Performance of urine, blood, and integrated metal biomarkers in relation to birth outcomes in a mixture setting

Abstract

Background: Studies on the health effects of metal mixtures typically utilize biomarkers measured in a single biological medium, such as blood or urine. However, the ability to evaluate mixture effects are limited by the uncertainty whether a unified medium can fully capture exposure for each metal. Therefore, it is important to compare and assess metal mixtures measured in different media in epidemiology studies.

Objectives: The aim of this study was to examine the mixture predictive performance of urine and blood metal biomarkers and integrated multi-media biomarkers in association with birth outcomes.

Methods: In our analysis of 847 women from the Puerto Rico PROTECT Cohort, we measured 10 essential and non-essential metals in repeated and paired samples of urine and blood during pregnancy. For each metal, we integrated exposure estimates from paired urine and blood biomarkers into multi-media biomarkers (MMBs), using intraclass-correlation coefficient (ICC) and weighted quantile sum (WQS) approaches. Using Ridge regressions, four separate Environmental risk scores (ERSs) for metals in urine, blood, MMB_ICC, and MMB_WQS were computed as a weighted sum of the 10 metal concentrations. We then examined associations between urine, blood, and multi-media biomarker ERSs and birth outcomes using linear and logistic regressions, adjusting for maternal age, maternal education, pre-pregnancy body mass index (BMI), and second-hand smoke exposure. The performance of each ERS was evaluated with continuous and tertile estimates and 95% confidence intervals of the odds ratio of preterm birth using area under the curve (AUC).

Results: Pb was the most important contributor of blood ERS as well as the two integrated multi-media biomarker ERSs. Individuals with high ERS (3rd tertile) showed increased odds of preterm birth compared to individuals with low ERS (1st tertile), with 2.8-fold (95% CI, 1.49 to 5.40) for urine (specific gravity corrected); 3.2- fold (95% CI, 1.68 to 6.25) for blood; 3.9-fold (95% CI, 1.72 to 8.66) for multi-media biomarkers composed using ICC; and 5.2-fold (95% CI, 2.34 to 11.42) for multi-media biomarkers composed using WQS. The four ERSs had comparable predictive performances (AUC ranging from 0.64 to 0.68) when urine is examined with specific gravity corrected concentrations.

Conclusions: Within a practical metal panel, measuring metals in either urine or blood may be an equally good approach to evaluate the metals as a mixture. Applications in practical study design require validation of these methods with other cohorts, larger panels of metals and within the context of other adverse health effects of interest.

Keywords: Manganese; Metals; Prenatal stress; Puerto Rico; Social support.

Figure 1.

Schematic plot of study design, sample size, and statistical methods for constructing and evaluating multi-media biomarker (MMB) and Environmental Risk Score (ERS).

Figure 2.

Bar graph of estimated urinary and blood biomarker weights for the MMBs using ICC approach and the WQS models of overall preterm birth. Larger weights indicate greater contributions of the original biomarkers to the MMBs.

Figure 3.

Heat map of weights for each individual metal biomarker extracted by the Ridge regression models regressing birth outcomes on urinary, blood, and two integrated multi-media biomarkers (MMB).

Figure 4.

Odds ratio (OR) of preterm birth associated with Environmental Risk Scores (ERSs) constructed for urine (SG-corrected), blood, and two integrated multi-media biomarkers (MMB). Effect estimates presented as OR for IQR increase in average exposure biomarker concentration. Models were adjusted for maternal age, maternal education, pre-pregnancy BMI, and exposure to secondhand smoking.

Figure 5.

Area under the curves for preterm birth according to environmental risk score (ERS) constructed for urinary (SG corrected), blood, and two integrated multi-media biomarkers (MMB).

Figure 6.

Odds ratio (OR) of preterm birth comparing the highest versus the lowest tertiles of individual metals and environmental risk scores (ERSs) constructed for urine, blood, and two integrated multi-media biomarkers (MMB) mixtures. Models were adjusted for maternal age, maternal education, pre-pregnancy BMI, and exposure to secondhand smoking.