Peek Behind the Paper: a smoking meta-analysis
In this interview, Melissa Bateson (Newcastle University, Newcastle, UK) discusses some of her epidemiological research concerning the effect of smoking on telomere lengths.
Melissa Bateson (Newcastle University; Newcastle, UK) exclusively peeks behind her study paper: Smoking does not accelerate leucocyte telomere attrition: a meta-analysis of 18 longitudinal cohorts.
Interview segments:
Please can you introduce yourself and your research?
As an ethologist, what led you to study the association between smoking and leukocyte attrition length?
What are you cautious of when assessing correlational data?
Can you explain the importance/contribution of real-world evidence to the outcomes of your study?
What have been some of your experiences employing real-world evidence and data in your other works?
What are some of the limitations to the study design you employed? Do you feel that your population sample was representative enough?
What are some of the implications of your research — for example, epidemiologically, in redefining biomarkers such as LTLs?
How do you think your use of real-world evidence will change in the future?
Please can you introduce yourself and your research?
I am an ethologist, meaning that I take a biological approach to the study of behavior. I am particularly interested in how exposure to various types of stress changes the phenotype of humans and other animal species. I’m particularly interested in the behavioural phenotype, but I’m also interested in somatic changes that might underlie changes in behaviour. The plasticity of behavior makes it one of the key ways animals adapt to their environment.
Understanding the biological impacts of stress is important because we need to understand how stress exposure affects individuals, with a view to tackling societal problems linked with stress — gambling, addictions, obesity and so on. Furthermore, this understanding would be useful to have validated markers of stress exposure that we could use to assess the lifetime experience of non-human animals in an animal welfare context.
As an ethologist, what led you to study the association between smoking and leukocyte attrition length?
I am currently very interested in the research suggesting that telomere length is an integrative biomarker of lifetime exposure to a variety of different types of stressor. In 1956, stress biologist Hans Selye, a professor at the Institute of Experimental Medicine and Surgery at the University of Montreal (Quebec, Canada), observed: “Every stress leaves an indelible scar, and the organism pays for its survival after a stressful situation by becoming a little older.”
This made me think that telomeres might be interesting from the perspective of finding an integrative biomarker of stress exposure. We know that telomeres shorten with chronological age. There is also evidence to suggest that telomere shortening may be accelerated by different kinds of stressor. I was recently involved in conducting a meta-analysis that demonstrated that, in humans, exposure to diverse stressors — including physical diseases, environmental hazards, nutrition, psychiatric illness, smoking, physical activity, psychosocial and socioeconomic exposures — is associated with shorter telomeres [2]. Whilst encouraging, this result was mainly based on cross-sectional studies that reported correlations.
I picked smoking for the current study mainly because there are more datasets available on telomere length and smoking than for any of the other types of stressor.
What are you cautious of when assessing correlational data?
The problem with correlational data is that, whilst correlations tell us that stress exposure and short telomeres are associated, they do not tell us why. Although, from what we know about telomere biology, it makes sense that stress might shorten telomeres, we cannot rule out alternative interpretations that having short telomeres causes stress exposure, or, that some other variable causes both stress and short telomeres. Therefore, we need to be very careful not to discount these alternatives.
Can you explain the importance/contribution of real-world evidence to the outcomes of your study?
Our study was the result of an international collaboration in which researchers from all over the world provided either access to the data that they had collected, or reanalyzed their data for the current study. It is a wonderful thing about modern science that such studies are becoming possible. It means that we can extract much more value from small individual data sets than was previously possible. The power of meta-analysis makes it obvious why it is important for researchers to make their raw data available for other researchers to re-analyze in the future.
What have been some of your experiences employing real-world evidence and data in your other works?
Although the current study was possible with a lot of hard work and collaboration, it was not easy to accumulate the data from the 18 different studies included. In the future, I would like to see all researchers moving towards making their datasets publicly available to bona fide scientists. All too often, it has been my experience that accessing existing data is surprising difficult. I think this is quite unethical — participants have given up their time to donate their blood and other data, and taxpayers have funded the research. It is our duty to make sure that the most value is gained from these data.
