Statistical Challenges in the Analysis of Biomarkers of Environmental Chemical Exposures for Perinatal Epidemiology
- 115 Downloads
Purpose of Review
Biomarkers are widely used in perinatal epidemiology to examine the health effects of environmental chemical exposures during pregnancy. These measurements take the form of chemical concentrations measured in blood, urine, or other biospecimens. Biomarkers have the advantage of providing objective estimates of chemical exposures from multiple sources. However, they are difficult to handle at the data analysis stage. We review recent trends and developments in the statistical analysis of biomarkers with particular emphasis on exposure assessment and multivariable modeling.
Six statistical challenges are presented in the recent literature: (1) the analysis of biomarkers that fall below the limit of detection, (2) adjustment for dilution-dependent sample variation, (3) handling repeated biomarker measurements within a single pregnancy, (4) accounting for heterogeneity in biomarker levels between chemicals within the same chemical class, (5) variable selection and shrinkage for biomarkers in the same class, and finally, (6) dimension reduction strategies including the sum-of-chemical approach.
The analysis of biomarkers of environmental chemical exposures remains immensely difficult, and the proper application of emerging statistical techniques requires input from experts in diverse disciplines. We highlight specific gaps in the literature where innovation in statistical methods is required.
KeywordsBiomarkers Environmental exposures Epidemiology Biostatistics Children’s health
The authors gratefully acknowledge input and thoughtful comments from Tim Takaro and Ryan Allen on an earlier version of the manuscript.
JH and LM drafted the manuscript with input from HZ and BB. All authors approved the final version of the manuscript.
This project was funded by a Catalyst Grant from the Canadian Institutes for Health Research (L-CIP-150736).
Compliance with Ethical Standards
Conflict of Interest
The authors declare that they have no conflicts of interest.
Human and Animal Rights and Informed Consent
This article does not contain any studies with human or animal subjects performed by any of the authors.
Papers of particular interest, published recently, have been highlighted as: • Of importance •• Of major importance
- 2.Govarts E, Nieuwenhuijsen M, Schoeters G, Ballester F, Bloemen K, De Boer M, et al. Birth weight and prenatal exposure to polychlorinated biphenyls (PCBs) and dichlorodiphenyldichloroethylene (DDE): a meta-analysis within 12 European Birth Cohorts. Environ Health Perspect. National Institute of Environmental. Health Sciences. 2012;120:162.Google Scholar
- 3.Lenters V, Portengen L, Rignell-Hydbom A, Jönsson BAG, Lindh CH, Piersma AH, et al. Prenatal phthalate, perfluoroalkyl acid, and organochlorine exposures and term birth weight in three birth cohorts: multi-pollutant models based on elastic net regression. Environ Health Perspect. 2016;124:365–72.PubMedCrossRefGoogle Scholar
- 6.Savitz DA, Wellenius GA. Exposure biomarkers indicate more than just exposure. Am J Epidemiol 2017; Available: https://academic.oup.com/aje/advance-article-abstract/doi/10.1093/aje/kwx333/4636591
- 10.Thomas DC. Statistical methods in environmental epidemiology: Oxford University Press; 2009.Google Scholar
- 12.Braun JM, Gray K. Challenges to studying the health effects of early life environmental chemical exposures on children’s health. PLoS Biol Public Library of Science. 2017;15:e2002800.Google Scholar
- 14.Rothman KJ, Greenland S, Lash TL. Modern epidemiology. 3rd. Philadephia: Lippincott Williams & Wilkins. 2008;Google Scholar
- 15.Weisskopf MG, Webster TF. Trade-offs of personal versus more proxy exposure measures in environmental epidemiology. Epidemiology. 2017;28: 635–643.Google Scholar
- 17.Hewett P, Ganser GH. A comparison of several methods for analyzing censored data. Ann Occup Hyg 2007;51: 611–632.Google Scholar
- 27.Rubin D. B,(1987) Multiple imputation for nonresponse in survey. New York: Wiley. New York: Wiley;Google Scholar
- 33.•• Chen Y-H, Ferguson KK, Meeker JD, McElrath TF, Mukherjee B. Statistical methods for modeling repeated measures of maternal environmental exposure biomarkers during pregnancy in association with preterm birth. Environ Health. 2015;14:9. This paper compared nine statistical models that are useful in assessing exposure and examining the dose-response relationship between repeated measures of maternal environmental exposure biomarkers and preterm birth. CrossRefPubMedPubMedCentralGoogle Scholar
- 38.Holland N, Huen K, Tran V, Street K, Nguyen B, Bradman A, et al. Urinary phthalate metabolites and biomarkers of oxidative stress in a Mexican-American cohort: variability in early and late pregnancy. Toxics. 2016;4 https://doi.org/10.3390/toxics4010007.
