点击显示 收起
1 From the Gertrude H Sergievsky Center, College of Physicians and Surgeons, and the Department of Epidemiology, Mailman School of Public Health, Columbia University; New York, NY (LK); the Department of Paediatrics and Child Health, Mandela School of Medicine, University of KwaZulu/Natal, Durban, South Africa (AC and DA); the Department of Immunology, University of Milano, Milano, Italy (DT and MC); the Department of Genetics, University of Trieste, Trieste, Italy (TR, LS, and SC); and the Genetic Service, Children's Hospital Burlo Garofolo, Trieste, Italy (SC).
2 Partially supported by an RC03/04 grant from the Instituto di Ricovero e Cura a Carattere Scientifico Burlo Garofolo (Trieste, Italy) and by a grant from the Italian Ministry of Research. 3 Reprints not available. Address correspondence to L Kuhn, Sergievsky Center, Columbia University, 630 West 168th Street, New York, NY. E-mail: lk24{at}columbia.edu.
ABSTRACT
Background: Mannose-binding lectin (MBL-2) allele variants are associated with deficiencies in innate immunity and have been found to be correlated with HIV infection in adults and children.
Objective: We tested whether MBL-2 variants among infants born to HIV-positive mothers have an increased susceptibility to HIV.
Design: MBL-2 allele variants were measured among 225 infants born to HIV-positive mothers enrolled in a trial in Durban, South Africa. Mothers of 108 infants were randomly assigned to receive vitamin A and ß-carotene supplementation and 117 to receive placebo. Infants were followed with regular HIV tests to determine rates of mother-to-child HIV transmission.
Results: A high proportion of infants were either homozygous (10.7%) or heterozygous (32.4%) for MBL-2 variants. MBL-2 variants within the placebo arm were associated with an increased risk of HIV transmission (odds ratio: 3.09; 95% CI: 1.21, 7.86); however, MBL-2 variants within the supplementation arm were not associated with an increased risk of transmission (P = 0.04; test of interaction). Among infants with MBL-2 variants, supplementation was associated with a decreased risk of HIV transmission (odds ratio: 0.37; 95% CI: 0.15, 0.91).
Conclusion: We observed what appears to be a gene-environment interaction between MBL-2 variants and an intervention with vitamin A plus ß-carotene that is relevant to mother-to-child HIV transmission.
Key Words: Mother-to-child HIV transmission • vitamin A supplementation • mannose-binding lectin • innate immunity
INTRODUCTION
Vitamin A initially looked promising as a means of reducing mother-to-child HIV transmission in the era before short-course antiretroviral drug regimens were shown to be effective. Although low maternal serum retinol concentrations were found to be associated with an increased risk of transmission to infants (1), none of the 3 studies (2-4) that tested vitamin A supplementation found it to be an effective intervention to reduce transmission. One of the studies even reported a significant increase in transmission via breastfeeding (5). In contrast, reductions in the morbidity and mortality of HIV-infected children after vitamin A supplementation have been observed consistently (6-9). In populations without widely disseminated HIV epidemics but with endemic low levels of vitamin A deficiency, supplementing young children with vitamin A significantly reduces childhood mortality (10).
Anticipating initially that vitamin A supplementation would not influence HIV transmission, we investigated retrospectively whether mannose-binding lectin (MBL-2) gene polymorphisms were associated with increased susceptibility to HIV among infants born to HIV-positive mothers enrolled in a randomized trial evaluating a vitamin A and ß-carotene intervention and vertical transmission in Durban, South Africa (2). We were interested in MBL-2 because studies in adults had observed an increased frequency of carriers of variant MBL-2 alleles among HIV-infected persons relative to uninfected control subjects (11-13). MBL-2 gene polymorphisms have also been shown to be associated with an increased risk of vertical transmission in Brazilian children (14) and with a more rapid progression to AIDS in Italian children (15). MBL is part of the complement system, which is involved in innate immune responses to pathogens before the activity of acquired immune responses. Polymorphisms in the MBL-2 gene reduce concentrations of functional MBL in serum and are associated with unexplained immunodeficiency and susceptibility to other infections (16, 17). We hypothesized that innate immune responses may be particularly important among neonates because their acquired immune responses are usually immature.
During the data analysis, we observed that associations between MBL-2 variants and susceptibility to HIV transmission were different between the mother-child pairs randomly assigned to receive vitamin A and ß-carotene intervention and those randomly assigned to receive placebo. We therefore expanded the study goals to examine gene-environment interactions between the nutritional intervention and MBL-2 variants in relation to mother-to-child HIV transmission.
