T-Lymphocyte Subsets and Prolonged Diarrhea in Young Children from Guinea-Bissau

Transcription

1 American Journal of Epidemiology Copyright O 1996 by The Johns Hopkins University School of Hygiene and PubOc Health AD rights reserved Vol. 143, No. 1 Printed in USA T-Lymphocyte Subsets and Prolonged Diarrhea in Young Children from Guinea-Bissau Kare Molbak, 1 Ida M. Lisse, 2 and Peter Aaby 1 In a community-based prospective study of conducted between 1987 and 1990, the rate of diarrhea was significantly associated percentage of CD8 T-lymphocytes and the CD4 : CD8 ratio. After adjustment for age and previous diarrhea, the relative incidence of diarrhea a duration of &7 days was 2.10 (95% confidence interval (CO ) in % CD8 T-cells and 3.41 (95% Cl ) in 2:30% CD8 T-cells (in comparison who had less than 20% CD8 cells) (p for trend = 0.004). There was a nonsignificant tendency for rates of diarrhea of >7 days to decrease according to increasing proportions of CD4 cells (p = 0.194). The authors found no significant association between T-cell subsets and diarrhea which resolved in 6 days. The association between the incidence of prolonged diarrhea and T-cell subset proportions could not be explained as a confounding effect of low weight, breastfeeding, or previous infection measles or Cryptosporidium. However, other prior infections or micronutrient deficiencies may explain the findings, and these host factors may be significant targets in intervention against diarrheal diseases. Am J Epidemiol 1996; 143: antigens, CD4; antigens, CD8; child; diarrhea; immunity, cellular; lymphocyte subsets; risk factors; T- lymphocytes Diarrhea in early childhood remains a major cause of morbidity and death in developing countries (1). Most episodes of diarrhea resolve in a week and may be effectively treated by oral rehydration and continued feeding (2). However, in the case of more prolonged illness, oral rehydration has little impact, the risk of developing or worsening malnutrition increases, and there is a substantial risk of death (3). Hence, persistent diarrhea, defined as diarrhea that has continued for at least 2 weeks (4), has in many areas emerged as a more important cause of serious illness and death than acute diarrhea (2, 5, 6). A series of papers (7) and reviews (2-4, 8) have recently addressed the topic of persistent diarrhea, including host risk factors for persistent diarrhea. After initial observations in Bangladesh showed that cutaneous anergy was associated increased diarrheal morbidity (9), two prospective studies evaluated this relation further. These investigations from Bangladesh (10, 11) and Received for publication August 18,1994, and in final form February 13, Abbreviation: Cl, confidence interval. 1 Epidemiology Research Unit, Danish Epidemiology Science Centre, Statens SeruminstiUit, Copenhagen, Denmark. 2 Department of Pathology, Hvidovre University Hospital, Copenhagen, Denmark. Reprint requests to Dr. Kara Molbak, Epidemiology Research Unit, Danish Epidemiology Science Centre, Statens Seruminstitirt, Artillertvej 5, DK-2300 Copenhagen S, Denmark. Peru (12) corroborated the finding that the risk of developing diarrhea is inversely associated delayed-type hypersensitivity reactions to standard skin-test antigens. As part of ongoing community studies of the longterm effects of measles, we conducted serologic surveys in which the T-cell subsets of were assessed by examination of blood smears from fingerprick blood (13, 14). A number of these were included in prospective diarrhea studies (5, 15-17), which permitted an investigation of the association between cellular immune function, as measured by CD4 and CD8 T-lymphocyte subsets, and the incidence of acute and prolonged diarrhea. MATERIALS AND METHODS Fieldwork The study was conducted in a peri-urban district, Bandim II, in the capital of Guinea-Bissau. The district has a population of approximately 10,000 who reside in 650 houses. From 301 randomly sampled houses, all born after June 1, 1984, were included in the morbidity study, which started in January 1987 and ended in April Children bom in or moving to these houses were also included. Children who moved in the area were followed up in their new homes. In addition, the study included all twins and 79

