Seeking the Longevity Equation
Whether the secret to long life turns out to be a positive attitude, lucky genes, or highly complex interactions among a multitude of factors over a lifetime, Anatoliy I. Yashin will likely capture those forces at work in a sophisticated mathematical model.
Yashin, the scientific director of DuPRI’s Center for Population Health and Aging, applies advanced mathematics to understanding the complexities of human aging. The new methods and models he has devised, and the results they generate, continually influence the course of research in the field of aging studies.
Indeed, in recent work, Yashin and his team have shown that although long-sought “longevity genes” remain elusive, hosts of genes with very small effects can, together, exert a powerful influence on a person’s healthy lifespan. Moreover, the most relevant genes might differ from one individual to another, depending on conditions in the environment.
Such challenges to conventional wisdom are typical of Yashin’s findings — the result of tackling some of the hottest topics in biology from an unusual direction. Originally trained in plasma physics at the Moscow Physical and Technical Institute, Yashin, in his quest to reveal the dynamic mechanisms behind aging and mortality, continues to come up with ever more creative ways of mining data for insights into what is perhaps the most complex system of all — the human body.
Demographers trying to understand aging have always known that it had to be a complicated and varied process, according to Yashin, and therefore mortality curves for whole populations reveal little about what preserves the health or causes the death of any individual in the group. But since the biology of that process was largely unknown — even to biologists — until fairly recently, population researchers continued trying to draw conclusions about mortality from group data.
Not satisfied with that approach, Yashin argued thirty years ago that to make meaningful predictions about mortality trends among real people required understanding the “Deviant Dynamics of Death in Heterogeneous Populations.” In a 1982 paper with that title, Yashin, then a senior researcher at the Academy of Sciences in Moscow, and his coauthor James W. Vaupel, an associate professor of policy at Duke (now, co-director of the DuPRI CPHA and founding director of the Max Planck Institute for Demographic Research), challenged their fellow demographers to take on this problem.
In brief, Yashin explains, the issue was that “Every individual has his own chances to live long or live short. It depends on genetic background, individual exposures, conditions, different families, etcetera, so everything influences chances of survival. And since there is a kind of distribution of individuals in the population, this variety of chances of death produced some interesting effects on mortality. If you change the distribution you will change the mortality rate.”
Yashin was drawn early in his career to analyzing healthcare policy questions in the Soviet Union — a line of inquiry that led to multi-year contracts pursuing health-related projects at the International Institute for Applied Systems Analysis in Vienna, where he first met and began collaborating with Vaupel. In 1990 when Vaupel, by then at the University of Minnesota, received a grant from the U.S. National Institutes of Health to study longevity in the oldest old, Yashin joined his friend in Minneapolis.
The pair set out to explore some of the intriguing questions raised by patterns within mortality curves, including the “population compositional influence on mortality rates” they had identified years earlier. Another mystery dated back to an observation in 1825 by Benjamin Gompertz that mortality rates in animals and people seemed to level off and sometimes even slow at advanced ages, hinting there might be something special about those who made it to very old age that allowed them to keep on going.
Vaupel had already devised the concept of individual frailty to describe such an intrinsic quality — whether an aspect of the environment or an inherited predisposition, for example — that would influence a person’s lifespan. To examine this quality in the oldest old more closely, they needed a model that could tease apart the genetic and environmental influences on longevity in individuals. They also needed a lot of data on twins. Yashin and Vaupel began working with Danish registry data, and eventually moved their operation to the University of Southern Denmark, in Odense, to work on determining which aspect of individual frailty — genes or environment — is more important to lifespan.
The correlated frailty model Yashin developed there with PhD student Ivan A. Iachine in 1995 remains widely used today by researchers trying to parse the relative influences of genes and environment. Applied to Scandinavian twins, it offered Yashin and Vaupel the surprising finding that genes were fairly unimportant to longevity — perhaps 25 percent of the picture. Genes, though, could explain about 50 percent of the differences in frailty between individuals. The intriguing result also begged a new set of questions: with or without fortunate genes, what had the very old done, or exposed themselves to, or avoided during their lifetimes that might account for their survival?
