The study cohort included 307 mother‒adolescent pairs divided into four groups according to maternal health behaviour during pregnancy: control group (n = 105): mothers who did not report any cigarette or alcohol use, nicotine-exposed group (n = 115): mothers who reported cigarette smoking only, dual-exposed group (n = 73): mothers who reported both cigarette smoking and alcohol use, and the alcohol-exposed group (n = 14): mothers who reported alcohol use only (Table 1). Compared with smoking mothers (U = 59.31 ± 17.62 kg, p < 0.05) and non-smoking mothers (U = 68.19 ± 17.53 kg), mothers who smoked cigarettes and consumed alcohol during pregnancy, the dual group, had significantly lower estimated pre-pregnancy weights (U = 52.18 ± 22.76 kg, p < 0.01). Furthermore, newborns in the dual exposed group were, on average, 379.89 g smaller (p < 0.05) than nonexposed newborns were (p < 0.01), but this difference became insignificant after a Bonferroni correction (p = 0.053). Similarly, newborns in the nicotine-exposed group were, on average, 298.16 g smaller (p < 0.05) and were born 9.8 weeks earlier than nonexposed newborns (p < 0.01) but also lost significance after a Bonferroni correction (p = 0.117) (Table 1). In addition, self-reported smoking was most prevalent in adolescents exposed to both maternal cigarette smoking and alcohol (11.9%) and alcohol only (38.5%).

Furthermore, cardiometabolic risk factors were more prevalent in girls as 22% were classified as abdominally obese, 28.1% had dyslipidaemia, 13.1% were overweight, 9.2% were obese, 12.6% had elevated SBP, 9.3% had elevated DBP and 13.1% were hypertensive (stage 1 and stage 2) compared to their male counterparts. However, only 4.1% of girls had elevated non-fasting glucose. In boys, 15.3% had abdominal obesity, 8.8% were overweight, 8.8% were obese, 8.0% had elevated SBP, 6.4% had elevated DBP and 9.6% were hypertensive (Stage 1 and 2). only 3.4% had elevated non-fasting glucose, and 23.3% had dyslipidaemia.

Table 2 shows the cardiometabolic measurements of 10–14-year-old adolescents. The adolescents in the control group had the highest prevalence rates of overweight (18.3%) and obesity (13.3%). However, when comparing BMI, no significant differences were found across exposure groups. Moreover, no significant differences in body length were observed across the exposure groups. Regarding weight, female adolescents in the control (U = 23.83 ± 11.00 mm) and dual-exposed groups (U = 42.01 ± 13.56 cm) weighed significantly more compared to boys. Also, girls in the control (U = 45.05 ± 12.78 cm) and dual-exposed groups (U = 20.44 ± 10.44 mm) had significantly higher SSFT compared to boys. In addition, Table 2 presents the anthropometric measurements, such as the SSFT, WC, DBP, triglycerides, non-fasting blood glucose and LDL cholesterol, which were not significantly different between the control group and exposure groups. However, SBP was significantly greater in the nicotine-exposed group (U = 117.92 ± 14.29 mmHg, p < 0.01) and the dual-exposed group (U = 115.25 ± 14.99 mmHg, p < 0.05) than in the control group (U = 110 ± 11.98 mmHg), as shown in Table 2. In addition, HDL cholesterol was significantly lower in the nicotine-exposed group than in the control group (U = 1.67 ± 0.79 mmol/L, p < 0.05).

We also compared all cardiometabolic measures in girls and boys separately across the exposure groups (Table 2). Girls in the control group had significantly higher anthropometric, adiposity and blood pressure measures, compared to boys. No significant differences were observed in cholesterol and non-fasting glucose levels across sex. Supplementary Table 1 shows the correlations between demographic characteristics, maternal behaviours and adolescent cardiometabolic measurements.

Table 3 presents the direct pathways in the mediation models, explaining the direct relationship between teratogen exposure during pregnancy and cardiometabolic risk factors in male and female adolescents. In the model, elevated or high SBP/DBP was combined as a variable to ensure a good fit for the model. Male adolescents exposed to maternal smoking were 6.09 times more likely to have low HDL cholesterol, after the adjustment for covariates (OR 6.090, 95% CI 1.447–25.622, p = 0.014). The significant association between maternal smoking and overweight/obesity suggested a reduced odds of overweight or obesity in girls (OR 0.280, 95% CI 0.082–0.954, p = 0.042) (Table 3).