2 80 Molbak et al. below 3 years of age who had experienced measles, even if they did not reside in one of the 301 selected houses (14). Field-workers collected information weekly on episodes of diarrhea (as defined by the mother) and child feeding patterns. A sequence of days diarrhea was regarded as one episode, provided that it was separated from previous episodes by at least 2 diarrhea-free days. Using portable spring scales and measuring boards, the field-workers measured each child's weight and length at intervals of approximately 3 months. Laboratory methods Capillary blood was collected in a heparinized Microtainer (Becton Dickinson, Rutherford, New Jersey), and on the same day the white blood cell count was estimated by a single experienced laboratory technician using a hemocytometer. Blood films for the hematologic examination were prepared, air-dried, wrapped in aluminum foil, and stored a desiccant at 10 C until transfer to Denmark (13). Immunocytochemical labeling was carried out monoclonal antibodies against CD4 and CD8 T-lymphocytes the alkaline phosphatase technique (13). Beginning in 1988, the alkaline phosphatase was conjugated the avidin-biotin complex (DAKO, Copenhagen, Denmark). Two hundred lymphocytes were counted for each of the immunolabeled slides. Differential counts were made visually, and the absolute values for each lymphoid subset were calculated. This method has been found to give T-cell subset values that are closely correlated results obtained by flow cytometry (13). Data analysis All out diarrhea on the day of blood sampling who were included in both the serologic surveys and the diarrhea surveillance were included in the present analysis. Children diarrhea were excluded, since they were not at risk of developing diarrhea. Thus, under 4 years of age were included, of which 317 (83.4 percent) were from the randomly selected group of 301 houses. Additionally, 50 were included because they had had measles and 13 because they were bom a twin. There was no difference in diarrhea rates between these three groups, which allowed us to analyze the data combined. A possible confounding effect of measles was nevertheless explored in the multivariate analysis. The data were analyzed by Cox proportional hazards regression (18). In the models, "survival" time was number of days free of diarrhea from the day of blood sampling to the day on which an episode of diarrhea was recorded (n = 305) or the child was lost to follow-up (n = 75). In the initial analysis, we categorized the results from the hematologic analysis into four groups by their quantiles and treated them as categorical variables, the highest group serving as the referent group. In addition, age (<12, 12-23, or t24 months) and year of investigation (1987 vs. after 1987) were included as categorical covariates. Further analysis included other possible determinants as covariates. Risk factors for acute and prolonged diarrhea of s7 days, respectively, were estimated in a model analogous to the competing risk models used in survival analysis different causes of exit (19). Hence, in the analysis of determinants for prolonged diarrhea of ^7 days, we censored the data at episodes a duration of less than 7 days, and in the analysis of determinants for acute episodes we censored at episodes of 2:7 days. All effects were expressed as the relative diarrhea rates, e p, where /3 is the regression coefficient in the multivariate analysis. The adequacy of the proportional hazards assumption was examined by plotting the log(-log) survival function. The computer program BMDP 21 (BMDP Statistical Software, Cork, Ireland) was used. RESULTS The study included who were followed for 428 child-months until their first episode of diarrhea or censoring. The median time to diarrhea was 24 days (median incidence, 4.2 episodes of diarrhea per 100 child-days). At entry, the had a median age of 18 months (range, 0-44 months); only seven were >35 months old. Table 1 shows the hematologic values and their association diarrhea in the initial analysis, which adjusted for age and year of investigation. Total lymphocyte count, CD8 cell percentage, and CD4: CD8 ratio were associated the rate of diarrhea. Since there was no natural trend in the association between lymphocyte count and risk of diarrhea, this finding was considered an artifact. In a following analysis, we examined possible confounders. Table 2 shows that who had experienced diarrhea in the month before entry, had been weaned, or had had cryptosporidiosis before entry had significantly increased rates. Previous diarrhea was of biologic importance in our attempts to determine the nature and causal direction of the association between T-cell subset values and diarrhea. Hence, further analysis was limited to the 289 for whom information could be obtained on both diarrhea in the month prior to blood sampling and T-cell subsets (table 3). After adjustment for age, year of study, and

