Am. J. Trop. Med. Hyg., 76(3), 2007, pp. 541–548 Copyright © 2007 by The American Society of Tropical Medicine and Hygiene

HELICOBACTER PYLORI INFECTION IS ASSOCIATED WITH SEVERE ANEMIA OF PREGNANCY ON PEMBA ISLAND, ZANZIBAR TAMER H. FARAG,* REBECCA J. STOLTZFUS, SABRA S. KHALFAN, AND JAMES M. TIELSCH Department of International Health, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, Maryland; Division of Nutritional Sciences, Cornell University, Ithaca, New York; Public Health Laboratory-Ivo de Carneri, Zanzibar, Tanzania

Abstract. Helicobacter pylori infection has recently been associated with iron deficiency and anemia in developed countries. To determine the association of H. pylori and anemia in a tropical region, we measured hemoglobin concentration (Hb) and H. pylori infection by the 13C urea breath test among 857 pregnant women attending antenatal care clinics. Parasitology, anthropometry, obstetric history, sociodemographic and dietary variables were also assessed. Logistic regression showed an odds ratio of 7.63 (95% confidence interval ⳱ 1.73–33.55) for H. pylori infection comparing women with and without severe anemia (Hb < 7 g/dL), controlling for hookworm infection, body mass index and parity, but only among women with a diet low in foods containing heme iron. Infection with H. pylori with low bacterial load was associated with lower Hb concentration while high bacterial load was associated with higher Hb concentration. Further research is needed to establish causality because high worldwide prevalence means that even a small associated risk would be of public health significance. MATERIALS AND METHODS

INTRODUCTION

Setting and study sample. A cross-sectional study examining the association between concurrent H. pylori infection and Hb levels was conducted on Pemba Island, Zanzibar, United Republic of Tanzania from April 13, 2004 to December 28, 2004. This study was nested in a trial of prevention and treatment of severe anemia in pregnancy. Pemba is a densely populated rural island with a tropical climate and ecology and is located approximately 40 km off the east coast of mainland Tanzania and north of Unguja, the main island of Zanzibar. Pembans subsist on agriculture and fishing and the major cash crop is cloves. The Bantu ethnic group is predominant, but there is a heavy Arab influence. The population is characterized by holo-endemic Plasmodium falciparum malaria, as well as endemic transmission of hookworm infection. Bioavailable iron in the diet is low. The diet consists largely of cassava and some rice, with only seasonal availability of vegetables and fish; red meat is largely unavailable or unaffordable.19 Women were recruited from eight antenatal health clinics in the northern part of Pemba. To be eligible for inclusion, the woman must have been enrolled in the parent trial and have clinically confirmed signs of pregnancy. To increase the number of women in higher and lower Hb level categories, we selected a stratified sample in which all women with Hb < 7 g/dL were recruited, but only a proportion of the women in the categories of Hb ⳱ 7–10.9 g/dL and Hb ⱖ 11.0 g/dL were recruited. Because of the stratified sampling procedure, 86% of women with Hb < 7 g/dL (n ⳱ 48) were selected compared with 65% of women with Hb ⳱ 7–7.9 g/dL (n ⳱ 70), 44% of women with Hb ⳱ 7–10.9 g/dL (n ⳱ 493), and 66% of women with Hb ⱖ 11 g/dL (n ⳱ 254). The study subsample was similar to the parent trial sample for all important indicators (Table 1). Venous blood was drawn from 328 women enrolling in 4 clinics (Wete, Jadida, Tumbe, and Konde) for a separate substudy currently under laboratory analysis. To confirm the results of the urea breath test (UBT), we selected a random sample of 31 women testing positive for H. pylori and a random sample of 31 women testing negative for H. pylori by the UBT and assessed their H. pylori exposure status by IgG enzyme immunoassay (EIA).

Worldwide, approximately 41% of women in sub-Saharan Africa, 23% in Latin America, 49% in Southeast Asia, and 60% in the Indian subcontinent of Asia are anemic.1 Moderate-to-severe maternal anemia has been associated with an increased risk of poor reproductive outcomes, including low birth weight and preterm birth deliveries.2 Anemia can contribute to hemorrhage and cardiac failure, two principal causes that comprise 25% of all maternal deaths.3 Severe anemia can be life threatening in settings where the health infrastructure is limited and childbirth at home is the norm. The case-fatality rate from severe anemia is high; 6.1% in one hospital in Zanzibar.4 Early detection and treatment of multiple causes of severe anemia may reduce the risk of blood transfusion and perinatal and maternal mortality. In poor populations in tropical regions, well-documented causes of anemia include dietary iron deficiency, vitamin A deficiency, the physiologic demands of pregnancy, and infectious diseases such as malaria and hookworm.5 Recent evidence suggests that Helicobacter pylori, a common bacterial infection of the gut, is associated with iron deficiency and anemia. It may be the most common bacterial infection worldwide, infecting almost half of people in developed countries and 80% of people in developing countries.6 Infection with H. pylori is strongly associated with peptic ulcer and gastric cancer.7 Several cross-sectional studies have found an association between H. pylori and low body iron stores8–13 and iron deficiency anemia13–16 and a reduced response to iron supplementation.17 Weyermann and others found that pregnant women infected with H. pylori had lower mean hemoglobin (Hb) level at the beginning of pregnancy and a greater decrease in the mean Hb level at the end of pregnancy.16 In Kenya, anemic children had a 2.5-fold higher proportion of elevated IgM antibody titers against H. pylori than non-anemic children.18

* Address correspondence to Tamer H. Farag, Department of International Health, Bloomberg School of Public Health, Johns Hopkins University, 615 North Wolfe Street, Room W5009, Baltimore, MD, 21205. E-mail: [email protected]

541

542

FARAG AND OTHERS

TABLE 1 Comparison of women enrolled in parent trial with those enrolled in this study Enrolled in parent trial

