Urinary BPA concentrations, specific gravity, and semen quality parameters from samples collected at the same clinic visit were available for 190 men. Of these men, a second urine sample was later collected from 78 of them, and a third urine sample was collected from 4 men. As a result, the total number of urine samples collected and analyzed for BPA was 272. The amount of time between consecutive urine samples ranged from 3 to 75 days, with a median (25th, 75th percentile) of 29 (27, 34) days. Demographic variables stratified by semen quality reference values are presented in . None of the demographic variables significantly differed between semen quality groups (p-values > 0.05). The distribution of semen quality, sperm motion, and sperm DNA damage measures are presented in . Because the comet assay was introduced into the study at a later time than the assessment of semen quality, DNA damage measures were only available for 132 of the men. The distribution of urinary BPA concentrations for the 194 urine samples collected on the same day as the semen sample are presented in . BPA was detected in 89% of urine samples analyzed in the study. The geometric mean BPA concentration in the present study (1.4 ng/mL) was significantly lower than the geometric mean concentration from adult men in the same age range (18-55 years) from the US general population reported in NHANES 2005-06 (2.3 ng/mL; p<0.00001). Among the 78 men from whom 2 urine samples were collected, BPA concentrations in the two samples were weakly correlated (Spearman r = 0.18; p-value = 0.10). Limiting this analysis to urine samples collected within the median duration between repeated urine sample collection within the same individual (29 days) did not significantly strengthen this correlation.
Demographic categories by semen parameters (N = 190 men)
Distribution of semen quality, sperm motion parameters, and comet assay-measured DNA damage (N = 190).
Distribution of uncorrected and SG-corrected BPA in urine collected from men at initial clinic visit in the present study, and in men 18-55 years of age from NHANES 2005-06 (ng/mL)
When considering semen quality, sperm motion parameters, and DNA damage measures as continuous variables, BMI was inversely associated with Tail% and abstinence period was positively associated with total sperm count and sperm concentration (p-values < 0.05). Geometric mean uncorrected urinary BPA concentration was also higher among men whose urine sample was collected in the afternoon (1.6 ng/mL) compared to men providing the urine sample in the morning (1.1 ng/mL; p=0.02). The difference in BPA concentrations between urine samples collected in the afternoon and morning was slightly lessened when comparing SG-corrected concentrations (1.9 ng/mL vs. 1.5 ng/mL; p=0.09).
Crude logistic and linear regression results (not shown) were similar to the adjusted results presented in and . All logistic/linear regression results in and were adjusted for urinary specific gravity, age, BMI, current smoking, abstinence period, and time of day of clinic visit (time of urine/semen sample collection). As shown in , there were no consistent relationships between urinary BPA concentrations and odds for below reference semen quality parameters, though odds ratios were somewhat elevated for below reference sperm concentration in statistical approach 1 (first column of odds ratios) and for below reference sperm motility and morphology in statistical approach 4 (fourth column of odds ratios). Results from statistical approach 3 (third column of odds ratios) differed as reduced odds were found, though the reported odds ratios were not statistically significant and considered unstable due to the reduced sample size.
Table 4 Adjusted odds ratios (95% confidence intervals) for below reference semen quality parameter associated with an interquartile range increase in urinary BPA concentration. Adjusted for specific gravity, age, BMI, abstinence period, current smoking, and (more ...)
Table 5 Adjusted linear regression coefficients (95% confidence interval) for change in semen quality or sperm DNA damage measure associated with an interquartile range increase in urinary BPA concentration. Adjusted for specific gravity, age, BMI, abstinence (more ...)
In multivariable linear regression models using only urinary BPA concentrations measured in the urine sample collected on the same day as the semen sample (statistical approach 1, first column of regression coefficients in ), we observed inverse associations between urinary BPA concentrations and sperm concentration, motility, morphology, VSL and VCL. For the median values of sperm concentration (64 million/mL), motility (49% motile), and morphology (7% normal) for the study population, the coefficients represent 23% (95%CI -40% to -0.3%), 7.5% (95%CI -17% to +1.5%), and 13% (95%CI -26% to -0.1%) declines in these parameters, respectively, for an IQR increase in urinary BPA concentration (IQR 0.8 to 2.5 ng/mL). Urinary BPA concentration was also positively associated with Tail%, where an IQR increase in concentration was associated with a 10% (95% CI 0.03% to 19%) increase in Tail% relative to the study population median. In sensitivity analyses, effect estimates from the multivariable models were similar when including only men with SG ≥ 1.01 and ≤1.03 were included (n=154 for semen quality parameters, n=105 for DNA damage measures; results not shown). Results were also similar when modeling SG-corrected BPA concentrations as the independent variable as compared to including SG as a covariate when modeling uncorrected BPA concentrations (not shown).
When urinary BPA concentrations were categorized into quartiles to assess potential nonlinear relationships, there was no evidence for elevated odds ratios for below reference semen quality parameters in logistic regression models (not shown). Conversely, in multivariable linear regression models of semen quality parameters as continuous dependent variables, there were suggestive trends for non-monotonic declines in sperm concentration, motility and morphology, and a significant increasing trend in Tail%, when categorizing urinary BPA concentration into quartiles (). BPA quartiles were not associated with total sperm count (not shown; p-value for trend = 0.57). There were also statistically suggestive, though non-monotonic, relationships between urinary BPA concentration quartiles and reduced VSL, VCL, and LIN (results not shown; p-values for trend ranged from 0.05 to 0.13).
Figure 1 Adjusted regression coefficients for a change in semen quality parameter or sperm DNA damage measure associated with increasing quartiles of urinary BPA concentration (n=190): a) sperm concentration (p-value for trend = 0.09); b) sperm motility (p-value (more ...)
Results from the multivariable regression models using geometric mean BPA concentrations from multiple urine samples per participant (statistical approaches 2 and 3, the second and third columns of effect estimates in and ), differed somewhat from the models using only urine samples collected on the same day as the serum samples. However, with the exception of the models for sperm concentration and motility using statistical approach 3, in which effect estimates changed in sign or “direction”, effect estimates were consistent though somewhat attenuated compared to statistical approach 1. Effect estimates obtained from approach 4 (using only BPA concentration from the same day as the semen sample and only from men with at least 2 urine samples) were also consistent with approaches 1, 2, and 3, with the exception of sperm concentration and motility effect estimates from approach 3. Wider confidence intervals were observed with approaches 3 and 4 as a result of the smaller sample size, and these effect estimates should be considered less stable.
Finally, we observed differences in effect estimates from multivariable linear regression models among (i.e., stratified by) men with (n = 114) or without (n = 76) at least one semen parameter below WHO reference values. When considering only BPA concentrations measured in urine samples collected on the same day as the semen sample (statistical approach 1), an IQR increase urinary BPA was associated with a 27% decline in sperm concentration (p = 0.048), a 6% decline in sperm motility (p = 0.2), and a 16% decline in sperm morphology (p = 0.04) among men with at least one semen parameter below WHO reference values compared to little change [-2% (p = 0.8), -2% (p = 0.6), and +1% (p = 0.9) for sperm concentration, motility and morphology, respectively], among men with all semen parameters above WHO reference values. These results were consistent when using geometric mean BPA concentrations (statistical approach 2), as the associations between urinary BPA and sperm concentration (p = 0.06), motility (p = 0.10), and morphology (p = 0.03) were stronger among men with at least one parameter below WHO reference compared to men with all parameters above WHO reference (all p-values > 0.4).