Exposure to aflatoxin and fumonisin in children at risk for growth impairment in rural Tanzania

Growth impairment is a major public health issue for children in Tanzania. The question remains as to whether dietary mycotoxins play a role in compromising children’s growth. We examined children’s exposures to dietary aflatoxin and fumonisin and potential impacts on growth in 114 children under 36 months of age in Haydom, Tanzania. Plasma samples collected from the children at 24 months of age (N = 60) were analyzed for aflatoxin B1-lysine (AFB1-lys) adducts, and urine samples collected between 24 and 36 months of age (N = 94) were analyzed for urinary fumonisin B1 (UFB1). Anthropometric, socioeconomic, and nutritional parameters were measured and growth parameter z-scores were calculated for each child. Seventy-two percent of the children had detectable levels of AFB1-lys, with a mean level of 5.1 (95% CI: 3.5, 6.6) pg/mg albumin; and 80% had detectable levels of UFB1, with a mean of 1.3 (95% CI: 0.8, 1.8) ng/ml. This cohort had a 75% stunting rate [height-for-age z-scores (HAZ) < −2] for children at 36 months. No associations were found between aflatoxin exposures and growth impairment as measured by stunting, underweight [weight-for-age z-scores (WAZ) < −2], or wasting [weight-for-height z-scores (WHZ) < −2]. However, fumonisin exposure was negatively associated with underweight (with non-detectable samples included, p = 0.0285; non-detectable samples excluded, p = 0.005) in this cohort of children. Relatively low aflatoxin exposure at 24 months was not linked with growth impairment, while fumonisin exposure at 24–36 months based on the UFB1 biomarkers may contribute to the high growth impairment rate among children of Haydom, Tanzania; which may be associated with their breast feeding and weaning practices. A detailed study design for the entire MAL-ED network of investigators is provided in the series of MAL-ED publications (MAL-ED Network Investigators, 2014). The Tanzania site within the MAL-ED network is Haydom, located in northcentral Tanzania approximately 300 km from Arusha. Haydom has a population of about 23,000 inhabitants of various ethnicities, and is a geographically diverse set of rural villages. This MAL-ED study area was chosen due to the comparatively high rates of growth impairment which is 55% among children aged 24 to 60 months according to a recent study (Psaki et al., 2014), the presence of Haydom Lutheran Hospital, and its clinics that support an efficient reproductive health system across the area (Ahmed et al., 2012; Mduma et al., 2014). In the present study, mother-infant dyads were recruited from communities within the Manyara Region of Haydom, Tanzania over a 2-year period beginning in November 2009. Determination of included study areas within the Manyara region is described in detail in Mduma et al. (2014). Location relative to Haydom Lutheran Hospital and child growth parameters were determining factors for inclusion of specific villages into the cohort. The Haydom cohort is a socioeconomically marginalized, rural community consisting primarily of subsistence farming households. Selection bias was minimized by defining a representative mother/child population and screening every mother for eligibility. Inclusion criteria included: mother aged 16 years or older, singleton pregnancy, intention to remain in the study area for at least 6 months following enrollment, and birth weight or enrollment weight of >1500 g. Exclusion criteria included diagnosis of congenital disease or severe neonatal disease (MAL-ED Network Investigators, 2014). IRB approval was obtained from the National Health Research Ethics Committee, which is part of the National Institute for Medical Research of Tanzania, and Michigan State University. In total, plasma samples collected at 24 months of age (N = 60) were utilized for measuring aflatoxin B1-lysine (AFB1-lys) biomarker concentrations and urine samples collected between 24 and 36 months of age (N = 94) were analyzed for urinary fumonisin B1 (UFB1) concentrations. Eighteen of the participants had both a plasma and a urine sample for the AFB1-lys and UFB1 analysis at the same timepoint of 24 months. AFB1-lys is a well-established biomarker of long-term dietary aflatoxin exposure during the past 2–3 months. Its concentrations were determined by liquid chromatography isotope dilution mass spectrometry (LC/MS) as described in Groopman et al. (2004) and McCoy et al. (2008). Briefly, plasma (200 μl) was vortexed with internal standard, 10 μl × 0.1 ng AFB1-D4-lys/ml, and pronase and incubated for 18 h at 37 °C. Samples were passed through solid-phase extraction (SPEs) columns and the eluent analyzed using ultra performance liquid chromatography – tandem mass spectrometer (UPLC-MS/MS). The parent ion for AFB1-D4-lys [(M+H)+, m/z 461.3] fragmented to yield a daughter ion at m/z 398.2. The AFB1-lys ion (m/z 457.2) fragmented to yield a daughter ion at m/z 394.1. This methodology had a limit of detection of 0.4 pg AFB1-lys/mg albumin and was run with quality control samples run in triplicate. Urinary fumonisin B1 (UFB1) has been proposed as an effective biomarker for dietary fumonisin exposure over the past 24 h, and is currently used worldwide for biomonitoring of human fumonisin exposure (van der Westhuizen et al., 2013). A significant