Enrolled in this study

Indicator

n

mean ± SD or %

n

mean ± SD or %

Wete clinic Jadida clinic Mzambarauni clinic Chwale clinic Kiungoni clinic Wingwi clinic Tumbe clinic Konde clinic Household motorcycle Household books Household latrine Costlier roofing* Costlier household lighting† Body mass index (kg/m2) Age (years) Gestational age (fundal height, cm)

1,678 1,678 1,678 1,678 1,678 1,678 1,678 1,678 1,678 1,677 1,676 1,674 1,675 1,678 1,678 1,678

12.1 15.0 9.8 10.8 10.3 15.4 12.8 13.3 4.9 11.3 42.4 42.5 21.1 23.3 ± 4.9 28.3 ± 6.9 24.8 ± 5.4

857 857 857 857 857 857 857 857 857 857 856 856 855 865 865 855

12.1 18.4 9.0 10.4 8.9 16.0 14.6 10.6 5.0 11.2 42.2 41.2 21.5 23.5 ± 5.9 28.1 ± 7.1 25.3 ± 5.6

* Corrugated iron sheeting, cement, or tile vs. roofing with any palm thatch. † Only glass or electric lighting vs. use any kerosene tin lighting.

Measurements. Helicobacter pylori infection status was measured by the 13C-UBT, which detects current infection by measuring urease activity. The sensitivity and specificity of this assay are both greater than 95%.20–22 Breath samples are taken before and after a 13C-labeled urea dose and the concentration of 13C-labeled carbon dioxide (13CO2) is measured and compared between them. The change in concentration of 13 CO2 is called delta over baseline (DOB), and is considered positive for infection at or above a 2.4 permil threshold. The BreathTek UBiT-IR300 portable infrared spectrophotometer (Meretek Diagnostics, Inc., Nashville, TN) was used to perform the assays on site in Pemba. Breath samples were collected in the clinic following the standard protocol. The sachets containing the 13C-labeled urea and citric acid test meal were shipped to Pemba by air and stored in a secure, dry environment. Because the sachets were designed to be stable, storage under warm conditions on Pemba would have no effect on the constituents (Meretek Diagnostics Inc., unpublished data). Breath bags were transported to the laboratory at the end of each day. Breath samples were tested within seven days after collection, as specified by the manufacturer. The UbiT-IR300 infrared spectrophotometer was located in an air-conditioned laboratory. Control gasses were tested every morning that samples were analyzed, and readings were always as expected. In addition, factory-standard stability tests were run; the results of repeated measurements of CO2 concentration were within the allowable coefficient of variation (1%). In a subset of samples, H. pylori exposure was also assessed by EIA using the automated Immulite 1000 System (Diagnostic Products Corporation, Los Angeles, CA), which has a sensitivity and specificity greater than 90%.23 As per the manufacturer’s protocol, results ⱖ 1.1 U/mL were considered positive, results < 0.9 U/mL were considered negative, and results ⱖ 0.9 U/mL but < 1.1 U/mL, were considered intermediate. Two aliquots of each sample were tested. Samples that were intermediate in one aliquot but positive in another were considered positive. Samples that were intermediate in one aliquot but negative in another were considered negative. In the interest of being conservative, if both aliquots were intermediate, the sample was considered positive.

Hemoglobin concentration was measured as per the parent trial protocol using a drop of capillary blood for 529 women, and a drop of venous blood for 328 women. A portable hemoglobinometer was used to make the measurements in the clinic (HemoCue AB, Angelholm, Sweden). Information was collected on mother’s age, parity, weight and height, gestational age based on fundal height, reproductive history, socioeconomic status, and diet by a one-week food frequency questionnaire. Based on previous research conducted on Pemba, we assessed socioeconomic status indicators known to be associated with anemia: occupation and educational level of the woman and her husband, type of roof, type of lighting, presence of household latrine, household ownership of a bicycle and/or books other than the Koran, and household ownership of a radio or motorcycle. Weight was measured in kilograms on a digital scale with 0.1-kg increments (Model H314, Tanita Corp., Tokyo, Japan). Height was measured in 0.1-cm increments against a wall-mounted tape measure. Local investigators were consulted to produce a foodfrequency questionnaire to include all foods commonly consumed on Pemba. The one-week food frequency questionnaire was designed to assess dietary sources of heme and non-heme iron (such as beef, chicken, beans, and spinach), non-heme iron chelators or inhibitors (such as tea and some plant-based foods), enhancers of non-heme iron absorption (such as foods rich in vitamin C), as well as caloric intake and sources of other micronutrients. A composite variable was created to indicate a diet high in foods containing heme iron; women eating four or more of either beef, goat, chicken, fish, whitebait, or other seafood (such as shrimp, crabs, octopus, or squid) were classified as having a high heme diet. Malaria infection status was assessed by standard World Health Organization methods on thick blood film slides made in the clinic. Ascaris, Trichuris, and hookworm were assessed by the Kato-Katz technique on stool samples collected within one week after the enrollment visit. Data management and analysis. Forms were entered into a Microsoft SQL-based (Microsoft Corp, Redmond, WA) database server. Regular manual and automated checks were performed to ensure that data were internally consistent and