3 T-Cell Subsets and Childhood Diantiea 81 TABLE 1. Median hematologlc values, 25th and 75th percentiles, and p values for an association between T- lymphocyte subsets and rates of diarrhea in from Guinea-Bissau, White Wood cell countt Lymphocyte % Total no. of lymphocytest CD4 cell % Total no of CD4 cellst CD8 cell % Total no. of CDS cellst CD4:CD8 cell ratio No. of cnwi'en Median value th-75th percentiles p value (or association diantiea* Likelihood ratio test from proportional hazards regression analysis, adjustment for age and year of examination. Hematologic values were categorized at their quantiles. t No. of cells x 109/liter. previous diarrhea, high proportions of CD8 cells had increased rates of diarrhea extending beyond 1 week, whereas there was no significant association between CD8 cell percentage and diarrhea of less than 7 days (p = ). Children >30 percent CD8 cells had a relative rate of 4.52 (95 percent confidence interval (CI) ) for diarrhea of at least 14 days' duration; the confidence interval was large because of a small number of persistent episodes (n = 12). Table 3 additionally shows a tendency for rates of diarrhea of 7 days to be lower in higher proportions of CD4 cells, although this relation was not statistically significant. The data were also analyzed in a Cox model only diarrhea episodes of &7 days' duration as the endpoint. Censoring was performed after a follow-up of 180 days or at a period of ^7 days missing information; episodes a duration of less than 7 days were ignored. The results were similar: There was an increased rate of diarrhea by increasing CD8 cell percentage (p = 0.037) and a 1.99 relative rate (95 percent confidence interval ; p = 0.028) in a CD4: CD8 ratio less than 1.0. Table 4 shows the results from a multivariate model that included diarrhea of at least 7 days as the response variable and adjusted for the confounding effect of previous diarrhea, weaning, previous infection Cryptosporidium (15, 16), and low weight-for-age. There were no significant first-order interactions between CD8 cell percentage and the other covariates, including prior measles. The table shows that the relative incidence of diarrhea of ^7 days was 2.09 in percent CD8 cells and 3.67 in &30 percent CD8 cells, in comparison who had less than 20 percent CD8 cells (p for trend = 0.004). DISCUSSION Studies from Bangladesh (9-11) and Peru (12) indicate that cell-mediated immune deficiency is an independent risk factor for diarrhea. In the study by Koster et al. (9), anergy was associated an 83 percent increased duration of diarrhea, and in the recent results reported by Baqui et al. (10, 11), the odds ratios were 1.55 (95 percent CI ) for all types of diarrhea and 2.09 (95 percent CI ) for persistent diarrhea in anergic. The statis- TABLE 2. Age-adjusted rate ratios for diantiea among from Guinea-Bissau, * Determinant Previous diarrhea}. Child weaned Previous cryptosporidiosis Welght-for-age z score i -21 Height-for-age z score <. -2i Previous measlesh Measles vaccination Male sex No. of characteristic No. of available data RRt % Clt * Determined by proportional hazards regression analysis, t RR, rate ratio; CI, confidence interval. % Diarrhea 30 days before entry into the study. No longer being breastfed. D Calculated before entry into the study using World Health Organization-US National Center for Health Statistics reference values (29). H Measles <360 days before entry into the study.

4 82 Molbak et al. TABLE 3. Rate ratios for diarrhea in relation to T-tymphocyte subset values among from Guinea-Bissau, * No. ol RRt,* All diarrhea (240 episodes) 95% Clt Diantiea of 7 days (54 episodes) RR* 95% Cl CD4 cell % < p = p = CD8 cell % < CD4:CD8 cell < ratio p = p = p = p = Proportional hazards regression including 289 Information on diarrhea In the month preceding blood sampling. t RR, rate ratio; Cl, confidence interval. i Adjusted for age at Wood sampling, year of study (1987 vs. after 1987), and diarrhea in the 30 days prior to blood sampling. p for trend (likelihood ratio test). tical model adjusted for the confounding effect of previous diarrhea and malnutrition. The present study from Guinea-Bissau suggests that high CD8 cell