Determinants of longevity : genetic, environmental
and medical factors
This review focuses on the determinants of longevity in the industrialized world, with emphasis on results from recently established data bases.
Strong evidenceis now available that demonstrates that in developed countries the maximum lifespan as well as the mean lifespan have increased substantially over the past
century.
There is no evidence of a genetically determined lifespan of around 85 years. On the contrary, the biggest absolute improvement in sur-
vival in recent decades has occurred amongst 80j year-olds.
Approximately one-quarter of the variation in lifespan in developed countries can ben attributed to genetic factors. The influence of both genetic and environmental factors on longevity can potentially be modified by medical treatment, behavioural changes and environmental improvements.
Introduction
The determinants of longevity might be expected to be well understood. The duration of life has captured
the attention of many people for thousands of years ; an enormous array of vital-statistics data are avail-
able for many centuries. Life-span is easily measured compared with other health phenomena, and in
many countries data are available on whole populations and not just study samples.
Knowledge concerning determinants of human longevity, however, is still sparse, and much of the little that is known has been learned in recent years. This review focuses on genetic, environmental and medical
factors as determinants of longevity in developed countries and discusses alternative paradigms con- cerning human longevity.
How should longevity be measured ?
Longevity can be studied in numerous ways ; key questions include the following. How long can a human live ? What is the average length of life ? Are the maximum and average lengths of life approaching limits ? Why do some individuals live longer than others ? In addressing these questions, it is useful to study the maximum lifespan actually achieved in various populations, the mean lifespan, and the variation in lifespan.
Estimating the maximum lifespan of human beings is simply a matter of finding a well-documented case
report of a person who lived longer than other well-documented cases. The assessment of mean lifespan
in an actual population requires that the study population is followed from birth to extinction.
An alternative approach is to calculate age-specific death rates at some point in time for a population, and then use these death rates to determine how long people would live on average in a hypothetical population in which these death rates prevailed over the course of the people’s lives.
This second kind of mean lifespan is generally known as life expectancy. The life expectancy of the Swedish population in 1996 is the average lifespan that would be achieved by the 1996 birth cohort if Swedish mortality rates at each age remained at 1996 levels for the entire future life
of this cohort.
Assessment of determinants of life expectancy and variation in lifespan amongst individuals rely on
demographic comparisons of different populations and on such traditional epidemiological designs as
follow-up studies of exposed or treated versus non- exposed or nontreated individuals. Designs from
genetic epidemiology – such as twin, adoption and other family studies – are useful in estimating the
relative importance of genes and environment for the variation in longevity.
Determinants of extreme longevity
Numerous extreme long-livers have been reported in various mountainous regions, including Georgia,
Kashmir, and Vilcabamba. In most Western countries, including the Scandinavian countries, excep-
tional lifespans have also been reported.
Examples are Drachenberg, a Danish–Norwegian sailor who died in 1772 and who claimed that he was born in 1626, and Jon Anderson, from Sweden, who claimed to be 147 years old when he died in 1729. There is
no convincing documentation for these extreme long- livers. When it has been possible to evaluate such
reports, they have proven to be very improbable
In countries, like Denmark and Sweden, with a long tradition of censuses and vital statistics, re-
markable and sudden declines in the number of extreme long-livers occur with the introduction of
more rigorous checking of information on age of death, as the result of laws requiring birth certificates,
the development of church registers and the establishment of statistical bureaus . This suggests
that early extreme long-livers were probably just cases of age exaggeration.
Today (March 1996), the oldest reported well-documented maximum lifespan for females is 121 years and for males 113 years . Both these persons are still alive. Analyses of reliable cases of
long-livers show that longevity records have been repeatedly broken over past decades ; this
suggests that even longer human lifespans may occur in the future.
There has been surprisingly little success in identi- nfying factors associated with extreme longevity. A
variety of centenarian studies have been conducted during the last half century. As reviewed by Seger-
berg , most of the earlier studies were based on highly selected samples of individuals, without
rigorous validation of the ages of reputed centenarians. During the last decade several more com- prehensive, less selected centenarian studies have
been carried out in Hungary , France , Finland and Denmark .