543

H. PYLORI INFECTION AND SEVERE ANEMIA

within expected values. Intermethod reliability between the UBT and Immulite EIA was assessed by calculating percent agreement and kappa statistic.24 Simple linear and simple logistic regressions were used to assess crude associations between variables. Logarithmic transformation was undertaken to normalize the distributions of some variables as needed; values less than 1 were replaced by a 0 after transformation. Multivariate logistic regression was used to assess the relationship of H. pylori infection to anemia. Likelihood ratio testing was used to select the model by adding variables to assess for confounding and interaction terms to assess for effect modification.25 Multivariate linear regressions were used to assess the association of infection to Hb concentration and the association of DOB to Hb and severe anemia. Model selection was done by nested Wald testing.26 All statistical analyses were performed using STATA version 9.0 (Stata Corporation, College Station, TX). The level of significance used in all statistical tests was 0.05. Ethical considerations. This study was reviewed and approved by the Committee on Human Research of the Bloomberg School of Public Health, Johns Hopkins University (Baltimore, MD) and the Zanzibar Health Research Council (Zanzibar, Tanzania). Women infected with H. pylori were not treated because no established clinical manifestation of H. pylori could be determined in the absence of gastroduodenoscopy.27 All women were given daily iron-folate supplements (60 mg of iron + 400 ␮g of folic acid, double for severely anemic women with Hb < 7 g/dL). RESULTS Of the 865 women enrolled, 8 (0.9%) were not included in the final analysis because of technical errors in the UBT test, yielding a total of 857 women. The final multivariate regression models controlling for hookworm fecal egg density included only the 803 women for whom usable stool samples were obtained. The prevalence of H. pylori infection was 17.5%, which was unexpectedly low (Table 2). When compared with the Immulite EIA, percent agreement was 92% and the kappa statistic was 0.8, which showed substantial to almost perfect intermethod agreement according to the criteria of Landis and Koch for interpreting the kappa statistic (Table 3).24 Women with H. pylori infection had a 1.85-fold higher odds of being infected with malaria (Table 4). Women infected with H. pylori were younger, had a higher body mass index, higher gestational age, and higher socioeconomic status (Table 4). The odds of infection with H. pylori were 2.42 times higher (95% confidence interval [CI] ⳱ 1.52–3.30) for women with costlier lighting in their houses. Presence of a household latrine and costlier household roofing were also associated with H. pylori, although with lower magnitude. Infection with H. pylori and severe anemia. Infection with H. pylori was not associated with severe anemia by simple logistic regression (Table 4), but stratification by the level of heme iron-containing foods in the diet suggested an effect modification. Among women in the high heme diet stratum, the odds ratio of infection was 0.74 (95% CI ⳱ 0.22–1.98) and showed little evidence of an association. Among women with a diet low in foods containing heme iron, the odds ratio of infection was large in magnitude (3.14) but with a wide con-

TABLE 2 Means or prevalences and 95% confidence intervals of outcome, exposure, and predictor variables Variable (n)

Mean ± SD or percent

Severe anemia (hemoglobin < 7 g/dL) (857) Hemoglobin concentration (g/dL) (857) Helicobacter pylori (857) Hookworm (infected/uninfected) (803) Geometric mean hookworm fecal egg density (eggs/gram) (262)* Malaria (infected/uninfected) (841) Household latrine (856) Household books (857) Household motorcycle (857) Household bicycle (8570 Husband has prestigious job (857)† Costlier lighting (855)‡ Costlier roofing (856)§ High tomato consumption (857)¶ High heme 4 (857)# Vitamin supplement use (856) Body mass index (kg/m2) (857) Parity (secunde or higher vs. primi or nulliparity) (857) Age (years) (856) Gestational age (fundal height, cm) (855)

5.5 9.9 ± 1.9 17.5 32.6 380.4 ± 4.2 7.4 42.2 11.2 5.0 65.8 23.3 21.5 41.2 52.2 80.5 7.9 23.3 ± 3.7 72.7 28.1 ± 6.9 25.3 ± 5.6

* Among women infected with hookworm. † Shopkeeper, civil servant, businessman. ‡ Only glass or electric lighting vs. use any kerosene tin lighting. § Corrugated iron sheeting, cement, or tile vs. roofing with any palm thatch. ¶ Five or more tomato foods int he past week. # At least one food containing heme ⱖ 4 times per week.

fidence interval (95% CI ⳱ 0.62–13.37). The P value for a test of interaction was 0.08. In the final multivariate model adjusting for hookworm fecal egg density, body mass index, and parity, H. pylori was strongly and statistically significantly associated with severe anemia, but only among women consuming a diet low in foods containing heme iron (Table 5). No other variables were found to confound or modify the association between H. pylori infection and severe anemia. Among women with a diet low in foods containing heme iron, we found an odds ratio of 7.63 (95% CI ⳱ 1.73–33.55) for H. pylori infection comparing women with and without severe anemia. Among women with a diet high in foods containing heme iron, the odds of being infected with H. pylori were 1.25 times higher (95% CI ⳱ 0.45–3.46) for severely anemic women than for non-severely anemic women. Because the confidence interval for the odds ratio OR in this stratum was wide and included 1.0, there was little evidence of an association. Infection with H. pylori and Hb concentration. Hemoglobin as a continuous outcome was not associated with H. pylori by crude analysis (␤ ⳱ 0.06 g/dL, 95% CI ⳱ −0.27–0.39). However, multiple linear regression of Hb on H. pylori infection status showed that among women with a diet low in foods TABLE 3 Cross-tabulation of Helicobacter pylori detection by Meretek urea breath test and Immulite IgG enzyme immunosorbent assay (EIA) Urea breath test

H. pylori IgG EIA

Yes No Total

Yes

No

Total

27 4 31

2 29 31

29 33 62

544

FARAG AND OTHERS

TABLE 4 Simple logistic regressions of Helicobacter pylori infection status or severe anemia (hemoglobin < 7 g/dL) status on exposure and predictor variables* H. pylori infection

Severe anemia status

Variable

OR (95% CI)

P

OR (95% CI)

P

H. pylori Hookworm (infected/uninfected) Log hookworm fecal egg density (eggs/gram) Malaria (infected/uninfected) Household latrine Household books Household bicycle Household motorcycle Husband has prestigious job† Costlier lighting‡ Costlier roofing§ High tomato consumption¶ High heme 4# Vitamin supplement use Body mass index (kg/m2) Multiparity** Age (years) Gestation age (fundal height, cm) (855)

– 0.71 (0.47–1.08) 0.94 (0.88–1.01) 1.85 (1.03–3.34) 1.52 (1.07–2.16) 1.68 (1.02–2.77) 0.96 (0.67–1.39) 2.14 (1.09–4.21) 0.96 (0.63–1.45) 2.42 (1.52–3.30) 1.64 (1.15–2.34) 1.38 (0.96–1.97) 1.07 (0.68–1.67) 1.24 (0.67–2.30) 1.04 (1.00–1.09) 0.65 (0.45–0.95) 0.97 (0.95–1.00) 1.04 (1.01–1.07)

