| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||
BOOK REVIEW |
Professor, Donald W. Reynolds Department of Geriatric Medicine The University of Oklahoma Health Sciences Center Oklahoma City, OK 73104
Longevity: The Biology and Demography of Life Span, by James R. Carey. Princeton University Press, Princeton, NJ, 2003, 278 pp., $75.00 (cloth), $29.95 (paper).
Books on aging range from those written for the general public (e.g., Clark, 1999; Hayflick, 1994), to those aimed either at students or professionals seeking a scientific overview and introduction to the literature (Holliday, 1995; Rose, 1991), to technical books directed at specific technical audiences (Finch & Kirkwood, 2000). Longevity: The Biology and Demography of Life Span, by James R. Carey, falls into the third category.
Carey is an entomologist who has spent more than a decade conducting large-scale studies using Mediterranean fruit flies (medflies). Anyone who has been involved with large-scale animal studies will be impressed with the careful design and execution of these experiments. Carey's extensive publication record in respected scientific journals reveals the quantity and quality of his work, and his obvious competence for writing a book. His book brings all of his research together in one place, along with some individual and global interpretations of its meaning and significance. The one warning I would give to a prospective reader is that the vast majority of this book is devoted to studies conducted on flies. Although this approach is reasonable, appropriate and understandable given the background of the author, it does require a motivated reader to make it to the big-picture chapters at the end of the book.
Aging can be examined from many perspectives. Carey uses a population-based biodemographic perspective. Demography provides the research issues and methods of analysis, while biology motivates his experimental studies and data interpretation. At the outset, Carey declares that his book is based on three "conceptual anchors:" (a) the "absence of species-specific life span limits," (b) the "mortality response of males and females is context-specific," and (c) the existence of "biodemographic linkages between longevity and reproduction" (pp. 7–8). There is another recurring theme permeating the book that should also be considered a conceptual anchor: "studying the causes of death often provides little insight into the nature of aging" (p. 3). It is these anchors, both stated and implied, that form the basis for my review of Carey's book.
There are numerous topics covered in the book that I found informative and useful. Carey provides a nice, although brief, summary of life table construction and interpretation of life table statistics. Through a detailed narrative and accompanying plots, he convincingly demonstrates that sex differences in age trajectories of mortality are far more complicated than the fact that females simply and consistently outlive males. This finding is consistent with studies of laboratory animals (dogs, mice, rats) in which significant differences in the response of males and females to insults (e.g., radiation, chemicals) have led to the two genders almost invariably being analyzed separately. Carey's medfly studies also make a persuasive and positive contribution to the scientific consensus that population mortality and even individual longevity are intimately linked to reproduction and the life history strategies that shape that reproductive biology.
Research on aging occurs at virtually all levels of biological organization, and real progress on this great unsolved problem in biology will only occur when an integration of the unique insights provided by each level of organization is achieved. This sentiment, though not explicitly stated in the book, is implied by Carey's observation that "aging-oriented research necessarily focuses on lower levels of biological organization such as the molecule and cell, longevity-oriented theory focuses on the level of organization considered by many biologists to be the quintessence of biological relevance—the whole organism" (p. 217).
Throughout the book Carey emphatically declares that his research demonstrates that limits do not exist for either the life span of individuals or the life expectancy of populations, populations don't possess characteristic schedules of age-specific death rates, and the Gompertz paradigm used by countless researchers to model mortality over the last 178 years should be rejected. These are serious claims because they have huge implications for such real-world issues as escalating medical costs, the solvency of age-based entitlement programs like Social Security, and whether human ingenuity should be directed toward manufacturing progressively more survival time or focused on improving the quality of life experienced by the rapidly growing number of individuals surviving into the post-reproductive period of the life span. The stakes are high, the biological and societal effects are real, the debate is intense, and a consensus on these issues (scientific or otherwise) is far from being achieved. Both the educational and technical value of this book suffer from the exclusion of opposing arguments, of which there are many, and the failure to place these important and contentious issues within the context of their long and rich history of intellectual discourse.