A few specific genetic factors have been found to be associated with extreme longevity. Takata et al.
found a significantly lower frequency of HLA-DRw9 amongst centenarians than in an adult control group in Japan, as well as a significantly higher frequency of HLA-DR1. The HLA-antigens amongst the Japanese centenarians are negatively associated with the presence of autoimmune diseases in the Japanese population, which suggests that the association with these genetic markers is mediated through a lower incidence of diseases.
More recently, both a French study and a Finnish study found a low prevalence of the e4 allele of apolipoprotein E amongst centenarians. The e4 allele has consistently been shown to be a risk factor both for coronary heart disease and for Alzheimer ’s dementia. In the French study , it was also found that centenarians had an increased prevalence of the DD- genotype of angiotensin-converting enzyme (ACE) compared with adult controls. This result is contrary to what was expected as the DD-genotype of ACE has been reported to be associated with myocardial infarction.
Only a few genetic association studies concerning extreme longevity have been published ;
Genes coding for :
Environment
Human leukocyte antigens HLA-DR (?)
Apolipoprotein E (?)
Angiotensin-converting enzyme (ACE) (?)
Year of birth
Smoking (?)
Alcohol (?)
Diet (?)
Dehydroepiandrosterone (DHEA) (?)
mortality.
Furthermore, these ‘ weak androgens ’have shown association with various objective and subjective health measurements in elderly persons.
.Clinical trials are now under way to test this medication and its influence on ageing. The discussed determinants of extreme lifespan are listed the associations described above need to be replicated
in future studies.
Amongst environmental factors, even such an obvious candidate as nonsmoking has failed to be
associated with extreme longevity. For example, the world ’s oldest man has smoked for most of his life
and still smokes cigars regularly [6]. This should not be taken as evidence of the harmlessness of smoking :
the influence of many risk factors attenuates when nstudied in older people.
For smoking, the underlying explanation is likely to be that individuals for whom
smoking was a severe health hazard (due to genetic make-up or other risk factors) have died well before the most advanced ages. The individuals surviving to extreme ages are particularly robust and either are
not very susceptible to the health hazards of smoking or have compensating characteristics .
Studies of alcohol consumption and diet through life, as well as physical activity, have also failed to show
consistent association with attainment of extreme age .
History provides numerous examples of medical treatments claimed to yield a long and healthy life.
However, none of these claims has been proven. In recent years, studies of humans have indicated that
blood level of dehydroepiandrosterone (DHEA) and its sulfate ester (DHEAS) are inversely correlated to
Determinants of mean lifespan
Amongst the lifespan measurements discussed here, mean lifespan is the one most important to social-
security and health-care systems in the western world. Demographers and epidemiologists agree that
life expectancy has been increasing during the last century in the industrialized world . The question has been whether this increase in life expectancy will continue. Until a few years ago, most researchers in demography, epidemiology and ger- ontology believed that life expectancy would not
increase much more .
This ‘ limited lifespan paradigm ’ goes back to Buffon,
in the 18th century [22], who hypothesized that a biological clock determines the maximum lifespan of an individual. The limited lifespan view has had many supporters over the years . The under-
lying axiom is that deaths at older ages are natural deaths due to intrinsic, intractable ageing processes,
whereas deaths at younger ages are generally extrinsic and potentially preventable.
The most prominent advocate of this hypothesis has been Fries . Fries claimed that there is a genetically determined upper limit to the lifespan of every individual. This limit varies from person to
person ; for the average person it is about 85 years.
Progress in living conditions and health would reduce death only at younger ages. The human survivorship
a From Caselli G, Mesle! F, Vallin J. Le triomphe de la medicine. Paris : Institut National d’E; tudes Demographiques, 1995 ; and Nordic.
Statistical Secretariat, eds. Yearbook of Nordic Statistics 1995. Copenhagen : Nordic Council of Ministers, 1995. curve would show increasing rectangularization with a dramatic drop in survival at around age 85.
A number of researchers have disputed this view and have argued that there is no biological or genetic
barrier to longer life expectancy [35–41]. They believe that mortality rates can be reduced at all ages
and that life expectancy could considerably exceed 85.
Until recently, it was impossible to test these alternative views because very few published life
tables provide age-specific data after age 85. Through a substantial effort during the last decade, death counts and population counts by individual year of age up to the highest ages and by individual year of time back at least 4 decades are now available in the Odense Archive of Population Data on Ageing for most
developed countries .
