001).
|
).
We analyzed the association between asymptomatic infection detected in stool samples (Table 4) and enteropathy as assessed on the day of investigation one month later. Organisms isolated during diarrheal episodes were not included because only participants who had been asymptomatic for at least one month (usually much longer) underwent small intestinal investigations. The VH was reduced in participants from whom C. rodentium had been isolated the month prior to biopsy (mean ± SD 3 239 ± 5 µm) compared with those free of this infection (257 ± 3.5 µm; P = 0.03), and permeability was higher in those who had had C. rodentium (mean = 0.101, IQR = 0.06–0.13) compared with those who had not (0.06, 0.05–0.09; P = 0.008). Crypt depth was increased in people with hookworm infection (mean ± SD = 193 ± 13 µm) compared with those without this infection (155 ± 2 µm; P < 0.001). The negative effect of C. rodentium infection on VH, the positive effect of C. rodentium on the L:R ratio, and of hookworm on CD were confirmed in logistic regression models.
|
DISCUSSION |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
By European and North American norms,16 mucosal absorptive capacity and permeability were abnormal, and these abnormalities were more severe in adults in the cohort than in controls of higher socioeconomic status, probably reflecting their greater exposure to environmental noxa. This is consistent with the effect of socioeconomic conditions reported at a population level by Menzies and others.16 In that study, a tropical population of high economic status (Qatar) had minimal or no increase in permeability, and this is supported by more recent data from Singapore (Lim S and Menzies I, unpublished data). The enteropathy therefore seems to be a consequence of contaminated environmental conditions rather than climate per se. The relationship between the severity of enteropathy and asymptomatic infection during the month prior to investigation indicates that exposure to organisms of modest pathogenicity may be an important environmental factor contributing to tropical enteropathy. Our study was not designed to detect the influence of diarrhea-causing organisms and we deferred participants who had experienced diarrhea until one month had elapsed.
There were two organisms notably associated with more severe enteropathy in biopsies taken one month later: Citrobacter rodentium and hookworm. The frequency of isolation of C. rodentium was surprisingly high and we took care to confirm the identity of the isolates using the API system read manually. However, this finding needs further confirmation in other geographic locations. These infections were not associated with diarrhea, nor with HIV, so it is unlikely that they merely behave as markers of a more generalized susceptibility to intestinal pathogens. Our analysis suggests that the effects of these organisms were specific, but an alternative possible explanation is that other organisms were isolated too infrequently to analyze their role with the same power. Indeed, the frequency of isolation of helminths is lower than in many tropical populations and our findings may be specific to this population. However, the seasonality of VH and Citrobacter incidence matched closely, giving further evidence to the idea that the association is specific. Enteropathy is well-recognized in patients with hookworm anemia,18 but C. rodentium is not known to be an intestinal pathogen in humans. It does cause intestinal damage in mice, a feature attributed to the bacterial expression of ß-intimin19,20 and host expression of Th1 cytokines.21 Whether isolates of C. rodentium from our population alter small intestinal structure and function through ß-in-timin expression is a question for further work. There are other intestinal infections which would be worth examining in such a study, including enterovirulent Escherichia coli and viruses, but time and resources did not permit their inclusion.