correlation in a positive dose-dependent manner was observed between dietary fumonisin exposure and the UFB1 levels in human populations (Riley et al., 2015). The analytical method used for UFB1 was a minor modification of a method described previously (Riley et al., 2012). Briefly, urine samples (2 ml) containing 10 ng of U-[13C34]-FB1 (33621 Sigma-Aldrich Corp. St. Louis, MO, USA 33621) were extracted for FB1 with C18-SPE cartridges. The loaded cartridges were eluted using 2 ml of 70% acetonitrile: 30% water made to 0.1% formic acid as previously described (Riley et al., 2012). The eluates were concentrated under N2 at room temperature, so that the final acetonitrile-to-water concentration (based on specific gravity) was 30%–70%, and approximately 0.1% formic acid. Quantitation was accomplished by LC/MS as previously described. Normalization of UFB1 concentrations was described in the Supplemental material. The limit of detection for UFB1 is 0.01 ng/ml. The detection limits for FB2 and FB3 are similar (Riley et al., 2012). MAL-ED trained staff members measured anthropometrics of children enrolled in the study on a monthly basis. Quality control measures included standardized techniques and instruments across study sites, and measurements were repeated on a subset of participants. Standard infant scales (SECA) were used to measure weight at the nearest 0.1 kg and length measuring board or non-stretch Teflon synthetic tape (SECA) was used to measure height at the nearest 0.1 cm, respectively. Socioeconomic status (SES) is a conceptualization of an individual’s, household’s or community’s access to resources and can be measured using various methodologies. Prior to initiation of recruitment and consent among participants, the MAL-ED network undertook preliminary research to determine an appropriate methodology to measure SES that could be applicable in a multi-country study. Determination of the critical variables and resources allowed the MAL-ED network to formulate an index of household SES called the Water/sanitation, Assets, Maternal education, and Income (WAMI) to be applied and comparable across multiple countries (Psaki et al., 2014). Components included in the WAMI index include improved access to water and sanitation, wealth measured by a set of assets, maternal education, and monthly household income. In the present study, the methodology used for combining these components into a WAMI score was conducted according to Psaki et al. (2014). Mothers or other adult household members were queried monthly (months 9–36 of the children’s lives) to collect the children’s 24-hour dietary recall data, which were used to derive food and nutrient intake information. For our analyses, we averaged monthly data points to produce estimated intakes for two age ranges: 16–24 months and 25–36 months. These data were analyzed in relation to mycotoxin biomarker concentrations. Using Haydom’s food composition tables, we quantified energy, macronutrient and micronutrient intakes including vitamin A, zinc, iron, folate, protein, animal protein, and protein from milk, meat, fish, poultry, eggs, and insects (Lukmanji et al., 2008). These variables were adjusted for average total energy intake (kcal), by calculation of the residuals from an ordinary least squares regression analysis (Willett and Stampfer, 1986). The residual values were used in all statistical analyses (MAL-ED Network Investigators, 2017a). Grain-based food items measured in the dietary recall questionnaire included rice, maize, wheat, millet, sorghum, and common beans; whereas chickpeas and mung beans are consumed at much lower frequencies. Maize makes up the main part of the children’s weaning foods (Kimanya et al., 2010). These data were averaged over the timepoints 16–24 months for aflatoxin and 24–36 months for fumonisin. The data were normally distributed (goodness of fit test, p < 0.27) and not adjusted for statistical analysis. The AFB1-lys data were not normally distributed, even following lognormal transformation. Therefore, all statistical analyses of AFB1-lys were conducted utilizing tests for non-parametric data. UFB1 was not normally distributed and had a significant portion of non-detectable values, therefore they were log (x + 1) transformed prior to statistical analysis. AFB1-lys, UFB1 and growth indicators were analyzed by univariate analysis with all possible confounding variables (dietary intake variables including plasma vitamin A, iron, zinc, protein and animal protein, folate, SES index, and gender). Anthropometric data (HAZ, WAZ and WHZ z-scores) were normally distributed and were not altered for statistical analysis. Wilcox-rank sum tests were used to analyze statistical differences between AFB1-lys or UFB1 concentrations and categorical data. Linear regression models were built with each growth indicator - HAZ, WAZ, and WHZ at 36 months of age - as dependent variables, and AFB1-lys or UFB1 concentrations as independent variables. A p value ≤ 0.05 (two-tailed) was considered statistically significant. All statistical analyses were conducted with JMP software version 13.1 (SAS Institute, Cary, NC, USA). In addition, both AFB1-lys and UFB1 concentrations were categorized by quartile, and assessed by z-scores using ANOVA followed by Tukey HSD tests to compare between quartiles and all other variables.