Maternal aflatoxin exposure during pregnancy and adverse birth outcomes in Uganda

Aflatoxins are toxic metabolites of Aspergillus moulds and are widespread in the food supply, particularly in low- and middle-income countries. Both in utero and infant exposure to aflatoxin B 1 (AFB 1 ) have been linked to poor child growth and development. The objective of this prospective cohort study was to investigate the association between maternal aflatoxin exposure during pregnancy and adverse birth outcomes, primarily lower birth weight, in a sample of 220 mother–infant pairs in Mukono district, Uganda. Maternal aflatoxin exposure was assessed by measuring the serum concentration of AFB 1 -lysine (AFB-Lys) adduct at 17.8 ± 3.5 (mean ± SD)-week gestation using high-performance liquid chromatography. Anthropometry and birth outcome characteristics were obtained within 48 hr of delivery. Associations between maternal aflatoxin exposure and birth outcomes were assessed using multivariable linear regression models adjusted for confounding factors. Median maternal AFB-Lys level was 5.83 pg/mg albumin (range: 0.71–95.60 pg/mg albumin, interquartile range: 3.53–9.62 pg/mg albumin). In adjusted linear regression models, elevations in maternal AFB-Lys levels were significantly associated with lower weight (adj-β: 0.07; 95% CI: −0.13, −0.003; p = 0.040), lower weight-for-age z-score (adj-β: −0.16; 95% CI: −0.30, −0.01; p = 0.037), smaller head circumference (adj-β: −0.26; 95% CI: −0.49, −0.02; p = 0.035), and lower head circumference-for-age z-score (adj-β: −0.23; 95% CI: −0.43, −0.03; p = 0.023) in infants at birth. Overall, our data suggest an association between maternal aflatoxin exposure during pregnancy and adverse birth outcomes, particularly lower birth weight and smaller head circumference, but further research is warranted. This was a prospective cohort study conducted in Mukono district, Uganda, from February to November 2017. Women were initially enrolled during their first prenatal visit at Mukono Health Center IV (MHC IV). Women qualified for the study if they were between 18 and 45 years old, resided within 10 km of MHC IV, carried a singleton pregnancy, and planned to remain in Mukono district throughout their pregnancy. Women were excluded if they were <18 years old or >45 years old, HIV‐positive (verified via routine rapid HIV test conducted at first prenatal visit), severely malnourished (defined as body mass index [BMI] <16.0 kg/m2), severely anaemic (defined as Hb < 7 g/dL), or planned to move away from Mukono district prior to delivery. Among 300 women screened, 258 met the inclusion criteria and were enrolled in the study. A sample size of 258 allowed for the detection of a relative risk of 2.0 within 50% of the true risk parameters, assuming 80% power, 0.05 significance, a 5% frequency of LBW, and 15% loss to follow up. Primary reasons for exclusion included multiple gestation and HIV infection. Of the enrolled 258 participants, 236 had a follow‐up visit conducted prior to delivery, 11 elected to drop out of the study, and 11 were considered lost to follow‐up as they moved away from Mukono district. Birth outcome data were collected within 48 hr for 232 infants, and 4 births were missed by the field team. Ten infants were stillborn, and two infants died before anthropometry measurements could be taken. Descriptive characteristics and birth outcome data were therefore analysed for 220 mothers and their infants. The study was approved by the Tufts Health Sciences Institutional Review Board in Boston, Massachusetts; the Mengo Hospital Research Ethics Committee in Kampala, Uganda; and the Uganda National Council for Science and Technology in Kampala, Uganda. Written consent was obtained from all participants prior to enrolment. Following enrolment, height (0.1‐cm precision; Infant/Child/Adult ShorrBoard, Shorr Production, Olney, MD, USA), weight (0.1‐kg precision; Seca 874, Hanover, MD, USA), mid‐upper arm circumference (MUAC; 0.1‐cm precision; tri‐coloured, nonstretch adult MUAC tape), and blood pressure (Omron 10 Series, Omron Healthcare, Kyoto, Japan) measurements were taken by the study nurse. All anthropometry measurements, including height, weight, and MUAC, were taken in triplicate and averaged. Maternal BMI was calculated using the formula BMI = kg/m2. Pulse pressure was calculated by taking the difference between the systolic and diastolic blood pressure. Haemoglobin measurements were taken using a portable hemoglobinometer (HemoCue Hb 301; HemoCue, Inc., Brea, CA, USA). Estimated date of delivery was assessed from an obstetric ultrasound examination by a trained technician. A venous blood draw (BD Vacutainer, Becton Dickinson, Durham, NC, USA) was performed by the phlebotomist at MHC IV, and blood samples were centrifuged at MHC IV for 5 min at speed of 4,000 rpm to separate serum. Finally, a questionnaire was administered by the study nurse that included questions related to demographics, prior pregnancies, health status, diet, food security (using the Household Food Insecurity Access Scale; Coates, Swindale, & Bilinsky, 2007), and water, hygiene, and sanitation practices. Diet was assessed using a diet diversity questionnaire that asked participants to respond yes/no to having eaten >50 specific foods in the previous 24 hr. Foods were selected based on their inclusion in the Ugandan Demographic and Health Survey, with minor modifications to account for the norms and preferences of the study site. Foods consumed by >10% of the participants are presented in Table S1. Responses were used to generate a Minimum Dietary Diversity for Women score, based on the number of food groups (0–10) consumed (Food and Agriculture Organization, 2016). Groups were considered (1) grains, white roots and tubers, and plantains; (2) pulses (beans, peas, and lentils); (3) nuts and