In Fig. 1, the first model revealed that the direct effect of maternal cigarette smoking during pregnancy on childhood SBP was statistically significant (b = 7.8593, SE = 2.4603, t = 3.1944, p = 0.0017, CI [2.9972,12.7215]), and remained significant after adjusting for the child’s current smoking, second-hand smoke exposure, family history of CVD, age, birthweight, BMI, physical activity, breastfeeding duration and gestational hypertension and diabetes. In addition, the effect of cigarette smoking during pregnancy on childhood BMI was significant (b = −1.0790, SE = 0,4821, t = −2.2382, p = 0.0259), but this effect became borderline insignificant after adjusting for the child’s current smoking, second-hand smoke exposure, sex, birthweight, mother’s education level and total minutes of moderate-high PA (b = −1.3074, SE = 0.7630, t = −1.7136, p = 0.0631). However, the indirect effect of BMI on SBP was significant, with a point estimate of b = 1.3892 (p = 0.0001) and a 95% bootstrap CI of [0.7903, 1.9693], suggesting that a 1-unit increase in BMI may lead to a 1.38 mmHg increase in SBP even after adjusting for the child’s current smoking, second-hand smoke exposure, sex, birthweight, mother’s education level, maternal smoking and alcohol use during pregnancy, and total minutes of moderate-high PA (Fig. 1). On the other hand, smoking during pregnancy had a significant direct effect on BW (b = −335.564, SE = 93,9464, t = −3,5719 p = 0.0001), suggesting a decrease of 335.56 g in BW resulting from maternal smoking. However, this difference was lost after adjustment for GA and maternal pre-pregnancy weight (b = −335.564, SE = 93.9464, t = −3.5719 p = 0.0001). In addition, BW was not associated with adolescent SBP (b = −261.4357, SE = 142.8085, t = −1.8307, p = 0.696) (Fig. 1).

Mediation analysis summary with unstandardized coefficients (b) Cigarette smoking during pregnancy → BMI/birthweight → average systolic blood pressure. X Independent variable, Y Dependent/Outcome variable, M Potential mediator. Solid lines indicate direct paths. Dashed lines indicate direct paths that involve a mediator

In the second model (Fig. 2), maternal cigarette smoking during pregnancy had no direct effect on adolescent overweight or obesity, but rather adolescent BMI mediated the effect of maternal cigarette smoking on overweight/obesity, therefore having an indirect effect with a point estimate of b = −0.9955 (p < 0.01) and a 95% bootstrapped CI of [−183.8902,11.8324] after adjustment for maternal pregestational weight and GA. Moreover, maternal cigarette smoking was not associated with BW after adjustment for maternal pregnancy weight and GA (b = −257.4036, SE = 148.1558, t = −1.7374, p = 0.0850). However, BW was significantly associated with BMI (b = 0.001, SE = 0.000, t = 1.976, p = 0.049) but not with adolescent overweight/obesity (Fig. 2).

Mediation analysis summary with unstandardized coefficients (b) Cigarette smoking during pregnancy → BMI/birthweight → overweight/obesity. X Independent variable, Y Dependent/Outcome variable, M Potential mediator. Solid lines indicate direct paths. Dashed lines indicate direct paths that involve a mediator

In the third mediation model (Fig. 3), maternal cigarette smoking during pregnancy had a direct effect on adolescent HDL-cholesterol, with a point estimate of b = −1.1545 (SE = 0.5796, t = −1.9919, p < 0.05) and a 95% bootstrapped CI of [−2.2906, −0.0185] after adjustment for age, weight, height and SSFT. However, no effect was observed in the mediation model when examining boys and girls separately. Moreover, low HDL cholesterol was not associated with the following mediators: recalled birthweight (b = 0.0007, SE = 0.0004, t = 1.7805, p = 0.0750), adolescent cigarette smoking (b = −0.0028, SE = 0.0292, t = −0.972, p = 0.9226) and household income (b = −0.0221, SE = 0.2013, t = −0.1097, p = 0.9127), except for total minutes of physical activity/week (b = −257.4036, SE = 148.1558, t = −1.7374, p = 0.0850). Although, maternal smoking was significantly associated with household income, and total minutes of physical activity was associated with low HDL cholesterol (Fig. 3).

Mediation analysis summary with unstandardized coefficients (b) Cigarette smoking during pregnancy → Total min physical activity/birthweight/Adolescent smoking (cig/day)/Household income → Low HDL (boys). X Independent variable, Y Dependent/Outcome variable, M Potential mediator. Solid lines indicate direct paths. Dashed lines indicate direct paths that involve a mediator