Calendar age is the major risk factor for common disease. It is therefore expected that understanding the aging process will eventually lead to promotion of better health conditions in elderly populations.

Such insight may be obtained by identifying the genetic determinants of familial and exceptional longevity and age-related disease. Research of these determinants has been performed in candidate gene, genome-wide association and linkage studies. Because exploration of the common variation in the genome did not explain much of the variation in the rate of aging and longevity, researchers in the field have only recently started to investigate the contribution of rare genetic variants to these traits.

The increased throughput and decreased costs of next generation sequencing (NGS) have resulted in a great deal of novel applications for sequencing sets of candidate genes, whole exomes, and whole genomes of individuals.

Most of the successful NGS applications are as yet those focused on genetic syndromes and cancers for which causal mutations are readily being identified.

In this book, we discuss the genetic and phenomic aspects of human aging research and the use of NGS data to identify genes relevant for age-related diseases and lifespan regulation, and to investigate the accumulation of somatic genetic variation during the course of life.

The epidemiology of longevity and exceptional survival, an ebook

The field of the “epidemiology of longevity” has been expanding rapidly in recent years. Several long-term cohort studies have followed older adults long enough to identify the most long-lived and to define many factors that lead to a long life span. Very long-lived people such as centenarians have been examined using case-control study designs. Both cohort and case-control studies have been the subject of genome-wide association studies that have identified genetic variants associated with longevity. With growing recognition of the importance of rare variations, family studies of longevity will be useful. Most recently, exome and whole-genome sequencing, gene expression, and epigenetic studies have been undertaken to better define functional variation and regulation of the genome. In this review, we consider how these studies are leading to a deeper understanding of the underlying biologic pathways to longevity.


aging 1.JPG

The Search for Longevity and Healthy Aging Genes: Insights From Epidemiological Studies and Samples of Long-Lived Individuals, an ebook

Genetic factors clearly contribute to exceptional longevity and healthy aging in humans, yet the identification of the underlying genes remains a challenge. Longevity is a complex phenotype with modest heritability. Age-related phenotypes with higher heritability may have greater success in gene discovery. Candidate gene and genome-wide association studies (GWAS) for longevity have had only limited success to date. The Cohorts for Heart and Aging Research in Genomic Epidemiology Consortium conducted a meta-analysis of GWAS data for longevity, defined as survival to age 90 years or older, that identified several interesting associations but none achieved genome-wide significance.

A recent GWAS of longevity conducted in the Leiden Longevity Study identified the ApoE E4 isoform as deleterious to longevity that was confirmed in an independent GWAS of long-lived individuals of German descent. Notably, no other genetic loci for longevity have been identified in these GWAS. To examine the conserved genetic mechanisms between the mouse and humans for life span, we mapped the top Cohorts for Heart and Aging Research in Genomic Epidemiology GWAS associations for longevity to the mouse chromosomal map and noted that eight of the ten top human associations were located within a previously reported mouse life-span quantitative trait loci.

This work suggests that the mouse and human may share mechanisms leading to aging and that the mouse model may help speed the understanding of how genes identified in humans affect the biology of aging. We expect these ongoing collaborations and the translational work with basic scientists to accelerate the identification of genes that delay aging and promote a healthy life span.

Keywords: Longevity, Genetics, Epidemiological studies
It is well established that both genetic factors and health-related behaviors influence survival to old age and survival to old age in good health. Longitudinal cohort studies demonstrate that lower levels of cardiovascular risk factors measured in midlife or early older years predict survival and healthy survival to 85 years of age (1,2) and beyond (3,4). Longevity has been observed to cluster within families so that parents and siblings of centenarians have a greater likelihood of attaining advanced age (5–7), and offspring of centenarians appear to have a delay in age-related disease (8,9). Studies of families clustered for longevity both in the United States and Europe (Long Life Family Study and Leiden Longevity Study) have demonstrated that offspring of long-lived participants has more favorable midlife risk factor profiles and less age-related disease (10,11). Similarly in the community-based Framingham Heart Study, adults with at least one parent surviving to old age have lower risk factor levels compared with individuals whose parents died younger and the risk factor advantage persists over time (12). The genetic contribution to longevity and human aging is likely to result from many genes each with modest effects. Some genes will likely affect longevity by increasing susceptibility to age-related disease and early death, whereas other genes are likely to slow the aging process itself leading to a long life. How genetic factors and their interaction with modifiable behavioral and environmental factors contribute to longevity remains unknown.

