McDermid, Joann M DipEpiBiostat, PhD*; Jaye, Assan DVM, PhD†; Schim van der Loeff, Maarten F MD, PhD†; Todd, Jim MSc‡; Bates, Chris DPhil§; Austin, Steve§; Jeffries, David PhD†; Awasana, Akum A BSc†; Whittle, Hilton C FMedSci†; Prentice, Andrew M PhD*
Understanding the basis for the wide variation in morbidity and mortality associated with HIV infection may improve clinical care and survival. Multiple studies have reported aberrations in iron status during HIV infection,1 and anemia is a strong and consistent predictor of morbidity and/or mortality independent of CD4 counts or viral load.2-4 However, considering iron status from only one direction-ie, insufficiency or anemia-may miss an important dimension of iron metabolism in HIV infection. Indeed, multiple lines of evidence suggest that elevated iron status is related to increased susceptibility to opportunistic infections and neoplasms, immunosuppression, and viral replication or mutation.1,5 Recently, a case-control study6 demonstrated that elevated iron status was associated with increased mortality among HIV-1-infected American women.
Iron status, with the exception of hemoglobin (Hb) concentration, is not assessed routinely or comprehensively in HIV. Identifying inexpensive and practical surrogate markers that predict progression or survival remains an important aspect of HIV clinical management because, unfortunately, regular monitoring of viral load and CD4 cells is economically prohibitive in some regions.7 Now, as antiretroviral medications become increasingly accessible worldwide, there are increased requirements to monitor HIV infection because the inability to adequately monitor HIV infection will continue to impair the overall quality of care attainable.8,9
In this study, the association between multiple indicators of iron status as predictors of all-cause mortality in a large, West African, HIV-seroprevalent clinical cohort was investigated. Relations are described for each indicator alone (to highlight its potential utility as a surrogate marker) and in the context of adjusting for multiple potential confounders (to provide insight into potential mechanisms).
The HIV Clinical Cohort (HIV-CC) is based at the Medical Research Council (MRC) Laboratories in The Gambia and recruits HIV-seroprevalent males and females (≥15 years; HIV seropositivity confirmed).10 As a national referral center for HIV infection, participants sought HIV testing either due to clinical signs of tuberculosis or sexually transmitted infections or because they were blood donors, commercial sex workers, or sexual partners of someone with HIV infection. Ethical procedures were followed in accordance with the ethical standards of the MRC/Gambian government, the London School of Hygiene and Tropical Medicine Ethics Committees, and the Helsinki Declaration of 1975, as revised in 2000. All subjects provided informed consent, were offered pre- and post-testing counseling, and were provided health care according to MRC guidelines. Symptomatic opportunistic infections were treated, and prophylactic treatment of opportunistic infections, including tuberculosis, was initiated in July 1999 when persons with an absolute CD4 count <500 cells/mm3 were offered prophylactic cotrimoxazole. At the time these data were collected, plasma viral load testing and antiretroviral treatment were not part of standard care, but Global Fund support for a national antiretroviral program was obtained in 2004.
HIV-CC subjects recruited from January 1, 1991, to December 31, 2001, were admissible if: ≥18 years (younger subjects were excluded due to the complexity of interpreting adolescent iron status in the context of a predominantly adult population); baseline (within 90 days of the first HIV-seropositive diagnosis) clinical data and blood samples were collected; archived baseline plasma was available for analyses; and, as a criterion for a complementary genetic study within the same subjects, a peripheral blood mononuclear cell or buffy coat sample had been archived.
The main outcome was defined as the time from the first HIV-seropositive diagnosis to all-cause mortality. This was ascertained at scheduled clinic appointments every 3 months, or when returning for symptomatic treatment. Subjects who failed to return were visited at their last known address and considered lost to follow-up (LFU) if their vital status was not determined. For these subjects, observation time was censored at the last date they were known to be alive, and for subjects who withdrew, at their date of refusal. The end of the cohort observation was June 1, 2002.