What are some of the limitations to the study design you employed? Do you feel that your population sample was representative enough?
Ideally, to definitively test a causal relationship, we need to do experiments. To show that stress shortens telomeres, we would need to randomly assign participants to different levels of stress exposure and measure the effect on their telomere lengths. Sadly, such studies are ethically difficult in humans; very few have ever been done and none have been done in the specific case of smoking.
In the absence of such experiments, longitudinal studies in which the same individuals are measured over time represent an improvement over cross-sectional studies. The rationale underlying the current study [2] was that if exposure to a stressor such as smoking shortens telomeres, then we ought to be able to see faster telomere shortening over time in smokers than in non-smokers. A number of individual longitudinal studies have been published examining this link, but unfortunately, the results are inconsistent. We therefore chose a meta-analytic approach, allowing us to combine the data from 18 different studies in a single analysis.
A strength of meta-analysis is that it allows you to test a hypothesis with much greater power, because you effectively have the combined sample size from all the component studies. A limitation of this approach is that it is necessary to simplify some aspects of the studies to make the analysis possible. For example, individual studies differed in the level of detailed information concerning participants smoking behavior. This forced us to simplify our independent variable to current smokers versus never smokers. All meta-analyses involve these kinds of trade-offs.
Further, although our population sample covered studies with participants from several countries, over four continents, all participants came from Westernized societies. Thus, it is possible that our results would not generalize to other human populations, though we have no reason to believe this.
Possibly the greatest limitation of what we have been able to do so far is the lack of data on telomere lengths from participants before they started smoking. If our hypothesis is correct and both short telomeres and likelihood of smoking are caused by stress experienced very early in life, then we would predict that children destined to become smokers would have shorter telomeres even before they started smoking. Sadly, we were unable to test this prediction, but we are intending to pursue this in future research.
What are some of the implications of your research — for example, epidemiologically, in redefining biomarkers such as LTLs?
The importance of our study is that it forces us to rethink the value of telomere length as a read-out of how our current lifestyles affect our bodies. We do not dispute the abundant evidence that smoking is bad for you, but we do dispute the evidence that telomere length is a good way of assessing the biological damage done by smoking. This may extend to the effects of other unhealthy behaviors and stressors. It was previously believed that telomere length responds dynamically to current adult behavior — shortening more when individuals engage in unhealthy activities such as smoking or are exposed to stressors of various types.
However, our study suggests that adult telomere length should be reinterpreted as a static biomarker that changes relatively little during adult life.
It may still be that telomere length is a good measure of stress experience in early life. We have published experimental data from animals showing that exposure to early-life adversity shortens telomeres [3].
How do you think your use of real-world evidence will change in the future?
I was trained as an experimental scientist; working with epidemiological data is a relatively new thing for me. Although I will continue to do my own experiments to test causal hypotheses, I see the meta-analysis of existing epidemiological datasets as being an extremely valuable complement. I very much hope to engage in more of this type of work in the future. The ‘open science’ movement is changing the way that we do science. The increasing availability of datasets for meta-analysis is very exciting to me; it allows me to test hypotheses on a much grander scale than is possible in a single laboratory experiment.
References:
[1] Pepper GV, Bateson M & Nettle D. Telomeres as integrative markers of exposure to stress and adversity: A systematic review and meta-analysis’, R. Soc. Open Sci.; 5: 180744 (2018);
[2] Bateson M, Aviv A, Bendix L et al. Smoking does not accelerate leucocyte telomere attrition: a meta-analysis of 18 longitudinal cohorts. R. Soc. Open Sci. 6: 190420 (2019);
[3] Nettle D, Andrews C, Reichert S et al. Early-life adversity accelerates cellular ageing and affects adult inflammation: Experimental evidence from the European starling. Sci. Rep.; 7: 40794 (2017).
Disclosures:
Bateson has no financial or competing interests to disclose.