  |
 |
 |
 |
 |
|||
 |
|||||||
 |
|
|||||||||||||||
|
Journal Publication |
|||||||||||||||||||||||||||||||||
Tumor
Necrosis Factor-a
Promoter Variant 2 (TNF2) Is Associated With Pre-term Delivery, Infant
Mortality, and Malaria Morbidity in Western Kenya: Asembo Bay Cohort Project
IX by Michael Aidoo,1 Peter D. McElroy,1 Margarette S. Kolczak,1 Dianne J. Terlouw,2,3 Feiko O. ter Kuile,1,2,3 Bernard Nahlen,1,2 Altaf A. Lal,1 and Venkatachalam Udhayakumar1* 1Division
of Parasitic Diseases, National Center for Infectious Diseases, Centers
for Disease Control and Prevention, Atlanta, GA Corresponding Author: Udhayakumar, Mail Stop F-12, Centers for Disease Control, 4770 Buford Highway, Atlanta, GA 30341-3724. E-Mail: vxu0@cdc.gov Grant sponsor: NCID / Emerging Infectious Disease Funds (V.U.) Received for publication 18 January 2001; accepted 7 March 2001
A polymorphism in the promoter region of the tumor necrosis factor-a (TNF-a) gene, with a guanine to adenine nucleotide change at position –308, TNF2 is associated with increased TNF-a production. TNF2 homozygotes have a higher risk of severe disease and/or death due to cerebral malaria and other infectious diseases. We investigated the impact of this allele on malaria morbidity and mortality in young children who participated in an immunoepidemiologic cohort study of malaria in an area of intense perennial Plasmodium falciparum trans-mission in western Kenya. A total of 1,048 children were genotyped. Poisson regression and Cox proportional hazards models were used to determine the relationship between TNF-308 variants and morbidity and mortality. The gene frequencies of the TNF1 and TNF2 alleles were 0.90 and 0.10, respectively. TNF2 homozygosity was associated with pre-term birth when compared with TNF1 homozygotes [relative risk (RR) 7.3, 95% CI, 2.85–18.9, P = 0.002) and heterozygotes (RR 6.7, 95% CI 2.0–23.0, P = 0.008). Among children born prematurely, the TNF2 allele was significantly associated with a higher risk of death in infancy compared with TNF1 (RR 7.47, 95% CI 2.36–23.6). The risk of death was higher among TNF2 homozygotes than among heterozygotes. The TNF2 allele was significantly associated with high density P. falciparum parasitemia (RR 1.11, 95% CI 1.0–1.24). Among low birth weight children, the TNF2 allele was associated with severe anemia (RR 2.16, 95% CI 1.17–4.01) and showed a trend toward a risk for severe malaria anemia (RR 1.99, 95% CI 0.89–4.46). These data suggest that TNF2 is a risk factor for pre-term birth and early child-hood mortality and malaria morbidity in children in this region. Further under-standing of the pathogenic mechanisms underlying this association is required.