– 0.11 0.08 0.04 0.02 0.04 0.83 0.03 0.83 < 0.01 0.01 0.08 0.78 0.49 0.05 0.03 0.03 0.02

1.12 (0.53–2.38) 4.03 (2.10–7.17) 1.31 (1.19–1.45) 0.27 (0.01–1.63) 0.45 (0.23–0.88) 0.73 (0.25–2.07) 1.63 (0.83–3.18) 0.40 (0.05–2.95) 0.38 (0.15–0.96) 0.86 (0.41–1.81) 0.59 (0.31–1.12) 0.41 (0.22–0.77) 0.78 (0.39–1.57) 0.24 (0.03–1.77) 0.86 (0.78–0.96) 2.68 (1.12–6.39) 1.00 (0.96–1.05) 1.00 (0.95–1.05)

0.76 < 0.01 < 0.01 0.20 0.02 0.55 0.16 0.37 0.04 0.68 0.10 0.01 0.49 0.16 0.01 0.03 0.94 0.93

* OR ⳱ odds ratio; CI ⳱ confidence interval. † Shopkeeper, civil servant, businessman. ‡ Only glass or electric lighting vs. use any kerosene tin lighting. § Corrugated iron sheeting, cement, or tile vs. roofing with any palm thatch. ¶ Three or more tomato foods in the past week. # At least one food containing heme ⱖ 4 times per week. ** Compared with nulliparity and primiparity.

containing heme iron being infected with H. pylori was associated with 0.88 g/dL lower Hb (95% CI ⳱ 1.65–0.10) when adjusted for hookworm fecal egg density, parity, body mass index, and heme diet. For women with a diet high in foods containing heme iron, H. pylori infection was associated with 0.03 g/dL higher Hb (95% CI ⳱ −0.33–0.40) and showed little evidence of an association. The P value for the test of interaction was 0.04. No other variables were found to confound or modify the association between H. pylori infection and Hb concentration in the multiple linear regression model. Helicobacter pylori bacterial load and Hb concentration. Urease activity as measured by DOB was used as a continuous measure of H. pylori bacterial load.28 Among women infected with H. pylori, H. pylori DOB was positively associated with high tomato consumption (odds ratio [log DOB] ⳱ 1.55, 95% CI ⳱ 1.09–2.21). Simple linear regression of Hb (g/dL) on DOB showed that a log increase in DOB was associated with a 0.50 g/dL (95% CI ⳱ 0.18–0.81) increase in Hb (Figure 1). The mean Hb concentration for uninfected women was 9.88 g/dL (95% CI ⳱ 9.74–10.02). We defined high and low DOB categories based on whether they were

associated with better or worse Hb relative to people who were uninfected. Low (2.4 ⱖ DOB < 20) and high (DOB ⱖ 20) bacterial load categories among the infected were defined as being values below and above the mean Hb level for uninfected women (Figure 1). When Hb concentration among all women was regressed against bacterial load categories, infection with low bacterial load (n ⳱ 74) was associated with 0.56 g/dL (95% CI ⳱ 0.12–1.01) lower Hb and infection with high bacterial load (n ⳱ 76) was associated with 0.66 g/dL (95% CI ⳱ 0.22–1.10) higher Hb. Among H. pylori-positive women, DOB was associated with a lower likelihood of being severely anemic (odds ratio [log DOB] ⳱ 0.39, 95% CI ⳱ 0.18–0.86). In multivariate regression, no other variables, including high tomato and high heme consumption, were found to confound or modify the relationships between hematologic outcome and DOB or bacterial load categories. DISCUSSION The prevalence of H. pylori infection was 17.5%, which was unexpectedly low, given that prevalence is 80–100% among

TABLE 5 Logistic regression model of severe anemia status (Hb < 7 g/dL) on Helicobacter pylori, with H. pylori × high heme consumption interaction term, n ⳱ 803* Variable

Beta

P

95% CI

OR

95% CI

H. pylori Ln (hookworm eggs/gram) Body mass index Multiparity† High heme 4‡ High heme 4 × H. pylori

2.03 0.25 −0.11 1.44 −0.06 −1.81

0.01 0.00 0.07 0.01 0.90 0.05

0.55–3.51 0.15–0.36 −0.23–0.01 0.33–2.56 −0.95–0.83 −3.59 to −0.03

7.63 3.63 0.89 4.23 0.94 –

1.73–33.55 3.20–4.16 0.79–1.01 1.39–12.89 0.39–2.29 – –

* Hb ⳱ hemoglobin; CI ⳱ confidence interval; OR ⳱ odds ratio. Hb < 7 g/dL (n ⳱ 42); Hb ⱖ 7 g/dL (n ⳱ 761). † Compared with nulliparity and primiparity. ‡ At least one food containing heme ⱖ 4 times per week.

H. PYLORI INFECTION AND SEVERE ANEMIA

FIGURE 1. Hemoglobin (g/dL) on Helicobacter pylori delta over baseline (DOB) in log scale among women infected with H. pylori with regression line (right) and mean hemoglobin level for uninfected women (left). CI ⳱ confidence interval.

neighboring populations.29–33 Although this finding is unusual, it is not without precedent. Some coastal adult populations in southeast Asia show prevalences between 0% and 32% in the context of high prevalence in surrounding populations.34–39 High concordance with the EIA for H. pylori IgG showed that the UBT was reliable in this population. Helicobacter pylori was strongly associated with severe anemia and a lower Hb level in this population, but only among women with a diet low in foods containing heme iron (Figure 2). Several candidate anemiagenic mechanisms exist for H. pylori. The most plausible among them are gastrointestinal blood loss, the anemia of chronic disease, bacterial sequestration of free iron, inhibited free iron absorption, and foodcobalamin malabsorption. The observed effect modification by a diet high in foods containing heme iron can be used as a framework to examine the plausibility of candidate mecha-

FIGURE 2. Effect modification by high heme diet* on the odds ratio (OR) for the association between infection with Helicobacter pylori and severe anemia (hemoglobin level < 7 g/dL)** *At least one food containing heme ⱖ 4 times per week. **Controlling for hookworm fecal egg density, body mass index, and parity.