Demographic data collected from study questionnaires included self-reported ethnicity, gender, and age. Baseline immunologic, virologic, and clinical data were obtained from HIV-CC databases. Absolute CD4 cell counts were measured by FACScan (Becton-Dickinson, Oxford, UK). Serological diagnosis was performed on screened serum (Wellcozyme HIV 1 + 2 or ICEHIV-1.O.2; Murex Diagnostics, Dartford, UK) and, if reactive, retested by type-specific enzyme-linked immunosorbent assay (ELISA) (HIV-1: HIV recombinant-1, Murex; HIV-2: Wellcozyme HIV-2, Murex), and a line immunoassay (LIA) (PeptiLav 1 to 2; Sanofi Diagnostics Pasteur, Marne-la-Coquette, France). Hb was estimated as part of standard clinical screening according to guidelines at MRC Laboratories. Body mass index (BMI) was calculated as body weight divided by height squared (kg/m2). Malaria infection was considered a potential confounder a priori, but data on this event are not recorded because low malaria parasite burdens are such a common annual occurrence. To account for possible effects of malaria coinfection, malaria status was estimated by proxy, using the month the blood sample was obtained as an indicator of malaria risk. In coastal regions of The Gambia, maximum malaria transmission occurs between August and November,11 and for the current study an additional range of 2 months was added.
Baseline heparinized plasma stored between −20°C and −80°C was used for all other parameters. Quality assurance samples established from pooled plasma and commercial quality controls were used. For all assays, the coefficients of variation for the quality control samples varied from a low of 4.5% (medium iron concentrations) to a high of 23.1% (low ferritin concentrations); however, all were within acceptable ranges. Due to baseline plasma volume limitations, the following indicators of iron status were given equal priority:
Circulating concentrations of soluble transferrin receptor (sTfR) are proportional to the membrane-associated TfR, with expression positively correlated with erythroid proliferation and inversely correlated with tissue iron stores. sTfR was measured by ELISA based on the microplate sandwich enzyme immunoassay technique using 2 different monoclonal antibodies specific for sTfR (R&D Systems, Abingdon, UK; lower detection limit = 3; upper = 80 nmol/L) with automatic calculation using an MRX Revelation Microplate Reader (Dynex Labsystems, Middlesex, UK).
Plasma ferritin correlates with the total amount of mobilizable storage iron and was measured by ELISA utilizing 1 rabbit antiferritin antibody for solid phase immobilization and a mouse monoclonal antiferritin antibody in the antibody (horseradish peroxidase) conjugate solution. Test samples were allowed to react simultaneously with the antibodies, resulting in the ferritin molecules being sandwiched between the solid phase and enzyme-linked antibodies (IBL Immuno-biological Laboratories, Hamburg, Germany, with BioRad LyphoCheck Immunoassay Controls, Bio-Rad Laboratories, Hemel Hempstead, UK; lower detection limit = 5; upper = 1000 μg/L) with automatic calculation.
Plasma iron measures the number of atoms of iron bound to the iron transport protein transferrin (Tf) and is an indication of iron in transit from the reticuloendothelial system to the bone marrow. Iron was measured using an endpoint assay (ABX Diagnostics, Shefford, UK; lower detection limit = 1.0 μmol/L) on the Cobas Bio Centrifugal Biochemical Analyzer (Roche Diagnostics Ltd., Garden City, UK), where iron was released from Tf by guanidine hydrochloride and reduced to Fe2+ by ascorbic acid. In this assay, bivalent iron forms a red-colored chelate complex from FerroZine (Hach Chemical Co., Ames, IA), and to prevent copper interference, cupric ions were bound to thiourea.
Tf has a primary role in transporting ferric iron from sites of absorption or Hb breakdown to iron-requiring cells and iron storage sites. The Tf estimate was based on mixing plasma with an antibody solution, with the resulting immune complexes measured by turbidimetry (ABX Diagnostics, Shefford, UK; lower detection limit = 0.40 g/L) using the Cobas Bio.
Transferrin saturation (TS) is the ratio of plasma iron to the iron-binding capacity. It is possible for values to exceed 100% if there is non-Tf bound iron present in the plasma, possibly representing ferritin iron. TS was calculated by TS (%) = [iron (μmol/L) × 100]/[Tf (μmol/L) × 2].12
The transferrin-receptor-ferritin index (TfR:F), a parameter reported to be of additional value in the discrimination between iron deficiency anemia and anemia of chronic disease (ACD), was calculated by TfR:F = log (sTfR/ferritin).13 For sTfR and ferritin values that exceeded the upper limit of detection, a value of the upper limit plus 1 was imputed.