Longevity and Healthy Aging Phenotypes: Definitions and Heritability

Longevity is often defined as age at death or survival to an exceptional age such as 90 years or older or 100 years or older. Because life expectancy has improved dramatically across birth cohorts since 1900, care must be taken when study designs compare long-lived with younger cohorts. Women live longer than men and make up a larger proportion of the older population especially at exceptional old ages. For example, among the original 5,209 Framingham Heart Study participants with follow-up through 2011, there are 43 centenarian women and only six centenarian men, whereas at study entry (1948–1953), 55% of participants were women. Men are more likely to attain extreme old age escaping common age-related disease, whereas women are more likely to attain 100 after surviving common morbidities (13). While these observations raise the hypothesis that genetic and environmental factors influence the path to longevity differently in men and women, whether genetic factors play a greater or lesser role in men than in women is an area of debate (3,14). In a study of centenarians (100–104 years), semisupercentenarians (105–109 years), and supercentenaians (110–119 years), there was a progressive delay in the onset of age-related disease and onset of physical and cognitive function impairment with increasing age (15). Whether genes that influence survival to these extreme ages also play a role in survival to age older than 90 years is unknown.

The genetic contribution to longevity (age at death) has been estimated using both large twin registries and population-based samples (Table 1). Most heritability estimates from twin registries range between 20% and 30% (16,17), whereas estimates from population-based samples are slightly lower, ranging from 15% to 25% (18,19), suggesting a significant but modest genetic contribution to the human life span. One study conducted among an ethnically diverse group suggests that genetic influences on life span may vary by ethnicity with heritabilities ranging from a low of 4% for African Americans to 29% and 26% for Caribbean Hispanics and Caucasians, respectively (21). Using data from the GenomeEUtwin project that included more than 20,000 Nordic twins, Hjelmborg and coworkers (42) noted that genetic effects on life span were minimal prior to 60 years of age, but genetic effects on life spans greater than 60 years of age were significant and constant to increasing with advancing age. Starting at about age 60 years, the relative recurrence risk of an individual living past a specified age given that his/her cotwin also lived past that age increased with increasing age cut point in both men and women so that at age 92, the recurrence risk was 4.8 and 1.8 in monozygotic and dizygotic male twins and 2.5 and 1.6 in monozygotic and dizygotic female twins. Notably, recurrence risks similar to men occurred in women at a 5- to 10-year older age perhaps reflecting the longer average longevity in women. Using the Framingham Heart Study cohorts, we explored whether genetic influences on life span increase with achievement of older ages by examining age at death as a dichotomous trait using a liability threshold model adjusting for sex and birth year (20). In contrast to the modest heritability estimate for continuous age at death (16%), heritability of surviving past 65 years and surviving past 85 years was substantial at 36% (p = 4.2 × 10−10) and 40% (p = 9.0 × 10−10), respectively. Thus, genetic effects appear to be greater for survival to more advanced ages. In the Framingham Heart Study cohorts, heritability appears to increase with each 10-year increment in survived age (65 years, 75 years, and 85 years) for men, but not women, again suggesting that genetic effects on aging may be more substantial for men than women (20).

Familiality of Aging Phenotypes

The longevity phenotype measures overall life span without consideration of health and physical or cognitive function and hence is a very heterogeneous phenotype that may be affected by many environmental and other nongenetic factors. The relative contribution of additive genetic effects may be greater for more homogeneous phenotypes that describe specific aspects of aging and in turn may result in greater success in gene discovery. The heritability of reproductive aging phenotypes is at least 50%, and heritability is even higher for age-related diseases such as osteoporosis (low bone mineral density) and Alzheimer’s disease (Table 1). Genetic association studies have been successful in identifying genetic variants for these aging phenotypes and have the potential to uncover new biologic insights into the associated underlying aging processes (43–49).