SUBJECTS AND METHODS
MBL-2 variants were examined among a subset of infants born to HIV-positive mothers enrolled in a randomized trial in Durban, South Africa, to evaluate vitamin A and ß-carotene supplementation as an intervention to reduce mother-to-child HIV transmission. HIV-positive women were recruited during pregnancy at 2 hospitals in Durban between June 1995 and June 1998 and were randomly assigned to receive intervention or placebo. The intervention consisted of a daily supplement containing 5000 IU retinyl palmitate and 30 mg ß-carotene, which began between 28 and 32 wk of gestation and a 200 000 IU dose of retinyl palmitate at delivery. An identically looking placebo was given on the same schedule (2). All HIV-positive pregnant women were counseled about the risks and benefits of infant feeding options, and some elected to avoid breastfeeding. Those who elected to breastfeed were encouraged to do so exclusively (18). Mothers and infants were followed after delivery with regularly scheduled study visits to determine the infection status of the child. In a final analysis of the trial (n = 632), no difference in transmission at 3 mo was observed between the intervention (20.3%) and placebo (22.3%) groups (2). All women provided written informed consent for participation in the trial, and all aspects of the study were approved by the Institutional Review Boards of the investigators' institutions. Neither the mothers nor the infants received antiretroviral drugs for prophylaxis or treatment because the study was completed before demonstration of the efficacy of short-course drug regimens for prevention of mother-to-child HIV transmission and before the availability of antiretroviral drugs for HIV treatment in the public sector in South Africa.
A subset of 250 of 632 infants enrolled in the trial had a dried blood spot specimen available at the completion of the study that was sent for genotyping after being stripped of any identifying information. No specimens from the mothers were available. Host DNA was adequately extracted from 225 of 250 samples by using Chelex-100 as described (19). The genotyped subset (n = 225) was similar to the total nongenotyped study population (n = 407) in terms of most clinical characteristics, including random assignment, maternal CD4 count, maternal age, parity, mode of delivery, site of recruitment, birth weight, infant sex, and HIV transmission rate; however, non-breastfeeders were significantly overrepresented in the genotyped subset.
Single nucleotide polymorphism genotyping of exon 1 of MBL-2 was performed as previously described (20) by using the following primers: forward primer 5-CATCAACGGCT-TCCCA-3 reverse primer 5-AACAGCCCAACACGTACCT-3. The 3 MBL-2 variants (Arg52Cys, Gly54Asp, and Gly57Glu) were grouped together as allele O, whereas the 3 wild-type alleles were grouped as allele A (21). Allele O has a dominant negative effect on MBL serum concentrations as a result of the incorrect assembly of MBL subunits in the collagen-like domain that makes the protein more vulnerable to degradation by metalloproteinases (22). In heterozygous individuals, the serum concentration of the protein is reduced 5–10 times, whereas in O/O homozygotes the concentration of the protein is undetectable (11, 23, 24). Amplification reactions were performed in a final volume of 25 µL with 1X SYBR Green I Amplification Master Mix (Euroclone, Milan, Italy), 150 pmol primer forward, 50 pmol primer reverse, and 10 ng genomic DNA. The cycling conditions were as follows: initial hold at 95 °C for 10 min, 95 °C for 30 s, and 60 °C for 1 min, which was repeated 45 times on the Rotor Gene 3000 (Corbett Research, Sydney, Australia). At the end of the polymerase chain reaction (PCR), the dissociation protocol included slow heating from 60 to 95 °C, with an increase of 0.2 °C at each step and an 8-s delay after each step. Melting curve profiles were obtained by using the dissociation software of the Rotor Gene 3000. Samples were analyzed in triplicate and in different PCR runs to test the reproducibility of this technique. The melting temperatures of the 3 MBL-2 genotypes were as follows: A/A (one peak 83.1 ± 0.1 °C), A/O (2 peaks: 82.6 ± 0.3 and 80.7 ± 0.1 °C), and O/O (one peak: 81.7 ± 0.1 °C). The genotypes of 100 selected samples from this study were confirmed by direct PCR sequencing of the first exon of MBL-2 gene containing the 3 polymorphisms. Samples characterized by the 3 MBL-2 variants (Arg52Cys, Gly54Asp, and Gly57Glu) always gave the same melting profile (with SD values as indicated above), independently by the position of the variation for the wild-type homozygous, the wild-type/mutated heterozygous, and the mutated/mutated homozygous. One compound heterozygous (54/57) subject was found by direct sequencing of MBL2 exon I (melting profile: 81.7, the same as the O/O genotype) and was included in this group. Infants were classified as having no MBL polymorphisms (genotype A/A), one copy of a polymorphic gene (heterozygotes A/O), or 2 copies of polymorphic genes (homozygotes O/O).