Cause of Death
What I described as Carey's undeclared fourth conceptual anchor lies at what I believe is the core of the disagreements that have polarized biodemographers and biogerontologists. Carey's conclusions (and those made by many influential demographers) about longevity and life expectancy are almost always derived from analyses that ignore cause of death. This is no surprise in Carey's case because in his own research it is not possible to determine with any degree of certainty why flies die. It is my opinion that ignoring cause of death ignores one of the most important variables that influence the duration of life of both individuals and populations—that is, biology. Aging and senescent death are interrelated biological phenomena, and understanding them requires an examination of the molecular and cellular pathogenesis that degrades functional integrity and homeostatic control, causes death, establishes a life span for individuals, and influences the vital statistics used to describe the population to which the individual belongs.
Recognizing that modern life expectancies were achieved by saving mothers from death during childbirth and saving children from the infectious diseases that killed the vast majority of their ancestors requires distinguishing among causes of death. The resulting shift of deaths from the young to the old was also accompanied by a transition from mainly exogenous (extrinsic) deaths (accidents, homicide, infectious disease, predation, starvation, suicide) to a mixture of exogenous and endogenous (intrinsic) deaths (genetic diseases—germ line and somatic—and the degenerative diseases of aging such as heart disease, cancer, and stroke). These two categories of death have vastly different etiologies.
Evolutionary theory tells us that the biology and life history of all organisms is a response to the persistent and extremely high levels of extrinsic mortality that have existed throughout the history of life on earth. Intrinsic mortality reflects both a built-in biology (genes, both harmful variants that cause diseases as well as normal variants involved in critical processes that influence longevity indirectly) and the unintended failure of that biology (degenerative diseases and disorders of aging) in bodies that were not designed for extended operation. Failure to distinguish between these fundamentally different inputs into mortality dynamics is at the heart of the disagreements that exist over present and predicted future patterns of aging, morbidity, mortality, and longevity.
Extended Operation
Although the absolute meaning of extended operation varies among species, it invariably implies an age beyond which the probability of continued survival is low. When thinking about when that age might be, it is essential to ignore the present and, instead, focus on the past. As recently as 1900, life expectancy at birth for women in the United States was around 50 years. Historical research estimates that life expectancy during the height of the Roman Empire was 20 years. What would it have been during the more biologically relevant timeframe of 130,000 years ago? No one knows for sure, but it is unlikely to have exceeded that which was experienced in the time of the Roman Empire. What is known with certainty is that low life expectancies mean high mortality pressures, particularly at younger ages. Thus, delayed reproduction in high mortality environments increases the risk of the ultimate biological penalty—death before reproduction.
In a world filled with easily accessible calories, girls today are giving birth at ages as young as 10 years. Although these births occur, both the mother and child often experience significant mortality and health risks. This implies that these mothers are bumping into the lower limit of the reproductive envelope for humans, and 12–14 years represents a reasonable estimate for the normal age of sexual maturation for humans. No matter where you go on the globe, the vast majority of human babies are born to parents (mothers and fathers) who are 35 years old or younger. Menopause—which occurs on average around 50—is theorized by evolutionary biologists to be beneficial by eliminating the historically high risk of death during childbirth so that the parents (especially the mother) can improve the survival probability and reproductive success of their existing offspring. This reproductive timeframe permits parents to be grandparents many times over by the age of 50. With life expectancies today approaching and in some cases exceeding 80 years, we have the historically recent and unprecedented phenomenon of human populations where most individuals in those populations survive beyond the ages required for reproduction and grandparenting. As such, humans in large numbers have entered into the relatively uncharted waters where evolution is completely indifferent to the consequences of gene expression (good, bad, or neutral). In the historical context of survival that molded our biology, it also means that humanity has already ushered itself well into the age ranges that must be considered extended operation.
The evidence that the age of extended operation of bodies is upon us is everywhere (Olshansky, Carnes, & Butler, 2003). Population aging has become a global phenomenon. In many countries, intrinsic causes of death have become a major contributor to the overall mortality burden. Geriatric syndromes have been formally recognized, and patients manifesting them have become commonplace. Medical schools are trying to train geriatricians to address the growing and unique needs of a rapidly growing elderly population. Scientists debate whether senescence (or generalized organ failure) should be a recognized cause of death. Medical costs are skyrocketing, and the solvency of age-based entitlement programs is being called into question. Mechanisms of aging and interventions for aging appear throughout the research descriptions of study sections of the National Institutes of Health. Anti-aging products represent a multibillion dollar global industry. In the collective, these phenomena are indicators that human ingenuity (dentistry, public health, medicine, and science) may have already reached a level of success that permits many people to approximate and some to even surpass their biological warranty periods (that is, their life span potential, what individuals can achieve under perfect conditions).