Epidemiological studies that have followed participants over adulthood and collected a wealth of information in a standardized fashion have been important sources for development of aging-related phenotypes. Alternative aging phenotypes include disease-free survival, preservation of high levels of function including maintenance of cognitive function (50) and avoidance of bone loss (51), and successful aging (reaching advanced age with intact cognitive ability, physical function, and social engagement) (52). An index of physiologic age developed in the Cardiovascular Health Study by combining data across multiple systems was found to be a better predictor of death and disability than age itself (53). A frailty phenotype defined by five criteria, including unintentional weight loss, exhaustion, weakness, low physical activity, and slow walking speed (54), is distinct from physical disability and itself is predictive of mortality and other adverse outcomes. Although the frailty phenotype was developed in the Cardiovascular Health Study sample, it was found to be applicable across diverse studies (55). Many components of the multidimensional aging phenotypes developed in longitudinal studies are heritable, such as weakness (defined using handgrip strength) and lower extremity function, suggesting the potential for a genetic contribution to the overall phenotype (26,31).

In the family-based Framingham Heart Study, we estimated the heritability of several of the age-related phenotypes including longevity, morbidity-free survival, physical function, and frailty as well as walking speed and handgrip strength (Table 1). For quantitative traits, we used the variance components model, and for dichotomous traits, we used the liability model implemented in the software Sequential Oligogenic Linkage Analysis Routines (56). For both, we defined heritability as the proportion of phenotypic variance due to additive genetic effects only. The heritability of the physical function and frailty phenotypes in the Framingham sample has not previously been reported. For physical disability, three items from the Rosow–Breslau Functional Health Scale (are you able to walk a half mile without help? are you able to walk up and down one flight of stairs without help? are you able to do heavy work around the house without help? [57]) and five items from the Katz Activities of Daily Living Scale (can you do the following five activities independently: dressing, bathing, eating, toileting, and transferring) (58) were used. Physical disability was defined as present if the participant was unable to do any of the items. We examined the presence of physical disability at age 75 years using the exam at which the participant was closest to and within 5 years from age 75 using both the original cohort and offspring samples. Among the 2,614 individuals included in the analysis, 42% reported physical disability at age 75, and the heritability was 44% (p = .0002).

We estimated heritability of frailty and two of its components handgrip strength and walking speed in the Framingham Offspring cohort participants who attended the last completed research examination (2005–2008) during which the short physical performance battery was administered including a timed 4-m usual paced and quick walk (59). Frailty was defined if three of the five criteria proposed by the Cardiovascular Health Study investigators were present and prefrailty if one to two criteria were present (54). The analysis was adjusted for age and sex and included only participants aged 60 and older. The prevalence of frailty and prefrailty among the 2,207 individuals in this sample was 5% and 41%, respectively, and the combined trait of prefrailty and frailty was modestly heritable (h2 = 19%, p = .05). The usual and fast paced walking times in participants aged 65 years and older were rank normalized to reduce skewness and adjusted for age, sex, body mass index, and height. In contrast to frailty, both the usual and quick walk had a substantial genetic contribution with heritabilities of nearly 40% (usual walk: h2 = 0.38, p = .0002; quick walk: h2 = 0.36, p = .0003).

Next, we estimated heritability of handgrip strength in all offspring participants (mean age 67, range 43–93 years). Handgrip strength was measured three times in each hand with a JAMAR dynamometer. The maximum of the six trials was used in the analysis. Consistent with reports from twin studies, handgrip strength adjusted for age and sex had a heritability of 38% (p = 5 × 10−15). Aging phenotypes are associated with varied heritabilities (Table 1), and thus, the genetic contribution to the phenotype may be quite modest. Populations that differ in environmental factors may produce different heritability estimates even if the genetic factors influencing the trait are the same. Hence, it is remarkable that the heritability estimates for many of the age-related phenotypes are similar. Longevity and age-related phenotypes with higher heritability are of higher priority for genetic association studies, as these phenotypes are more likely to result in multiple genetic associations.

Genetic Association Studies

Genome-wide association studies (GWAS) test genetic variants across the entire genome for association with a phenotype and have proven highly successful for discovery of novel genes and biologic pathways involved in many common complex conditions (Table 2). However, few GWAS of longevity have been conducted to date. The Framingham Heart Study 100K project was the first investigation of the GWAS approach for longevity and aging traits (76). The project was relatively small in size including just 1,345 Framingham participants from the largest 310 families and limited in coverage of the genome as the genotyping was conducted with the 100K Affymetrix GeneChip. Modest associations between longevity (defined as age at death) and single nucleotide polymorphisms (SNPs) in or near FOXO1a, a gene important for life span in animal models, as well as other candidate genes were observed but failed to reach genome-wide statistical significance. Results of this investigation are considered hypothesis-generating and remain to be replicated.