545

nisms because the evidence would favor those that are biologically consistent. However, speculation should be tempered by the limitations of the one-week food frequency questionnaire, which can only be used to describe the relative frequency but not the adequacy of consumption within this population. As with all food frequency questionnaires, it describes consumption within a specific period of time and is assumed to reflect habitual consumption although consumption can change over time, including over the course of pregnancy, when eating habits can change. Gestational age was controlled for in analysis, however, and found not to confound any of the multivariate regression results. Infection with H. pylori tends to be either antral (adjacent to the duodenum) or to involve both the antrum and the body of the stomach (pan-gastritis). With antral gastritis, gastrinproducing G cells in the antrum are stimulated to overproduce gastrin, and the enterochromaffin-like and parietal cells of the gastric body largely escape damage, which results in increased acidity and increased risk of duodenal ulcer. With pan-gastritis, damage to the gastric mucosa can be so severe that acid-producing cells are destroyed, which leads to achlorhydria, gastric metaplasia, gastric atrophy, and gastric ulcer.40 Because H. pylori causes dramatic changes to the gut, the way in which free iron and heme are absorbed from the gut lumen are central to the evaluation of the plausibility of a candidate anemiagenic mechanism. Free iron is absorbed in the duodenum and upper jejunum, and to a lesser extent in the stomach, ileum, and colon. Free iron is available in the diet in its trivalent ferric form and must be converted to bivalent ferrous iron by a combination of ferrireductase, ascorbic acid, and gastric acid before it can be transported into the enterocyte by membrane divalent metal transporter 1.41 Unlike free iron, heme iron found in Hb and myoglobin in meat is readily absorbed by heme carrier protein 1 in the absence of promoters by enterocytes in the duodenum,42,43 where it joins the common intracellular iron pool.44 Whether originating from free or heme iron, cellular iron is exported from the enterocyte to the plasma by means of a membrane protein, ferroportin.45 The role of subclinical infection or inflammation as a cause of anemia, the so-called anemia of chronic disease,46 is mediated by cytokines, which have been shown to inhibit erythropoietin secretion,47 block its action on the bone marrow,48 and upregulate the acute phase protein hepcidin, which regulates serum iron levels and absorption of iron from the gut.49 Hepcidin binds to ferroportin, inducing receptor-ligand internalization and lysosomal degradation, and thereby sequestering iron within the enterocyte.50 Because absorptive enterocytes perform their function for only two days before being sloughed from the gut wall, iron locked into the cells is thought to be shed into the stool.51 Hepcidin-mediated inhibition of iron absorption would therefore not discriminate between heme and non-heme sources because the iron lost would come from the common intracellular pool. Atrophic body gastritis caused by H. pylori leads to lowered production of intrinsic factor, which is produced by parietal cells, and is necessary to bind cobalamin (vitamin B12) so that it can be absorbed in the ileum.52 Helicobacter pylori is implicated in pernicious anemia, low serum vitamin B12associated anemia, and low serum vitamin B12 in adults.53 It is possible that this loss could be overcome by increased frequency of consumption of vitamin B-12 from a meat diet rich

546

FARAG AND OTHERS

in foods containing heme iron. The oral administration of cobalamin is in fact a clinical option for treating foodcobalamin malabsorption.52 Helicobacter pylori induced destruction of parietal cells in the gastric body mucosa could also lead to decreased HCl output.54 Indeed, H.pylori-infected anemic patients have a dramatically higher gut pH.55 The consumption of foods high in heme iron could bypass the specific H. pylori-induced inhibition of free iron absorption. In a case series of H. pylori-associated refractory anemia, doses of radioactive 59Fe did not appear in the erythron or the stool, indicating that iron was sequestered in the intestinal lumen.56 Lactoferrin (LF), a free-iron sequestering member of the transferrin family regulated by the immune response to infection and secreted into external body fluids, has been found to support the in vitro growth of H. pylori,57 which produces an LF-binding protein that is expressed only in ironstarved media and is increased in biopsy specimens of anemic H. pylori-infected persons.58 Within H. pylori, iron is stored in the non-heme protein bacterial ferritin,59 which would be lost from the gut wall and shed into the feces during rapid bacterial turnover.58 Because iron originating from heme would not be captured by LF, it would be free to be absorbed into the enterocyte. Because they act in a manner that is biologically consistent with an effect modification by a diet high in foods containing heme iron, food-cobalamin malabsorption, inhibition of free iron absorption and bacterial capture, and shedding of free iron are supported by our observations. . However, gastrointestinal blood loss is not supported because iron lost from blood would be from a common corporal pool, and would therefore not distinguish between dietary iron source. Similarly, the anemia of chronic disease is not supported because hepcidin-induced inhibited iron efflux would not discriminate by the heme or non-heme origin or dietary iron. As a measure of urease activity in the stomach, DOB is increasingly recognized as a quantitative measure of H. pylori gastric bacterial load.60 For H. pylori-positive women, DOB was positively associated with Hb concentration and severely anemic women were observed to have lower DOB values than non-severely anemic women. Among all women, being infected by H. pylori with low bacterial load was associated with lower Hb, whereas high H. pylori bacterial load was positively associated with Hb concentration. Helicobacter pylori–induced pan-gastritis involving the gastric body can lead to metaplasia and gastric atrophy, gradually eliminating the very microecologic niche exploited by the bacterium and reducing bacterial load.61,62 In contrast, high bacterial load may reflect H. pylori infection in the context of a relatively intact gastric body with parietal cells that are capable of producing HCl and cobalamin. The high bacterial load may also enhance HCl production by overstimulating gastrin release from the antrum. Interestingly, two of the anemiagenic mechanisms that were supported by the observed effect modification by heme diet, food-cobalamin malabsorption and hampered reduction of ferric to ferrous iron, are caused by destruction of cells in the gastric body. We speculate that the negative association between low bacterial load H. pylori infection and Hb indicates that H. pylori causes anemia by inducing gastric body damage that results in food-cobalamin malabsorption and/or inhibited free iron absorption. We further speculate that high

