Assessment of iron status using plasma transferrin receptor in pregnant women with and without human immunode ciency virus infection in Malawi

Transcription

1 (2000) 54, 872±877 ß 2000 Macmillan Publishers Ltd All rights reserved 0954±3007/00 $ Assessment of iron status using plasma transferrin receptor in pregnant women with and without human immunode ciency virus infection in Malawi RD Semba 1 *, N Kumwenda 1, DR Hoover 2, TE Taha 1, L Mtimavalye 3, R Broadhead 3, W Eisinger 1, PG Miotti 4 and JD Chiphangwi 3 1 Departments of Ophthalmology and Epidemiology, Johns Hopkins University Schools of Medicine and Hygiene and Public Health, Baltimore, Maryland, USA; 2 Department of Statistics, Rutgers University, Piscattaway, New Jersey, USA; 3 Departments of Obstetrics and Gynaecology and Paediatrics and Child Health, College of Medicine, University of Malawi, Blantyre, Malawi; and 4 National Institute of Allergy and Infectious Diseases, Bethesda, Maryland, USA Background: Although anemia is highly prevalent during pregnancy and is common during human immunode ciency virus (HIV) infection, anemia and iron status have not been well characterized in HIV-infected pregnant women. Objective: To gain insight into iron status in HIV-infected pregnant women using plasma transferrin receptor and related indicators of anemia. Study design: Plasma transferrin receptor, ferritin, a 1 -acid glycoprotein, C-reactive protein and hemoglobin concentrations were measured in pregnant women, gestational age 18 ± 28 weeks, seen in an urban antenatal clinic in Blantyre, Malawi. Results: The prevalence of anemia among 662 HIV-positive and 190 HIV-negative pregnant women was 73.1% and 50.0%, respectively (P <0.0001). Among HIV-positive and HIV-negative women, median plasma transferrin receptor concentrations were 24.4 and 24.1 nmol=l (P ˆ 0.5), respectively, and median plasma ferritin concentrations were 17.8 and 20.8 mg=l (P <0.05), respectively. There was a large overlap in plasma transferrin receptor concentrations among women with and without anemia. Using the combination of hemoglobin and ferritin as a standard, the sensitivity and speci city of plasma transferrin receptor in diagnosing iron de ciency anemia was estimated at 45.9% and 68.1%, respectively. Conclusion: The use of plasma transferrin receptor concentrations as an indicator of iron de ciency anemia may be limited in pregnant women with chronic in ammation and infection. Sponsorship: The National Institutes of Health (HD32247, HD30042, HIVNET contract N01-AI ), the Fogarty International Center, and the United States Agency for International Development (Cooperative Agreement HRN-A ). Descriptors: pregnancy; iron; iron de ciency anemia; transferrin receptor; ferritin; hemoglobin; human immunode ciency virus; Malawi (2000) 54, 872±877 Introduction Iron de ciency is a major public health problem among pregnant women worldwide and is associated with adverse pregnancy outcome (Murphy et al, 1986; Scholl & Hediger, 1994; Allen, 1993). In developing countries, 35 ± 75% of pregnant women may be affected by anemia, (World Health Organization, 1992), and a large proportion of anemia is thought to be due to iron de ciency (Yip, 1997). Anemia is a common hematological complication of human immunode ciency virus (HIV) infection, and anemia has been implicated as a risk factor for mortality *Correspondence: Dr RD Semba, 550 North Broadway, Suite 700, Baltimore, MD 21205, USA. rdsemba@jhmi.edu Guarantor: RD Semba. Contributors: RD Semba, PG Miotti and JD Chiphangwi were responsible for the concept and planning of the study. Field work was supervised by N Kumwenda and TE Taha, and clinical work was carried out by L Mtimavalye and R Broadhead. D Hoover provided statistical expertise, and W Eisinger conducted laboratory analyses. All co-authors contributed to the writing of the paper. Received 3 February 2000; revised 25 August 2000; accepted 28 August 2000 during HIV infection (Moore, 1999). In some African cities, the proportion of pregnant women with HIV infection found in antenatal clinics has reached 20 ± 40% (UNAIDS, 1998). Although anemia is common in HIVinfected pregnant women, it is