Manufactured Survival Time
More people than ever before are surviving to ages where deaths due to intrinsic causes dominate. This growing importance of intrinsic causes of death means that longevity gains in the future will require more than battling microbes and reducing maternal mortality. Although deaths from intrinsic causes have been successfully delayed, it is not clear whether the age at onset of these diseases has been delayed, or whether the progression of their pathogenesis has been slowed. To date, not one intrinsic cause of death has been eliminated. Despite a 30-year declared war on cancer, cancer still exists. Has modern medicine simply become better and better at managing the manifestations of aging and age-related diseases? Unless we learn how to manipulate the fundamental biology of humans, it will become progressively more difficult to manufacture additional survival time (Olshansky, Carnes, & Desesquelles, 2001).
Although no credible scientist believes that aging can be stopped or reversed, there are many (including me) who are optimistic that the rate of biological aging can be slowed. However, the intensity of the ongoing anti-aging debates reveals how divided the scientific community is over this issue. If humans are already achieving or surpassing the expiration dates of their biological warranty periods, then an ability to usher people even further beyond their inherent life span potential has serious implications. Living older longer would be a disaster on both a personal and a societal level unless life extension is accompanied by either a reduction in or much enhanced medical management of the nonfatal diseases and disorders that so dramatically degrade quality of life in the elderly population. Even under the most optimistic scenarios, reducing both mortality and morbidity simultaneously will be a daunting task because instead of adding decades to the lives of children, it will require adding healthy years to the lives of people who have already survived seven or eight decades.
Gompertz Rejected
One of the most emphasized results from Carey's large-scale medfly studies described in his book is the deceleration of death rates observed at older ages—a phenomenon that has also been observed in human populations. This topic deserves discussion because the deceleration has been used by Carey and other demographers to reject the Gompertz paradigm of mortality.
Benjamin Gompertz was an English actuary who in 1825 developed an equation to describe age patterns of mortality that is still used today. Theory tells us that the Gompertz equation applies when the failure of the weakest link in a complex system causes the entire system to fail. Between 1950 and 1970, the famous biogerontologist (and president of the Gerontological Society of America) George Sacher (1956) developed a physiological theory of radiation injury and aging based on the Gompertz distribution. Sacher envisioned physiological oscillations occurring around a mean homeostatic state lying between lethal boundaries. Death occurred whenever an aging or injury-induced oscillation crossed the lethal boundary. Arguing over a 178-year-old formula seems pretty absurd until you realize that the Gompertz function has been used so often and for so long that it has taken on almost mystical importance in the study of mortality. Overturning this paradigm is a very big claim.
Carey rejects Gompertz because the slowing of the increase (deceleration) in death rates at older ages that he and others have observed appears to violate Gompertzian mortality behavior. Gompertz has an exponential form, so death rates rise exponentially on an arithmetic scale or linearly on a semilog scale. This means that Gompertzian mortality can never decelerate. According to Carey, mortality deceleration at older ages "forces demographers and gerontologists to rethink the idea that senescence can be operationally defined and measured by the increase in mortality rates with age" (p. 180). I would agree. The increase in mortality rates used to estimate the actuarial rate of aging makes little sense. However, it makes little sense not because it violates Gompertz, but because the extrinsic mortality that contaminates the all-cause mortality schedule, and varies among populations, obscures and distorts the increase in the underlying, and more relevant, intrinsic mortality rate (Carnes, Olshansky, & Grahn, 1996).
What Carey has neglected to point out in both his book and his research articles is that Gompertz himself (and numerous scientists since him) recognized that his formula does not apply to mortality at older ages (Olshansky & Carnes, 1997). Although this lessens the significance of Carey's rejection of Gompertz and his claim to have overturned a paradigm, it remains true that Gompertz does not work at older ages for the very reason that Carey and others before him have noted. This, however, is not the end of this story; we must now delve into muddier waters.