bacterial load H. pylori infection enhances Hb concentration by leaving the gastric body intact and increasing the production of HCl, which promotes iron absorption. Randomized trials are needed to test whether treatment of H. pylori leads to improved hematologic outcome. Further research is needed to determine the potential role of H. pylori bacterial load in anemiagenesis. Because of the high worldwide prevalence of this infection, even a small associated risk for anemia would be of public health significance. If H. pylori is found to cause anemia, public health interventions must be found to address this infection on the population level, especially in developing countries where clinical options are limited. Received August 1, 2006. Accepted for publication November 24, 2006. Acknowledgments: We thank the women who participated in the study. Tamer H. Farag designed and implemented the study, interpreted the results, and wrote the manuscript. Tamer H. Farag and James M. Tielsch performed the analysis. Rebecca J. Stoltzfus and James M. Tielsch participated in study design, project planning, interpretation of results, and manuscript review. Sabra S. Khalfan participated in the design and implementation of the data collection and field management systems and manuscript review. As the senior investigator, James M. Tielsch oversaw all aspects of the study. Financial support: This study was supported by a grant from the Bill and Melinda Gates Foundation. Meretek Diagnostics, Inc. (Nashville, TN) loaned us the UBiT-IR300 portable infrared spectrophotometer and provided the 13C-labeled urea test kits at cost. Disclosure: None of the authors had a conflict of interest. Authors’ addresses: Tamer H. Farag and James M. Tielsch, Department of International Health, Bloomberg School of Public Health, Johns Hopkins University, 615 North Wolfe Street, Room W5009, Baltimore, MD, 21205, Telephone: 410-236-4280, Fax: 410-955-2029, E-mails: [email protected] and [email protected]. Rebecca J. Stoltzfus, Division of Nutritional Sciences, Cornell University, 120 Savage Hall, Ithaca, NY 14853, Telephone: 607-255-7671, Fax: 607-255-1033, E-mail: [email protected]. Sabra S. Khalfan, Public Health Laboratory–Ivo de Carneri, Chake-Chake, Pemba Island, Zanzibar, Tanzania, Telephone: 255-777-424690, E-mail: [email protected].

REFERENCES 1. Stoltzfus RJ, 2003. Iron deficiency: global prevalence and consequences. Food Nutr Bull 24: S97–S101. 2. Stoltzfus RJ, Mullany L, Black RW, 2005. Iron Deficiency Anaemia. Comparative Quantification of Health Risks: Global and Regional Burden of Disease Attributable to Selected Major Risk Factors. Geneva: World Health Organization. 3. Rush D, 2000. Nutrition and maternal mortality in the developing world. Am J Clin Nutr 72: 212S–240S. 4. Alonzo GM, Menendez C, Font F, Kahigwa E, Kimario J, Mshinda H, Tanner M, Bosch-Capblanch X, Alonso PL, 2000. Cost-effectiveness of iron supplementation and malaria chemoprophylaxis in the prevention of anaemia and malaria among Tanzanian infants. Bull World Health Organ 78: 97– 107. 5. Dreyfuss ML, Stoltzfus RJ, Shrestha JB, Pradhan EK, LeClerq SC, Khatry SK, Shrestha SR, Katz J, Albonico M, West KP Jr, 2000. Hookworms, malaria and vitamin A deficiency contribute to anemia and iron deficiency among pregnant women in the plains of Nepal. J Nutr 130: 2527–2536. 6. Go MF, 2002. Review article: natural history and epidemiology of Helicobacter pylori infection. Aliment Pharmacol Ther 16 (Suppl 1): 3–15. 7. NIH Consensus Conference, 1994. NIH consensus development panel on Helicobacter pylori in peptic ulcer disease. JAMA 272: 65–69. 8. Yip R, Limburg PJ, Ahlquist DA, Carpenter HA, O’Neill A, Kruse D, Stitham S, Gold BD, Gunter EW, Looker AC, Par-

H. PYLORI INFECTION AND SEVERE ANEMIA

9.

10.

11.

12.

13. 14.

15. 16.

17.

18.

19.

20.

21. 22. 23. 24. 25. 26. 27. 28.