unclear what proportion of anemia might be attributed to iron de ciency. The diagnosis of iron de ciency is often made using hemoglobin and plasma ferritin, as low plasma ferritin concentrations are considered to be consistent with the depletion of body iron stores (Cook et al, 1974). In the presence of infection and in ammation, plasma ferritin concentrations can rise during the acute phase response, (Mast et al, 1998) making it dif cult to diagnose iron de ciency based upon ferritin and hemoglobin alone. Serum or plasma transferrin receptor concentrations are considered a sensitive indicator of tissue iron availability, (Ahluwalia, 1998) and studies have suggested that serum or plasma transferrin receptor is not affected by chronic disease, liver disease, infection, malnutrition or in ammation (Ferguson et al, 1992; Nielson et al, 1994; Kuvibidila et al, 1996). Serum or plasma transferrin receptor concentrations have been reported to be a sensitive and speci c marker of iron de ciency during pregnancy, (Carriaga et al,

2 1991; A Ê kesson et al, 1998) although this indicator has not been examined widely in populations at high risk for iron de ciency. The aims of the present study were to characterize plasma transferrin receptor, ferritin and hemoglobin concentrations in a group of pregnant women during the second trimester of pregnancy in Malawi and to gain further insight into anemia in pregnant women with and without HIV infection. Subjects and methods Subjects Pregnant women from 18 to 28 weeks gestation were seen at the antenatal clinic of the Queen Elizabeth Central Hospital in Blantyre, Malawi, from November 1995 through December The Queen Elizabeth Central Hospital is the main hospital for Blantyre, a city of approximately inhabitants. Pregnant women received instructions in prenatal care, AIDS education, HIV testing, pre- and post-test HIV counseling, physical examination and treatment for sexually transmitted diseases and malaria. Height and weight were recorded, and gestational age was estimated using the last reported menstrual period. A venous blood samples were drawn at screening and one week later at enrollment. All women received daily oral iron and folate from enrollment until delivery. Half the women were randomized to receive daily oral vitamin A, 3 mg retinol equivalent per day from enrollment until delivery (Semba, 1997). The study of anemia was done within the context of a clinical trial of vitamin A supplementation during pregnancy to improve maternal and infant health outcomes (Semba, 1997). Written, informed consent was obtained from all study participants. The study protocol was approved by four different independent ethical review committees: the Johns Hopkins School of Medicine; the Malawi Health Sciences Research Committee, Government of Malawi; the National Cancer Institute, National Institutes of Health; and the Ministry of Health and Population, Government of Malawi. Final approval was given by the Of ce for Protection from Research Risk, National Institutes of Health, Bethesda, Maryland. Laboratory methods Serum was separated for the presence of HIV-1 antibody using enzyme-linked immunosorbent assay (EIA) (Wellcozyme, Wellcome Diagnostics, Dartford, Kent, UK, and Genetic Systems EIA, Seattle, WA, USA). Both EIAs were required to be positive for a woman to be considered HIV- 1-positive. Immunoblotting (Bio-Rad Laboratories, Hercules, CA, USA) was used to con rm HIV-1 status in women with equivocal HIV-1 testing by EIA. One week later, blood samples were drawn by venipuncture, and plasma was separated immediately and frozen at 770 C. A complete blood count including hemoglobin was performed using an automated cell counter (Coulter, Hialeah, FL, USA). Anemia was de ned as a hemoglobin concentration <110 g=l, and severe anemia was de ned as hemoglobin concentration <70 g=l, according to standard guidelines. (World Health Organization, 1993). Intermediate cut-off points for hemoglobin were given in order to provide more descriptive information about the lower end of the distribution of hemoglobin level (Stoltzfus, 1997). Plasma was kept frozen at 770 C until time of the analyses. Pooled