Key Words: polymorphism; malaria; premature birth Tumor necrosis factor-a (TNF-a) is a pro-inflammatory cytokine known to be essential in the control of many intracellular infectious agents in humans [Beutler and Grau, 1993]. Elevated levels of TNF-a have, however, been implicated in the pathogenesis of infectious diseases such as malaria [Kwiatkowski et al., 1990], meningococcal disease [Westendorp et al., 1995; Riordan et al., 1996], schistosomiasis [Amiri et al., 1992], and some autoimmune diseases [Wilson et al., 1995]. In addition, several studies have shown that increased TNF-a levels in placental and fetal tissues are associated with spontaneous abortions and pre-term delivery [Girardin et al., 1990; Shaarawy et al., 1997; Arntzen et al., 1998; Romero et al., 1998; Wenstrom et al., 1998]. A single nucleotide polymorphism from guanine (G) in the normal TNF1 allele to adenine (A) in the TNF2 variant at position –308 relative to the transcription start site of the TNF-a gene is associated with increased TNF-a production [Wilson et al., 1997]. The TNF2 allele is associated with lepromatous leprosy [Roy et al., 1997], scarring trachoma infections [Conway et al., 1997], mucocutaneous leishmaniasis [Cabrera et al., 1995], and a higher risk of death in meningococcal disease [Nadel et al., 1996]. Homozygosity for this rare TNF2 allele has been associated with a four-fold increased risk of cerebral malaria and seven times greater likelihood of death after a cerebral malaria attack [McGuire et al., 1994]. In another study, TNF2 was found to be associated with severe malaria outcomes (multiple-organ dysfunction syndrome) and impaired consciousness due to infectious diseases other than malaria [Wattavidanage et al., 1999]. Although these studies identified TNF2 as a risk factor for morbidity and mortality after infection with different pathogens, the overall impact of this allele on child survival is not known. Furthermore, the impact of this allele on the severity of malaria is unknown in areas with intense malaria transmission such as western Kenya, where severe anemia and not cerebral malaria is the most common manifestation of severe malaria. Taking advantage of a longitudinal birth cohort (Asembo Bay Cohort) in western Kenya, we retrospectively determined the relationship between the TNF2 allele and all-cause mortality and malaria-specific morbidity. Here, we report an association of TNF2 with birth outcome and infant survival. In addition, the impact of this allele on malaria-associated morbidity in this population is described. This study was a retrospective investigation using DNA from participants en-rolled in a mother-infant cohort study, the Asembo Bay Cohort Project (ABCP) con-ducted in western Kenya. More than 95% of the residents in this area belong to Luo ethnic group. The methodology and cohort design of the ABCP were described else-where [Bloland et al., 1999]. In the study, infants were followed up from birth to 5 years of life with mortality and monthly malaria-associated morbidity data collected. Briefly, pregnant women identified through monthly census were enrolled after in-formed consent was obtained. Within 24 hours of delivery, the infant’s birth weight was obtained, and gestational age was assessed using the Dubowitz score. Thereaf-ter, trained village monitors visited each mother-infant pair every fortnight for clinical observation and to administer a questionnaire on disease symptoms and any medication received. Monthly capillary blood samples were taken by finger or heel prick from infants for determination of malaria parasitemia and hemoglobin levels. Aliquots of peripheral blood were stored in liquid nitrogen and used as a source of DNA. Only children from singleton births and the first child of mothers who contributed more than one birth to the cohort were included in the study. There were 1,247 infants who met this criterion in the cohort, and DNA samples were available from 1,077 infants for genetic analysis. Genomic DNA was obtained from whole blood using the Puregene and Capture column DNA extraction kits (Gentra Systems, Minneapolis, MN) as directed by the manufacturer. From 1,077 DNA samples, we were able to genotype 1,048 samples. Twenty-nine samples could not be polymerase chain reaction (PCR)–amplified. The TNF –308 was typed as described by Wilson et al. [1992]. Briefly, primers AL1566 (5¢-AGG CAA TAG GTT TTG AGG GCC AT-3¢) and AL1567 (5¢-TCC TCC CTG CTC CGA TTC CG-3¢) were used to amplify a 107-bp region of the TNF-a promoter region spanning the polymorphic site. Fifty to 100 ng of DNA was