Association of longitudinal body mass index trajectories with phenotypic age acceleration: a cross-sectional study based on growth mixture modeling

Abstract

To examine the association between body mass index (BMI) trajectories, early and recent BMI changes, and phenotypic age acceleration (PhenoAgeAccel), addressing inconsistent findings in previous studies on weight change and aging. Data from the National Health and Nutrition Examination Survey from 2005 to 2018 were used, selecting participants aged 50 years and older. A growth mixture model was employed to identify BMI trajectories. The association between different BMI trajectories and PhenoAgeAccel was assessed using linear and multinomial logistic regression models. The nonlinear effects of BMI changes were identified through threshold effect analysis. Among 5404 participants, the four BMI trajectories identified were as follows: stable weight (29.07%), midlife weight gain (24.31%), late-life weight gain (32.22%), and chronic obesity (14.41%). The chronic obesity group exhibited the most significant elevations in PhenoAgeAccel, indicating they were phenotypically older compared to other groups (β = 4.34, 95% confidence interval 3.67-5.02). Early BMI changes of less than 6% were associated with being phenotypically younger (β = - 5.06, P = 0.029), whereas increases exceeding 6% were linked to being phenotypically older (β = 2.83, P < 0.001). The key threshold for recent BMI changes was 2%; changes below this level were associated with being phenotypically younger, while those exceeding this threshold were linked to being phenotypically older (P < 0.001). This cross-sectional study suggests that individuals with long-term chronic obesity tend to be phenotypically older, whereas those with stable body weight are more likely to be phenotypically younger.

Keywords: BMI; Growth mixture model; Phenotypic age acceleration; Threshold effect; Trajectory.

Fig. 1

Screening flow chart

Fig. 2

Mean BMI trajectories across different time points by trajectory class identified using the growth mixture model. Mean BMI at four time points: 25 years ago, 10 years ago, 1 year ago, and baseline, categorized into four trajectory classes. Trajectory Classes 0, 1, 2, and 3 are represented by orange, red, pink, and magenta, respectively. BMI, body mass index

Fig. 3

Density plot of body mass index (BMI) distribution and its variation. A Density plot of BMI distributions at four time points: 25 years ago (BMI25), 10 years ago (BMI10), 1 year ago (BMI1), and baseline (BMI). The plot illustrates the BMI value distribution over time, demonstrating shifts in BMI patterns with the age of the participants. B Density plot of BMI changes in early (BMIearly) and recent (BMIrecent) phases. The early phase represents BMI change from 25 to 10 years ago, while the recent phase represents BMI change from 10 years ago to baseline. This plot highlights the differences in BMI change dynamics across the two phases

Fig. 4

The distribution of phenotypic age acceleration across different body mass index (BMI) trajectory classes. Violin plot displaying the distribution of phenotypic age acceleration (PhenoAgeAccel) across four BMI trajectory classes: stable weight, midlife weight, late-life weight, and chronic obesity. P-values indicate the significance of differences between the groups, with comparisons made using appropriate statistical tests

Fig. 5

The association between body mass index (BMI) change and PhenoAgeAccel with the restricted cubic spline (RCS) function. Association between early-stage BMI change (BMIearly, A) and recent-stage BMI change (BMIrecent, B) with PhenoAgeAccel. The models use five knots located at the 5 th, 28 th, 50 th, 72nd, and 95 th percentiles. The Y-axis represents the beta coefficient for PhenoAgeAccel relative to the 50 th percentile of BMI change. Adjustments are made for age, sex; race; education level; poverty-income ratio (PIR); smoking status; drinking status; and history of hypertension, diabetes, and cancer or malignancy