The ELISA was performed using 96-well high binding micro titer plates coated with coating antigen cAg06 , washed with 10mM phosphate buffered saline + 0.05% Tween 20 , blocked with 0.5% bovine serum albumin in PBS. All standards, quality control samples and urine samples were prepared in 10% MeOH in PBS prior to the assay procedure. A 10-point calibration curve was prepared for each assay using a serial 1:5 dilution from the highest standard solution. A blank 0 ng/mL standard in the same 10% MeOH/PBS solution was also prepared. Wells containing instrument blanks received a 50 uL aliquot of PBST, while all remaining wells received a 50 uL aliquot of anti-rabbit 3PBA antibody #294 diluted 1:7000 in PBST. The goat anti-rabbit IgG-horseradish peroxidase conjugate was diluted 1:3000 in PBST. The substrate solution in dimethylsulfoxide per 25 mL of acetate buffer, pH 5.5 was added and color development was stopped after 15 min with 2 M H2SO4. The absorbance was measured using a Vmax micro plate reader in dual wavelength mode at 450 nm – 650 nm. Urinary creatinine concentrations were determined using the methods described in Ahn et al. 2011 .All urine samples, blank samples and QC samples were analyzed in triplicate. The final concentrations were calculated from the means of the triplicate values. If one of the triplicate values fell below the LOD it was not considered in the calculation of the mean concentration. Mean values that fell between the LOD and the LOQ were retained in all further statistical analyses. For each set of extractions,horticulture rack one urine sample was randomly chosen for QC analysis and was spiked to a level of 10 ng/mL 3PBA before extraction in order to verify that there were no matrix effects from the sample extract and to check method accuracy. Five different urine samples were also extracted and analyzed in duplicate at different time points in the analysis to verify the method precision.
For further validation of the ELISA method, and in order to try to quantify any cross reactivity that may have occurred, a set of six urine samples was sent to Emory University in Atlanta, GA where they were analyzed by high performance liquid chromatography-tandem mass spectrometry using a well established method .Summary statistics for both the volume based and creatinine adjusted 3PBA data were calculated. For concentrations below the limit of detection , an imputed value was assigned equal to the LOD divided by the square root of 2 . To determine predictive variables from the questionnaire data to include in a multivariate analysis, linear regression with both the log-transformed creatinine concentration and the variable of interest, referred to as the bivariate comparisons, were performed for each variable with 3PBA concentration. For each type of pesticide application, we created a continuous variable that represented the number of applications per year as well as a categorical variable indicating if that type of pesticide had been applied. Additional variables were also created summing pesticide applications across different application types. Food diaries were translated to English, and individual food items were grouped into ten different food categories: Fruit, Vegetables, Legumes, Meats, Snack/Processed Foods, Dairy, Beverages, Grains, Mixed Foods , and Other. Each category also had subcategories for specific, popular food items. For example, in the Dairy category, subcategories included Milk, Cheese and Other Dairy. Two variables were created for each food category and sub-category. First a categorical variable that indicated if the participant had consumed an item from a given category or sub-category. Second, a continuous variable was created with the number of servings in each category or sub-category. Volume based 3PBA concentrations were log-transformed to better approximate a normal distribution and regressed against the log-transformed creatinine concentrations and the individual questionnaire variable . Two regression analyses were conducted for each variable, one including only data from the mothers and one including only data from the children . There were a number of variables related to individual measures of home disrepair that were significant in the bivariate analysis. Because many of these measures of disrepair were correlated, item analysis was performed to select and evaluate the internal consistency of a set of items for a summative scale score.
The resulting Home Disrepair Score is computed by summing the water damage, leaks, carpet damage, worn spots or holes in the counters and rotten wood indicators and has good internal consistency in our sample . Multivariate regression was then performed to evaluate which questionnaire variables were most predictive of the urinary 3PBA concentrations. Three models were fit, one with the data from both the mothers and the children, one with data only from the children, and one with data only from the mothers. As an alternative, we also ran the models using the metabolite concentrations directly adjusted by the creatinine concentration. All statistical analyses were performed using SAS version 9.2 .The resulting Home Disrepair Score is computed by summing the water damage, leaks, carpet damage, worn spots or holes in the counters and rotten wood indicators and has good internal consistency in our sample . Multivariate regression was then performed to evaluate which questionnaire variables were most predictive of the urinary 3PBA concentrations. Three models were fit, one with the data from both the mothers and the children, one with data only from the children, and one with data only from the mothers. As an alternative, we also ran the models using the metabolite concentrations directly adjusted by the creatinine concentration. All statistical analyses were performed using SAS version 9.2 .Women in this study ranged in age from 23 to 51 years old; they had very low educational levels, with 46% having only a 6th grade education or lower; they were almost all married and lived in homes with 4 or more residents . Pest problems were common with 59% reporting insect problems, 43% using pesticides indoors and 35% applying pesticides outside . A Spearman rank correlation analysis was performed to see if there were associations between the two measures of home disrepair with pesticide application. Multiple significant correlations were observed , suggesting that poor housing conditions do lead to higher rates of pesticide application. For each set of three triplicates, the sample standard deviation of the urinary 3PBA concentration was computed. The average 3PBA concentration was 2.51 with an average standard deviation of 0.42 ng/mL. Less than 10% of the samples had a triplicate that fell below the LOD resulting in that value being dropped from the calculation of the mean.