amplified in a PCR mixture containing 0.2 mM of each primer, 1 ´ concentration of PCR buffer (PE Applied Biosystems, Foster City, CA, 1.25 U Taq polymerase (PE Applied Biosystems), and 200 mM of each deoxyribonucleoside triphosphate (PE Applied Biosystems), in a final volume of 100 mL. PCR cycling conditions were as described by Wilson et al. [1992]. The amplified DNA fragment appears as a 107-bp band, which, when treated with NcoI restriction enzyme, produces an 87- and a 20- bp band in the presence of TNF1 but is undigested and remains a 107-bp band in the presence of TNF2. Data were analyzed using univariate and multivariate Poisson regression models to determine the relationship between TNF –308 variants and morbidity. Survival analysis (Cox proportional hazard regression) was used to determine the relationship between TNF2 and mortality. Covariates for statistical control included mother’s survival (dead or alive) and educational level (<5 or ³5 years), and the child’s gestational age (<37 or ³37 weeks), birth weight (<2,500 or ³2,500 g), and gender. Prematurity and low birth weight were defined as less than 37 weeks’ gestation and less than 2,500 g, respectively. Severe anemia was defined as hemoglobin level <6 g/dL and severe malarial anemia was defined as hemoglobin level <6 g/dL with a P. falciparum parasitemia of >10,000 parasites/mL of blood. These regression models were fit using the SAS statistical package (SAS Institute, Cary, NC) and Epi Info 6 (Centers for Disease Control, Atlanta, GA) was used to calculate risk ratios and P-values. The statistical differences in the mean birth weights were determined using Tukey test for multiple comparisons. We genotyped 1,048 DNA samples for the TNF –308 polymorphism. The gene frequencies of the TNF1 and TNF2 alleles were 0.90 and 0.10, respectively. The frequencies of different TNF –308 genotypes in this population are shown in Table I. Our data analysis showed that the TNF2 homozygote genotype is significantly associated with pre-term birth when compared with TNF1 homozygotes (RR 7.3, 95% CI 2.85–18.9) and TNF1/2 heterozygotes (RR 6.7, 95% CI 2.0–23.0) (Table I). The mean birth weight was lower among pre-term babies when compared with babies born at term; however, the difference was significantly different only for TNF1 homozygotes (P = 0.001) and TNF2 homozygotes (P = 0.004) but not for TNF1/2 heterozygotes (P = 0.12). Among babies born pre-term, TNF2 homozygotes had the lowest mean birth weights, but the difference was not significantly different when compared with TNF1 homozygotes (P = 0.79) or TNF1/2 heterozygotes (P = 0.80). The risk of being born with low birth weight was also higher for TNF2 homozygotes when compared with TNF1 homozygotes, but the difference was not significant (Table I). We also compared the risk of intrauterine growth retardation (IUGR) among the three TNF-a promoter variants but did not identify any (Table I). Survival analysis was performed to determine whether the TNF2 allele was associated with mortality. Using multivariate analyses, which showed a strong interaction between gestational age and TNF2, we found that the TNF2 allele was associated with increased mortality among children born prematurely (RR 7.47, 95% CI 2.36– 23.6) (Fig. 1A). Among premature infants, the risk of childhood death was higher for TNF2 homozygotes (TNF2 homozygotes versus TNF1 homozygotes: RR 70.9, 95% CI 18.2–277.2) than for TNF1/2 heterozygotes (TNF1/2 heterozygotes versus TNF1 homozygotes: RR 3.43, 95% CI 0.82–14.36). There was no association of the TNF2 allele with mortality among children born at term (³37 weeks) (RR 0.86, 95% CI 0.58–1.27) (Fig. 1B). In this study area, P. falciparum malaria accounts for a significant portion of mortality among children younger than 2 years. Therefore, we compared common manifestations of malaria morbidity among the TNF –308 genotypes (Table II). Potential confounding by gender, mother’s education level, gestational age, birth weight, and the mother’s survival status were assessed. Our data showed that TNF2 was marginally associated with a higher incidence of high-density (>10,000/mL) P. falciparum blood infections (RR 1.11, 95% CI 1.00–1.24, P = 0.043). This risk was similar for TNF2 homozygotes and TNF1/2 heterozygotes (data not shown). Among low birth weight children there was a significant association of the TNF2 allele with increased frequency of severe anemia (hemoglobin <6 g/dL) episodes (RR 2.16, 95% CI 1.17–4.01) as well as severe anemia episodes with positive P. falciparum parasitemia (RR 2.16, 95% CI 1.11–4.19). However, the