As part of the overall trial, maternal CD4 cells were enumerated by flow cytometry (EPICS Profile 2 Flow Cytometer; Coulter, Miami, FL) at enrollment, and a preintervention maternal serum sample that was protected from light during processing and storage was tested for retinol concentration by reversed-phase HPLC (25). Women were followed through delivery, and data on obstetrical events and infant birth characteristics were collected. Blood was drawn from the infant at birth, 1 wk, 6 wk, and 3 mo and then at 3-mo intervals to 15 mo or until 3 mo after the complete cessation of breastfeeding. Samples collected before the child was 9 mo of age were tested for HIV RNA by PCR (Roche Molecular Systems, Branchburg, NJ) and samples collected later were tested for HIV antibodies (Abbott Laboratories, Chicago, IL). A child was defined as HIV-infected if HIV RNA was detected in any sample or if HIV antibody was detected at 15 mo of age. All infections were confirmed in a second sample unless the child died or was lost to follow-up before a second sample could be collected.
Chi-square tests were used to test whether genotypes were in Hardy-Weinberg equilibrium and for other comparisons of proportions. The Breslow-Day test for homogeneity of the odds ratio (OR) was used to test for interactions between random assignment and MBL-2 variants on transmission. Multivariable analysis to investigate confounding was performed by using logistic regression with inclusion of interaction terms where appropriate. The cutoffs defining low preintervention maternal serum retinol and CD4 counts were determined by the value that identified the lowest third of the distribution within the study population. All analyses were conducted by using SAS version 9.1 (SAS Institute Inc, Cary, NC).
RESULTS
Characteristics of the mother-child pairs included in the analysis are displayed in Table 1. The distribution of MBL-2 genotypes among 225 infants born to HIV-positive mothers did not differ significantly from expectations based on the Hardy-Weinberg equilibrium (P = 0.20). A relatively high proportion (10.7%) of infants was homozygous for the variant allele (O/O); 32.4% of the infants were heterozygous for the variant allele (A/O), and 56.9% of the infants had the normal genotype (A/A).
View this table:
TABLE 1. Characteristics of 255 HIV-positive mothers and their infants with mannose-binding lectin (MBL-2) genotype data recruited into a trial of vitamin A + ß-carotene supplementation to prevent mother-to-child HIV transmission in Durban, South Africa1
Of the mother-child pairs randomly assigned to the placebo group, 24.0% of the HIV-infected mothers of infants with wild-type genes transmitted HIV compared with 40.0% of infants with heterozygous (A/O) and 41.7% with homozygous (O/O) MBL-2 polymorphisms. In contrast, of the mother-child pairs randomly assigned to the vitamin A supplementation group, there was no evidence of a trend toward increased transmission and MBL-2 variants (Table 2). We formally tested whether there was homogeneity of the main treatment effect by genotype (ie, evidence of an interaction between genotype and treatment) in pairwise comparisons. Because there was significant heterogeneity of the treatment effect in a comparison of infants homozygous wild type (A/A)and infants heterozygous for MBL-2 variants (A/O) (P = 0.05), but no significant heterogeneity between infants heterozygous (A/O) and homozygous (O/O) for MBL-2 variants (P = 0.77), and because transmission rates did not differ significantly between A/O and O/O infants in either the placebo or the vitamin A group, we combined infants either homozygous or heterozygous for MBL-2 variant genes into one category for comparison with infants with wild-type genes. There was significant heterogeneity of the association between MBL-2 variants (O/O and A/O combined) and transmission by randomized group (P = 0.04), ie, an interaction between treatment and genotype. Thus, findings among infants within the placebo arm, but not in the vitamin A plus ß-carotene arm, were consistent with the hypothesis that MBL-2 variants may be a susceptibility marker for HIV acquisition.
