"How long can I live?" my six-year-old daughter asked me when I returned home from my father's funeral in India. My father had lived until eighty-five—a vigorous, preternaturally healthy octogenarian at eighty-two—until he had spiraled inexplicably into a ferocious form of dementia that took his life in three years.
At first glance, a simple question. Plot a smooth line through the generally increasing arc of average life expectancy over the last century, and you should arrive, quite easily, at the answer. A quick web search reveals the number: about 78.8 years for a child born in 2010 (my daughter), up by nearly twelve or fifteen years since the 1940s.
But that, of course, was not her question. "How long can I live?" is a more capricious puzzle; it asks us to solve a very different sort of conundrum—not the average human life expectancy, but its outer limits. Is there a limit to human longevity? When Jeanne Louis Calment, a woman living in Arles, France, died on 4th August 1997 at 122 years, she achieved the longest human lifespan documented on record (reassuringly, she claimed that she was not particularly athletic, and ate nearly a kilogram of chocolate every week). Will someone from my daughter's cohort live longer than Calment?
Last month, a paper published in the scientific journal Nature attempted to answer the question and ignited an unexpected storm of controversy. The authors of the study—Xiao Dong, Brandon Milholland and Jan Vijg, all at Albert Einstein College in New York—began by acknowledging an important but little-appreciated fact: In animals, genetic manipulations, combined with artificial environments, can increase the outer limits of lifespan so dramatically that these animals can live two or three life-cycles longer than their wild counterparts. The crucial piece of logic is this: Since previously unknown genes and environments had intersected to contribute to the "off-the-chart" lifespans of these animals, might there by some human gene-environment combination—some unknown mix of gene variants and a unique environment—that might extend our lifespans beyond known limits? If so, the smooth, rising arcs of average human lifespan would not predict the limits of our longevity—just as the average lifespan of a wild mouse tells us virtually nothing about how long a genetically manipulated mouse can live in a lab. In short, the outer limit of human longevity might leap forward without boundaries, while the average lifespan might demonstrate only moderate increases.
How, then, might we determine the outer limits of human longevity? What if we looked year by year and determined the maximum reported age of death for that particular year? Say the oldest reported death in 1974 was 110 years; in 1975, 112 years; in 1976, 113 years—and so forth. What if we plotted the lines through these points? Would the maximum reported age of death keep rising inexorably, foreboding a human Methuselah? Or would the maximum age plateau, signaling a limit to longevity? Might it even fall, suggesting a contraction in this feature of humans? To answer these questions, the authors of the Nature paper used data from the International Database on Longevity to determine maximum age of death, and plotted these dots across three decades from the 1970s to 2010. As expected, perhaps, the spray of dots that first appeared on the graph spurted forth with a generally upward trend. In 1970, for instance, the maximum reported age of death was 110. By 1990, that age was 114. By the mid-1990s, it reached 115.
But then an odd thing happened to the plot: The data began to move in curious, unexpected ways. The maximum age of death rose to its peak in 1997, with Jeanne Calment's death at age 122. Two years later, in 1999, an American woman named Sarah Knauss died in Pennsylvania at age 119. And then the age of death dipped downwards, clustering again around 115 between 1999 and 2010.
To Vijg's team, the spray of dots suggested a V shaped curve: a line that rose continuously through the 1970s to the 1990s, reached its peak with Calment and Knauss, and then broke and flattened and dipped downwards from that point towards 115 or 116. The implication was evident: The downward slope of the V suggested that a natural limit to human lifespans had been reached. "There is a natural limit to human lifespan of about 115 years old," Vijg's team concluded in a discussion of the paper in Nature. "There will still be occasional outliers like Calment, but … the probability of a person exceeding 125 in any given year is less than 1 in 10,000." "The limit is surprising," Vijg continued. "If anything you would have expected more Jeanne Calments in recent years, but there aren't."
Vijg's conclusions generated a brisk response on the blogosphere. To other readers, the same spray of dots suggested quite the opposite interpretation. When Philipp Berens and Tom Wallis, researchers at University of Tübingen in Germany, examined the same data points used in Vijg's paper, they found that they could plot a perfectly reasonable straight line through it—a line that seemed about as plausible given the data. Far from plateauing or falling, their line (also plotted with statistical software, given a somewhat different set of underlying assumptions) rose continuously. Their conclusion was precisely the opposite: If the trend-line continued to move upwards, then it evidently suggested that the limits of human lifespan had not been reached. Yet when they replotted the lines having deepened their analysis with a much larger database of maximum age at death, with points that extended backwards to the early 1900s, Berens and Wallis found that the plateauing effect noted by Vijg reappeared.
So which of these various representations is correct? There is, oddly, no simple way to know. The interpretation of statistical data, especially when points are sparse, lies, strangely, in the eye of the beholder. Or, more accurately perhaps, it lies in mathematical assumptions that the "beholder" chooses to make. Are Knauss and Calment outliers, or do they truly represent some outermost boundary of longevity? If you make one series of assumptions, it turns out, the line of longevity plateaus and falls (a la Vijg). Change the assumptions, and the same data suggests a rising line (a la Berens and Wallis). Add points of data to one end of the curve, and the line might seem to plateau and fall again. Change the parameters, and yet another kind of line can be plotted.
And superposed on these statistical uncertainties is the final uncertainty of all: human intervention. What if the manipulations achievable in lab animals were extended to humans? The transplantation of stem cells, the creation of artificial organs, radical metabolic reprogramming achievable through new diets, the possibility of entirely novel environments to decelerate aging and decay—all these might challenge even the most stringent attempt to model the future of longevity based on the past. Technology, uniquely, swallows its own history. The doctor who tried to opine on the limits of antimicrobial therapy on the eve of Penicillin's discovery turned, overnight, into history's fool.
Imagine, then, a world in which a few men and women reach 160 years of age. If such a radical extension of lifespan occurred in my daughter's generation, say, then the person in question would be living in the 2170s. As a point of comparison, a 160-year-old man or woman alive today would have been born in 1856; she would have lived through the assassination of Abraham Lincoln, the birth of the commercial lightbulb, both World Wars and the college graduation of her great-great-great grandchild. (When Sarah Knauss became the world's oldest living person in 1998, her daughter, also alive, was reportedly 96 years old.)
Such an extension of human longevity would allow us to answer scientific and medical questions that lie outside current human experience. What organs degenerate under such extreme pressures of age, and what remains? Can organs, once replaced, be replaced again and again? Can tissue be regenerated from aging stem cells, or do we need harvests of certain tissues earlier in life to maintain these cells as "backups" or repositories of information? Does our concept of kinship—of familial bonds—change across four generations? What kinds of memories remain etched in the brain for 160 years?
Most importantly, as we accumulate a trove of data—genomes, behaviors, environments, diets—from men and women living at the extremities of age, could we envision using this information to radically extend the average lifespan of humans?
For now, to avoid the pitfalls of foolishness, I told my daughter what seemed like the most even-handed interpretation: "As long as you think you could live, darling."