increased risk associated with severe malaria anemia (hemoglobin <6 g/dL with >10,000 P. falciparum parasites/ mL of blood) (RR 1.99, 95% CI 0.89–4.46) among children with TNF2 was not significant (Table II). Using a community-based longitudinal study, we have shown that TNF2 homozygosity is associated with premature birth, and among those born prematurely, the TNF2 allele is associated with increased mortality. Previous reports showed that high levels of TNF-a in amniotic fluid are associated with pre-term delivery [Shaarawy and Nagui, 1997; Arntzen et al., 1998; Romero et al., 1998; Wenstrom et al., 1998]. One of the mechanisms by which higher levels of TNF-a could accelerate pre-term labor is through up-regulation of cyclooxygenase (COX)-2 enzyme–mediated prostaglandin production [Perkins and Kniss, 1997; Swaisgood et al., 1997]. In addition, TNF-a is known to enhance the production of matrix metalloproteinases that can degrade collagen in fetal membranes and maternal tissues [So et al., 1992] and may lead to premature rupture of the membranes and cervical dilation [Roberts et al., 1999]. This interpretation is based on the assumption that TNF2 is associated with increased TNF-a production as shown previously [Conway et al., 1997; Wilson et al., 1997; Louis et al., 1998; Warzocha et al., 1998]. This is the first study with a large enough sample size to show that TNF2 in the fetus is associated with pre-term birth. Because this association was found only among TNF2 homozygotes, by inference, the mothers of these children must have carried at least one TNF2 allele. In a previous study, Dizon-Towson et al. [1997] observed that TNF2 in neonates and mothers was not significantly associated with pre-term birth. Conversely, Roberts et al. [1999] reported that the maternal TNF2 allele was significantly associated with pre-term birth due to premature rupture of membranes and not with idiopathic pre-term delivery (another category of pre-term birth due to spontaneous contractions without any known antecedent pathology). Because premature rupture of membranes is associated with an infectious etiology in many cases, the findings of Roberts et al. [1999] and our results are consistent with the hypothesis that the increased TNF-a production due to the TNF2 allele could have contributed to pre-term birth. In this study, TNF2 homozygosity was associated with a twofold increased risk of low birth weight. Although this increased risk was not a significant one, this observation is consistent with the fact that often pre-term birth is associated with a low birth weight. Overall, the lack of statistically significant difference in the mean birth weights and IUGR between the three TNF –308 genotypes suggests that TNF2 is an independent risk factor for pre-term birth. Our study shows for the first time that the TNF2 allele is associated with all-cause mortality among infants born prematurely. The risk of mortality was much higher for TNF2 homozygotes than for heterozygotes. In the Gambian study [McGuire et al., 1994], the TNF2 homozygotes but not TNF1/2 heterozygotes had a higher risk for cerebral malaria and its associated mortality. In other studies, TNF2 has been found to be a risk factor for mortality due to meningococcal disease [Nadel et al., 1996] and septic shock [Mira et al., 1999]. The most common childhood illnesses in this study area are malaria, upper respiratory infection, gastrointestinal infections, and pneumonia. Because we did not have data on disease-specific mortality in this cohort, it was not possible to link TNF2 associated deaths with any particular cause. Although in the Gambian study, the TNF2 genotype was associated with an increased risk of mortality among children with cerebral malaria, we believe that such a risk is most probably not the cause of TNF2-associated deaths in this study due to the following reasons. First, cerebral malaria is rare in our study population in western Kenya [Snow et al., 1997] and, second, most of the deaths among the premature infants with TNF2 occurred in the first 6 months of life, a time period during which cerebral malaria rarely occurs in children living in endemic areas. In previous studies conducted in areas with seasonal malaria transmission, it had been reported that TNF2 was not associated with severe malaria anemia [McGuire et al., 1994, 1999]. On the contrary, in this study, we found that TNF2 was associated with severe anemia among low birth weight children. It is also important to point out that the lack of statistical significance between TNF2 and severe malarial anemia