The comprehensive data bases in this archive provide very strong support
for the ‘ mortality reduction paradigm ’ : evidence
from 27 countries show a dramatic decline in death
rates at older ages . In fact, the older the age
category the greater the absolute reduction in death
rate . The number of centenarians has roughly
doubled every 10 years in developed countries over
the past half century, and most of this rapid rise can
be attributed to the marked increase in survival from
age 80 to age 100 .
Rates of mortality im-
provement at older ages in countries with the longest
life expectancy are not, on average, slower than
rates of improvement in countries with shorter life
expectancies [44]. Rates of improvement for women,
who generally enjoy substantially longer life ex-
pectancy than men, are more rapid than for men.
These findings suggest that life expectancy in de-
veloped countries is not currently coming close to an
upper limit.
Why is life expectancy increasing ?
Genetic factors
have been shown to play an important role in
determining lifespans in various animals. For
example, a single mutation in the nematode Caenor-
habditis elegans doubles lifespan [49]. In humans,
gross genetic abnormalities like trisomy have proven
the influence of genetic factors in shortening lon-
gevity and, as discussed above, some genetic factors
appear to be associated with exceptional longevity.
However, the dramatic change in life expectancy
within a few generations cannot be due to change in
the genetic composition of the population.
Life expectancy is clearly influenced by environ-
mental and public-health factors. Up until 1950 or
so, most of the gain in life expectancy was because of
reductions in infant, childhood, and early adult
mortality, as the result of rising standards of living,
public-health interventions and medical develop-
ments that reduced deaths from infectious diseases.
Since 1950, however, the increase in life expectancy
in developed countries is largely attributable to
reductions in mortality at older ages [50]. Thus, the
key question is : why are death rates declining, even
at and, indeed, especially at advanced ages ?
It is well known that socio-economic status,
education and occupation influence mortality pat-
terns throughout life .
However, the causal
pathways by which social status, wealth, education
and occupation influence mortality are not well
understood. Part of the problem is that it is difficult to
unravel the intricate web of causality linking socio-
economic factors and longevity. For instance, both
wealth and health at older ages are influenced by
both wealth and health at younger ages and, through
both genetic and environmental influences, by the
wealth and health of one ’s parents.
A person who is
healthy when young may get a better education and
a better first job. The health in youth may carry over
to health (and longevity) in old age, which may also
be influenced by education and occupation.
In today’s western world, cigarette smoking is the
most prominently known environmental factor that
influences lifespan. In a large study of English doctors
it was found that the age at which half of cigarette
smokers had died was 8 years less than the age for
nonsmokers, and for heavy cigarette smokers
10 years less .
The influence of cigarette smoking on longevity is probably mainly mediated through an
increased risk for cardiovascular and lung diseases, as well as for cancer. Declining cigarette smoking as
well as changes in diet and physical activity may be important factors in rising life expectancy in at least
some countries.
Recent research has confirmed the increased risk associated with excessive alcohol consumption, but has also showed reduced mortality amongst individuals with moderate daily wine consumption compared to teetotallers . Comprehensive population studies of height and weight have shown that
high and low body-mass-index (weight divided by
height squared) is associated with increased mortality.
A long-standing debate concerns the extent to
which intrauterine and perinatal exposures influence
late life morbidity and hence lifespan.
In particular, nutrition and infection early in life have been claimed
to be a strong determinant of late life mortality. In a
series of papers Barker and co-workers have reported
a follow-up of the individuals found in a large UK source of detailed birth and infant records from the
beginning of the century . These studies have indicated that low growth rates in utero and during infancy are associated with adverse health outcomes at adult ages.
In particular, a strong
relationship between reduced growth early in life
and high death rates for cardiovascular diseases has
been reported. Based on these observations, Barker
and co-workers have proposed ‘ the fetal-origins
hypothesis ’, which asserts that a baby’s nourishment
before birth and during infancy programme its
susceptibility to diseases and mortality later in life. A
number of studies, however, have failed to support
this hypothesis.
For example, Swedish and Danish studies have shown no increased mortality in adult-
hood amongst large samples of twins, although
twins experience severe growth retardation in utero
during the third trimester .
A significant proportion of the increase in life
expectancy might be attributable to medical progress.