seeds; (4) dairy; (5) meat, poultry, and fish; (6) eggs; (7) dark green leafy vegetables; (8) other vitamin A‐rich fruits and vegetables; (9) other vegetables; and (10) other fruits. Infant anthropometry data, including length (0.1‐cm precision; Infant/Child/Adult ShorrBoard, Shorr Production, Olney, MD, USA), weight (0.1‐kg precision; Seca 874, Hanover, MD, USA), and head circumference (0.1‐cm precision; flexible measuring tape), were assessed within 48 hr of delivery. All anthropometry measurements were taken in triplicate and averaged. Head circumference was measured as the largest possible occipital‐frontal circumference. AFB1 (> 98% purity), albumin determination reagent bromocreosol purple, and normal human serum were purchased from Sigma Aldrich Chemical Co. (St. Louis, MO, USA). Pronase (25 kU, Nuclease‐free) was purchased from Calbiochem (La Jolla, CA, USA). Protein assay dye reagent concentrate and protein standards were purchased from Bio‐Rad Laboratories Inc. (Hercules, CA, USA). Mixed mode solid phase extraction cartridges were purchased from the Waters Corp. (Milford, MA, USA). Authentic AFB‐Lys was synthesized as previously described (Sabbioni, Skipper, Büchi, & Tannenbaum, 1987). All other chemicals and solvents used were of highest grade commercially available. Midgestation maternal aflatoxin exposure was assessed using the serum AFB‐Lys adduct biomarker. Serum samples were transported on dry ice to the Wang laboratory at the University of Georgia, Athens, USA, and analysed with a high‐performance liquid chromatography (HPLC)‐fluorescence method. This included measurement of albumin and total protein concentrations for each sample, digestion with protease to release amino acids, concentration and purification of the AFB‐Lys adduct, and finally separation and quantification by HPLC (Qian et al., 2013a; Qian, Tang, Liu, & Wang, 2010). Specifically, thawed serum samples were inactivated for possible infectious agents via heating at 56°C for 30 min, followed by measurement of albumin and total protein concentrations using modified procedures as previously described (Qian et al., 2013b). A portion of each sample (approximately 150 μL) was digested by pronase (pronase: total protein, 1:4, w: w) at 37°C for 3 hr to release AFB‐Lys. AFB‐Lys in digests were further extracted and purified by passing through a Waters MAX solid phase extraction cartridge, which was preprimed with methanol and equilibrated with water. The loaded cartridge was sequentially washed with 2 ml water, 1 ml 70% methanol, and 1 ml 1% ammonium hydroxide in methanol at a flow rate of 1 ml/min. Purified AFB‐Lys was eluted with 1 ml 2% formic acid in methanol. The eluent was vacuum‐dried with a Labconco Centrivap concentrator (Kansas City, MO, USA) and reconstituted for HPLC‐fluorescence detection. The analysis of AFB‐Lys adduct was conducted in an Agilent 1200 HPLC‐fluorescence system (Santa Clara, CA, USA). The mobile phases consisted of buffer A (20 mM NH4H2PO4, pH 7.2) and buffer B (100% Methanol). The Zorbax Eclipse XDB‐C18 reverse phase column (5 micron, 4.6 × 250 mm) equipped with a guard column was used (Agilent, Santa Clara, CA, USA). Column temperature was maintained at 25°C during analysis, and a volume of 100 μL was injected at a flow rate of 1 ml/min. A gradient was generated to separate the AFB‐Lys adduct within 25 min of injection. Adduct was detected by fluorescence at maximum excitation and emission wavelengths of 405 and 470 nm, respectively. Calibration curves of authentic standard were generated weekly, and the standard AFB‐Lys was eluted at approximately 13.0 min. The limit of detection was 0.2 pg/mg albumin. The average recovery rate was 90%. The AFB‐Lys concentration was adjusted by albumin concentration. Quality assurance and quality control procedures were maintained during analyses, which included simultaneous analysis of one authentic standard in every 10 samples and two quality control samples daily. Furthermore, following completion of the laboratory analysis, sets of three samples were selected and pooled into 11 intraday reproducibility samples, which were analysed twice on the same day by the same analyst, and 11 interday reproducibility samples, which were analysed on different days by different analysts, to demonstrate laboratory precision and sampling reproducibility. All statistical analyses were performed using STATA 15 software (Stata Corps, College Station, TX, USA). Variables were first assessed for outliers and normality. Because of their skewed distribution, AFB‐Lys levels were natural log (ln) transformed prior to all analyses. Weight, length, and head circumference measurements were converted to z‐scores for WAZ, LAZ, WLZ, and HCZ using the World Health Organization standards. Outliers were defined as −6 > WAZ > +5, −5 > WLZ > +5, −6 > LAZ > +6, and −5 > HCZ > +5 based on the World Health Organization’s recommendation for biologically implausible values and were excluded from analysis. (Group, 2006). Enrolment characteristics for mothers were calculated and presented as mean ± SD. Pearson’s correlation coefficients were calculated to assess the relationship between maternal characteristics and ln AFB‐Lys levels and between maternal characteristics and infant birth weight. T tests were used to compare maternal ln AFB‐Lys levels by foods consumed in the 24‐hr dietary recall. Associations between ln maternal AFB‐Lys levels and infant birth characteristics were assessed using unadjusted and adjusted linear regression models. Covariates with a bivariate association with infant birth weight (p‐value < 0.10) were included in the adjusted models except in cases of collinearity with other covariates. For all adjusted models, the absence of multi‐collinearity was verified using variance inflation factor. For all analyses, p < 0.05 was considered statistically significant.