Characteristics of the study participants
Table 1 displayed the characteristic of participants included in the analysis. 417 males and 434 females were finally enrolled. Females had higher BMI values, higher percentage of college or AA degree, college graduate or above compared with males. Additionally, males intended to exercise more and the proportion of current-smoker and former-smoker was significantly higher than those in females. While no significant differences were found in age, race, family poverty-income ratio (PIR), the prevalence of hypertension, and diabetes between men and women.
Table 1
Characteristics of participants in NHANES 2013–2014.
Variables
|
Male (n = 417)
|
Female (n = 434)
|
p value
|
Age, years, mean ± SD
|
47.24 ± 17.71
|
48.10 ± 16.73
|
0.469
|
Race, n (%)
|
|
|
0.581
|
Mexican American
|
66 (15.83%)
|
83 (19.12%)
|
|
Other Hispanic
|
34 (8.15%)
|
41 (9.45%)
|
|
Non-Hispanic White
|
181 (43.41%)
|
184 (42.40%)
|
|
Non-Hispanic Black
|
75 (17.99%)
|
66 (15.21%)
|
|
Other race-including multi-racial
|
61 (14.63%)
|
60 (13.82%)
|
|
BMI (kg/m2), mean ± SD
|
28.21 ± 6.19
|
29.37 ± 7.56
|
0.015
|
Education level, n (%)
|
|
|
0.046
|
Less than 9th grade
|
43 (10.83%)
|
34 (8.15%)
|
|
9 − 11th grade
|
64 (16.12%)
|
53 (12.71%)
|
|
High school grade/GED or equivalent
|
99 (24.94%)
|
89 (21.34%)
|
|
Some college or AA degree
|
105 (26.45%)
|
148 (35.49%)
|
|
College graduate or above
|
86 (21.66%)
|
93 (22.30%)
|
|
Family PIR (%), n (%)
|
|
|
0.204
|
< 1
|
107 (27.44%)
|
129 (31.54%)
|
|
≥ 1
|
283 (72.56%)
|
280 (68.46%)
|
|
Physical activity, n (%)
|
|
|
< 0.001
|
Never
|
225 (53.96%)
|
293 (67.51%)
|
|
Moderate
|
77 (18.47%)
|
80 (18.43%)
|
|
Vigorous
|
115 (27.58%)
|
61 (14.06%)
|
|
Smoking, n (%)
|
|
|
< 0.001
|
Never
|
162 (38.85%)
|
218 (50.23%)
|
|
Current smoker
|
149 (35.73%)
|
146 (33.64%)
|
|
Former smoker
|
106 (25.42%)
|
70 (16.13%)
|
|
Drinking
|
91 (21.82%)
|
51 (11.75%)
|
< 0.001
|
Hypertension, n (%)
|
135 (32.37%)
|
151 (34.79%)
|
0.455
|
Diabetes, n (%)
|
40 (9.59%)
|
56 (12.90%)
|
0.127
|
Abbreviations: BMI: body mass index; PIR: Poverty-income ratio.
|
Distribution of aldehydes and sex steroid hormones
Significant differences were observed in the distribution of serum aldehydes and sex hormones by sex. Females had lower concentrations of butyraldehyde (0.53 vs. 0.58 ng/mL), heptanaldehyde (0.49 vs. 0.52 ng/mL), isopentanaldehyde (0.49 vs. 0.58 ng/mL) and propanaldehyde (2.04 vs. 2.15 ng/mL) compared with those in males. Whereas the levels of benzaldehyde and hexanaldehyde were similar among males and females. In terms of sex steroid hormones, males had remarkably higher levels of total testosterone (423.94 vs. 23.02 ng/dL), lower estradiol (24.11 vs. 51.23 pg/mL) and lower SHBG levels (43.92 vs. 73.25 nmol/L) than females (Table 2).
Table 2
Serum concentration of aldehydes and sex steroid hormones, stratified by sex.
