Article

The effect of lipid adjustment on the analysis of environmental contaminants and the outcome of human health risks.

Division of Epidemiology, Statistics and Prevention Research, Eunice Kennedy Schriver National Institute of Child Health and Human Development, NIH, Rockville, MD, USA.
Methods in molecular biology (Clifton, N.J.) (Impact Factor: 1.29). 01/2009; 580:371-81. DOI: 10.1007/978-1-60761-325-1_20
Source: PubMed

ABSTRACT Past literature on exposure to lipophilic agents such as organochlorines (OCs) is conflicting, posing challenges for the interpretation of their potential human health risks. Since blood is often used as a proxy for adipose tissue, it is necessary to model serum lipids when assessing health risks of OCs. Using a simulation study, we evaluated four statistical models (unadjusted, standardized, adjusted, and two-stage) for the analysis of polychlorinated biphenyls (PCBs) exposure, serum lipids, and health outcome risk. Eight candidate true causal scenarios, depicted by directed acyclic graphs, were used to illustrate the ramifications of misspecification of underlying assumptions when interpreting results. Biased results were produced when statistical models that deviated from the underlying causal assumptions were used with the lipid standardization method found to be particularly prone to bias. We concluded that investigators must consider biology, biological medium, laboratory measurement, and other underlying modeling assumptions when devising a statistical model for assessing health outcomes in relation to environmental exposures.

Download full-text

Full-text

Available from: Audrey Jane Gaskins, Sep 10, 2014
0 Followers
 · 
102 Views
  • Source
    • "and Gaskins and Schisterman (2009) "
    [Show abstract] [Hide abstract]
    ABSTRACT: In clinically aggressive diseases, patients experience pathophysiological changes that often alter concentrations of lipids and environmental lipophilic factors; such changes are related to disease signs and symptoms. The aim of the study was to compare the effects of correcting for total serum lipids (TSL) and other clinical factors on the odds of mutations in the K-ras oncogene by organochlorine compounds (OCs), in logistic models, in 103 patients with exocrine pancreatic cancer (EPC) using a causal directed acyclic graph (DAG) framework. Results and likelihood of bias were discussed in the light of possible causal scenarios. The odds of K-ras mutated EPC was associated with some TSL-corrected OCs, including p,p′-DDT (p-value: 0.008) and polychlorinated biphenyl 138 (p-trend: 0.024). When OCs were not corrected by TSL, the OR of a K-ras mutation was significant for p,p′-DDT (p-trend: 0.035). Additionally adjusting for cholestatic syndrome increased the ORs of TSL-corrected OCs. When models were adjusted by the interval from first symptom to blood extraction (ISE), the ORs increased for both TSL-corrected and uncorrected OCs. Models with TSL-corrected OCs and adjusted for cholestatic syndrome or ISE yielded the highest ORs. We show that DAGs clarify the covariates necessary to minimize bias, and demonstrate scenarios under which adjustment for TSL-corrected OCs and failure to adjust for symptoms or ISE may induce bias. Models with TSL-uncorrected OCs may be biased too, and adjusting by symptoms or ISE may not control such biases. Our findings may have implications as well for studying environmental causes of other clinically aggressive diseases.
    Chemosphere 11/2014; 114:219–225. DOI:10.1016/j.chemosphere.2014.04.066 · 3.50 Impact Factor
  • Source
    • "In the first set of models ( Model A ) adjustment was performed for gender , BMI , serum tri - glycerides and cholesterol . According to the discussion by ( Gaskins and Schisterman , 2009 ) , we adjusted for serum lipids in the statis - tical analysis rather than normalizing the POP levels for lipids . In the next set of models ( Model B ) we further adjusted for education ( three levels ) , exercise habits ( four levels ) and smoking habits ( cur - rent smoking ) . "
    [Show abstract] [Hide abstract]
    ABSTRACT: Background: Persistent organic pollutants (POPs) have been suggested to be linked to obesity. We have previously shown that less-chlorinated PCBs were positively related to fat mass, while highly-chlorinated PCBs were inversely related to obesity. Objective: The aim of the present evaluation is to investigate the relationship between retrospective assessed life-time change in body weight (20-70years) with circulating POP levels measured at age 70years. Methods: 1016 subjects aged 70years were investigated in the Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUSs) study. 16 PCBs and 3 OC pesticides were analyzed using HRGC/HRMS. Current body weight was measured and participants self-reported their weight at age 20. Results: The average estimated weight change over 50years was 14.4kg. Both the sum of OC pesticide concentrations (4.3kg more weight gain in quintile 5 vs. quintile 1, p<0.0001) and the sum of the less-chlorinated PCBs were positively related to the estimated weight change (3.7kg more weight gain in quintile 2 vs. quintile 1, non-linear relationship p=0.0015). In contrast, the sum of concentrations of highly-chlorinated PCBs were inversely related to estimated weight change (8.4kg less weight gain in quintile 5 vs. quintile 1, p<0.0001). Conclusion: High levels of OC pesticides and the less-chlorinated PCBs at age 70 were associated with a pronounced estimated weight change over the previous 50years. However, the opposite was seen for highly-chlorinated PCBs. Differences in mode of action, toxicokinetics, non-linear relationships and reverse causation might explain these discrepancies.
    Chemosphere 08/2012; 90(3). DOI:10.1016/j.chemosphere.2012.07.051 · 3.50 Impact Factor
  • Source
    • "The results of analyses that modeled lipid-adjusted transnonachlor and oxychlordane produced results similar to those presented. We included serum lipids as a covariate in our model rather than using lipid-standardized chlordane concentrations because the latter may be prone to bias, depending on the underlying mechanism of the chlordane–lipid disease association (Gaskins and Schisterman 2009). All models included trans-nonachlor or oxychlordane as the main exposure and were adjusted for total PCBs, p,p´-DDE, triglycerides, and cholesterol as a priori selected variables. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The etiologies of the male urogenital anomalies--cryptorchidism and hypospadias--are poorly understood. Given positive associations between chlordane isomers and testicular germ cell tumors, it is reasonable to assume that chlordanes might also be associated with other testicular dysgenesis syndrome disorders, namely cryptorchidism and hypospadias. To examine whether exposure to in utero chlordane is related to cryptorchidism and hypospadias, we evaluated levels of chlordane derivatives, trans-nonachlor and oxychlordane, among pregnant women enrolled in the Collaborative Perinatal Project (CPP). From 1959 to 1965, the CPP enrolled pregnant women at 12 U.S. medical centers. We analyzed serum trans-nonachlor and oxychlordane levels measured in third-trimester serum from the mothers of 217 sons with cryptorchidism, 197 sons with hypospadias, and 557 sons with neither condition. Adjusted odds ratios (ORs) and 95% confidence intervals were calculated using conditional logistic regression. The quartile-specific ORs for cryptorchidism or hypospadias show no notable associations with trans-nonachlor or oxychlordane. Further, there were no significant trends with increasing quartile of maternal trans-nonachlor or oxychlordane level in either cryptorchidism or hypospadias (p-trend all > 0.45). The results do not support an association between chlordane levels and cryptorchidism or hypospadias. It is unlikely that current chlordane exposure is related to the development of either anomaly, given that serum chlordane levels at the time of sample collection, the early 1960s, were considerably higher than levels at present.
    Environmental Health Perspectives 03/2012; 120(3):478-82. DOI:10.1289/ehp.1103936 · 7.98 Impact Factor
Show more