kinson AJ, Nobmann ED, Petersen KM, Ellefson M, Schwartz S, 1997. Pervasive occult gastrointestinal bleeding in an Alaska native population with prevalent iron deficiency. Role of Helicobacter pylori gastritis. JAMA 277: 1135–1139. Parkinson AJ, Gold BD, Bulkow L, Wainwright RB, Swaminathan B, Khanna B, Petersen KM, Fitzgerald MA, 2000. High prevalence of Helicobacter pylori in the Alaska native population and association with low serum ferritin levels in young adults. Clin Diagn Lab Immunol 7: 885–888. Milman N, Rosenstock S, Andersen L, Jorgensen T, Bonnevie O, 1998. Serum ferritin, hemoglobin, and Helicobacter pylori infection: a seroepidemiologic survey comprising 2,794 Danish adults. Gastroenterology 115: 268–274. Berg G, Bode G, Blettner M, Boeing H, Brenner H, 2001. Helicobacter pylori infection and serum ferritin: A populationbased study among 1806 adults in Germany. Am J Gastroenterol 96: 1014–1018. Baggett HC, Parkinson AJ, Muth PT, Gold BD, Gessner BD, 2006. Endemic iron deficiency associated with Helicobacter pylori infection among school-aged children in Alaska. Pediatrics 117: e396–e404. Cardenas VM, Mulla ZD, Ortiz M, Graham DY, 2006. Iron deficiency and Helicobacter pylori infection in the United States. Am J Epidemiol 163: 127–134. Choe YH, Lee JE, Kim SK, 2000. Effect of Helicobacter pylori eradication on sideropenic refractory anaemia in adolescent girls with Helicobacter pylori infection. Acta Paediatr 89: 154– 157. Choe YH, Kwon YS, Jung MK, Kang SK, Hwang TS, Hong YC, 2001. Helicobacter pylori-associated iron-deficiency anemia in adolescent female athletes. J Pediatr 139: 100–104. Weyermann M, Rothenbacher D, Gayer L, Bode G, Adler G, Grab D, Flock F, Brenner H, 2005. Role of Helicobacter pylori infection in iron deficiency during pregnancy. Am J Obstet Gynecol 192: 548–553. Mahalanabis D, Islam MA, Shaikh S, Chakrabarty M, Kurpad AV, Mukherjee S, Sen B, Khaled MA, Varmund SH, 2005. Haematological response to iron supplementation is reduced in children with asymptomatic Helicobacter pylori infection. Br J Nutr 94: 969–975. Siekmann JH, Allen LH, Watnik MR, Nestel P, Neumann CG, Shoenfeld Y, Peter JB, Patnik M, Ansari AA, Coppel RL, Gershwin ME, 2003. Titers of antibody to common pathogens: relation to food-based interventions in rural Kenyan schoolchildren. Am J Clin Nutr 77: 242–249. Stoltzfus RJ, Chwaya HM, Montresor A, Albonico M, Savioli L, Tielsch JM, 2000. Malaria, hookworms and recent fever are related to anemia and iron status indicators in 0- to 5-y old Zanzibari children and these relationships change with age. J Nutr 130: 1724–1733. Roth DE, Taylor DN, Gilman RH, Meza R, Katz U, Bautista C, Cabrera L, Velapatino B, Lebron C, Razuri M, Watanabe J, Monath T, 2001. Posttreatment follow-up of Helicobacter pylori infection using a stool antigen immunoassay. Clin Diagn Lab Immunol 8: 718–723. Stone MA, 1999. Non-invasive testing for Helicobacter pylori. Postgrad Med J 75: 74–77. Vaira D, Vakil N, 2001. Blood, urine, stool, breath, money, and Helicobacter pylori. Gut 48: 287–289. von der WT, Honarvar N, Langhorne J, 1996. Gene-targeted mice lacking B cells are unable to eliminate a blood stage malaria infection. J Immunol 156: 2510–2516. Landis JR, Koch GG, 1977. The measurement of observer agreement for categorical data. Biometrics 33: 159–174. Hosmer DW, Lemeshow S, 2000. Applied Logistic Regression. New York: John Wiley & Sons, Inc. Weisberg S, 1985. Applied Linear Regression. New York: John Wiley & Sons, Inc. Consensus Panel, 1994. Helicobacter pylori in peptic ulcer disease. NIH Consens Statement 12: 1–23. Perri F, Clemente R, Pastore M, Quitadamo M, Festa V, Bisceglia M, Li BM, Lauriola G, Leandro G, Ghoos Y, Rutgeerts P, Andriulli A, 1998. The 13C-urea breath test as a predictor of intragastric bacterial load and severity of Helicobacter pylori gastritis. Scand J Clin Lab Invest 58: 19–27.

547

29. Lindkvist P, Asrat D, Nilsson I, Tsega E, Olsson GL, Wretlind B, Giesecke J, 1996. Age at acquisition of Helicobacter pylori infection: comparison of a high and a low prevalence country. Scand J Infect Dis 28: 181–184. 30. Nabwera HM, Nguyen-Van-Tam JS, Logan RF, Logan RP, 2000. Prevalence of Helicobacter pylori infection in Kenyan schoolchildren aged 3-15 years and risk factors for infection. Eur J Gastroenterol Hepatol 12: 483–487. 31. Fernando N, Holton J, Zulu I, Vaira D, Mwaba P, Kelly P, 2001. Helicobacter pylori infection in an urban African population. J Clin Microbiol 39: 1323–1327. 32. Glupczynski Y, Bourdeaux L, Verhas M, DePrez C, DeVos D, Devreker T, 1992. Use of a urea breath test versus invasive methods to determine the prevalence of Helicobacter pylori in Zaire. Eur J Clin Microbiol Infect Dis 11: 322–327. 33. Aoki K, Kihaile PE, Zhao WY, Zhang XH, Castro M, Disla M, Nyambo TB, Misumi J, 2005. Comparison of prevalence of chronic atrophic gastritis in Japan, China, Tanzania, and the Dominican Republic. Ann Epidemiol 15: 598–606. 34. Uyub AM, Raj SM, Visvanathan R, Nazim M, Aiyar S, Anuar AK, Mansur M, 1994. Helicobacter pylori infection in northeastern peninsular Malaysia: evidence for an unusually lowprevalence. Scand J Gastroenterol 29: 209–213. 35. Raj SM, Yap K, Haq JA, Singh S, Hamid A, 2001. Further evidence for an exceptionally low prevalence of Helicobacter pylori infection among peptic ulcer patients in north-eastern peninsular Malaysia. Trans R Soc Trop Med Hyg 95: 24–27. 36. Huang SS, Hassan AK, Choo KE, Ibrahim MI, Davis TM, 2004. Prevalence and predictors of Helicobacter pylori infection in children and adults from the Penan ethnic minority of Malaysian Borneo. Am J Trop Med Hyg 71: 444–450. 37. Goh KL, Parasakthi N, 2001. The racial cohort phenomenon: seroepidemiology of Helicobacter pylori infection in a multiracial South-East Asian country. Eur J Gastroenterol Hepatol 13: 177–183. 38. Tokudome S, Soeripto, Triningsih FX, Ananta I, Sadao S, Koyonori K, Susumu A, Kosaka H, Ishikawa H, Azuma T, Moore MA, 2005. Rare Helicobacter pylori infection as a factor for the very low stomach cancer incidence in Yogyakarta, Indonesia. Cancer Lett 219: 57–61. 39. Atisook K, Kachinthorn U, Luengrojanakul P, Tanwandee T, Pakdirat P, Puapairoj A, 2003. Histology of gastritis and Helicobacter pylori infection in Thailand: a nationwide study of 3776 cases. Helicobacter 8: 132–141. 40. Goodwin CS, 1997. Helicobacter pylori gastritis, peptic ulcer, and gastric cancer: clinical and molecular aspects. Clin Infect Dis 25: 1017–1019. 41. Gunshin H, Mackenzie B, Berger UV, Gunshin Y, Romero MF, Boron WF, Nussberger S, Gollan JL, Hediger MA, 1997. Cloning and characterization of a mammalian proton-coupled metal-ion transporter. Nature 388: 482–488. 42. Hallberg L, Bjorn-Rasmussen E, Howard L, Rossander L, 1979. Dietary heme iron absorption. A discussion of possible mechanisms for the absorption-promoting effect of meat and for the regulation of iron absorption. Scand J Gastroenterol 14: 769– 779. 43. Shayeghi M, Latunde-Dada GO, Oakhill JS, Laftah AH, Takeuchi K, Halliday N, Khan Y, Warley A, McCann FE, Hider RC, Frazer DM, Anderson GJ, Vulpe CD, Simpson RJ, McKie AT, 2005. Identification of an intestinal heme transporter. Cell 122: 789–801. 44. Raffin SB, Woo CH, Roost KT, Price DC, Schmid R, 1974. Intestinal absorption of hemoglobin iron-heme cleavage by mucosal heme oxygenase. J Clin Invest 54: 1344–1352. 45. Abboud S, Haile DJ, 2000. A novel mammalian iron-regulated protein involved in intracellular iron metabolism. J Biol Chem 275: 19906–19912. 46. Weiss G, 1999. Iron and anemia of chronic disease. Kidney Int Suppl 69: S12–S17. 47. Jelkmann W, 1998. Proinflammatory cytokines lowering erythropoietin production. J Interferon Cytokine Res 18: 555–559. 48. Davis D, Charles PJ, Potter A, Feldmann M, Maini RN, Elliott MJ, 1997. Anaemia of chronic disease in rheumatoid arthritis: in vivo effects of tumour necrosis factor alpha blockade. Br J Rheumatol 36: 950–956.