Gompertz Resurrected?
As Strehler and Mildvan (1960) noted, non-Gompertzian mortality trajectories (including either decelerations or accelerations or both) can arise from mixtures of subpopulations, each of which adheres to Gompertz. It is easy to envision how human populations should be comprised of a heterogeneous collection of subgroups (gender, genetic, environmental, behavioral, socioeconomic). In the demography literature, the subgroup argument used to explain deceleration is referred to as demographic selection. The people who survive to older ages are those who successfully ran the gauntlet of mortality that killed people at younger ages, and would (ipso facto) have lower mortality rates. Thus, as the shorter-lived subgroups drop out, the surviving population that becomes progressively dominated by longer-lived subgroups will transition toward the lower mortality characteristics of the long-lived subgroups. When subgroups are ignored, this mixture scenario will produce the mortality deceleration in all-cause mortality that Carey and others observe at older ages.
Carey clearly recognizes this phenomenon, and reports that a dozen Gompertz subpopulations gave a good fit to his medfly data. However, just when it seemed that Gompertz was resurrected, Carey quotes literature (Peters, 1991) which contends that selection arguments are "tautological," "the under-lying argument is circular," the "explanations are not open to falsification," and concepts that explain everything, explain nothing (p. 37). At another point in the book, Carey opines that "one is tempted to classify individuals into groups displaying various degrees of vitality or frailty which expresses itself in both longevity and level of reproductive activity" (p. 157). After reading this, I was no longer sure whether Carey accepts or rejects Gompertz, or whether he accepts or rejects heterogeneity arguments. I'm no philosopher or logician, but heterogeneity is one of the defining biological characteristics of sexually reproducing species, and I cannot envision how populations of humans or any other sexually reproducing species could be anything other than a collection of heterogeneous subgroups (Carnes & Olshansky, 2001).
Why is the subgroup debate important? Once again, we have a phenomenon with important biological and demographic implications that can be interpreted in completely opposite ways. In purely mathematical projections, decelerations of death rates can become declines which when extrapolated into the future will lead to extreme estimates of life expectancy at birth (e.g., 100 years or more). An age range of deceleration, declining or flat death rates can also be interpreted as a timeframe when the grim reaper takes a break. Under this scenario, life spans could be significantly extended if people who would normally die at younger ages could somehow be ushered to this age range.
The Gompertzian subgroup argument produces a much different interpretation. As mentioned above, such a mixture of Gompertz subgroups can produce a deceleration of death rates. In fact, the deceleration will become a decline in death rates as attrition of the long-lived subgroups begins to reveal the longest-lived subgroup. However, when all that is left is the longest-lived subgroup, mortality rates will once again conform to a Gompertzian trajectory of exponential increase (on an arithmetic scale). In the former conceptual framework that ignores subgroups, mortality declines at older ages are a sign of an unlimited potential for future gains in life expectancy. In the latter conceptual framework, the deceleration of death rates at older ages is a harbinger of increasing mortality rates, and the emergence of people who represent the extreme end of the distribution for human longevity. Which, if any, of these scenarios are realistic has significant implications for predictions of future patterns of aging, health, and longevity.
Mortality Signatures
Throughout this review, I have emphasized the importance of distinguishing between intrinsic and extrinsic mortality (Carnes & Olshansky, 1997). The importance of partitioning causes of death into biologically meaningful categories for the analysis of mortality has an unbroken intellectual history dating back to William Makeham (1867), the father of competing risk theory. We applied our partitioning to mortality and pathology data for 13 strains of laboratory mice (inbreds, and hybrids) used in studies spanning a 40-year timeframe of declining death rates from infectious disease ascribed to progressive improvements in animal husbandry (Carnes, Olshansky, & Grahn, 1996). All-cause mortality trajectories differed dramatically across time within strains, while the age-specific mortality trajectories for intrinsic causes of death (referred to as intrinsic mortality signatures) were, without exception, statistically indistinguishable. The remarkable constancy of intrinsic mortality signatures across time for mouse strains exhibiting a wide range of mortality dynamics was interpreted as strong support for the existence of signatures.