However, it is hard to disentangle the effect of the
simultaneous improvement in living conditions and medical treatment. In the industrialized world, the classical example is tuberculosis incidence and mor- ntality, which has declined steadily through the first 8 decades of this century with no increased decline after the introduction of antibiotics after World War II .
Because many new treatments are being intro-
duced following evidence of increased survival in a
clinically controlled trial, it seems likely that medical
treatment has, at least in recent decades, reduced
death rates at older ages and that such improvements
will continue. An example of a promising devel-
opment is that large randomized trials have demon-
strated that fibrinolytic therapy can reduce mortality
in patients with suspected acute myocardial infarction (AMI) .
Improved treatment of frequent and potentially fatal diseases, such as AMI and
serious infections, can be expected to increase life
expectancy. Furthermore the emergence of geriatric
medicine, with emphasis on both prevention and
intervention initiatives amongst elderly persons, is
likely to reduce the number of premature deaths
amongst elderly persons .
# 1996 Blackwell Science Ltd Journal of Internal Medicine 240 : 333–341
Table 3 Determinants of mean lifespan in the industrialized
world
Genes
Environment
Medicine
Probably multiple
Year of birth
Socio-economic status
Education
Occupation
Smoking
Alcohol
Body-mass index
Diet (?)
Physical activity (?)
Intra-uterine conditions (?)
Probably multiple
In the USA, an activist approach towards medical treatment of elderly persons seems to be more prevalent than in Europe and Japan. This is a possible explanation of the recent finding that survival after age 80 in the USA is considerably greater than in Europe and Japan . If this explanation is verified,
then more intensive medical treatment of the elderlym may lead to substantial increases in mean lifespans.
The discussed determinants of extreme lifespan are
listed in Table 3.
Determinants of variation in lifespan
Whatever mean lifespans are, some individuals live
longer than others. Why ? The correlation in lifespan
between different kinds of relatives can shed light on
the importance of genetic and environmental factors
in variation of lifespan. Folk wisdom suggests that if
you want to live long you should choose two long-
lived parents. Family studies, however, have gen-
erally found only small correlations of between 0.01
and 0.15 in lifespan between parents and offspring
whilst the correlation between siblings has been
found to be considerably higher, in the range of
0.15–0.35 . Correlation within families
can be due either to genetic factors or to shared
environment. The finding of a higher correlation
between siblings compared to the parent–offspring
correlation could either be due to a higher degree of
shared environment amongst siblings (perhaps be-
cause they belong to the same generation) or to
nonadditive genetic factors.
The concept of nonadditive genetic factors can be
described by an example. Assume that a gene is
associated with longevity ; for simplicity, suppose
two different alleles, A and A , exist for this gene.
depends on the other allele, that is to say the effect of
same genotype as either parent. In such a situation,
nonadditive genetic factors are not transmitted from
one generation to another but they will generate
some similarity between siblings. More generally,
the interaction between genes is not necessarily
limited to one locus but can also be between different
gene loci.
Nonadditive genetic factors are of special interest
in studies of longevity. In 1930, Fisher described
what he called the fundamental theory of natural
selection for ‘ fitness traits ’ . Fisher pointed out
that reproduction is the key feature in evolution and
hence there has been a strong selection pressure on
fitness traits such as survival to reproduction. Such
selection pressure tends to minimize variance in
additive genetic factors.
If a specific additive genotype is very important to survival to reproduction, there
will be a tendency for only those who have this genotype to pass on their genes to the next generation. After many generations of selection, all
individuals in the population will have the genotype
(except for new mutations). Therefore, there will be
little or no variation in additive genetic factors for
fitness traits but there will still be variation because
of nonadditive genetic factors. Longevity might be
such a trait if it is highly correlated with the chance
of surviving to reproduction.
The correlation in lifespan described above – with
higher correlation between siblings compared with
parent offspring correlations – suggests the influence
of nonadditive genetic factors. To disentangle the
effect of shared genetic and environmental factors,
twin and adoption studies are required.
Both twin and adoption studies have demonstrated a strong genetic component of premature death. The only adoption study published shows a correlation between Danish adoptees and their biological parents, especially for death due to infection and vascular causes. In contrast, death from cancer
appears to be influenced by the family environment
. A Swedish twin study showed that death from
coronary heart diseases was influenced by genetic
factors both in women and in men, but that the
influence of genetic effects on cardiovascular mor-
tality decreased at older ages .