human plasma was analyzed in each run of the laboratory analyses in order to assess between-assay coef cient of variation (CV). Plasma ferritin was measured by enzyme-linked immunosorbent assay (Human Ferritin Enzyme Immunoassay Test Kit, American Laboratory Products Company, Windham, NH, USA) with a CV ˆ 5.7%. Plasma ferritin <12 mg=l was considered consistent with iron de ciency (Ali et al, 1978). Because some recent studies suggest that ferritin <30 mg=l may be a better indicator for iron storage (Mast et al, 1998; Van den Broek 1998a), this cut-off point was also included in the analyses. Plasma transferrin receptor was measured using an enzyme-linked immunosorbent assay (Quantikine IVD, Human stfr Immunoassay, R & D Systems, Minneapolis, MN, USA) with a between-assay CV ˆ 5.8%. Plasma transferrin receptor concentration was considered greater than normal range if >28.1 nmol=l, according to manuacturer's guidelines. Plasma a 1 -acid glycoprotein (AGP) was measured using radial immunodiffusion (Bindarid, The Binding Site, Birmingham, UK) with a betweenassay CV ˆ 3.5%. C-reactive protein (CRP) was measured by enzyme-linked immunosorbent assay (Virgo CRP 150, Hemagen Diagnostics, Waltham, MA, USA) with a between-assay CV ˆ 4.1%. Plasma AGP was considered to be elevated if >1 g=l (Semba et al, 2000, Christian et al, 1998) and plasma CRP was considered to be elevated if >5mg=l (Semba et al, 2000). Women were de ned in this paper as having elevated acute phase protein if they had elevated plasma AGP and=or CRP levels. In HIV-infected women, plasma HIV-1 load was measured using quantitative reverse transcriptase polymerase chain reaction (Roche Amplicor Monitor, Branchberg, NJ, USA), and these assays were run and validated in the AIDS Clinical Trials Group reference laboratory at Johns Hopkins Hospital (Dr Brooks Jackson). Statistical analysis Comparisons between groups were made using Wilcoxon rank-sum test for continuous variables. Chi-square tests or Fisher's exact tests were used to compare categorical variables between groups. Spearman correlation was used to examine the relationship between selected variables. Results The study included 900 pregnant women (697 HIV-positive and 203 HIV-negative) in the second trimester of pregnancy. The median age (25th, 75th percentiles) of the 697 HIV-positive women was 23 (20, 26) y compared with 22 (20, 27) y in the 203 HIV-negative women (P ˆ 0.31). Median gestational age (25th, 75th percentiles) of HIVpositive women was 24 (20, 26) compared with 24 (20, 26) in HIV-negative women (P ˆ 0.92). Body mass index (wt=ht 2 ; 25th, 75th percentiles) was 22.3 (20.8, 23.8) in HIV-positive women compared with 22.7 (21.3, 24.1) in HIV-negative women (P <0.02). Among HIV-positive women, median HIV load (25th, 75th percentiles) was (8 163, ) copies=ml. The analysis of hemoglobin and iron indicators was restricted to the 662 HIV-positive women and 190 HIV-negative women who had both hemoglobin and plasma transferrin receptor measurements. Median hemoglobin concentrations were signi cantly lower among HIV-positive compared with HIV-negative pregnant women, and the prevalence of anemia was higher among HIV-positive than HIV-negative pregnant women (Table 1). The median concentrations of the acute phase 873

3 874 Table 1 Indicators of iron and nutritional status and acute phase proteins in pregnant women in Malawi a Variable HIV-positive women HIV-negative women Hemoglobin (g=l) 102 (92, 110) [662] 109 (100, 118) [190] b Hemoglobin <110 g=l (%) 73.1 [662] 50.0 [190] c Hemoglobin <100 g=l (%) 42.6 [662] 24.2 [190] c Hemoglobin <90 g=l (%) 20.1 [662] 8.4 [190] c Hemoglobin <80 g=l (%) 7.5 [662] 2.1 [190] d Hemoglobin <70 g=l (%) 1.9 [662] 0.5 [190] Ferritin mg=l 17.8 (8.8, 34.9) [657] 20.8 (11.0, 46.6) [187] d Ferritin <12 mg=l (%) 34.8 [657] 28.3 [187] Ferritin <30 mg=l (%) 68.9 [657] 60.4 [187] d Hemoglobin <110 g=l and Ferritin <12 mg=l (%) 27.5 [657] 17.1 [187] d Hemoglobin <110 g=l and Ferritin <30 mg=l (%) 49.6 [657] 31.5 [187] c Transferrin receptor (nmol=l) 24.4 (18.8, 33.7) [662] 24.1 (19.7, 31.0) [190] Transferrin receptor >28.1 nmol=l (%) e 38.7 [662] 31.9 [190] a 1 -acid glycoprotein (g=l) 0.75 (0.58, 0.92) [643] 0.52 (0.43, 0.61) [190] b a 1 -acid glycoprotein >1g=l (%) 19.7 [643] 2.1 [188] c C-reactive protein (mg=l) 5.9 (2.5, 12.7) [643] 2.5 (1.4, 6.5) [188] b C-reactive protein >5mg=l (%) 56.7 [643] 30.4 [188] b a Median (25th, 75th percentiles) for continuous variables; n in brackets. b,c,d Signi cantly different between the two groups: b P <0.0001; c P <0.001; d P <0.05 (Wilcoxon rank sum test for continuous variables, chi-square for categorical variables). e Plasma transferrin receptor concentration considered above normal range (Quantikine IVC, Human stfr Immunoassay, R&D Systems, Minneapolis, MN). Figure 1 Frequency distribution of plasma transferrin receptor concentrations in HIV-positive and HIV-negative pregnant women in Malawi (Mantel-Haenszel chi-square, P ˆ 0.12). proteins, AGP and CRP, were higher among HIV-positive compared with HIV-negative pregnant women. Median plasma transferrin receptor concentrations were not signi cantly different between HIV-positive and HIV-negative pregnant women, and the frequency distribution of plasma transferrin receptor concentrations did not differ between these two groups (Figure 1). Indicators of iron status were compared between HIVpositive and HIV-negative women, excluding women who had elevated acute phase proteins (AGP >1g=l and=or CRP >5mg=l), because ferritin is a positive acute phase reactant and cannot be reliably used to indicate depletion of iron stores during an acute phase response. Median hemoglobin levels were signi cantly lower and the prevalence of anemia was signi cantly higher among HIV-positive than HIV-negative women (Table 2). Median plasma ferritin concentrations were signi cantly lower among HIV-positive women compared with HIV-negative women. There were no signi cant differences in plasma transferrin receptor concentrations between HIV-positive and HIV-negative women, although there was a higher proportion of HIVpositive women than HIV-negative women with plasma transferrin receptor concentrations above the normal range. The concentrations of plasma transferrin receptor were compared with plasma ferritin and hemoglobin concentrations, excluding women who had elevated acute phase proteins. The women were divided into four groups: (1) hemoglobin <110 g=l and ferritin <12 mg=l, considered to represent iron de ciency anemia; (2) hemoglobin <110 g=l and ferritin >12 mg=l, considered to represent anemia with normal iron stores; (3) hemoglobin >110 g=l and ferritin >12 mg=l, considered to represent depletion of iron stores without impaired erythropoiesis; and (4) hemoglobin >110 g=l and ferritin >12 mg=l, considered to represent no anemia and normal iron status (Ahluwalia, 1998). The medians, quartiles and range of plasma transferrin receptor concentrations suggest a large overlap between groups, even between women in group 1 with iron de ciency anemia and women in group 4 without anemia and with normal iron status (Figure 2a). Medians, quartiles, and 5th and 95th percentiles of plasma transferrin receptor concentrations are also shown for women in four groups, as above, except that a cut-off of ferritin <30 mg=l was used (Figure 2b). Among women without elevated acute phase proteins, there was considerable overlap in plasma transferrin receptor concentrations between women with iron de ciency anemia (hemoglobin <110 g=l and ferritin <12mg=l) and women with no anemia and normal iron status (hemoglobin >110g=l and ferritin >12 mg=l), and no clearly discernable cut-off for plasma transferrin receptor was noted between these two groups. To make provisional estimations of sensitivity and speci city of plasma transferrin receptor as an indicator of iron de ciency anemia, a cut-off of 28.1 nmol=l was used, as this represents the 97.5 percentile of plasma transferrin receptor in a normal reference population as provided by the manufacturer (Quantikine IVC, Human stfr Immunoassay, R&D Systems, Minneapolis, MN, USA). For this estimation, a `gold standard' for iron de ciency anemia was de ned as hemoglobin <110 g=l and ferritin <12 mg=l among women with-