548

FARAG AND OTHERS

49. Nicolas G, Chauvet C, Viatte L, Danan JL, Bigard X, Devaux I, Beaumont C, Kahn A, Vaulont S, 2002. The gene encoding the iron regulatory peptide hepcidin is regulated by anemia, hypoxia, and inflammation. J Clin Invest 110: 1037–1044. 50. Nemeth E, Tuttle MS, Powelson J, Vaughn MB, Donovan A, Ward DM, Ganz T, Kaplan J, 2004. Hepcidin regulates cellular iron efflux by binding to ferroportin and inducing its internalization. Science 306: 2090–2093. 51. Ganz T, 2005. Hepcidin–a regulator of intestinal iron absorption and iron recycling by macrophages. Best Pract Res Clin Haematol 18: 171–182. 52. Bolaman Z, Kadikoylu G, Yukselen V, Yavasoglu I, Barutca S, Senturk T, 2003. Oral versus intramuscular cobalamin treatment in megaloblastic anemia: a single-center, prospective, randomized, open-label study. Clin Ther 25: 3124– 3134. 53. Perez-Perez GI, 1997. Role of Helicobacter pylori infection in the development of pernicious anemia. Clin Infect Dis 25: 1020– 1022. 54. Dale A, Thomas JE, Darboe MK, Coward WA, Harding M, Weaver LT, 1998. Helicobacter pylori infection, gastric acid secretion, and infant growth. J Pediatr Gastroenterol Nutr 26: 393–397. 55. Capurso G, Lahner E, Marcheggiano A, Caruana P, Carnuccio A, Bordi C, Delle FG, Annibale B, 2001. Involvement of the corporal mucosa and related changes in gastric acid secretion characterize patients with iron deficiency anaemia associated with Helicobacter pylori infection. Aliment Pharmacol Ther 15: 1753–1761.

56. Barabino A, Dufour C, Marino CE, Claudiani F, De Alessandri A, 1999. Unexplained refractory iron-deficiency anemia associated with Helicobacter pylori gastric infection in children: further clinical evidence. J Pediatr Gastroenterol Nutr 28: 116– 119. 57. Nakao K, Imoto I, Ikemura N, Shibata T, Takaji S, Taguchi Y, Misaki M, Yamauchi K, Yamazaki N, 1997. Relation of lactoferrin levels in gastric mucosa with Helicobacter pylori infection and with the degree of gastric inflammation. Am J Gastroenterol 92: 1005–1011. 58. Annibale B, Capurso G, Delle FG, 2003. The stomach and iron deficiency anaemia: a forgotten link. Dig Liver Dis 35: 288– 295. 59. Bereswill S, Greiner S, van Vliet AH, Waidner B, Fassbinder F, Schiltz E, Kusters JG, Kist M, 2000. Regulation of ferritinmediated cytoplasmic iron storage by the ferric uptake regulator homolog (Fur) of Helicobacter pylori. J Bacteriol 182: 5948–5953. 60. Kobayashi D, Eishi Y, Ohkusa T, Ishige I, Suzuki T, Minami J, Yamada T, Takizawa T, Koike M, 2002. Gastric mucosal density of Helicobacter pylori estimated by real-time PCR compared with results of urea breath test and histological grading. J Med Microbiol 51: 305–311. 61. Goodwin CS, 1997. Helicobacter pylori gastritis, peptic ulcer, and gastric cancer: clinical and molecular aspects. Clin Infect Dis 25: 1017–1019. 62. Chen X, Haruma K, Kamada T, Mihara M, Yoshihara M, Sumii K, Kajiyama G, 2000. Factors that affect results of the 13C urea breath test in Japanese patients. Helicobacter 5: 98–103.