Most studies of lifespan in twins have been based
on populations in which many twins were still alive,
in which there was only a short period of observation,
and in which no information was available about the
zygosity (fraternal or identical) of the twins . A recent study by Herskind et al. followed more than 2800 twin-pairs with known zygosity, born between 1870 and
1900, from birth to death.
This study showed that
approximately one-quarter of the variation in lifespan
in this population could be attributed to genetic
factors, whilst the remaining three-quarters was
due to nonshared environmental factors. As expected
from the fundamental theory of natural selection,
the genetic factors were found to be nonadditive.
Twin and adoption studies do not shed light on
which environmental and genetic factors influence
the variation in lifespan. It seems plausible that the
candidates are the same as those described above as
determinants of mean lifespan (for example smoking,
nutrition, medical treatment, socio-economic condi-
tions and various candidate genes).
Probably, a large number of environmental and genetic factors interact
to determine lifespan. A recent analysis by our group suggests that only 1 % of the variation in lifespan can
be explained by the variation in the genetic factor apolipoprotein E, the most potent ‘ longevity gene ’
yet uncovered in humans.
It should be noted that the one-quarter or so of the
variation in lifespan due to genetic factors is not
necessarily beyond intervention. Insight into the
function of genes can immediately open up possi-
bilities for modifying adverse genetic effects if gene–
environment interaction plays a part in the aetiology.
A classic example of such an interaction is phenyl-
ketonuria, an autosomal recessive trait with elevated
blood and urine phenylalanine, caused by a defect in
a locus, on chromosome 12, coding for phenylalanine
hydroxylase. Normal development is obtained in
homozygous individuals with a diet low in phenyl-
alanine and supplemented with tyrosine. Mental
retardation appears if the individuals consume an
ordinary diet .
Another example is the Smith–Lemli–Opitz syn-
drome (SLOS), first described in 1964. SLOS is an
autosomal recessive disorder with variable expres-
sivity. Craniofacial anomalies are predominant, in
addition to limb and genital anomalies, failure to
A newborn homozygous for the autosomal recessive trait phenylketonuria
thrive and mental retardation. SLOS is thought to be
the second most common autosomal recessive dis-
order amongst white North Americans after cystic
fibrosis, with a carrier frequency of 1–2 %. Recent
research has shown that SLOS results in an error in
cholesterol synthesis. Insight into the aetiology of
this genetically determined disease has induced a
medical treatment that may reduce some of the
postnatal symptoms. The treatment is a diet high in
cholesterol .
Conclusion
Knowledge of the determinants of human longevity
is still sparse. On the population level, a number of
factors associated with mortality can be identified.
On the individual level, however, these associations
are too weak to be reliable predictors of a person ’s
lifespan. This can be illustrated by a case history. A
Danish woman, who was born in 1890 and who
grew up in a poor family, was sent away from home
when she was 17 because she had severe tuberculosis
that was a potential hazard to the rest of the family.
The infection was treated with an operation. At the
age of 48 she got breast cancer, first in one breast
and 6 years later in the other, and had both breasts
removed. In her 80s she had three minor strokes.
The woman herself told this story, which was verified,
in 1996 shortly after her 105th birthday.
The prospects for deeper understanding of the
determinants of longevity may, however, be good.
Rapid progress in genetics may add considerably to our understanding of survival. Because processes of
ageing in such species as yeast, worms, insects, and rats are similar in some ways to processes of ageing
for humans, advances in experimental gerontology may prove to be informative
Demographic and epidemiological studies of human populations may also play an important role
as more reliable and more extensive data are collected and analysed. Mortality is changing in different
regions and countries : life expectancy is declining in Russia and parts of Eastern Europe, and rapidly
increasing in Japan and some other developed countries. Mortality change in Denmark is following
a different pattern than in the other Nordic countries
. In many countries, the changes in mortality rates vary substantially at different ages.