Variables
|
Male (n = 417)
|
Female (n = 434)
|
p value
|
Aldehyde (ng/mL), median (Q1-Q3)
|
|
|
|
Benzaldehyde
|
1.16 (0.74–1.80)
|
1.29 (0.80–1.89)
|
0.118
|
Butyraldehyde
|
0.58 (0.45–0.73)
|
0.53 (0.38–0.71)
|
0.003
|
Heptanaldehyde
|
0.52 (0.44–0.62)
|
0.49 (0.42–0.57)
|
< 0.001
|
Hexanaldehyde
|
2.20 (1.77–2.73)
|
2.19 (1.76–2.68)
|
0.997
|
Isopentanaldehyde
|
0.58 (0.38–1.14)
|
0.49 (0.33–1.02)
|
0.008
|
Propanaldehyde
|
2.15 (1.66–2.74)
|
2.04 (1.51–2.59)
|
0.016
|
Sex steroid hormones
|
|
|
|
Total testosterone (ng/dL),
mean(SD), median (Q1-Q3)
|
423.94 (170.14),
400.00 (297.00-526.83)
|
23.02 (15.8),
19.00 (13.20-27.91)
|
< 0.001
|
Estradiol (pg/mL),
mean (SD), median (Q1-Q3)
|
24.11 (9.40),
22.90 (17.80–28.60)
|
51.23 (71.51),
21.40 (5.17–67.30)
|
< 0.001
|
SHBG (nmol/L),
mean (SD), median (Q1-Q3)
|
43.92 (24.91),
39.60 (27.25–55.30)
|
73.25 (46.12),
64.19 (42.25–90.72)
|
< 0.001
|
Abbreviations: SHBG: sex hormone binding globulin;
|
Association between aldehydes and sex hormones in males and females
Independent association between aldehydes and sex hormones stratified by sex were evaluated through multivariate linear regression models after adjusting potential covariates, including age, race, BMI, education level, family PIR, physical activity, drinking, smoking, hypertension, diabetes. Each kind of aldehydes were equally divided into 4 quartiles for further analyses: Benzaldehyde: Q1 (< 0.77 ng/mL), Q2 (0.77–1.26 ng/mL), Q3 (1.26–1.84 ng/mL), Q4 (≥ 1.84 ng/mL); Butyraldehyde: Q1 (< 0.41 ng/mL), Q2 (0.41–0.56 ng/mL), Q3 (0.56–0.73 ng/mL), Q4 (≥ 0.73 ng/mL); Heptanaldehyde: Q1 (< 0.43 ng/mL), Q2 (0.43–0.51 ng/mL), Q3 (0.51–0.60 ng/mL), Q4 (≥ 0.60 ng/mL). Hexanaldehyde : Q1 (< 1.77 ng/mL), Q2 (1.77–2.19 ng/mL), Q3 (2.19–2.72 ng/mL), Q4 (≥ 2.72 ng/mL); Isopentanaldehyde: Q1 (< 0.35 ng/mL), Q2 (0.35–0.54 ng/mL), Q3 (0.54–1.11 ng/mL), Q4 (≥ 1.11 ng/mL); Propanaldehyde: Q1 (< 1.57 ng/mL), Q2 (1.57–2.09 ng/mL), Q3 (2.09–2.65 ng/mL), Q4 (≥ 2.65 ng/mL). Figure 2 showed that none of these 6 kinds of aldehydes were associated with total testosterone levels. Compared with the lowest quartile of bytyraldehyde, quartile 2 in butyraldehyde was negatively associated with the concentration of estradiol (β=-20.59, 95% CI: -38.30 to -2.88, p = 0.0233, p trend = 0.187) (Fig. 3). Propanaldehyde was also inversely related to SHBG levels in male participants (Q2 vs. Q1: β=-8.13, 95% CI: -14.92 to -1.33, p = 0.0197; Q4 vs. Q1:-7.79, 95% CI: -14.91 to -0.67, p = 0.0326. p trend = 0.096) (Fig. 4). No other significant differences were observed between aldehyde exposure and sex hormones in males and females. Furthermore, the above-mentioned aldehydes and sex hormones were selected and displayed through curve fitting analysis more intuitively. As presented in Figure S1, butyraldehyde was negatively associated with estradiol in females, while propanaldehyde was negatively associated with SHBG in males.
