Fig. 2
Association between dietary lignans and metabolic risk factors and potential mediation effect by enterolactone and enterolactone-predicting species score. *P < 0.05 for Beta (SEE) of the change of metabolic risk factors associated with per standard deviation (SD) changes of log-transferred dietary lignans; Model 1: Generalized linear mixed-effects regressions adjusted for repeated measurements (participant ID as random intercept), age (year), energy intake (kcal/day), alcohol (g/day), smoking (currently smoking cigarettes or not), physical activity (METs-h/week), using of antibiotics (yes vs. no), consumed any probiotics (yes vs. no), body mass index at age 21 (kg/m2) and fecal sample characteristics (6 category groups from hard to soft stool). Model 2: Model 1 further adjusted for enterolactone related species score. Model 3: Model 2 further adjusted for plasma enterolactone level. Mediation effect: proportion of dietary lignan effects potentially mediated by enterolactone species score (orange) or plasma enterolactone (green). P for interactions between lignans and enterolactone species score were 0.35 for metabolic score, 0.66 for BMI, 0.38 for HbA1c, 0.46 for HDL_C, 0.33 for TG, 0.49 for TC and 0.33 for CRP; P for interactions between lignans and enterolactone were 0.58 for metabolic score, 0.81 for BMI, 0.06 for HbA1c, 0.95 for HDL_C, 0.36 for TG, 0.33 for TC and 0.42 for CRP