- 39.•• O’Brien KM, Upson K, Cook NR, Weinberg CR. Environmental chemicals in urine and blood: improving methods for creatinine and lipid adjustment. Environ Health Perspect. 2016;124:220–7. Using directed acyclic graphs and simulations, this study compared adjustment methods for dilution-dependent creatinine and serum lipids and recommended the novel method for urine dilution and traditional method for serum lipids biomarkers. PubMedCrossRefGoogle Scholar
- 42.Costanza MC, Cayanis E, Ross BM, Flaherty MS, Alvin GB, Das K, Morabia A Relative contributions of genes, environment, and interactions to blood lipid concentrations in a general adult population. Am J Epidemiol academicoupcom; 2005;161: 714–724.Google Scholar
- 43.• Calafat AM. Contemporary issues in exposure assessment using biomonitoring. Curr Epidemiol Rep. 2016;3:145–53. This review described factors that affect biomarkers of exposure such as biomarker selection, variability in biomarker concentrations, biomarker collection, and storage issues. CrossRefPubMedPubMedCentralGoogle Scholar
- 44.• Fisher M, Arbuckle TE, Mallick R, LeBlanc A, Hauser R, Feeley M, et al. Bisphenol A and phthalate metabolite urinary concentrations: daily and across pregnancy variability. J Expo Sci Environ Epidemiol. 2015;25:231–9. This study assessed chemical variability in urine sample over time using intraclass correlation coefficients (ICCs) and showed that multiple biomarker of exposure measurements collected throughout the pregnancy better represent average exposures across pregnancy compared to single spot measure. CrossRefPubMedGoogle Scholar
- 45.•• Perrier F, Giorgis-Allemand L, Slama R, Philippat C. Within-subject pooling of biological samples to reduce exposure misclassification in biomarker-based studies. Epidemiology. 2016;27:378–88. This study, using simulation, showed that as intraclass correlation coefficients (ICCs) of chemicals decrease, the number of biospecimens needed to limit bias increases and that within-subject pooling of biospecimens before assaying the chemical can reduce exposure misclassification without increasing assay costs. CrossRefPubMedPubMedCentralGoogle Scholar
- 46.• Vernet C, Philippat C, Calafat AM, Ye X, Lyon-Caen S, Siroux V, et al. Within-day, between-day, and between-week variability of urinary concentrations of phenol biomarkers in pregnant women. Environ Health Perspect. 2018;126:037005. This study showed that, during pregnancy, phenol biomarkers have a strong within-day variability and that one biospecimen does not represent the average exposure of phenol throughout the pregnancy. CrossRefPubMedPubMedCentralGoogle Scholar
- 49.Frederiksen H, Kranich SK, Jørgensen N, Taboureau O, Petersen JH, Andersson A-M. Temporal variability in urinary phthalate metabolite excretion based on spot, morning, and 24-h urine samples: considerations for epidemiological studies. Environ Sci Technol. 2013;47:958–67.CrossRefPubMedGoogle Scholar
- 52.Zeger SL. Invited commentary: epidemiologic studies of the health associations of environmental exposures with preterm birth. Am J Epidemiol 2012;175: 108–110; discussion 111–3.Google Scholar
- 55.•• Agier L, Portengen L, Chadeau-Hyam M, Basagaña X, Giorgis-Allemand L, Siroux V, et al. A systematic comparison of linear regression-based statistical methods to assess exposome-health associations. Environ Health Perspect. 2016;124:1848–56. This simulation study evaluated the performance of various linear regression-based statistical methods in exposome studies where it is difficult to discern between true predictors and correlated exposures. CrossRefPubMedPubMedCentralGoogle Scholar
- 62.Gelman A, Park DK. Splitting a predictor at the upper quarter or third and the lower quarter or third. Am Stat Taylor Francis. 2009;63:1–8.Google Scholar
- 64.Harrell FE. Regression modeling strategies: with applications to linear models, logistic regression, and survival analysis. Springer Science & Business Media; 2013.Google Scholar
- 67.Schisterman EF, Little RJ. Opening the black box of biomarker measurement error. Epidemiology 2010;21 Suppl 4: S1–S3.Google Scholar
- 68.•• Patel CJ. Analytic complexity and challenges in identifying mixtures of exposures associated with phenotypes in the exposome era. Curr Epidemiol Rep. 2017;4:22–30. This review described the challenges in identifying co-occurring exposures and discussed machine-learning and data analytics methods useful in narrowing down correlated exposures. CrossRefPubMedPubMedCentralGoogle Scholar
- 69.•• Weisskopf MG, Seals RM, Webster TF. Bias amplification in epidemiologic analysis of exposure to mixtures. Environ Health Perspect. 2018;126:047003. This study found that bias increases when outcome model included highly correlated mixtures or more mixture components. CrossRefPubMedPubMedCentralGoogle Scholar
- 71.Keil AP, Daza EJ, Engel SM, Buckley JP, Edwards JK. A Bayesian approach to the g-formula. Stat Methods Med Res. 2017;962280217694665Google Scholar
- 74.James G, Witten D, Hastie T, Tibshirani R. An introduction to statistical learning: with applications in R. Springer New York; 2013.Google Scholar
- 83.Carrico C, Gennings C, Wheeler DC, Factor-Litvak P. Characterization of weighted quantile sum regression for highly correlated data in a risk analysis setting. JABES Springer US. 2014;20:100–20.Google Scholar
- 87.Hernán MA, Robins JM. Causal inference. CRC Boca Raton, FL:; 2010.Google Scholar