Association between aldehydes and sex steroid hormones in pre-menopause and post-menopause
In females, considering menopausal status as a covariate, the association between aldehydes and sex hormones were evaluated in pre-menopausal and post-menopausal women, respectively. The potential confounding factors were the same as before. As shown in Fig. 5, the highest quartile of isopentanaldehyde was associated with 7.95ng/dL lower of total testosterone levels compared with the lowest concentration of isopentanaldehyde in pre-menopausal women (95% CI: -15.62 to -0.27, p = 0.0437, p trend = 0.038). Positive associations were observed between isopentanaldehyde and estradiol in post-menopause (Q3 vs. Q1: β = 3.53, 95% CI: 0.08 to 6.97, p = 0.0470, p trend = 0.274), propanaldehyde and estradiol in pre-menopause (Q3 vs. Q1: β = 28.88, 95% CI: 0.83 to 56.94, p = 0.0448, p trend = 0.926) (Fig. 6). However, none of these aldehydes were significantly independently associated with SHBG levels in both pre-menopausal and post-menopausal women (Fig. 7). Then, curve fitting analyses were performed to display the relationship between the above-mentioned aldehydes and sex hormones stratified by menopausal status (Figure S2). Isopentanaldehyde was negatively associated with estradiol in pre-menopause, while isopentanaldehyde and propanaldehyde were positively associated with estradiol in post- and pre-menopausal women.
Association between aldehyde and sex steroid hormones in males, stratified by age
Then, multivariate linear regression models were performed to evaluate age-specific effect on sex hormones associated with quartiles of aldehydes after adjusting the above cofounders in male participants and the results were presented in forest plots. Compared with the lowest quartile of benzaldehyde, multivariate linear regression analysis for total testosterone revealed β (95% CI) was − 155.7 (-248.0 to -23.34) for the highest quartile among male participants over 60. A significant decrease in total testosterone was found in the second quartile of butyraldehyde among males under 40 (β=-93.52, 95% CI: -181.16 to -5.87, p = 0.0386, p trend = 0.621). Similarly, significant decrease in total testosterone were found in the fourth quartile of heptanaldehyde in males aged less than 40 (β=-81.04, 95% CI: -157.97 to -4.1, p = 0.0412, p trend = 0.025) (Fig. 8). In males under 40, the third quartile of heptanaldehyde was negatively associated with estradiol (β=-4.47, 95% CI: -8.64 to -0.30, p = 0.0376, p trend = 0.046). Besides, the fourth quartile of hexanaldehyde (β = 4.96, 95% CI: 0.63 to 9.29, p = 0.0266, p trend = 0.011) in males aged 40–60, the third quartile of hexanaldehyde (β = 6.91, 95% CI: 1.01 to 12.82, p = 0.0244, p trend = 0.791) and the third quartile of isopentanaldehyde (β = 6.84, 95% CI: 0.57 to 13.11, p = 0.0357, p trend = 0.048) in males aged over 60 were all positively associated with estradiol concentrations compared with individual in the corresponding lowest quartiles (Fig. 9). With regard to SHBG, only butyraldehyde in the second quartile in males under 40 (β=-9.99, 95% CI: -18.01 to -1.97, p = 0.0161, p trend = 0.230), propanaldehye in the second (β=-22.69, 95% CI: -40.45 to -4.93, p = 0.0144, p trend = 0.083) and fourth quartile (β=-25.60, 95% CI: -48.20 to -3.00, p = 0.0293, p trend = 0.083) were inversely independently associated with the concentrations of SHBG in males over 60. Other differences were not statistically significant in the remaining subgroups irrespective of age and aldehyde type (Fig. 10). Additionally, Figure S3 presented the relationship between the above-mentioned aldehydes and sex hormones in adult males with different age through curve fitting analysis, and the associations between aldehydes and sex hormones were consistent with the results from multivariate linear regression analysis.