4 Table 2 Indicators of iron status in pregnant women in Malawi, excluding individuals with elevated acute phase proteins a 875 Variable b HIV-positive women HIV-negative women Hemoglobin (g=l) 103 (94, 111) [267] 108 (101, 118) [130] c Hemoglobin <110 g=l (%) 70.0 [267] 52.3 [130] d Hemoglobin <100 g=l (%) 38.2 [267] 23.8 [130] e Hemoglobin <90 g=l (%) 14.6 [267] 9.2 [130] Hemoglobin <80 g=l (%) 3.3 [267] 1.5 [130] Hemoglobin <70 g=l (%) 1.5 [296] 0.77 [130] Ferritin mg=l 14.1 (7.6, 28.8) [266] 17.5 (9.3, 41.8) [128] d Ferritin <12 mg=l (%) 41.3 [266] 34.3 [128] Ferritin <30 mg=l (%) 76.3 [266] 64.0 [128] e Hemoglobin <110 g=l and ferritin <12 mg=l (%) 31.9 [266] 21.0 [128] e Hemoglobin <110 g=l and ferritin <30 mg=l (%) 54.8 [266] 35.1 [128] d Transferrin receptor (nmol=l) 25.2 (18.8, 32.8) [267] 23.5 (19.7, 29.9) [130] Transferrin receptor >28.1 nmol=l (%) f 38.5 [267] 29.3 [130] g a Excluding women with AGP >1g=l and=or CRP >5mg=l. b Median (25th, 75th percentiles) for continuous variables; n in brackets. c,d,e Signi cantly different between the two group: c P <0.0001; d P <0.001; e P <0.05 (Wilcoxon rank sum test for continuous variables, chi-square for categorical variables). f Plasma transferrin receptor concentration considered above normal range (Quantikine IVC, Human stfr Immunoassay, R & D Systems, Minneapolis, MN). g w 2 test, P ˆ out elevated acute phase proteins. Using this de nition, the sensitivity and speci city of plasma transferrin receptor in diagnosis of iron de ciency anemia was 45.9% and 68.1%, respectively. Alternatively, if iron de ciency anemia were de ned as hemoglobin <110 g=l and ferritin <30 mg=l among the same women, the sensitivity and speci city of plasma transferrin receptor in diagnosis of iron de ciency was 45.7% and 69.2%. Spearman correlations were calculated between plasma transferrin receptor and ferritin concentrations including all women (r c ˆ ± 0.08, P <0.02), and excluding women with elevated acute phase proteins (r c ˆ ± 0.11, P <0.02). Among HIV-positive women, hemoglobin was negatively correlated with transferrin receptor (r c ˆ ± 0.41, P <0.0001), AGP (r c ˆ , P <0.0001), CRP (r c ˆ ± 0.13, P <0.0001), and plasma HIV load (r c ˆ ± 0.11, P <0.0001), respectively. Ferritin was positively correlated with AGP (r c ˆ 0.33, P <0.0001) and CRP (r c ˆ 0.17, P <0.0001), respectively. Discussion Figure 2 Transferrin receptor concentrations in pregnant women in Malawi, using ferritin cut-off of 12 mg=l (a) and 30 mg=l (b). Women with elevated acute phase proteins (AGP >1 g=l and=or CRP >5mg=l) excluded. Box plots indicate median, 25th and 75th percentiles. Vertical lines indicate range. This study suggests that the prevalence of anemia is high among pregnant women in urban and periurban Blantyre, Malawi, affecting nearly three-quarters of HIV-positive women and one-half of HIV-negative women during the second trimester. These ndings are consistent with another study in the same study population which reported a high prevalence of anemia and a higher risk of anemia noted among HIV-positive women (Van den Broek et al, 1998b). In rural Malawi, anemia appears to be even more prevalent, as it was recently reported among over 90% of pregnant women (Verhoeff et al, 1999). Severe anemia (hemoglobin < 70 g=l) occurred in only 2% of HIV-positive and 0.5% of HIV-negative pregnant women, which is lower than the prevalence reported among pregnant women in rural Malawi and in Nepal, but higher than the prevalence noted among pregnant women in Shanghai and Peru (Stoltzfus, 1997; Verhoeff et al, 1999). Of the women with anemia in the present study, it appears that about one-third to one-half were anemic because of iron de ciency, as re ected by use of plasma ferritin and hemoglobin concentrations as indicators. Because plasma ferritin