Intrinsic mortality signatures were also consistent with the genetic constancy claimed by the companies that sell laboratory animals. Finally, our intrinsic/extrinsic partition was nearly identical to the endogenous/exogenous partition recommended by Shryock and Siegel (1975) for comparing mortality among geographically separated human populations in their highly regarded book The Methods and Materials of Demography. In both cases, the expectation is that populations separated by either time or location possess a common pattern of intrinsic mortality that is embedded within and obscured by age patterns of extrinsic mortality that vary between and among populations.
Carey repeatedly rejects mortality signatures, and concludes that "Olshansky and Carnes (1997) are incorrect in their assertion that species possess a characteristic signature, and thus an irreducible mortality component defined by Tuljapurkar and Boe (1998) as a component of mortality that will never be reduced by human intervention" (p. 181). One fundamental problem with Carey's description of our assertion is that it is not our assertion. Our papers specifically and repeatedly emphasize that intrinsic mortality signatures only remain stable over time and location in the absence of medical intervention. Although laboratory animals receive excellent veterinary care, no heroic measures are taken to save their lives when they fall ill. Humans, on the other hand, seek medical care when they are sick, and the treatment of intrinsic disease processes (heart disease, cancer, and stroke) is commonplace. Our recognition that intrinsic mortality can be delayed led us to coin the term manufactured survival time (Olshansky & Carnes, 1997) to describe the survival time arising from medical interventions that save the lives of people who would otherwise have died at younger ages. Further, the idea that mortality is intractable, especially at older ages, is inconsistent with evolutionary theories of senescence. The absence of selection (evolutionary neglect) in the post-reproductive period of the life course means that there are loopholes in the biological contract of life that can be and are exploited to manufacture additional survival time (Carnes, Olshansky, & Grahn, 2003).
Another reason for Carey's rejection of mortality signatures was his experimental finding that "both the level and the trajectory of mortality can be substantially modified by changing an environmental variable such as diet" (p. 142). Two issues underlie this interpretive disagreement. As just noted, Carey mistakenly considers intrinsic mortality signatures as immutable. They are not. Just as medical interventions can have a favorable impact on intrinsic mortality signatures, experimentally induced insults like radiation have a detrimental impact on them (Carnes, Grahn, & Hoel, 2003). Carey works with total mortality because pathology examinations on millions of flies are impossible. As such, it is not possible to know whether the mortality modifications that Carey observes occur within the intrinsic or extrinsic component of mortality, and arguments could be made for either. For example, starvation-induced deaths are extrinsic events, but the caloric restriction of obese rats would be hypothesized to have a beneficial impact on mortality risks for intrinsic causes of death. In the absence of cause of death information, there is nothing about Carey's results that justifies the rejection of species-specific mortality trajectories.
Why debate whether intrinsic mortality signatures exist or not? One demographic model of human data predicts that life expectancies in excess of 100 years will be observed in the not distant future (Oeppen & Vaupel, 2002). Intrinsic mortality signatures establish lower limits to age-specific death rates, and lead to theoretical upper limits on life expectancy that are more in the range of 85–90 years. These estimated limits fall even lower when reasonable and unavoidable levels of extrinsic mortality are reintroduced to create a total mortality schedule. Although often unstated, a significant and contentious implication of a limit on the life expectancy of a population is that limits must also exist for the life spans of the individuals who make up the population.
Duration of Life
Considering or ignoring cause of death also has an important role in the debates over the duration of life for individuals. Is nature or nurture the primary determinant of longevity? This issue is usually couched in terms of heritability. Heritability is a concept originally developed by geneticists to help them monitor the progress of artificial selection experiments designed to enhance attributes of agricultural importance in plants and animals. They wanted to know how much of the variation in a quantitative trait of interest (e.g., milk or egg production, protein content, corn kernel count) was genetic, as opposed to environmental. Carey declares that the heritability of individual life span is small, and "contrary to popular myth, parental age of death appears to have minimal prognostic significance for offspring longevity" (p. 187). Of course if the heritability of longevity is small, then behavior and lifestyle must be relatively more important, and it is comforting to think that the duration of our lives is within our control, rather than dictated by our genes. Historically, estimates of the heritability of longevity have been all over the place. Current estimates fall within the 30–40% range, which is not small and may be an underestimate.