Given the potential confounding effect of clinical or subclinical infection, the acute phase response (APR) reactant α1-antichymotrypsin (ACT) was measured using the Cobas Bio by a nephelometric assay (DakoCytomation, Ely, UK; lower detection limit = 0.05 g/L) on the basis of the specific reaction between ACT and its antibody.
The iron status relationship was assessed in 2 parts. First, all subjects were categorized according to tertiles (Tf, TfR:F) or cut-off points as follows: TS was dichotomized according to the commonly reported value of 55% that is associated with elevated iron status.14,15 Ferritin was categorized into 3 groups that approximated cut-off points reported for an African population (ie, gender-nonspecific limits associated with iron-deficiency anemia, <12 μg/L; normal iron status, 12 to 300 μg/L; and iron overload, >300 μg/L16,17), plus a fourth group representing ferritin concentrations above the upper limit detection, >1000 μg/L. Then sTfR was categorized into lower-than-normal sTfR concentrations, <10.6 nmol/L; the normal reference range for Blacks living at low altitudes, 10.6 to 29.6 nmol/L;17,18 greater-than-normal concentrations, >29.9 to 80.0 nmol/L; plus a fourth group representing sTfR concentrations above the upper limit of detection, >80 nmol/L.
A second analysis was restricted to include only subjects classified as having elevated iron status compared to normal iron status subjects. Normal ranges were based on values reported for healthy subjects (not specific to Africans, unless specified),17-19 with the following exceptions: TfR:F, by calculating normal limits reported for sTfR and ferritin (gender- and age-specific);17,18 ferritin, by using the cut-off for iron deficiency anemia for the lower limit of normality and the cut-off for iron overload in men and women combined for the upper limit.16,17 Elevated iron status was defined according to values reported under conditions of iron overload and/or in medical conditions associated with ACD in which iron parameters are elevated.14,15,17,18 For Tf, values associated with iron overload/ACD have not been commonly reported in the literature, so values below the lower limit of the normal referent range19 were classified as elevated (ie, Tf concentration decreases as iron status increases). For TfR:F, a similar assay and reference range was not used in other reports; therefore, TfR:F limits were obtained using the elevated iron status values for sTfR and ferritin individually as follows: elevated iron status according to sTfR, <10.6 nmol/L; elevated iron status for ferritin for African men and women combined, >300 μg/L.17,18 Thus, elevated iron status according to TfR:F was set at <−1.45 = log10 (10.6/300) (ie, TfR:F decreases as iron status increases). These classifications are summarized in Table 1.
Survival analyses were conducted by examining mortality rates (MR) as well as unadjusted and adjusted Cox regression models. These models were based on the following assumptions: All effects were assessed using multiplicative models with differences estimated on a log scale; the outcome was estimated as survival time from the first HIV-seropositive diagnosis; the explanatory variables were estimated by plasma sTfR, iron, ferritin, Tf, TS, and/or TfR:F; and a priori confounding variables included age, gender, self-reported ethnicity, BMI,20 plasma ACT concentration, malaria status by proxy, absolute CD4 cell count, and HIV type (although malaria status and ethnicity were not observed to be strong confounders of these data and were excluded from multivariate models). Lastly, Hb, a known independent predictor of mortality, was examined in an unadjusted model to confirm a consistent association in these data; thereafter, multivariate models were adjusted for Hb.
Statistical analyses were conducted using Intercooled STATA 9.2 (STATA, College Station, TX) and SPSS 14.0 (SPSS, Chicago, IL).