View this table:
TABLE 2. HIV transmission rates by mannose-binding lectin (MBL-2) variants in 225 infants born to HIV-positive mothers randomly assigned to receive vitamin A + ß-carotene or placebo to reduce transmission
We next investigated whether confounding by other risk factors for HIV transmission may have spuriously inflated the association between MBL-2 variants and transmission within the placebo group. Adjustment for maternal CD4 count strengthened the association between MBL-2 variants (A/O or O/O) and transmission (OR: 3.09; 95% CI: 1.21, 7.86; P = 0.018) within the placebo arm. With further adjustment for low birth weight and nonexclusive breastfeeding simultaneously with maternal CD4 count, there continued to be a significant association between MBL-2 variants and susceptibility to HIV transmission (OR: 2.97; 95% CI: 1.12, 7.83; P = 0.03). Adjustment for these variables among infants randomly assigned to the vitamin A plus ß-carotene group failed to uncover any trend toward an association between MBL-2 variants and HIV transmission (OR: 0.93; 95% CI: 0.34, 2.55).
Other risk factors did not display differential associations with transmission by random assignment (ie, no significant interactions). Lower maternal CD4 count, low birth weight, and nonexclusive breastfeeding were each associated with HIV transmission in the placebo and intervention groups separately within this genotyped subset, as they were in the overall study population.
The vitamin A plus ß-carotene intervention was associated with a significant reduction in HIV transmission among infants with variant MBL-2 alleles (OR: 0.37; 95% CI: 0.15, 0.91; P = 0.03). In those with normal MBL-2 alleles, the intervention was not associated with a reduction in transmission (OR: 1.25; 95% CI: 0.56, 2.78).
We also investigated whether associations between MBL-2 variants and HIV transmission were modified by preintervention maternal serum retinol and CD4 counts. In the placebo group, MBL-2 variants were significantly associated with increased transmission (P = 0.01) if the mothers' serum retinol concentrations were within the lowest third of the population (<22.4 ng/mL). A nonsignificant trend in this same direction existed only if serum retinol concentrations were higher (Table 3). The interaction term was of borderline significance (P = 0.09). In the intervention group, there was a significant interaction (P = 0.02) between preintervention maternal CD4 counts and MBL-2 variants. Specifically, if the mothers' preintervention CD4 counts were within the lowest third of the population (<335 counts), MBL-2 variants were significantly associated with lower rates of transmission (P = 0.02; Table 3). The pattern was similar by preintervention serum retinol, but the interaction term was not significant.
View this table:
TABLE 3. HIV transmission rates by mannose-binding lectin (MBL-2) variants stratified by preintervention maternal serum retinol concentrations and CD4 counts within the placebo and vitamin A + ß-carotene arms
DISCUSSION
We observed that, in the absence of intervention, infants with MBL-2 genetic variants were more likely to acquire HIV from their mothers than were infants with normal MBL-2 genes in this South African population. Previous studies in adults and children (among populations in Tanzania, Denmark, Finland, Italy, Brazil, and Gabon) have also reported increased susceptibility to HIV infection attributable to MBL-2 variants (11-15, 26, 27); one study in Colombia did not observe an association (28).
MBL-2 allele variants cause deficiencies in the production of MBL protein (29). MBL is an important component of the innate immune system that can bind to a wide range of carbohydrate ligands on the surface of many pathogens and to activate the lectin pathway of the complement cascade (29, 30). The highly glycosylated nature of the HIV envelope provides a potential opportunity for attack by the innate immune system, and MBL has an affinity for the types of carbohydrate that are abundant on gp120 (31). Although MBL binds to HIV and opsonizes it, how MBL might affect HIV control in vivo remains poorly understood (31, 32). Although there is relative consistency in the association between MBL-2 variants and susceptibility to HIV infection, associations with disease progression after infection are contradictory (11, 33, 34). Genotype may be more informative than phenotype in investigating disease progression, because interpretation of serum concentrations of MBL, which is considered to be an acute phase protein, is complicated by chronic activation characteristic of established HIV infection.
Each of the 3 single base-pair substitutions at codons 52, 54, and 57 in exon 1 of the MBL-2 gene (jointly designated as O) causes low serum concentrations of MBL and appears to have been independently selected after the divergence of African and non-African populations. This suggests some advantage of the variant (29). Although the specific mutation comprising the variant differs between populations, MBL-2 variants are common in almost all populations studied (29). The gene frequency that we observed (0.27) was consistent with the frequencies reported for other sub-Saharan African populations who generally have the mutation at codon 57 (13, 29). Children of HIV-positive mothers might be expected to have a higher frequency of MBL-2 variants than their background population if MBL-2 variant status affects transmission to adults. Because young children with MBL-2 variants are vulnerable to respiratory tract infections (35), the overrepresentation of these variants among children of HIV-positive mothers may increase their risk of early mortality in addition to the risk of HIV transmission.
