Biomarkers of maternal environmental enteric dysfunction are associated with shorter gestation and reduced length in newborn infants in Uganda

Background Adverse birth outcomes, including preterm birth and stunting at birth, have long-term health implications. The relation between adverse birth outcomes and chronic, asymptomatic gastrointestinal inflammation (environmental enteric dysfunction – EED) is poorly understood. Objective We aimed to examine the relation between maternal EED and adverse birth outcomes in a sample of pregnant Ugandan women and their newborn infants. Design We conducted a prospective cohort study in Mukono, Uganda. A total of 258 pregnant women were enrolled at their first prenatal visit (~18 weeks of gestation). EED was measured by urinary lactulose:mannitol (L:M) ratio and serum concentrations of antibodies to the bacterial components flagellin and LPS. Covariates were obtained from survey data collected at 2 time points. Associations were assessed through the use of unadjusted and adjusted simple linear regression models. Results Complete birth outcome data were recorded for 220 infants within 48 h of delivery. Mean ± SD gestational age was 39.7 ± 2.1 wk, and 7% were born preterm. Mean ± SD length and length-for-age z score (LAZ) at birth were 48.1 ± 3.2 cm and -0.44 ± 1.07, respectively. L:M ratio was not associated with any birth outcome. In adjusted models, higher concentrations of natural log-transformed anti-flagellin immunoglobin G (IgG) and anti-LPS IgG were significantly associated with shorter length of gestation (β: -0.89 wk; 95% CI: -1.77, -0.01 wk, and β: -1.01 wk; 95% CI: -1.87, -0.17 wk, respectively) and with reduced length (β: -0.80 cm; 95% CI: -1.55, -0.05 cm, and β: -0.79 cm; 95% CI: -1.54, -0.04 cm, respectively) and LAZ at birth (β -0.44 z score; 95% CI: -0.83, -0.05, and β: -0.40 z score; 95% CI: -0.79, -0.01, respectively). Conclusion Maternal anti-flagellin and anti-LPS IgG concentrations in pregnancy, but not L:M ratio, were associated with shorter gestation and reduced infant length at birth. Further research on the relation between maternal EED and birth outcomes is warranted. We performed a prospective cohort study between February and November 2017 in Mukono District, Central Region, Uganda. Mukono is a semiurban district situated 20 km east of the capital city, Kampala. The study was based at Mukono Health Center IV (MHC IV), a public outpatient health facility located in the center of Mukono Town. Pregnant women were recruited during their first prenatal visit at MHC IV, which occurred at ∼18 weeks of gestation. Eligible women were 18–45 y of age, residing within 10 km of Mukono Town, and carrying a singleton pregnancy. Women were excluded from the study if they were <18 or >45 y old, HIV positive (verified via routine rapid HIV test conducted at the first prenatal visit), severely malnourished [defined as BMI (kg/m2) <16.0], severely anemic (defined as hemoglobin <7 g/dL), or planning to move away from Mukono District before delivery. The study was approved by the Tufts Health Sciences Institutional Review Board in Boston, MA; the Mengo Hospital Research Ethics Committee in Kampala, Uganda; and the Uganda National Council for Science and Technology in Kampala, Uganda. Before enrollment, written consent in either Luganda or English was obtained from each participant. Participation in the study involved 4 visits over a 4–6-mo period: enrollment visit, immediately after the first prenatal visit (MHC IV); L:M test visit, within 1 wk of the first visit (participant's residence); follow-up visit, 3 wk before the expected date of delivery (also at the participant's residence); and postdelivery visit, within 48 h of delivery (either at the participant's residence, MHC IV, or another health facility). An ultrasound scan was performed by a trained professional at MHC IV to both confirm a singleton pregnancy and determine participants’ estimated date of delivery. Hemoglobin was measured with a portable hemoglobinometer (HemoCue Hb 301; HemoCue, Inc., Brea, CA). A venous blood draw was performed by the phlebotomist at MHC IV (BD Vacutainer, Becton Dickinson, Durham, NC). Systolic and diastolic blood pressure (DBP) measurements were taken with a digital upper arm blood pressure monitor (Omron 10 Series, Omron Healthcare, Kyoto, Japan). All anthropometry measurements were performed in triplicate and the mean was used for analysis. Weight was measured to the nearest 0.1 kg with the use of a digital weight scale (Seca 874, Hanover, MD). Height was measured to the nearest 0.1 cm with the use of a portable, rigid height board (Infant/Child/Adult ShorrBoard, Shorr Production, Olney, MD). These measurements were used to calculate BMI. Midupper arm circumference (MUAC) was measured to the nearest 0.1 cm with the use of a standard tricolored, nonstretch adult MUAC tape. 