Baseline characteristics for 1362 HIV-seropositive subjects are summarized in Table 2. Approximately half were female (52.7%), and 75% were between 18 and 40 years (median: 33 years; interquartile range [IQR] = 28 to 40). The mean BMI at entry was 18.5 kg/m2 (standard deviation [SD] ± 3.3) for men and 20.2 kg/m2 (SD ± 4.6) for women (P = 0.001), whereas the median ACT concentration (0.44 g/L; IQR = 0.33 to 0.68) was indicative of many subjects experiencing an APR (defined by ACT >0.40 g/L21). The predominant HIV type was HIV-1 (67.3%), and, at enrollment, many subjects were already experiencing considerable immunosuppression (median absolute CD4 cell count: 234 cells/mm3). This statistic alone did not reveal the considerable degree of variation in baseline absolute CD4 counts (IQR = 83 to 468 cells/mm3), nor differences by HIV type or gender. Subjects who were HIV-1 seropositive (absolute CD4 cells/mm3 mean ± SD: HIV-1, 304 ± 329; HIV-2, 435 ± 457; HIV-Dual, 489 ± 495; P < 0.001) or male (absolute CD4 cells/mm3 mean ± SD: male, 249 ± 276; female, 434 ± 437; P < 0.001) were more likely to enroll with a greater degree of immunosuppression.
Iron status at baseline was comprehensively estimated from 5 directly measured variables (sTfR, iron, Tf, ferritin, Hb) plus 2 multiparameter indices (TS, TfR:F) (Table 3). Summarizing the overall iron status based on the result of a single iron status parameter is difficult. For example, the data (mean ± SD or median [IQR]; P) for Hb (male, 99 g/L ± 25; female, 100 ± 22 g/L; 0.444) or iron (male, 8.3 μmol/L [5.4 to 12.5]; female, 8.6 μmol/L [6.0 to 12.8]; 0.403) suggest a population at risk of anemia. Conversely, Tf (male, 1.43 g/L ± 0.54; female, 1.81 ± 0.67 g/L; <0.001), TfR:F (male, −1.09 ± 0.60; female, −0.44 ± 0.87; <0.001) and ferritin (proportion high or very high ferritin: male, 62%; female, 38%; <0.001) suggest that elevated iron status is predominant. At the same time, TS (male, 31.0% [21.3 to 44.3]; female, 25.8% [17.3 to 37.4]; <0.001) and sTfR (proportion within normal range: male, 62.7%; female, 62.1%; 0.054) are indicative of normal iron status.
Over an 11.5-year observation period, the median survival time was 1.1 years (IQR = 0.2 to 3.1), with a maximum survival time of 9.8 years and LFU of 21.6%. In this cohort, 713 (52.3%) of subjects died, giving an overall MR of 25.9/100 person-years (PY) of cohort observation (95% CI = 24.1 to 27.8). As expected, MR differed by HIV type, with the greatest MR associated with HIV-1 (MR/100 PY [95% CI]: HIV-1, 31.7 [29.0 to 34.5]; HIV-2, 18.4 [16.0 to 21.1]; HIV-Dual, 13.2 [6.6 to 26.3]).
Table 4 summarizes the MR and the unadjusted and adjusted hazard ratio (HR) for each iron status indicator plus absolute CD4 counts separately. Examination of the MR data suggests that there are considerable differences in a dose-response manner for all iron status indices and absolute CD4 counts, with the largest effect observed for different ferritin concentrations (lowest vs. highest ferritin, MR/100 PY: <12 μg/L, 7.3 vs. 1000 μg/L, 121.1). The unadjusted Cox regression analyses mirrored these findings, and the majority of associations were statistically significant, with the exception of sTfR. In partially adjusted models that did not include Hb, it is evident that there is considerable reduction in the magnitude of the HR (compared to the unadjusted HR) for all iron status variables and for absolute CD4 counts. Despite the reduction, Tf, ferritin, TfR:F, Hb, and absolute CD4 counts remained significant predictors of all-cause mortality in HIV infection. When Hb was added to the models, the HR were slightly modified in either direction, but the interpretations remained unchanged. Overall, it can be summarized that all iron status indices, with the exception of sTfR, are significantly associated with mortality, and that Tf, ferritin, TfR:F and Hb are strong predictors of mortality in fully adjusted models.