Recoveries of the fortified urine samples ranged from 67 to 111% with an average of 82 ± 12%. The LOD of this analysis was estimated to be 0.1 ng of 3PBA in 1 mL urine. The limit of quantitation was determined to be 2 ng/mL. The percent difference between concentrations in duplicate aliquots of selective urine samples ranged from 3.6 to 28% with an average of 14%. Six urine samples were also analyzed by HPLC-MS/MS to further validate the ELISA method. The square of the correlation coefficient between the 3PBA concentrations from the two laboratory methods for the six samples tested was R2 = 0.934, and the %D ranged from 7.9 to 30.6% with an average of 25%, with the ELISA resulting in higher concentrations in four of the samples, and lower concentrations in two of the samples. Urinary 3PBA concentrations in our study were detected in 80% of all samples with a range of 0.3–13 ng/mL .However, adjustment for urinary creatinine resulted in a significantly higher concentration of urinary metabolites in children than in mothers . We calculated the correlation of urinary 3PBA concentrations between mothers and children. Urinary 3PBA concentrations from mothers and children in the same household were positively correlated for both volume based and creatinine adjusted concentrations .Variables included in the multivariate analysis were based upon the results of the bivariate analysis. The Home Disrepair Score, derived from the combination of multiple questionnaire items,vertical grow system was significant in the bivariate analysis only for mothers, while the Inside Housing Conditions score, derived from the staff evaluation during the MICASA follow-up interview, was significant only for the children. Because these two scores were designed to measure similar housing characteristics, both scores along with the Outdoor Spray pesticide use variable from the MICASA baseline questionnaire and the log-transformed creatinine concentrations were included in all three multivariate models described below. Multivariate models assessed factors associated with 3BPA concentrations in the combined sample , children only and mothers only . Both the Home Disrepair Score and Outdoor Spray were positive significant predictors of urinary 3PBA levels in the total study population model, which included logtransformed creatinine, the Home Disrepair Score, Outdoor Spray, Inside Housing Conditions and a Mother/Child variable .
The model restricted to children included food diary variables significant in the bivariate model: Apple , Milk , All Meat and Cereal as well as the log-transformed creatinine, the Home Disrepair Score, Outdoor Spray and Inside Housing Conditions. In this model Outdoor Spray and Inside Housing Conditions were marginally significant positive estimators of urinary 3PBA concentration. Cereal Total, while marginally significant, was negatively associated with urinary 3PBA in the children only data. In the mother only model we included the food diary variables Eggs , Beans , Grapes , Chicken , and Cereal as well as log-transformed creatinine, the Home Disrepair Score, Outdoor Spray and Inside Housing Conditions. The Home Disrepair Score , Outdoor Spray , and Cereal Total were all significant positive estimators of urinary 3PBA levels in the mothers. The models with the metabolite concentrations directly adjusted for creatinine resulted in similar associations .We assessed the exposure to pyrethroid pesticides in 105 women and 103 children in a farm worker population by laboratory measurements of the metabolite 3PBA in urine samples and by questionnaire data. This population had a relatively low educational level, with only about half of the adult women participating in our study reporting a 6th grade education or higher, in contrast to the 85% of U.S. adults who have a high school diploma . The results from the ELISA method used to analyze urinary 3PBA concentrations in this study were validated by a more traditional instrumental method at the Emory University in Atlanta, GA. The suitability of this newer analytical technique for use in biological monitoring of 3PBA is further corroborated by Chuang et al. who analyzed over 100 urine samples and showed high correlation between ELISA and GC/MS data, with the square of the linear correlation coefficient R2 = 0.906 and no significant between the two methods of analysis for any given sample . Children had higher concentrations of this metabolite than their mothers. This result is consistent with multiple pesticide exposure studies and most likely has to do with differences in behavior that leads to higher non-dietary ingestion in children . Once pyrethroid pesticides have entered the home, the carpets and cushioned furniture can act as repositories for pesticides . High levels of pesticides in carpet dust is a particular concern for young children who, due to their continual exploration of their environments, spend a large amount of time on the floor and have increased hand to mouth activity, resulting in increased exposure to the pollutants through dermal and non-dietary ingestion routes . The median concentrations in the National Health and Nutrition Examination Survey , a population based sample, collected from 1999 to 2000 were 0.30 and 0.26 ug/g creatinine for children and adults , respectively, and only increased slightly to 0.33 and 0.30 ug/g creatine for children and adults, respectively, in samples collected from 2001 to 2002 . Median urinary 3PBA concentrations of NHANES samples collected from 2007 to 2008 increased a bit more to 0.42 and 0.38 ug/g creatinine for children and adults, respectively . In the Children’s Total Exposure to Persistent Pesticides 2000 to 2001 study, a population based sample of preschool children living in Ohio, the median level was 0.32 ug/g creatinine for children aged 1–5 years . The 2004 Casa y Campo study, a community-based project aiming to reduce pesticide exposure among farm workers and their families in eastern North Carolina, reported median concentrations of 3PBA in children aged 1–6 years of 0.15 ug/g creatinine .