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 (HFIAS)] (27), and water, sanitation, and hygiene practices (see Supplemental File 1 for the complete questionnaire). Within 1 wk of the enrollment visit, a household visit was conducted to perform a L:M dual sugar absorption test. Participants with diarrhea (≥3 loose stools/d) in the last 2 wk had their test rescheduled for a different day. After urination to void the bladder and an observed 1-h fast, participants consumed a 50-mL solution containing 5 g of lactulose (Lactulose Solution; Mckesson, San Francisco, CA) and 2 g of mannitol (D-mannitol powder; Sigma-Aldrich, St. Louis, MO) completely dissolved in sterile water. Urine was collected for a period of 4 h in a 2-L plastic collection bottle containing 0.05 mL of 50% thimerosal (Sigma-Aldrich, St. Louis, MO) as a preservative. Water intake was permitted ad libitum 1 h after ingestion of the solution, and women were encouraged to drink a minimum of 500 mL of water during the test to ensure sufficient urine output. A final urine sample was collected at the 4-h time-point and total urine volume was measured to the nearest 1.0 mL with the use of a graduated cylinder in the field. Samples were frozen at −20°C at the MHC IV laboratory before being transferred to a −80°C freezer in Kampala. A second household visit was conducted 3 wk before participants’ estimated delivery date, which consisted of a follow-up survey with questions related to pregnancy risk factors. In addition, weight and MUAC measurements were taken following identical procedures to those used at the enrollment visit. Finally, participants were asked to provide a sample of water from their drinking water storage container for the purposes of a water quality test. Infant characteristics (live birth, date and time of delivery, sex, weight, length, and head circumference) were collected within 48 h of delivery. Birth weight was measured to the nearest 0.1 kg with the use of a digital weigh scale (Seca 874, Hanover, MD), and birth length was measured to the nearest 0.1 cm with the use of a portable, rigid height board (ShorrBoard, Shorr Production, Olney, MD). Head circumference was measured to the nearest 0.1 cm with a flexible measuring tape. All anthropometry measurements were taken in triplicate and averaged. In the case of a stillbirth or neonatal death, only birth date, time, and infant sex were recorded. Urine samples were analyzed for concentrations of lactulose and mannitol with the use of previously described HPLC methods at the Shulman Laboratory at Baylor College of Medicine (28). The L:M ratio was calculated by dividing the urinary lactulose concentration by the urinary mannitol concentration. Lactulose (%LE) and mannitol excretion ratios were calculated from the measured amount of each in urine (concentration × total urine volume) relative to the initial dose of each sugar. Blood samples were centrifuged at 1900 × g for 5 min at room temperature, and serum was divided into aliquots in 2.0-mL clear plastic cryovials. Anti-flagellin and anti-LPS Ig concentrations (IgA and IgG) were measured at the Gewirtz Laboratory at Georgia State University via previously described ELISA methods (19). Concentrations of serum biomarkers are reported as optical density units throughout. Water quality was assessed in the field through the use of a compartment bag test (Aquagenx, Chapel Hill, NC). One hundred mL of drinking water from each household was mixed with an Escherichia coli chromogenic growth medium and incubated for a period of 48 h inside a sealed plastic bag containing 5 compartments of varying volumes. Risk categories were determined by noting which, if any, compartments changed from yellow to green/blue during the incubation period and matching that to a most probable number table based on the WHO guidelines (29). Because there is no safe level of E. coli contamination, a dichotomous (safe/unsafe) water variable was created, corresponding to no E. coli detected and any E. coli detected, respectively. Based on studies of maternal idiopathic inflammatory bowel diseases and adverse birth outcomes, the RR of maternal EED and subsequent preterm birth was assumed to be 2.0 (30, 31). Assuming 80% power, 0.05 significance, a 5% frequency of preterm birth, and 15% loss to follow-up, a sample size of 258 allowed for the detection of an RR of 2.0 within 50% of the true risk parameters. All analyses were carried out with the use of STATA 15 software (Stata Corps, College Station, TX). Weight and length measurements were converted to z scores for weight-for-age (WAZ), LAZ, and weight-for-length (WLZ) with the use of the WHO standards (1). Outliers were defined as WAZ <−6 or >+5, WLZ <−5 or >+5, and LAZ <−6 or >+6. On this basis, 7 observations were excluded from WLZ analyses; none of the other measures had any outliers. Stillbirth was defined as fetal death after 20 weeks of gestation, and preterm birth was defined as a live birth before 37 weeks of gestation. Low birth weight (LBW) was defined as weighing <2500 g at birth, and SGA was defined as birth weight for gestational age <10th percentile via sex-specific INTERGROWTH-21st standards (32). Stunted and wasted at birth were defined as LAZ <−2 and WHZ <−2, respectively. Before analysis, distributions of biomarker values were assessed for outliers and normality. Because of their skewed distribution, L:M ratios, %LE, and serum biomarkers were natural log transformed (ln) in all regression models. Associations between EED biomarkers (continuous, independent variables) and birth outcomes [i.e., gestational age, length, and LAZ (continuous, primary outcomes) and weight, WAZ, WLZ, and head circumference (continuous, secondary outcomes)] were assessed via simple unadjusted and adjusted linear regression models. For adjusted models, covariates were selected via bivariate analyses, with gestational age at birth and birth length as dependent variables, and a P value <0.25. Before inclusion in the model, correlation among covariates was assessed through the use of Pearson correlation coefficients, and the absence of multicollinearity was verified with the use of the variance inflation factor. t Tests were used to assess differences in maternal biomarker means for infants born stillborn, preterm, LBW, SGA, stunted, and wasted compared with not. Pearson correlations were calculated to evaluate agreement between ln L:M ratio and ln serum biomarkers. In all cases, statistical significance was determined by a P value <0.05.