Table 5 summarizes the iron status relationship for the subset of subjects with elevated iron status compared to subjects with normal iron status. The MR uniformly demonstrates that elevated iron status is associated with greater all-cause mortality, with the largest difference apparent for the combined iron status index (normal vs. highly elevated iron status MR: 5.6/100 PY vs. 88.3/100 PY). This is further supported by Kaplan-Meier survival curves (only selected shown) and log-rank tests (all analyses P < 0.001) for Tf concentration (Fig. 1A-B) and for the combined iron status index (Fig. 1C-D). Importantly, these results graphically demonstrate that there are clear differences in survival according to iron status, whether including all subjects from the time of HIV diagnosis or only including subjects surviving >2 years (ie, longer-term prediction). The unadjusted analyses for all iron status indicators universally support the hypothesis that elevated iron status is associated with a significantly greater probability of all-cause mortality compared to normal iron status, with the exception of sTfR (HR [95% CI]; P: elevated iron status versus normal referent range: 1.22 [0.89 to 1.67]; 0.224), which was not statistically significant. The greatest effect was observed when all iron status indicators were combined (combined iron status index, highly elevated vs. normal: 11.26 [7.36 to 17.24]; <0.001). Singly, the greatest magnitude of effects were apparent for Tf (elevated vs. normal: 4.32 [3.47 to 5.37]; <0.001), ferritin (elevated vs. normal: 3.57 [3.03 to 4.22]; <0.001); TfR:F (elevated vs. normal: 3.45 [32.97 to 4.02]; <0.001) and iron (elevated vs. normal: 3.07 [1.76 to 5.37]; <0.001).
In Table 5, adjusting for multiple confounders reduced sample sizes (due to missing data on baseline characteristics) and the magnitude of effects were diminished. Essentially, the partially adjusted models (without CD4) and the fully adjusted models (with CD4) produced similar results, demonstrating that elevated iron status is associated with increased mortality in HIV infection independent of absolute CD4 counts. In the partially adjusted models, elevated iron status as estimated by Tf, ferritin, TS, or the combined iron status index was associated with significantly greater mortality, with a trend observed for TfR:F. After adjustment for absolute CD4 cell counts, Tf (elevated vs. normal: 1.77 [1.30 to 2.42]; <0.001), ferritin (elevated vs. normal: 1.40 [1.07 to 1.83]; 0.014), and the combined iron status index (highly elevated vs. normal: 2.20 [1.16 to 4.18]; 0.016) remained statistically significant predictors of increased mortality, with trends apparent for TfR:F and TS.
This study demonstrates that elevated iron status estimated from multiple iron status indicators is associated with an increased probability of mortality in HIV-infected adults, and this effect is evident when considered with and without adjustment for multiple baseline characteristics. Like others,3,4,22 we show that lower Hb is associated with greater mortality. However, we also provide evidence that estimating Hb alone cannot reveal the complete picture of iron status in HIV infection.
Only limited evidence indicating that elevated iron status is associated with an increased probability of mortality in HIV infection has been reported. In a small HIV-seroincident cohort of thalassemia subjects, mean serum ferritin was associated with mortality,23 and de Monye and colleagues24 found that mortality was associated with higher degrees of bone marrow iron storage among antiretroviral-naive subjects. Early termination of a clinical trial of dapsone plus 60 mg iron protoxalate (vs. aerosolized pentamidine) for Pneumocystis jiroveci pneumonia was necessary due to increased mortality,25 and, more recently, a small case-control study of HAART-naive American women reported a modestly significant association between elevated iron status (TfR:log10 ferritin) and mortality.6 Like Gordeuk et al,6 observations from the current study did not indicate a strong association between sTfR and mortality. This may be a result of the dual correlation that exists between sTfR and tissue iron storage levels and the degree of erythropoiesis, with the latter possibly the stronger determinant of sTfR concentration.26 In addition, whereas most studies have concluded that sTfR concentration is unaffected by the APR,27 in populations with inflammation there is evidence that the ratio of sTfR to ferritin may be a better indicator of iron status than either alone.28 Finally, it is possible that sTfR is a better indicator of iron status under iron-deficient compared to iron-elevated circumstances.26,29,30 Building on the collective evidence from these studies, the current study both complements and extends these findings by using multiple iron status parameters to comprehensively estimate iron status; interpreting findings both in unadjusted models and in the context of multiple confounding variables; utilizing a cohort design with a large sample size to ascertain the temporal relationship of these associations over an extended period of observation; and including both genders at all stages of infection.