There was a lack of correlation between architectural and functional measures, and a lack of parallelism in their associations with other variables. Permeability was higher in women, but no sex difference was evident in architectural parameters. It was clear that CD was higher in HIV-infected adults irrespective of CD4 cell counts, but permeability was increased only in the group with low CD4 cell counts. This dissociation has previously been documented in HIV-infected adults living in London.22 Correlations between the temporal changes in structural and functional parameters were also weak. Although there is some evidence that permeability predicts architecture in jejunal biopsies,23 more recent evidence suggests that reduction in permeability precedes restoration of villous architecture in patients with celiac disease after gluten elimination.12 One explanation for the weak correlation between structure and function may be that functional studies provide information about the whole jejunum and ileum, beyond the biopsy site in the proximal jejunum. Interestingly, the correlation between VH and xylose recovery was closer than that with rhamnose or 3-O-methyl-D-glucose recovery, which might reflect the fact that xylose is largely absorbed in the proximal jejunum. Another explanation for the lack of correlation could be that the dynamics of villous and crypt architecture are determined by different cellular and molecular processes than those that control absorptive capacity and permeability. Permeability reflects tight junction function, which is under the control of at least one newly described molecule, zonulin.24
In this population, VH and CD were positively correlated at baseline, over time, and seasonally, but the process which leads to these parallel changes is unclear. T cell activation in vitro induces either an enteropathy characterized by villous atrophy and crypt hyperplasia or a mucosal destruction,25 and the same pattern is seen in celiac disease.15 T cell activation was associated with reduced VH and increased CD observed in Zambia compared with a South African black population without tropical enteropathy.26 However, in our cohort, the positive correlation between VH and CD suggests that T cell activation is not the principal determinant of the variation in enteropathy within this population. In animal models, starvation alters function,27 and undernutrition leads to true mucosal atrophy (i.e., reduced VH and CD in parallel). In humans in the short term, starvation can induce reduced absorptive capacity and increased permeability.28,29 Malnutrition in children is seasonal.30 We have previously reported that CD and body mass index were positively correlated in adults with advanced acquired immunodeficiency syndrome (AIDS)-related diarrhea and malnutrition,31 but further work is required to confirm and explain this finding.
Infection with HIV was associated with reduced VH and increased CD, but not with permeability or absorption. This is consistent with previous morphometric analysis in HIV-infected adults with or without AIDS.32,33 However, xylose recovery and permeability were altered when CD4 cell counts were reduced, again consistent with previous studies.34 It remains to be established whether the mucosal changes seen in HIV infection are due to HIV itself, to immunologic changes,35 or to secondary opportunistic infection, but it is well-established that anti-retroviral therapy leads to improvement in many of the pathologic changes observed.36 Our data suggest that in this population the earliest significant change in the mucosa of HIV-infected adults is increased CD.
Received October 2, 2003. Accepted for publication December 8, 2003.
Acknowledgments: We gratefully acknowledge the dedicated work of Emmanuel Kunda, Rosemary Banda, Vera Yambayamba, Coillard Kaunga, John Samson Mbewe, Stayner Mwanamakondo, Rose Soko, and Miriam Banda in the clinical team and the endoscopy unit. We are also very grateful to our study participants.
Financial support: This study was supported by The Wellcome Trust (grant 056481).
Authors’ addresses: Paul Kelly, Department of Medicine, University of Zambia School of Medicine, University Teaching Hospital, Lusaka, Zambia and Department of Adult and Paediatric Gastroenterology, Bart’s and The London School of Medicine, Turner Street, London E1 2AD, United Kingdom. Ian Menzies and Roger Crane, Department of Biochemistry, King’s College Hospital, London SE5 9RS, United Kingdom. Isaac Zulu, James Mwansa, Victor Mudenda, and Max Katubulushi, Department of Medicine, University of Zambia School of Medicine, University Teaching Hospital, Lusaka, Zambia. Carole Nickols, Roger Feakins, and Steve Greenwald, Department of Histopathology and Morbid Anatomy, Bart’s and The London National Health Service Hospital Trust, London E1 1BB, United Kingdom. Michael Farthing, St. George’s Hospital Medical School, London SW17 0RE, United Kingdom.
Reprint requests: Paul Kelly, Department of Adult and Paediatric Gastroenterology, Bart’s and The London School of Medicine, Turner Street, London E1 2AD, United Kingdom. Telephone: 44-20-7882-7191, Fax: 44-20-7882-7192, E-mail: m.p.kelly{at}qmul.ac.uk.
|
REFERENCES |
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  B S Ramakrishna, S Venkataraman, and A Mukhopadhya Tropical malabsorption Postgrad. Med. J., December 1, 2006; 82(974): 779 - 787. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Galpin, M. J Manary, K. Fleming, C.-N. Ou, P. Ashorn, and R. J Shulman Effect of Lactobacillus GG on intestinal integrity in Malawian children at risk of tropical enteropathy Am. J. Clinical Nutrition, November 1, 2005; 82(5): 1040 - 1045. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
= VH; 
= crypt depth; Yr = year.