episodes could be simply due to the small sample size because there was a definite trend toward the increased relative risk (RR 1.99, 95% CI 0.89–4.46). In another study conducted in a holoendemic malaria transmission area in Tanzania [Stirnadel et al., 1999], infants with TNF2 had higher but statistically insignificant mean P. falciparum parasite densities when compared with TNF1. Consistent with the Tanzanian study, we also found that TNF2 was only marginally associated with a risk of high density P. falciparum parasitemia. Our estimates showed that less than 2% of the high-density parasitemia episodes could be attributed to the TNF2 allele (attributable risk 0.019). Overall, the results from the morbidity analysis did not pro-vide any evidence to suggest malaria as the definitive cause of the association of TNF2 with mortality among pre-term babies. Caution, however, is needed in the interpretation of these observations because all children with positive blood films were treated with antimalarial drugs. The drug intervention could potentially lead to an underestimation of the association of TNF2 with severe malarial anemia. It is not clear why TNF2 is associated with infant mortality among infants born pre-term. Prenatal exposure to TNF-a has been suggested to be a risk factor for respiratory distress syndrome [Speer, 1999]. It is possible that premature children born with TNF2 may be at a high risk of developing life-threatening complications such as pneumonia, septic shock, or other complications of unknown etiology, especially when they have respiratory and or other organ systems that are not fully developed at birth. This hypothesis is consistent with our observation that TNF2 homozygotes had a several-fold increase in the risk of death compared with TNF1/2 heterozygotes. Recently, Mira et al. [1999] demonstrated that TNF2 individuals were more susceptible to septic shock and were also more likely to die from it. In Sri Lanka, TNF2 was found to be associated with severe complications resulting from malaria as well as impaired consciousness associated with bacterial and viral infections [Wattavidanage et al., 1999]. Clearly, the TNF2 allele seems to be associated with severe disease caused by a variety of infectious agents, and it is plausible that abnormal levels of TNF-a could trigger a pathway that eventually leads to death. Further studies are required to test such a hypothesis. Despite the observed association, TNF2 may be only causally associated with mortality and that other closely linked genetic factors may play a role. The TNF2 allele, which is located in the MHC class III region, is in linkage disequilibrium with some HLA class I and II antigens [Wilson et al., 1993]. We did not determine this linkage in our study; however, previous studies showed that TNF2-associated deaths among cerebral malaria [McGuire et al., 1994] and septic shock patients [Mira et al., 1999] were independent of linked HLA alleles. There are at least two other single nucleotide polymorphisms in the TNF-a gene promoter region. The TNF-238A allele was found to be associated with severe malaria anemia [McGuire et al., 1999] and TNF-376A was associated with cerebral malaria [Knight et al., 1999]. However, TNF2 is not in linkage with any of these two polymorphisms [Knight et al., 1999; McGuire et al., 1999]. In conclusion, this study shows that TNF2 homozygosity is associated with pre-term birth and the TNF2 allele is a risk factor for infant mortality among those born prematurely. A marginal association of TNF2 with high-density P. falciparum parasitemia and severe anemia was also found. Further studies are needed to fully understand the mechanisms underlying these risks in infancy and possibly in the unborn child. We thank Kenya Medical Research Institute (KEMRI) for approving the publication of this research. We also thank Lee Kubersky for technical support, D.J. Perkins for comments and suggestions, the study participants for their willingness to participate in the study, and the CDC/KEMRI ABCP staff for their support. This study was funded by the NCID/Emerging Infectious Disease Fund (V. Udhayakumar, PI). Michael Aidoo was supported in part by ASM/NCID Postdoctoral Research Fellowship Program. D.J. Terlouw and F.O. ter Kuile were supported by a grant from The Netherlands Foundation for the Advancement of Tropical Diseases Re-search. Use of trade names are for identification purposes only and do not imply endorsement by the Public Health Service or by the U.S. Department of Health and Human Services.
|
|||||||||||||||||||||||||||||||||
