The question

The length of a human lifespan is controlled by a combination of individual health and environmental factors. Twin and family-based studies have concluded that over 15% of variation in human lifespan can be explained by genetics1. The importance of genetic factors is further supported by recent large-scale genome-wide association studies (GWAS), which have identified more than 20 loci that are linked to human lifespan2. However, while sufficiently powerful to link common genetic variants to the risk of human traits and diseases, GWAS usually lack the resolution needed to identify causal genes and associated mechanisms. For example, genetic changes that result in phenotypic changes of larger impact, such as protein-truncating variants (PTVs), tend to be poorly captured on GWAS genotyping arrays. PTVs often arise from premature stop codons, frameshifts or aberrant splicing, and can lead to partial or complete loss of function of the respective gene. It is therefore reasonable to hypothesize that PTVs may have an effect on aging traits such as lifespan, although little knowledge in this area is currently available.

The discovery

The UK Biobank (UKB) is an unprecedented cohort of over 500,000 participants with detailed health and lifestyle information3. We and others have recently complemented the UKB with exome-sequencing results4. Unlike GWAS, exome-wide studies have the advantage of allowing one to conduct association testing for rare variants with individually high effect sizes, including PTVs that are predicted to disrupt the function of encoded proteins. Using a survival model, we tested how exome-wide PTVs, as well as each PTV in each gene individually, affected the lifespan of over 350,000 UKB participants and their parents.

Our analysis identified four genes in which the burden of PTVs was associated with human lifespan at exome-wide significance at the population level: BRCA2, BRCA1, TET2 and ATM (Fig. 1a,b). All of these genes have previously established roles in various cancers, which we confirmed by performing phenome-wide association studies across over 4,000 UKB phenotypes. Interestingly, for TET2 and ATM, our results propose a somatic origin of lifespan-reducing PTVs, consistent with mutations in these genes underlying clonal hematopoiesis of indeterminate potential, which supports clonal hematopoiesis of indeterminate potential and somatic variants as being a potentially important determinant of human lifespan. Notably, our results overlapped only modestly with established lifespan-associated GWAS loci, indicating that exome studies can yield additional insights into the genetic underpinnings of human aging traits.

Fig. 1: Exome-based identification of genes associated with survival.
figure 1

a, Manhattan plot of gene-based PTV-burden survival testing in a discovery cohort of 238,239 UKB participants with survival as an outcome. Each point represents a single gene. The red dashed line indicates the exome-wide significance threshold of P < 4.17 × 10−7. Different chromosomes are shown with alternating blue and green colours. b, Kaplan–Meier curves for BRCA2 PTV survival analysis in female UKB participants (left) and their mothers (right). Each data point represents a right-censored observation. Numbers under the graphs indicate UKB participants with (bottom row; in blue) or without (top row; in orange) PTVs in BRCA2, in respective age categories. The shaded areas represent the 95% confidence interval. © 2022, Liu, J. Z. et al., CC BY 4.0.

Full size image

The implications

Our results show that PTVs, which often lead to loss of gene function, are associated with human lifespan at the population level and highlight distinct genes as targets for therapeutic and societal intervention. Importantly, we demonstrate limited overlap between the results of our exome study and loci identified to be associated with lifespan in GWAS, which supports the value of rare-variant-based analyses to complement our understanding of the genetic basis of human traits and diseases.

Our results reflect the demographics of the UKB and may therefore not extrapolate to all populations. For instance, owing to a lack of available data from other ethnicities, our analyses were based on individuals of white European origin and therefore might not translate to those of non-European ancestry. Future work should involve replication of our results in larger, more-diverse cohorts to study the genetic basis of aging in diverse ancestral origins. Nonetheless, an advantage of our analytical strategy is its applicability to thousands of other traits and diseases captured in the UKB, which include many common diseases with an onset later in life.

Moreover, as we have witnessed during the SARS-CoV-2 pandemic, life expectancy in a population is not static, but is instead strongly influenced by the environment or the implementation of healthcare measures5. Future similar studies linking genetics with life expectancy will need to take such dynamics and global differences into account.

Heiko Runz, Biogen, Cambridge, MA, USA

Expert opinion

“This is an important analysis encompassing the largest monolithic collection of sequence data, longevity (directly measured and inferred) and associated phenotypic consequences. It addresses a simple but fundamental question — what is the relationship between PTV burden and lifespan? The pathways implicated in the analysis are expected, though the specific associations have in some cases not been empirically observed”. Frederick Dewey, Foresite Labs, MA, USA

Behind the paper

This research would not have been possible without the willingness of UKB participants and researchers to share data. Moreover, several pharmaceutical companies needed to team up to generate exome-sequencing data in a precompetitive setting and agree to make results freely available to the wider research community (more information of UKB exome data, including access, is discussed in ref. 4). A remarkable insight since initial submission of our manuscript was how rapidly life expectancy can drop, both globally and in the UK, as a consequence of a pandemic5, with trajectories at time of publication now fortunately rebounding in many countries. Conversely, a truly encouraging insight from our study is how life expectancy increased in female UKB participants who carry BCRA2 and BRCA1 PTVs within just a single generation, most probably owing to the introduction of efficient screening programs and treatment strategies for breast and ovarian cancer. H.R.

From the editor

“Large-scale GWAS of lifespan have led to the identification of over 20 loci. In this study, Heiko Runz and colleagues report the first whole-exome sequencing analysis of individual and parental survival in over 350,000 UK Biobank participants, identifying four exome-wide significant genes in which the burden of protein-truncating variants is associated with lifespan. By focusing on coding variants, the study fills an important gap in the aging and longevity genetics landscape.” Sebastien Thuault, Chief Editor, Nature Aging.