The variety and speed of mortality change provides excellent opportunities for identifying underlying mechanisms and causal factors if appropriate data are gathered, especially data on the specific charac- teristics of individuals. The development, particularly in the Nordic countries, of large registers of health- related information about individuals – including such special populations as twins, adoptees, and centenarians, as well as the general population– may permit significant breakthroughs in knowl-
edge about the determinants of the duration of life.
and medical factors
This review focuses on the determinants of longevity in the industrialized world, with emphasis on results from recently established data bases.
Strong evidenceis now available that demonstrates that in developed countries the maximum lifespan as well as the mean lifespan have increased substantially over the past
century.
There is no evidence of a genetically determined lifespan of around 85 years. On the contrary, the biggest absolute improvement in sur-
vival in recent decades has occurred amongst 80j year-olds.
Approximately one-quarter of the variation in lifespan in developed countries can ben attributed to genetic factors. The influence of both genetic and environmental factors on longevity can potentially be modified by medical treatment, behavioural changes and environmental improvements.
Introduction
The determinants of longevity might be expected to be well understood. The duration of life has captured
the attention of many people for thousands of years ; an enormous array of vital-statistics data are avail-
able for many centuries. Life-span is easily measured compared with other health phenomena, and in
many countries data are available on whole populations and not just study samples.
Knowledge concerning determinants of human longevity, however, is still sparse, and much of the little that is known has been learned in recent years. This review focuses on genetic, environmental and medical
factors as determinants of longevity in developed countries and discusses alternative paradigms con- cerning human longevity.
How should longevity be measured ?
Longevity can be studied in numerous ways ; key questions include the following. How long can a human live ? What is the average length of life ? Are the maximum and average lengths of life approaching limits ? Why do some individuals live longer than others ? In addressing these questions, it is useful to study the maximum lifespan actually achieved in various populations, the mean lifespan, and the variation in lifespan.
Estimating the maximum lifespan of human beings is simply a matter of finding a well-documented case
report of a person who lived longer than other well-documented cases. The assessment of mean lifespan
in an actual population requires that the study population is followed from birth to extinction.
An alternative approach is to calculate age-specific death rates at some point in time for a population, and then use these death rates to determine how long people would live on average in a hypothetical population in which these death rates prevailed over the course of the people’s lives.
This second kind of mean lifespan is generally known as life expectancy. The life expectancy of the Swedish population in 1996 is the average lifespan that would be achieved by the 1996 birth cohort if Swedish mortality rates at each age remained at 1996 levels for the entire future life
of this cohort.
Assessment of determinants of life expectancy and variation in lifespan amongst individuals rely on
demographic comparisons of different populations and on such traditional epidemiological designs as
follow-up studies of exposed or treated versus non- exposed or nontreated individuals. Designs from
genetic epidemiology – such as twin, adoption and other family studies – are useful in estimating the
relative importance of genes and environment for the variation in longevity.
Determinants of extreme longevity
Numerous extreme long-livers have been reported in various mountainous regions, including Georgia,
Kashmir, and Vilcabamba. In most Western countries, including the Scandinavian countries, excep-
tional lifespans have also been reported.
Examples are Drachenberg, a Danish–Norwegian sailor who died in 1772 and who claimed that he was born in 1626, and Jon Anderson, from Sweden, who claimed to be 147 years old when he died in 1729. There is
no convincing documentation for these extreme long- livers. When it has been possible to evaluate such
reports, they have proven to be very improbable
In countries, like Denmark and Sweden, with a long tradition of censuses and vital statistics, re-
markable and sudden declines in the number of extreme long-livers occur with the introduction of
more rigorous checking of information on age of death, as the result of laws requiring birth certificates,
the development of church registers and the establishment of statistical bureaus . This suggests
that early extreme long-livers were probably just cases of age exaggeration.
Today (March 1996), the oldest reported well-documented maximum lifespan for females is 121 years and for males 113 years . Both these persons are still alive. Analyses of reliable cases of
long-livers show that longevity records have been repeatedly broken over past decades ; this
suggests that even longer human lifespans may occur in the future.
There has been surprisingly little success in identi- nfying factors associated with extreme longevity. A
variety of centenarian studies have been conducted during the last half century. As reviewed by Seger-
berg , most of the earlier studies were based on highly selected samples of individuals, without
rigorous validation of the ages of reputed centenarians. During the last decade several more com- prehensive, less selected centenarian studies have
been carried out in Hungary , France , Finland and Denmark .