Lending some support to the Framingham 100K longevity investigation, a genome-wide linkage study looking for chromosomal regions linked to successful aging in the Amish Study identified a linkage region near one of the SNPs associated with age at death (52). In 2007, more than 9,300 Framingham Heart Study participants were genotyped with the Affymetrix 500K mapping array plus 50K gene centric supplemental array as part of the National Heart, Lung, and Blood Institute’s SNP Health Association Resource project (, accessed March 9, 2012). At the same time, multiple large population-based longitudinal cohort studies in the United States and Europe with richly phenotyped participants planned to conduct genome-wide genotyping. Thus, in 2008, the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) Consortium was formed to facilitate GWAS meta-analyses and replication opportunities to enhance gene discovery for many phenotypes (77).

The CHARGE aging and longevity working group conducted a meta-analysis of GWAS results for longevity defined as survival to age 90 and older from four cohort studies (Age, Gene/Environment, Susceptibility-Reykjavik Study, the Cardiovascular Health Study, the Framingham Heart Study, and the Rotterdam Study) (64). The CHARGE collaboration permitted the assembly of one of the largest samples of long-lived individuals with genome-wide genotyping available at that time (1,836 individuals achieved aged ≥90 years). The comparison group was drawn from the same cohorts and included only deceased participants to ensure that no individual achieved longevity. The investigation detected 273 SNP associations for longevity that achieved p < .0001, but none of the associations achieved genome-wide significance (p < 5 × 10−8). In the next stage of the discovery analysis, among the 24 strongest independent SNP associations in the CHARGE meta-analysis, 16 SNPs were successfully genotyped in the Leiden Longevity Study and the Danish 1905 cohort, and one SNP near the MINPP1 gene was associated with longevity with p = 6.8 × 10−7 in the combined stage 1 and stage 2 discovery samples.

The minor (less frequent) allele was associated with a lower odds of achieving longevity (odds ratio 0.8). MINPP1 is a highly conserved gene involved in cellular proliferation. The CHARGE aging and longevity working group now includes investigators from over 15 cohort studies permitting future investigations of even larger samples of long-lived individuals with genome-wide genotyping and additional aging phenotypes that may improve our power to detect age-related genetic variation and provide support to our initial findings. We have subsequently conducted a meta-analysis of GWAS data from nine studies in more than 25,000 individuals aged 55 years and older for two age-related phenotypes, all-cause mortality and survival free of major disease and death (63). Although none of the SNP associations for either phenotype achieved genome-wide significance, 14 independent SNPs were associated with mortality, and 8 independent SNPs were associated with event-free survival. The SNPs were in or near genes highly expressed in the brain, genes involved with neural function, and genes associated with a variety of age-related diseases. Thus, our findings suggest that neural processes may be important in regulating aging.

Genetic Association Studies for Human Longevity

A GWAS conducted in 403 nonagenarians from the Leiden Longevity Study, and 1,670 younger population controls identified 62 SNPs associated with longevity at p < 1 × 10−4 (61). Successful genotyping of 58 of these SNPs was conducted in three independent studies: the Rotterdam Study, Leiden 85-plus Study, and Danish 1905 cohort. A meta-analysis of the 58 SNPs in all four studies that included more than 4,000 nonagenarians and 7,500 younger controls identified only one genome-wide significant SNP rs2075650 in TOMM40 at chromosome 19q13.32 close to the ApoE gene. The minor allele was associated with lower odds of longevity (odds ratio 0.71, p = 3.4 × 10−17). No other SNPs were associated with longevity. SNP rs2075650 was noted to be in moderate linkage disequilibrium with rs429358, the SNP that defines the ApoE E4 isoform and in very low linkage disequilibrium with rs7412, the SNP that defines the ApoE E2 isoform. In conditional analysis, with all three SNPs in the model, rs2075650 was no longer associated with longevity, whereas the minor allele of rs429358 had a deleterious effect on longevity, and rs7412 a protective effect leading the authors to conclude that rs2075650 effect on longevity was most likely mediated through the isoforms of the ApoE gene. A case-control GWAS conducted in 763 German nonagenarians and centenarians, and 1,085 controls (mean age 60 years) identified rs4420638 near the APOC1 gene and replicated the finding in an independent sample (60).