percentages, low CD4 cell percentages, and thus low CD4: CD8 ratios have increased rates of diarrhea episodes extending for at least 1 week. The rate of persistent diarrhea, which has been arbitrarily defined as diarrhea that lasts for at least 14 days (4), may be even higher. It is important to emphasize that we studied relatively mild alterations in the T-cell subset values. The similarity to the findings from Bangladesh is nevertheless striking. In addition, our study corroborates research suggesting that prior illness is a risk factor for diarrhea (2, 20). In particular, the association between high CD8 cell percentages and increased morbidity of prolonged diarrhea cannot be regarded as one of cause and effect unless possible sources of confounding and bias have been considered carefully. The lack of interaction between prior measles and T-cell subset values suggests that the representativeness of the population was unaffected by the inclusion of the prior measles. Furthermore, the multivariate analysis adjusted for age, lactation, previous diarrhea, previous infection measles or Cryptosporidiwn, and malnutrition. These factors were all considered biologically important in examining the association. However, nutritional factors such as specific micronutrient deficiencies that are imperfectly related to body size may have an impact on the regulation of cell-mediated immune response. Vitamin A, zinc, and iron are known to be related to the immune system (21). Thus, a recent study from Indonesia showed that xerophthalmia had lower proportions of CD4 cells and higher proportions of CD8 cells. After vitamin A supplementation, CD4 cell percentages increased and CD8 cell percentages decreased in comparison a placebo group (22). In addition, vitamin A deficiency has been found to be a risk factor for persistent diarrhea (23). Unfortunately, in the present study, no data on micronutrient status were available for determination of whether the association T- lymphocyte subsets reflected a specific deficiency. A number of infections may be associated low CD4: CD8 ratios, including infections human immunodeficiency virus, measles, Epstein-Barr virus, cytomegalovirus, and human T-cell leukemia virus II (24, 25). Human immunodeficiency virus infection is rare in in Guinea-Bissau, and is therefore unlikely to explain our findings (26, 27). Adjustment for possible confounding effects of prior measles, measles vaccination, and infection Cryptosporidiwn did not have a significant impact on the results. Unfortunately, we did not have information on malaria. However, it is our experience that although CD4 percentage may be inversely correlated the den-

5 T-Cell Subsets and Childhood Diantiea 83 TABLE 4. Multrvariate analysis of risk factors for diarrhea of 27 days' duration among from Guinea-Bissau, * Covartate CD8 cell % < Age (months) at analysis Diarrhea episode in previous month No Yes Breastfeeding Partial Weaned Previous cryptosporidiosis No Yes Weight-for-aget z score > -2 z score <, -2 No. of characteristic Rate ratio % confidence Interval Proportional hazards regression analysis including 276 complete data. t Calculated using World Health Organizatjon-US National Center for Health Statistics reference values (29). sity of malaria parasites, there is no association between malaria density and CD8 percentage or CD4: CD8 ratio (28). It is therefore unlikely that the present findings can be explained by a confounding effect of malaria parasitemia, though further studies are warranted. Pathologic studies have suggested a link between impaired cell-mediated immunity, chronic diarrhea, and damage to the intestinal mucosa; the pathology of this damage includes an increased number of intraepithelial CD8 lymphocytes (3). Further understanding may have significant implications for our knowledge of host risk factors for the development of persistent diarrhea. If the findings are related to specific extraneous factors, such as nutritional deficiencies or concomitant infections, these may be important targets in interventions against persistent diarrhea. Secondly, if confirmatory studies support the finding that mild alterations in the regulation of cellular immune response have a major impact on morbidity, a number of more fundamental questions regarding the function of the heterogeneous group of CD8 cells may be posed. ACKNOWLEDGMENTS This study was supported by the Science and Technology for Development Programme, Commission of the European