5 876 can be elevated during the acute phase response, this estimate of the proportion of anemic women who had iron de ciency is probably a conservative estimate. Overall, median plasma transferrin receptor concentrations did not differ signi cantly between HIV-positive and HIV-negative pregnant women, nor was the frequency distribution of plasma transferrin receptor concentrations signi cantly different by HIV status. These data suggest that HIV infection does not adversely in uence tissue iron availability as re ected by plasma transferrin receptor concentrations. Although other studies have shown that plasma transferrin receptor is a speci c marker of iron de ciency during pregnancy, (Carriaga et al, 1991; A Ê kesson et al, 1998), this study shows that there was a great deal of overlap in the distribution of plasma transferrin receptor concentrations between pregnant women with and without anemia. Overlap in serum transferrin receptor concentrations has also been described among lactating women with and without anemia in Zaire (Kuvibidila et al, 1994). In a study in which bone marrow aspirates were used as the standard for comparison, serum ferritin with a cut-off point of 30 mg=l was found to be the best single indicator of storage iron, whereas serum transferrin receptor had lower sensitivity and speci city (Van den Broek et al, 1998a). A limitation of the present study is that a combination of hemoglobin and ferritin was used as the standard for comparison, instead of bone marrow biopsy, which is generally considered the gold standard. Bone marrow biopsy is invasive and expensive (Ahluwalia, 1998), and a battery of laboratory tests, including hemoglobin, ferritin, erythrocyte protoporphyrin, transferrin saturation and serum iron are often used to diagnose iron de ciency anemia. Most of these indicators are affected by in ammation and the acute phase response, making the diagnosis dif cult (Ahluwalia, 1998). In sub-saharan Africa, the most common causes of anemia are considered to be iron de ciency, folate de ciency, malaria, sickle-cell disease and AIDS (World Health Organization, 1992). Among anemic women seen in antenatal clinics in Dar-es-Salaam, Tanzania, iron de ciency and malaria were the two leading causes of anemia (Massawe et al, 1999). During HIV infection, the etiology of anemia is often dif cult to determine, as HIV infection itself, malnutrition, malignancies, infections and hemolysis have all been identi ed as factors which may contribute to anemia (Kreuzer & Rockstroh, 1997). Malaria is endemic in the study population, and a limitation of this study is that blood smears were not taken in all women to assess possible malaria parasitemia. The role of the anemia of chronic disease (Means & Krantz, 1992) in the pathogenesis of anemia in HIV-infected women still remains a major gap in knowledge. There were signi cant negative correlations between hemoglobin concentrations and acute phase proteins, and these may re ect an association between anemia and the in ammation associated with infectious diseases. Hemoglobin concentrations also had a negative correlation with plasma HIV load, suggesting that more advanced HIV disease is associated with lower hemoglobin concentrations. Plasma ferritin and acute phase proteins had a strong positive correlation, which is consistent with the observation that plasma ferritin is a positive acute phase reactant (Mast et al, 1998). Plasma transferrin receptor had relatively poor sensitivity and speci city in diagnosing iron de ciency anemia in this group of women from sub-saharan Africa. These ndings are contrary to earlier studies which suggested that plasma transferrin receptor was not affected by the anemia of chronic disease (Ferguson et al, 1992; Nielson et al, 1994; Kuvibidila et al, 1996). Further studies are needed to identify additional laboratory indicators of iron de ciency anemia in groups of individuals who have a heavy burden of concurrent infection. Acknowledgements ÐWe thank the mothers who participated in this study, the staff of the Johns Hopkins Project and Queen Elizabeth Central Hospital, the Malawi Health Sciences Research Committee, Anne Willoughby and Robert Nugent, National Institute for Child Health and Human Development, and Kenneth Bridbord, Fogarty International Center, the National Institutes of Health, Brooks Jackson, Department of Pathology, the Johns Hopkins School of Medicine. References Ahluwalia N (1998): Diagnostic utility of serum transferrin receptors measurement in assessing iron status. Nutr. Rev. 56, 133 ± 141. AÊ kesson A, Bjellerup P, Berglund M, Bremme K & Vahter