The role of adolescent lifestyle habits in biological aging: A prospective twin study

Reviewer #1 (Public Review):

Previous studies have linked several lifestyle-related factors, such as body mass index and smoking, alcohol use with accelerated biological aging measured using epigenetic clocks, however, most of them focused on single lifestyle factors based on cross-sectional data from older adults. The current study has a couple of major strengths: it has a decent sample size, lifestyle was measured longitudinally during puberty and adolescence, it looked at the effect of multiple lifestyle measures collectively, it looked at multiple epigenetic clocks, and due to the data from twins, it could examine the contribution of genetic and environmental influences to the outcomes. I have a couple of comments that are mainly aimed at improving the clarity of the methods (e.g. how was multiple testing correction done, how did the association model account for the clustering of twin data, how many samples were measured on 450k vs EPIC and were raw or pre-QC'd data supplied to the online epigenetic age calculator), and interpretation of findings (why were 2 measures of Dunedin PACE of aging used, how much are results driven by BMI versus the other lifestyle factors, and the discussion on shared genetic influences should be more nuanced; it includes both pleiotropic effects and causal effects among lifestyle and biological ageing).

Reviewer #2 (Public Review):

Kankaanpää and colleagues studied how lifestyle factors in adolescence (e.g., smoking, BMI, alcohol and exercise) associate with advanced epigenetic age in early adulthood.

Strengths:

The manuscript is very well written. Although the analyses and results are complex, the authors manage very well to convey the key messages.
The twin dataset is large and longitudinal, making this an excellent resource to assess the research questions.
The analyses are advanced including LCA capitalizing on the strength of these data.
The authors also include a wider range of epigenetic age measures (n=6) as well as a broader range of lifestyle habits. This provides a more comprehensive view that also acknowledges that associations were not uniform across all epigenetic age measures.

Weaknesses:

The accuracy of the epigenetic age predictions was moderate with quite large mean absolute errors (e.g., +7 years for Horvath and -9 years for PhenoAge). Also, no correlations with chronological age are presented. With these large errors it is difficult to tease apart meaningful deviations (between chronological and biological age) from prediction error.

The authors claim that 'the unhealthiest lifestyle class, in which smoking and alcohol use co-occurred, exhibited accelerated biological aging...'. However, this is only partially true. For example, PhenoAge was not accelerated in lifestyle class C5. Similarly, all classes showed some degree of deceleration (not acceleration) with respect to DunedinPACE (Figure 3D). The large degree of heterogeneity across different epigenetic age measures needs to be acknowledged.

The authors claim that 'Practically all variance of AAPheno and DunedinPACE common with adolescent lifestyle was explained by shared genetic factors'. However, Figure 4 suggest that most of the variation (up to 96%) remained unexplained and genetics only explained around 10-15% of total variation. The large amount of unexplained variation should be acknowledged.

Author Response

Reviewer #1 (Public Review):

Previous studies have linked several lifestyle-related factors, such as body mass index and smoking, alcohol use with accelerated biological aging measured using epigenetic clocks, however, most of them focused on single lifestyle factors based on cross-sectional data from older adults. The current study has a couple of major strengths: it has a decent sample size, lifestyle was measured longitudinally during puberty and adolescence, it looked at the effect of multiple lifestyle measures collectively, it looked at multiple epigenetic clocks, and due to the data from twins, it could examine the contribution of genetic and environmental influences to the outcomes. I have a couple of comments that are mainly aimed at improving the clarity of the methods (e.g. how was multiple testing correction done, how did the association model account for the clustering of twin data, how many samples were measured on 450k vs EPIC and were raw or pre-QC'd data supplied to the online epigenetic age calculator), and interpretation of findings (why were 2 measures of Dunedin PACE of aging used, how much are results driven by BMI versus the other lifestyle factors, and the discussion on shared genetic influences should be more nuanced; it includes both pleiotropic effects and causal effects among lifestyle and biological ageing).

Thank you for the encouraging comments and important suggestions.

Reviewer #2 (Public Review):

Kankaanpää and colleagues studied how lifestyle factors in adolescence (e.g., smoking, BMI, alcohol and exercise) associate with advanced epigenetic age in early adulthood.

Strengths:

The manuscript is very well written. Although the analyses and results are complex, the authors manage very well to convey the key messages.

The twin dataset is large and longitudinal, making this an excellent resource to assess the research questions.

The analyses are advanced including LCA capitalizing on the strength of these data.

The authors also include a wider range of epigenetic age measures (n=6) as well as a broader range of lifestyle habits. This provides a more comprehensive view that also acknowledges that associations were not uniform across all epigenetic age measures.

Weaknesses:

The accuracy of the epigenetic age predictions was moderate with quite large mean absolute errors (e.g., +7 years for Horvath and -9 years for PhenoAge). Also, no correlations with chronological age are presented. With these large errors it is difficult to tease apart meaningful deviations (between chronological and biological age) from prediction error.

The authors claim that 'the unhealthiest lifestyle class, in which smoking and alcohol use co-occurred, exhibited accelerated biological aging...'. However, this is only partially true. For example, PhenoAge was not accelerated in lifestyle class C5. Similarly, all classes showed some degree of deceleration (not acceleration) with respect to DunedinPACE (Figure 3D). The large degree of heterogeneity across different epigenetic age measures needs to be acknowledged.

The authors claim that 'Practically all variance of AAPheno and DunedinPACE common with adolescent lifestyle was explained by shared genetic factors'. However, Figure 4 suggest that most of the variation (up to 96%) remained unexplained and genetics only explained around 10-15% of total variation. The large amount of unexplained variation should be acknowledged.

Thank you for the encouraging comments and important notes.

We have now acknowledged that the standard deviations of epigenetic age estimates were high (lines 409-418). Due to the narrow age range of this study, the correlations between chronological age and epigenetic age estimates were weak. We aimed to overcome these weaknesses and calculated the epigenetic age estimates using recently developed principal component (PC)-based clocks, which are shown to improve the reliability and validity of epigenetic clocks (Higgins-Chen et al., 2022). In our data, the standard deviations of epigenetic age estimates were similar or even higher compared with those obtained with the original clocks, but the correlations between epigenetic age acceleration measures assessed with different clocks were consistently higher when PC-based epigenetic clocks were used. Importantly, the observed associations with the adolescent lifestyle behavior patterns did not substantially change.

Moreover, we have now more carefully reported and interpreted the results obtained using different epigenetic aging measures and acknowledged their heterogeneity (lines 459-467).

Figure 4 presents the genetic and environmental influences on biological aging shared with adolescent lifestyle and biological aging. There are also unique genetic and environmental influences on biological aging not shown in the figure. Therefore, the unexplained variation in biological aging was not that large. Most of the total variation in biological aging was explained by the genetic factors unique to biological aging. We have now clarified the description of the estimation of genetic and environmental influences (lines 283-300) and the presentation of the results (lines 437-449).

References:

Higgins-Chen, A. T., Thrush, K. L., Wang, Y., Minteer, C. J., Kuo, P.-L., Wang, M., Niimi, P., Sturm, G., Lin, J., Moore, A. Z., Bandinelli, S., Vinkers, C. H., Vermetten, E., Rutten, B. P. F., Geuze, E., Okhuijsen-Pfeifer, C., van der Horst, M. Z., Schreiter, S., Gutwinski, S., … Levine, M. E. (2022). A computational solution for bolstering reliability of epigenetic clocks: implications for clinical trials and longitudinal tracking. Nature Aging, 2(7), 644–661. https://doi.org/10.1038/s43587-022-00248-2