By 1996 Vaupel and Yashin were at the MPIDR in Rostock, Germany, and shifting their focus to the interactions between genes and environment over a lifetime that might shed light on extreme longevity. They realized that longitudinal studies could provide the kind of long-term data on health histories and mortality they would need, but wouldn’t contain everything required to understand the life course of individuals.
“We understood that in order to study aging, we should take into account the wealth of knowledge about aging and lifespan that is accumulated in the field,” Yashin says. “It is not represented in specific data, but people have found connections between phenomena of aging, for example, stress resistance tends to decline with age, which means that the same stress that could be tolerated when we were young could be very dangerous when we're old.” Similarly important, he adds, are adaptive capacity — the body’s ability to adjust its functioning in response to stressors, as well as so-called allostatic load, essentially the burden of chronically making those adjustments.
The quadratic hazard model Yashin developed to incorporate those kinds of unobserved variables, along with longitudinal data, made it possible to analyze a much richer picture of the forces influencing health and lifespan. With this new approach, Yashin and his colleagues have demonstrated the importance of stress and other exposures on how we age and even found evidence for the controversial phenomenon of hormesis — the idea that a little bit of stress, even in the form of a toxin, can produce beneficial effects in the body.
The model also allowed Yashin and his team to show in detail the interactions of various objective biomarkers — such as blood glucose, blood pressure, or weight, with one another, as well as their individual fluctuations over time. Among the important insights these studies revealed is that norms change over a lifetime — for instance, a healthy body mass index for an 80-year-old might be the same BMI that person had at 30, but that is not the ideal at age 15 or 50. Moreover, at the same BMI, the 80-year-old and the 30-year-old have very different states of health, so the biomarker itself is not very informative. The arc of such health measures over the course of that person's life can be very telling, however.
“If you look at the age trajectory of blood pressure,” Yashin explains, “it maxes then declines, and then if we see one individual has a high rate of decline, and another has a slower one, and we know those who have slow decline live longer, then the slope of decline is a kind of risk factor. So when we describe these dynamic changes in terms of mathematical equations, the parameters of these equations not only describe chances of death, they also characterize biological changes in the individual.”
Yashin calls the original quadratic hazard model, which continued to evolve into the stochastic process models he is applying in many of his current projects, “the kind of methodological breakthrough that allowed us to go further, and to merge together data from different worlds...like population data, demography, longitudinal and some biological and genetic data.” That original model built on a long history of collaboration between Yashin and Duke mathematicians Max Woodbury and Kenneth G. Manton, both innovators who established Duke’s reputation for applying advanced mathematics in population studies. When an opportunity to join the Duke faculty opened up in 2003, Yashin moved to the U.S. with his wife and CPHA colleague, Svetlana V. Ukraintseva, a geneticist who was also studying aging at MPIDR at the time, as well as Konstantin Arbeev, a specialist in statistical modeling. The team now includes Eric Stallard, Kenneth Land, Alexander Kulminski, Igor Akusevich, Irina Kulminskaya and Deqing Wu.
Putting the Pieces Together
The latest models developed by Yashin’s team enable them to perform aging studies with unprecedented breadth and depth. By adapting the stochastic approach to jointly analyze huge and disparate longitudinal datasets, such as data from the multigeneration Framingham Heart Study, Medicare records and Duke’s own National Long Term Care Survey, the group is continuing to explore the interconnected forces that influence how any one person will age. Their work demands an interdisciplinary approach, Yashin says, “to put all these ideas together.”
The team’s methods can be considered a form of systems biology — a cutting-edge field enabled by modeling techniques borrowed from the physical sciences. And their projects can touch on some of the most fundamental questions in biology. Recent studies have examined, for instance, the relationship between genes involved in aging and those activated in cancer cells, and whether evolutionary trade-offs explain connections between seemingly unrelated diseases — such as the apparent postponement of cancer risk seen in people with a gene variant that raises the risk of developing Alzheimer’s disease.
But their goal remains a very pragmatic one. Discovering how and why the human body ages will help researchers “find the optimal balance between postponing aging and postponing decline,” Yashin says. That, in turn, will help policymakers to predict the population's needs and help individuals to live the longest, happiest and healthiest lives possible.