Inclusion of current exercise, smoking, alcohol consumption, red meat consumption, parity, and spontaneous abortion history in the analyses did not affect the relationships between creatinine-adjusted aMT6s and 8-oxodG or 8-oxoGua (data not shown). They are therefore not included in the results discussed below.
Table provides the descriptive statistics for subjects by status: mother, father, and daughter. The age ranges (years) were similar for the mothers and fathers: 43.5 - 80.7 and 46.1 - 80.9, respectively. The age range for the daughters was 18.6 - 51.6. The mean ages were as follows: mothers - 59.6; fathers - 62.4; and daughters - 32.9. Mean 8-oxoGua urinary excretion levels were 6-8 times that of mean urinary 8-oxodG levels. The fathers' aMT6s/Cr values were generally lower than the mothers' values, although the range was nearly identical. For example, the 25th-75th percentile ranges were different: 33.3 - 80.3 for the mothers and 15.6 - 49.9 for the fathers.
Descriptive Statistics by Mothers, Fathers, and Daughters
For puposes of possible comparison with published studies which used creatinine-adjusted 8-oxodG and/or 8-oxoGua values, the creatinine-adjusted means (standard deviations) in nmol/mmol are as follows: (1) 8-oxodG: Mothers - 1.5 (0.53); Fathers - 1.3 (0.64); Oldest Daughters - 1.3 (0.49); (2) 8-oxoGua: Mothers - 11.7 (11.43); Fathers - 10.3 (10.68); Oldest Daughters - 9.0 (8.17).
Table provides pairwise comparisons of aMT6s/Cr, 8-oxodG, and 8-oxoGua values between mothers, fathers, and daughters. The mothers and the daughters had significantly higher mean aMT6s/Cr values than the fathers. In addition, the daughters had a significantly higher mean aMT6s/Cr than their mothers. Fathers had a significantly higher urinary 8-oxodG mean than did the mothers (p < 0.001), while their mean 8-oxoGua level was marginally (p = 0.08) higher.
Paired Comparisons of Mothers, Fathers, and Daughters: Creatinine-Adjusted aMT6s, 8-oxodG, and 8-oxoGua.
Pearson correlations were also calculated between urinary 8-oxodG and 8-oxoGua for the mothers, fathers, and daughters, separately. The correlations were, respectively, 0.10 (p = 0.50), 0.19 (p = 0.26), and 0.71 (p < 0.0001). Thus, only the daughters' urinary 8-oxodG and 8-oxoGua correlation was significantly different from 0.
Table provides the analytic regression results of urinary 8-oxodG and 8-oxoGua on aMT6s/Cr alone and on aMT6s/CR, age, and BMI combined for mothers, fathers, and oldest daughters. (The results for all daughters, using a regression model which includes correlations between sisters, were similar to the results for the oldest daughters.) The aMT6s/Cr parameter estimates (regression line slope) and significance levels were nearly identical with and without age and BMI (or weight) in the model in each analysis. In the analyses for the mothers and fathers who were older than the oldest daughter (Table ), the results were similar, except that specific p-values were lower.
Regressions of Urinary Creatinine-Adjusted aMT6s, Age, and BMI on 8-oxodG and 8-oxoGua by Mother, Father, and Daughter
Regressions of Urinary Creatinine-Adjusted aMT6s, Age, and BMI on 8-oxodG and 8-oxoGua by Mothers and Fathers Older than the Oldest Daughter
The primary multivariate analysis results are as follows:
1. among all mothers, there was
a. a statistically significant (p = 0.02) decrease in 8-oxodG as endogenous overnight urinary aMT6s/Cr increased for models with or without BMI or weight,
b. a marginally significant (p < 0.08) increase in 8-oxodG as weight increased (data not shown), but not BMI;
2. among the mothers older than the oldest daughter (age 51.6),
a. the significance levels of aMT6s/CR fell to 0.009 and 0.008 for 8-oxodG with BMI or weight in the model, respectively;
b. BMI was marginally positively associated (p = 0.097), while weight was significantly positively associated (p = 0.04; data not shown) with 8-oxodG;
3. among all fathers, there was
a. a marginally significant (p = 0.063 and 0.065 with BMI or weight in the model, respectively) decrease in 8-oxoGua as endogenous overnight urinary aMT6s/Cr increased;
4. among the fathers older than the oldest daughter (age 51.6),
a. the inverse association between 8-oxoGua and aMT6s/CR was statistically significant (p = 0.03 for either BMI or weight in the model);
5. among the oldest daughters, there was a. a marginally significant (p < 0.07 and < 0.08 for BMI and weight, respectively) increase in 8-oxoGua as age increased.
Table provides the correlations and percent of variation explained (R2) between aMT6s/Cr and 8-oxodG (mothers) and 8-oxoGua (fathers). The percent of variance explained by aMT6s/Cr pertains to the univariate regression and is 12.4% and 8.4% for the mothers' 8-oxodG and fathers' 8-oxoGua, respectively.
Correlations and R2 for Urinary Creatinine-Adjusted aMT6s and 8-oxodG (Mothers) and 8-oxoGua (Fathers)
Figure and Figure provide scatter plots of aMT6s/Cr and 8-oxodG (mothers) and 8-oxoGua (fathers) with the univariate regression lines added. Note that the differences between the slope estimates and p-values for aMT6s/Cr are minimal for the models with and without age and BMI (Tables and ) or age and weight. A straight line regression is appropriate for the mothers' 8-oxodG data, based on the scatter plot (Figure ). However for the fathers' 8-oxoGua data, the scatter plot (Figure ) shows there to be 3 very influential data points: the three points in the upper left-hand corner (low aMT6s/Cr and high 8-oxoGua). These 3 points represent 7.5% of the data and 2 outliers have already been eliminated, so it can certainly be argued that it is not appropriate to discard these three observations.
Scatter Plot of Mothers' Total Nocturnal Urinary 8-oxodG (nmol) versus Total Nocturnal Urinary Creatinine-Adjusted aMT6s (ng/mg), with Regression Line. The slope of the regression line is -0.025 (p = 0.01).
Scatter Plot of Fathers' Total Nocturnal Urinary 8-oxoGua (nmol) versus Total Nocturnal Urinary Creatinine-Adjusted aMT6s (ng/mg), with Regression Line. The slope of the regression line is -0.77 (p = 0.07).
However, we conducted further linear regression analyses on the fathers' 8-oxoGua data transformed in four ways: natural logarithm, square root, reciprocal, and logistic. The logistic transformation is log [y/(1+y)], where 'log' is the natural logarithm function. These transformations are often used to reduce or eliminate the effect of outliers on statistical results. All four analyses, using the models with age and BMI or with age and weight, showed a marginally significant inverse association between the transformed 8-oxoGua and aMT6s/Cr values. The p-values were very close to the p-values for the untransformed regression model. For example for the models with age and BMI, the p-values were 0.063 (untransformed), 0.066 (log transform), 0.068 (square root transform), 0.069 (logistic transform), and 0.069 (reciprocal).