My logic for why the heritability of longevity may be underestimated is as follows (Carnes et al., 1999). The free radical hypothesis of aging remains a central paradigm in the field of aging. The energy produced by mitochondria made the evolution of eukaryotic cells and complex organisms possible. Unfortunately, the biochemical pathways in mitochondria that produce predictable amounts of energy, also generate predictable amounts of free radicals. Free radicals are nothing new; the very earliest life forms developed efficient mechanisms for dealing with the free radicals generated from ionizing events that occurred when radiation from the sun (in an environment without the benefit of an ozone layer) interacted with cellular water. The remarkable, but imperfect surveillance and repair mechanisms inherited from life's origins allow genetic damage in the somatic cells of our body to accumulate over time.
Today, entire journals are devoted to the putative diseases linked to free radical damage—everything from cancers to Type II diabetes. How significant a factor is free radical damage? The evidence is admittedly anecdotal, but I believe the mortality characteristics of inbred mice provide some insight to this question. A survivorship curve is nothing more than a curve that starts at 100% (or 1.0) when all the mice are alive, and then declines to zero over the course of about 1,200 days when the last mouse dies. If any creature can achieve a square survivorship curve (i.e., all mice die on the same day), it ought to be a population of genetically identical mice. It doesn't happen though; the survivorship curve (intrinsic mortality) for genetically identical mice looks just like the one for genetically heterogeneous hybrid mice (Carnes & Olshansky, 2001). The difference between the observed intrinsic survivorship curve for inbred mice and the hypothetical square curve may provide a measure of the contribution to mortality made by stochastic events. Further, the genetic consequences (possibly measurable) of the predictable production of free radicals should be considered as genetic as the genes responsible for the metabolic pathways that generate them. Shifting this variance component into the genetic compartment would probably increase the heritability of longevity up to 50%, which leads to the reasonable conclusion that genes and environment are coequal partners in determining the observed variation in life span.
Longevity: Lamarckian or Darwinian?
Although I agree with Carey that there is "not a single number that can be assigned to the life-span limit of species" (p. 233), I would not use mortality deceleration to justify this conclusion. While I believe that almost all scientists reject fixed limits to both life expectancy and duration of life, I do not think very many will reject probabilistic limits. Single numbers never make sense for species whose members are unique in virtually every way that uniqueness can be measured. Carey views maximal length of life, a single number metric, as "one of the most compelling concepts in demography and gerontology" (p. 186). I see it as: (a) a world record held by a single individual that will eventually be surpassed; (b) a single age that has virtually no measurable impact on measures of central tendency (e.g., life expectancy at birth); and (c) an unattainable age of little relevance to the vast, vast majority of individuals who will never achieve extreme longevity.
Carey sees the "post-Darwinian tails" as a way to escape the "putative limits imposed by the evolutionary theory of aging" (p. 217). In my view, a tail does not wag the dog; the dog wags the tail. Carey believes that longevity extension is a "self-reinforcing process," and that science and technology will lead to "the decommissioning of natural selection so that individuals can make choices conducive to engineering a long life for themselves and their children" (p. 219). He argues further that "we are now in an era of emerging ability to control our actuarial destiny in response to the desire in humans throughout history to live comfortably and to delay death" (p. 220).
What Carey, in my view, seems to ignore is that scientific and technological progress is Lamarckian—such developments are fundamentally based on directed change toward intended goals that can be extremely rapid. Biological evolution, on the other hand, is Darwinian—directionless change occurring in small increments at an incredibly slow pace, where the concept of progress has no meaning. As such, modernity confronts us with a uniquely human dilemma. We have Darwinian bodies, adapted for prehistoric environments, living in an artificial Lamarckian world where nearly everyone survives beyond the ages required for evolutionary success (reproduction before death), and the majority of people are achieving or surpassing the expiration dates of their biological warranty periods. This is our biological reality. Changing that reality in order to achieve significant "longevity extension" and a new "actuarial destiny" will require transcending the biological legacy for humans that emerged with the origin of our species. It may eventually happen, and if it does, I hope the life-extending technologies anticipated by Carey for the future will achieve what those of the past and present have been unable to achieve for most people—an extension of youthful health and vigor rather than a prolongation of the frailty and disability of old age.
References
| ||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