Several hypothetical mechanisms may explain why elevated iron status is associated with increased mortality in HIV infection.1 First, there may be an altered immune response, immunosuppression, or enhanced viral replication due to oxidative stress resulting from excess iron status. Second, any factor that increases viral replication could lead to selection of a more virulent or pathogenic HIV strain. Beck et al31 demonstrated that greater pathology due to oxidative stress was associated with significant increases in mRNA levels of proinflammatory Th2 response cytokines and chemokines and decreases in the anti-inflammatory cytokines interferon-γ and IL2 in a murine model infected with influenza. Third, elevated iron has been associated with a greater likelihood of certain opportunistic infections (Candida, Pneumocystis jiroveci, and Mycobacterium24,32,33) and neoplasms (Kaposi sarcoma34,35), possibly via an altered host-pathogen competition for essential iron.
Although the above mechanisms assume a direct role for iron, it is also possible that the markers of iron status estimated in this study are proxy indicators for a mechanism unrelated to the actions of iron itself. One such possibility is that these markers are reflecting the APR and therefore an association with clinical or subclinical infection processes. In this study, models were adjusted for the APR reactant ACT; however, the inclusion of only a single marker of APR (in this study, due to limited baseline plasma quantities) likely resulted in residual confounding of this complex association. It should also be noted that hepatocellular damage contributes to elevated serum/plasma iron status in the absence of increased iron stores. Whereas hepatic damage secondary to alcohol intake would be highly unlikely in this predominantly Muslim population (>90%) because alcohol abstinence is typical, liver tests were not performed as part of this study. Interpretation of these liver test data would be difficult due to the multiple etiologies in advanced HIV infection and in the absence of established reference values for the study population. Importantly, these uncertainties do not negate the potential utility of iron status markers as prognostic indicators of mortality in HIV infection; however, they do highlight the need for mechanistic studies designed to investigate these issues.
On the basis of the results from this study, it is apparent that estimating iron status has strong prognostic value in the clinical management of HIV infection, but would it have value additional to what is already being utilized? Although this question can be answered only partially in the absence of data on viral load, it could be argued that if information is available for only a single indicator in the absence of complete demographic and clinical data (ie, the unadjusted analyses), estimating ferritin or Tf concentration is more informative than absolute CD4 cell counts or Hb concentration alone. Though measuring Hb is common, the findings from this study indicate that monitoring iron status to identify subjects at both ends of the iron status spectrum is advisable. Ideally, as shown in this study, multiple indicators of iron status should be measured simultaneously to provide the most accurate indication of overall iron status. In resource-limited settings, this may be unrealistic; thus, based on the current study findings, ferritin and Tf would be the strongest individual predictors.
These findings also highlight the problem of defining what is best practice for the management of anemia in HIV infection. In resource-limited areas, preventative or corrective actions typically involve prescribing or self-supplementing with oral iron supplements, and in severe cases of anemia, blood transfusions. Is such a policy warranted, especially in areas with high HIV prevalence? It is if the confirmed etiology of anemia is due to dietary iron insufficiency or increased requirements secondary to pregnancy, lactation, or blood loss. In reality, limited resources do not allow investigation, and the etiology often remains unknown. A small Kenyan study has shown that iron supplementation did not increase plasma HIV-1 viral load;36 however, firm conclusions cannot be made on the basis of this single study using low-dose iron supplements (60 mg of elemental iron as ferrous dextran twice weekly for 4 months). In a recently reported phase 3 clinical trial restricted to American female injection drug users, daily micronutrients with iron (18 mg) did not increase HIV RNA levels over 1 year.37 Again, broad conclusions remain limited, given the restricted eligibility criteria and the likelihood this population was iron-deficient.38 In vitro evidence39 demonstrated that HIV-1 viral replication is affected by excess iron, and Moore et al40 has shown that blood transfusions compared to erythropoietin therapy are associated with accelerated mortality, even early in the course of HIV infection. Studies designed to address this clinical equipoise are needed, especially because the mechanism responsible for the elevated iron and mortality association may not be causally associated with viral load.