Reviews have concluded that consequences of MBL deficiency are highly dependent on the immune status of the individual tested (29). This conclusion is supported by the fact that MBL-2 variants are common; hence, most MBL-deficient individuals are healthy. This conclusion is also supported by the fact that most associations between MBL-2 variants and susceptibility to infectious disease have been observed in populations with coexisting conditions that affect immune function, eg, malignancies, cystic fibrosis, and HIV (17, 21, 36, 37). Furthermore, the association between MBL-2 variants and acute respiratory tract infections among children was confined to the period between 6 and 18 mo of age, which corresponds to the vulnerable period when passively acquired maternal antibodies have begun to wane and the child's own adaptive immunity is still developing (35). Our observation of an apparent enhanced effect of MBL-2 variants on transmission when the mothers' serum retinol concentrations or CD4 counts were low (markers of more advanced disease) supports this "two-hit" model, which suggests that some maternal impairment may be necessary for the compromising effects of MBL-2 variants to be measurable. Because there is redundancy within the immune system, defective MBL-mediated innate immunity may be compensated by alternative defense strategies. An elegant study showed that the classic antibody-mediated pathway can compensate for impairments in MBL-deficient individuals (30). Thus, it may not be surprising that we observed an interaction between MBL-2 variants and the nutritional intervention because vitamin A is the classic "antiinfective" micronutrient.
Vitamin A deficiency is common in HIV-infected pregnant women (38), and low serum retinol is associated in nonexperimental studies with increased mucosal viral shedding in breast milk and cervicovaginal secretions (39, 40), mother-to-child HIV transmission (1), and infant mortality (41). However, experimental studies of vitamin A have found it to be an ineffective intervention to reduce vertical HIV transmission (2-4). A possible explanation for the discrepancies may be confounding by inflammatory consequences of chronic HIV infection. A transient decrease in serum retinol generally occurs during the acute phase response to infections and trauma; the decrease is greater as the severity of the infection increases (indicated, for example, by increases in concentrations of C-reactive protein) and corresponds with decreases in retinol binding protein—a negative acute phase protein and the principal transport for delivering vitamin A from liver stores to peripheral tissues (10). Thus, serum retinol concentrations in HIV-infected individuals may not be accurate markers of vitamin A tissue stores, but may instead reflect more active viral infection. A high viral load, low CD4 counts, and acute phase proteins were found to correlate with low retinol concentrations in HIV-infected women, and vitamin A supplementation failed to correct vitamin A deficiency in the presence of strong acute phase reactions (42).
Our study results should be viewed as hypothesis-generating given the small size and lack of a priori hypothesis about a gene-environment interaction. It will also be important to investigate maternal genotype. Because of a lack of samples, we were unable to do this. We also assumed a dominant model, and larger sample sizes will be needed to distinguish whether there are differences between heterozygous and homozygous carriers. Nevertheless, our data suggest that among those who have some deficiency in their acute phase response, in this case as a result of MBL-2 allele variants, vitamin A plus ß-carotene supplementation may partially counteract the increased risk of transmission. However, for some individuals with normal MBL-2 genes, such nutritional interventions may be harmful. It may be possible, although it may not be feasible from a programmatic point of view, to identify and target individuals or groups who might benefit from supplementation. It is intriguing why experimental studies in which HIV-infected adults were supplemented with vitamin A have observed no changes in viral load or CD4 counts (43, 44) nor any significant clinical benefits in terms of progression (45); however, experimental studies in HIV-infected children have consistently observed clinical benefits of vitamin A (6-9). Genetic factors that influence HIV transmission may make infected children nonrepresentative genetically from their background populations.
Understanding the immunogenetic factors of vertical HIV transmission is important, even in the current era of effective antiretroviral therapy (46). A better understanding of immunogenetics may help with the development of interventions to prevent transmission through breastfeeding and of effective vaccines and may expand our knowledge of developmental immunology. Additional investigations of the deficits in innate immunity in conjunction with nutritional deficiencies may be useful to understanding these multifactorial processes.
ACKNOWLEDGMENTS
We thank the other members of the South African Vitamin A Group (K Pillay, E Spooner, HM Coovadia, G Sinclair, N Mngqundaniso, K Uebel, I Coetzee, J Moodley, and D Moodley) for their assistance in the conduct of the clinical trial.
LK, AC, SC, and MC were responsible for the study design and interpretation. LK conducted the data analysis. AC and DA recruited and cared for the patients. SC, TR, LS, and DT were responsible for genotyping. None of the authors had a conflict of interest to declare.
REFERENCES