This finding was also fully explained by linkage disquilibrium with ApoE E4 isoform confirming the prior report. These results are intriguing as the ApoE gene is one of only two candidate genes with consistent evidence for association with longevity in humans (73). The ApoE E4 isoform has been linked to elevated cholesterol, cardiovascular disease, age-related cognitive decline, and dementia. The ApoE E4 isoform is more strongly associated with Alzheimer disease than longevity and other conditions. Homozygosity for the Apo E E4 allele confers up to a 15-fold risk for Alzheimer’s disease in whites and an 8-fold risk in African Americans compared with the most common ApoE genotype (E3/E3 [78]). Thus, ApoE may influence longevity through premature atherosclerosis and age-related diseases. Notably, the CHARGE study did not observe genome-wide significant associations between the ApoE gene region and longevity. However, neither of the two SNPs (rs429358 and rs7412) that define the ApoE E4 polymorphism nor any strong proxies appear on any of the chips used by the CHARGE consortium studies. In the CHARGE meta-analysis, the odds of living past age 90 years associated with the minor allele of rs2075650 was 0.85 ( p = .046); hence, the effect is consistent with the prior reports.

FOXO3a was first noted to be associated with longevity in a candidate gene study conducted in male centenarians of Japanese descent (65) and subsequently replicated in diverse samples of centenarians and long-lived individuals (66–69). Remarkably neither the Leiden or German studies nor the CHARGE GWAS identified an association between longevity and FOXO3a. Finally, a recent GWAS of 410 long-lived individuals and 553 young controls from Southern Italy identified an SNP in an intron of the CAMKIV gene among the top associations. This association was replicated in a sample of 116 long-lived and 160 young controls (p < 10−4 discovery analysis, joint replication analysis p = 1.7 × 10−6) (62). Interestingly, in vitro work suggests that this gene activates proteins in candidate genes for longevity (AKT, SIRT1, and FOXO3a). About 300 genetic variants in 30 genes in the insulin/insulin-like growth factor 1 (IGF-1) signaling pathway were genotyped in older women participating in the Study of Osteoporotic Fracture. Replication studies were conducted in the Cardiovascular Health Study and an Ashkenazi Jewish Centenarian Study (70). SNPs in two genes in this pathway (AKT1 and FOXO3a) were significantly associated with human life span. A better understanding of the biological mechanisms by which the FOXO3a and AKT1 variants influence human longevity will be important to development of interventions to delay aging (79).

Individuals with a family history of longevity have lower mortality for most age-related diseases (80). Therefore, researchers tested the hypothesis that this may be due to the absence of susceptibility alleles for common diseases in the Leiden Longevity Study and the Leiden 85 Plus Study. They examined whether the long-lived individuals had fewer copies of 30 alleles discovered through GWAS to be associated with coronary artery disease, cancer, and type 2 diabetes compared with a younger comparison group (81). Notably, no difference in the number of risk alleles was detected, suggesting that at least in these populations, survival to old age is not determined by the absence of risk alleles identified to date for age-related disease.

The effort to identify genes that affect longevity through candidate gene and GWAS studies has had only modest success to date. This is likely due to a combination of factors including the heterogeneity of the phenotype, the influence of environmental and dietary factors, which vary widely across populations and the relatively small sample sizes of published longevity GWAS. Many of the successful GWAS with many replicating genome-wide significant signals have sample sizes of more than 10,000, and some published studies of quantitative traits such as age at menarche have had sample sizes of more than 80,000 (43). It is clear that defining more homogeneous phenotypes through age-related traits such as age at menopause and bone mineral density leads to greater success in identifying aging-related genes. Another potential explanation for the lack of identification of risk variants for longevity is that the bulk of the genetic effects are due to rare variants or structural variation in the genome. The GWAS chips used to date have focused on common SNP variants, which typically do not tag rare variants well. Recent work within the CHARGE consortium studies suggests that copy number variants are associated with mortality (82). With the advent of low-cost exome and whole genome sequencing as well as higher-density SNP chips with 5 million or more variants, we will soon have the opportunity to determine whether rare or structural variants explain a substantive proportion of the heritability of longevity and other aging traits.

Join 25,000 people in helping redefine health with health concierge, healthy aging and precision medicine.