Limitations inherent in this study are common to retrospective cohort study designs. We were restricted to baseline data that had been collected with sufficient accuracy and completeness since 1991. Although it would have been desirable, detailed data on baseline clinical status, including subclinical infections, were unavailable. The proportion LFU was expected for this population and research setting, although it differed by iron status and baseline characteristics. Further investigation that compared models under the assumption that all subjects LFU did not experience the outcome and a hypothetical situation where all subjects LFU did experience the outcome (at the date of cohort exit) suggests these results were not strongly biased by LFU. Lastly, although there were very little missing data on the main predictors, missing data on baseline characteristics reduced the sample size for multivariate analyses.
In summary, these results indicate that iron status is a predictor of all-cause mortality among HIV-seropositive West African adults, and, for the first time in a large cohort of both men and women, elevated iron status is shown to be strongly associated with increased mortality. These associations are independent of the risk of dying associated with immunosuppression (absolute CD4 count), the APR related to clinical or subclinical infection (ACT), viral differences due to HIV type, and other known or hypothesized predictors of mortality (gender, age, BMI, Hb). Although these findings imply that iron status may be a valuable prognostic marker of mortality in HIV infection, they also caution there may be negative implications in providing supplemental iron or transfusion therapy for anemia of unknown etiology in HIV infection. Finally, this study has broader implications for understanding predictors of HIV mortality in a global context. The clinical situation of these subjects reflects that of the majority of people living with HIV today: a population without access to antiretroviral or prophylactic medications or extensive clinical investigations including routine viral load monitoring, and at risk of poverty-related illnesses unrelated to HIV infection.
We are grateful to the study participants and the clinical, data management, and laboratory staff for their contributions to the MRC HIV Clinical Cohort over the years and to those who assisted in this research.
1. McDermid JM, Prentice AM. Iron and infection: effects of host iron status and the iron-regulatory genes haptoglobin and NRAMP1 (SLC11A1) on host-pathogen interactions in tuberculosis and HIV. Clin Sci (Lond)
2. Sullivan PS, Hanson DL, Chu SY, et al. Epidemiology of anemia in human immunodeficiency virus (HIV)-infected persons: results from the multistate adult and adolescent spectrum of HIV disease surveillance project. Blood
3. Moore RD. Human immunodeficiency virus infection, anemia, and survival. Clin Infect Dis
4. Mocroft A, Kirk O, Barton SE, et al. Anaemia is an independent predictive marker for clinical prognosis in HIV-infected patients from across Europe. EuroSIDA study group. AIDS
5. Gordeuk VR, Delanghe JR, Langlois MR, et al. Iron status and the outcome of HIV infection: an overview. J Clin Virol
6. Gordeuk VR, Onojobi G, Schneider MF, et al. The association of serum ferritin and transferrin receptor concentrations with mortality in women with human immunodeficiency virus infection. Haematologica
7. Balakrishnan P, Solomon S, Kumarasamy N, et al. Low-cost monitoring of HIV infected individuals on highly active antiretroviral therapy (HAART) in developing countries. Indian J Med Res
8. Tucker T, Yeats J. Laboratory monitoring of HIV-after access to antiretroviral drugs, the next challenge for the developing world. S Afr Med J
9. Stephenson J. Cheaper HIV drugs for poor nations bring a new challenge: monitoring treatment. JAMA
10. Schim van der Loeff MF, Jaffar S, Aveika AA, et al. Mortality of HIV-1, HIV-2 and HIV-1/HIV-2 dually infected patients in a clinic-based cohort in The Gambia. AIDS
11. Hill AV, Allsopp CE, Kwiatkowski D, et al. Common west African HLA antigens are associated with protection from severe malaria. Nature
12. Milman N, Thomsen H, Mathiassen B. Serum ferritin, iron status and plasma ascorbic acid in 40- to 49-year- old males in the Faroe Islands. Scand J Clin Lab Invest
13. Malope BI, MacPhail AP, Alberts M, et al. The ratio of serum transferrin receptor and serum ferritin in the diagnosis of iron status. Br J Haematol
14. Mainous AG 3rd, Gill JM, Carek PJ. Elevated serum transferrin saturation and mortality. Ann Fam Med
15. Mainous AG 3rd, Wells B, Carek PJ, et al. The mortality risk of elevated serum transferrin saturation and consumption of dietary iron. Ann Fam Med
16. Khumalo H, Gomo ZA, Moyo VM, et al. Serum transferrin receptors are decreased in the presence of iron overload. Clin Chem
17. Gibson RS. Principles of Nutritional Assessment
. New York: Oxford University Press; 1990.