Community (grants TS and TS3-OO6O); the University of Copenhagen, Danish Council for Development Research (grant 104.Dan.8/1114); and the Danchurchaid primary health care project in Guinea-Bissau. REFERENCES 1. Bern C, Martines J, de Zoysa I, et al. The magnitude of the global problem of diarrhoeal disease: a ten-year update. Bull World Health Organ 1992;70: Black RE. Persistent diarrhea in of developing countries. Pediatr Infect Dis J 1993;12: Lima AAM, Guerrant RL. Persistent diarrhea in : epidemiology, risk factors, pathophysiology, nutritional impact, and management Epidemiol Rev 1992;14: Persistent diarrhoea in in developing countries: memorandum from a WHO meeting. Bull World Health Organ 1988;66: M0lbak K, Aaby P, Ingholt L, et al. Persistent and acute diarrhoea as the leading causes of child mortality in urban Guinea Bissau. Trans R Soc Trap Med Hyg 1992,86: Victora CG, Huttly SR, Fuchs SC, et al. Deaths due to dysentery, acute and persistent diarrhoea among Brazilian infants. Acta Paediatr Suppl 1992;381: Black RE, ed. Persistent diarrhea in of developing countries: proceedings of a symposium, Mombasa, Kenya, January Acta Paediatr Suppl 1992;381: Bhan MK, Bhatnagar S. Persistent diarrhea in infants and young. (Editorial). Indian Pediatr 1993;30: Koster FT, Palmer DL, Chakraborty J, et al. Cellular immune competence and diarrheal morbidity in malnourished Bangladeshi : a prospective field study. Am J Clin Nutr 1987;46: Baqui AH, Black RE, Sack RB, et al. Malnutrition, cellmediated immune deficiency, and diarrhea: a communitybased longitudinal study in rural Bangladeshi. Am J Epidemiol 1993; 137: Baqui AH, Sack RB, Black RE, et al. Cell-mediated immune deficiency and malnutrition are independent risk factors for persistent diarrhea in Bangladeshi. Am J Clin Nutr 1993;58: Black RE, Lanata CF, Lazo F. Delayed cutaneous hypersensivity: epidemiologic factors affecting and usefulness in predicting diarrheal incidence in young Peruvian. Pediatr Infect Dis J 1989;8: Lisse IM, Whittle H, Aaby P, et al. Labelling of T cell subsets under field conditions in tropical countries: adaptation of the immuno-alkaline phosphatase staining method for blood smears. J Immunol Methods 1990;129: Aaby P, Lisse IM, M0lbak K, et al. No persistent T-lymphocyte immunosuppression or increased mortality after measles infection: a community study from Guinea-Bissau. Pediatr Infect Dis J (in press). 15. M0lbak K, H0jlyng N, Gottschau A, et al. Cryptosporidiosis in infancy and childhood mortality in Guinea-Bissau, West Africa. BMJ 1993;307: M0lbak K, Wested N, H0jlyng N, et al. The etiology of early

6 84 Melbak et al. childhood diarrhea: a community study from Guinea-Bissau. J Infect Dis 1994;169: M0lbak K, Gottschau A, Aaby P, et al. Prolonged breastfeeding, diarrhoea] disease, and survival of in Guinea- Bissau. BMJ 1994;308: Cox DR, Oakes D. Analysis of survival data. London, England: Chapman and Hall Ltd., David HA, Moeschberger ML. The theory of competing risk. London, England: Charles Griffin and Company, Sazkwal S, Bhan MK, Bhandari N, et al. Evidence for recent diarrhoeal morbidity as a risk factor for persistent diarrhoea: a case-control study. Int J Epidemiol 1991;20: Beisel WR. Single nutrients and immunity. Am J Clin Nutr 1982;35(suppl): Semba RD, Muhilal, Ward BJ, et al. Abnormal T-cell subset proportions in vitamin-a-deficient. Lancet 1993;341: Mahalanabis D, Alam AN, Rahman N, et al. Prognostic indicators and risk factors for increased duration of acute diarrhoea and for persistent diarrhoea in. Int J Epidemiol 1991;20: Laurence J. T-cell subsets in health, infectious disease, and idiopathic CD4+ T lymphocytopenia. Ann Intern Med 1993; 119: Kiepiela P, Coovadia HM, Coward P. T helper cell defect related to severity in measles. Scand J Infect Dis 1987; 19: Poulsen A-G, Kvinesdal BB, Aaby P, et al. Prevalence of and mortality from human immunodeficiency virus type 2 in Bissau, West Africa. Lancet 1989; 1: Poulsen A-G, Kvinesdal B, Aaby P, et al. Lack of evidence of vertical transmission of human immunodeficiency virus type 2 in a sample of the general population in Bissau. J Acquir Immune Defic Syndr 1992;5: Lisse IM, Aaby P, Whittle H, et al. A community study of T lymphocyte subsets and malaria parasitaemia. Trans R Soc Trop Med Hyg 1994;88: National Center for Health Statistics. Growth curves for birth-18 years, United States. Washington, DC: US GPO, (Vital and health statistics, series 1, no. 165) (DHEW publication no ).