M (1998): Serum transferrin receptor: a speci c marker of iron de ciency in pregnancy. Am. J. Clin. Nutr. 68, 1241 ± Ali MAM, Luxton AW & Walker WHC (1978): Serum ferritin concentration and bone marrow iron stores: a prospective study. Can. Med. Assoc. J. 118, 945 ± 946. Allen LH (1993): Iron-de ciency anemia increases risk of preterm delivery. Nutr. Rev. 51, 49 ± 52. Carriaga MT, Skikne BS, Finley B, Cutler B & Cook JD (1991): Serum transferrin receptor for the detection of iron de ciency in pregnancy. Am. J. Clin. Nutr. 54, 1077 ± Christian P, Schulze K, Stoltzfus RJ & West KP Jr (1998): Hyporetinolemia, illness symptoms, and acute phase protein response in pregnant women with and without night blindness. Am. J. Clin. Nutr. 67, 1237 ± Cook JD, Lipschitz DA, Miles LEM & Finch CA (1974): Serum ferritin as a measure of iron stores in normal subjects. Am. J. Clin. Nutr. 27, 681 ± 687. Ferguson BJ, Skikne BS, Simpson KM, Baynes RD & Cook JD (1992): Serum transferrin receptor distinguishes the anemia of chronic disease from iron de ciency anemia. J. Lab. Clin. Med. 119, 385 ± 390. Kreuzer KA & Rockstroh JK (1997): Pathogenesis and pathophysiology of anemia in HIV infection. Ann. Hematol. 75, 179 ± 187. Kuvibidila S, Yu LC, Ode DL, Warrier RP & Mbele V (1994): Assessment of iron status of Zairean women of childbearing age by serum transferrin receptor. Am. J. Clin. Nutr. 60, 603 ± 609. Kuvibidila S, Warrier RP, Ode D & Yu L (1996): Serum transferrin receptor concentrations in women with mild malnutrition. Am. J. Clin. Nutr. 63, 596 ± 601. Massawe SN, Urassa EN, Mmari M & Ronquist G (1999): The complexity of pregnancy anemia in Dar-es-Salaam. Gynecol. Obstet. Invest. 47, 76 ± 82. Mast AE, Blinder MA, Gronowski AM, Chumley C & Scott MG (1998): Clinical utility of the soluble transferrin receptor and comparison with serum ferritin in several populations. Clin. Chem. 44, 45 ± 51. Means RT Jr & Krantz SB (1992): Progress in understanding the pathogenesis of the anemia of chronic disease. Blood 80, 1639 ± Moore RD (1999): Human immunode ciency virus infection, anemia, and survival. Clin. Infect. Dis. 29, 44 ± 49. Murphy JF, O'Riordan J, Newcombe RG, Coles EC & Pearson JF (1986): Relation of haemoglobin levels in rst and second trimesters to outcome of pregnancy. Lancet 1, 992 ± 994. Nielson OJ, Andersen LS, Hansen NE & Hansen TM (1994): Serum transferrin receptor levels in anaemic patients with rheumatoid arthritis. Scand. J. Clin. Lab. Invest. 54, 75 ± 82. Scholl TO & Hediger ML (1994): Anemia and iron-de ciency anemia: compilation of data on pregnancy outcome. Am. J. Clin. Nutr. 59, 492S ± 501S. Semba RD (1997): An overview of the potential role of vitamin A in mother-to-child transmission of HIV-1. Acta. Paediatr. Suppl. 421, 107 ± 112. Semba RD, Muhilal, West KP Jr, Natadisastra G, Eisinger W, Lan Y & Sommer A (2000): Hyporetinolemia and acute phase proteins in children with and without xerophthalmia. Am. J. Clin. Nutr. 72, 146 ± 153.

6 Stoltzfus RJ (1997): Rethinking anaemia surveillance. Lancet 349, 1764 ± UNAIDS (1998): AIDS epidemic update: December Joint United Nations Programme on HIV=AIDS and the World Health Organization, Geneva. Van den Broek NR, Letsky EA, White SA & Shenkin A (1998a): Iron status in pregnant women: which measurements are valid? Br. J. Haematol. 103, 817 ± 824. Van den Broek NR, White SA & Neilson JP (1998b): The relationship between asymptomatic human immunode ciency virus infection and the prevalence and severity of anemia in pregnant Malawian women. Am. J. Trop. Med. Hyg. 59, 1004 ± Verhoeff FH, Brabin BJ, Chimsuku L, Kazembe P & Broadhead RL (1999): An analysis of the determinants of anaemia in pregnant women in rural Malawi Ð a basis for action. Ann. Trop. Med. Parasitol. 93, 119 ± 133. World Health Organization (1992): The prevalence of anaemia in women: a tabulation of available information. Document WHO=MCH=MSM= World Health Organization (1993): Prevention and management of severe anaemia in pregnancy. Document WHO=FHE=MSM=93.5. Yip R (1997): The challenge of improving iron nutrition: limitations and potentials of major intervention approaches. Eur. J. Clin. Nutr. 51(Suppl 4), S16 ±