18. Allen J, Backstrom KR, Cooper JA, et al. Measurement of soluble transferrin receptor in serum of healthy adults. Clin Chem
19. Fields-Gardner C, Thomson CA, Rhodes SS. A Clinician's Guide to Nutrition in HIV and AIDS
. Chicago: The American Dietetic Association; 1997.
20. van der Sande MA, Schim van der Loeff MF, Aveika AA, et al. Body mass index at time of HIV diagnosis: a strong and independent predictor of survival. J Acquir Immune Defic Syndr
21. Gomo E, Vennervald BJ, Ndhlovu PD, et al. Reference values and predictors of white blood cell subset counts: a cross-sectional study among HIV seronegative pregnant women in Zimbabwe. Eur J Obstet Gynecol Reprod Biol
22. Sullivan P. Associations of anemia, treatments for anemia, and survival in patients with human immunodeficiency virus infection. J Infect Dis
. 2002;185(Suppl 2):S138-S142.
23. Salhi Y, Costagliola D, Rebulla P, et al. Serum ferritin, desferrioxamine, and evolution of HIV-1 infection in thalassemic patients. J Acquir Immune Defic Syndr Hum Retrovirol
24. de Monye C, Karcher DS, Boelaert JR, et al. Bone marrow macrophage iron grade and survival of HIV-seropositive patients. AIDS
25. Salmon-Ceron D, Fontbonne A, Saba J, et al. Lower survival in AIDS patients receiving dapsone compared with aerosolized pentamidine for secondary prophylaxis of Pneumocystis carinii pneumonia. J Infect Dis
26. Beguin Y. Soluble transferrin receptor for the evaluation of erythropoiesis and iron status. Clin Chim Acta
27. Cook JD. The measurement of serum transferrin receptor. Am J Med Sci
28. Punnonen K, Irjala K, Rajamaki A. Serum transferrin receptor and its ratio to serum ferritin in the diagnosis of iron deficiency. Blood
29. Halonen P, Mattila J, Suominen P, et al. Iron overload in children who are treated for acute lymphoblastic leukemia estimated by liver siderosis and serum iron parameters. Pediatrics
30. Worwood M. Serum transferrin receptor assays and their application. Ann Clin Biochem
31. Beck MA, Nelson HK, Shi Q, et al. Selenium deficiency increases the pathology of an influenza virus infection. FASEB J
32. Lounis N, Truffot-Pernot C, Grosset J, et al. Iron and Mycobacterium tuberculosis infection. J Clin Virol
33. Douvas GS, May MH, Crowle AJ. Transferrin, iron, and serum lipids enhance or inhibit Mycobacterium avium replication in human macrophages. J Infect Dis
34. Offermann MK, Lin JC, Mar EC, et al. Antioxidant-sensitive regulation of inflammatory-response genes in Kaposi's sarcoma cells. J Acquir Immune Defic Syndr Hum Retrovirol
35. Simonart T, Noel JC, Andrei G, et al. Iron as a potential co-factor in the pathogenesis of Kaposi's sarcoma? Int J Cancer
36. Olsen A, Mwaniki D, Krarup H, et al. Low-dose iron supplementation does not increase HIV-1 load. J Acquir Immune Defic Syndr
37. Semba R, Ricketts E, Mehta S, et al. Effect of micronutrients and iron supplementation on hemoglobin, iron status, and plasma hepatitis C and HIV RNA levels in female injection drug users: a controlled clinical trial. J Acquir Immune Defic Syndr
38. Semba RD, Shah N, Strathdee SA, et al. High prevalence of iron deficiency and anemia among female injection drug users with and without HIV infection. J Acquir Immune Defic Syndr
39. Traore HN, Meyer D. The effect of iron overload on in vitro HIV-1 infection. J Clin Virol
. 2004;31(Suppl 1):S92-S98.
40. Moore RD, Keruly JC, Chaisson RE. Anemia and survival in HIV infection. J Acquir Immune Defic Syndr Hum Retrovirol
© 2007 Lippincott Williams & Wilkins, Inc.