A few specific genetic factors have been found to be associated with extreme longevity. Takata et al.
found a significantly lower frequency of HLA-DRw9 amongst centenarians than in an adult control group in Japan, as well as a significantly higher frequency of HLA-DR1. The HLA-antigens amongst the Japanese centenarians are negatively associated with the presence of autoimmune diseases in the Japanese population, which suggests that the association with these genetic markers is mediated through a lower incidence of diseases.
More recently, both a French study and a Finnish study found a low prevalence of the e4 allele of apolipoprotein E amongst centenarians. The e4 allele has consistently been shown to be a risk factor both for coronary heart disease and for Alzheimer ’s dementia. In the French study , it was also found that centenarians had an increased prevalence of the DD- genotype of angiotensin-converting enzyme (ACE) compared with adult controls. This result is contrary to what was expected as the DD-genotype of ACE has been reported to be associated with myocardial infarction.
Only a few genetic association studies concerning extreme longevity have been published ;
Genes coding for :
Environment
Human leukocyte antigens HLA-DR (?)
Apolipoprotein E (?)
Angiotensin-converting enzyme (ACE) (?)
Year of birth
Smoking (?)
Alcohol (?)
Diet (?)
Dehydroepiandrosterone (DHEA) (?)
mortality.
Furthermore, these ‘ weak androgens ’have shown association with various objective and subjective health measurements in elderly persons.
.Clinical trials are now under way to test this medication and its influence on ageing. The discussed determinants of extreme lifespan are listed the associations described above need to be replicated
in future studies.
Amongst environmental factors, even such an obvious candidate as nonsmoking has failed to be
associated with extreme longevity. For example, the world ’s oldest man has smoked for most of his life
and still smokes cigars regularly [6]. This should not be taken as evidence of the harmlessness of smoking :
the influence of many risk factors attenuates when nstudied in older people.
For smoking, the underlying explanation is likely to be that individuals for whom
smoking was a severe health hazard (due to genetic make-up or other risk factors) have died well before the most advanced ages. The individuals surviving to extreme ages are particularly robust and either are
not very susceptible to the health hazards of smoking or have compensating characteristics .
Studies of alcohol consumption and diet through life, as well as physical activity, have also failed to show
consistent association with attainment of extreme age .
History provides numerous examples of medical treatments claimed to yield a long and healthy life.
However, none of these claims has been proven. In recent years, studies of humans have indicated that
blood level of dehydroepiandrosterone (DHEA) and its sulfate ester (DHEAS) are inversely correlated to
Determinants of mean lifespan
Amongst the lifespan measurements discussed here, mean lifespan is the one most important to social-
security and health-care systems in the western world. Demographers and epidemiologists agree that
life expectancy has been increasing during the last century in the industrialized world . The question has been whether this increase in life expectancy will continue. Until a few years ago, most researchers in demography, epidemiology and ger- ontology believed that life expectancy would not
increase much more .
This ‘ limited lifespan paradigm ’ goes back to Buffon,
in the 18th century [22], who hypothesized that a biological clock determines the maximum lifespan of an individual. The limited lifespan view has had many supporters over the years . The under-
lying axiom is that deaths at older ages are natural deaths due to intrinsic, intractable ageing processes,
whereas deaths at younger ages are generally extrinsic and potentially preventable.
The most prominent advocate of this hypothesis has been Fries . Fries claimed that there is a genetically determined upper limit to the lifespan of every individual. This limit varies from person to
person ; for the average person it is about 85 years.
Progress in living conditions and health would reduce death only at younger ages. The human survivorship
a From Caselli G, Mesle! F, Vallin J. Le triomphe de la medicine. Paris : Institut National d’E; tudes Demographiques, 1995 ; and Nordic.
Statistical Secretariat, eds. Yearbook of Nordic Statistics 1995. Copenhagen : Nordic Council of Ministers, 1995. curve would show increasing rectangularization with a dramatic drop in survival at around age 85.
A number of researchers have disputed this view and have argued that there is no biological or genetic
barrier to longer life expectancy [35–41]. They believe that mortality rates can be reduced at all ages
and that life expectancy could considerably exceed 85.
Until recently, it was impossible to test these alternative views because very few published life
tables provide age-specific data after age 85. Through a substantial effort during the last decade, death counts and population counts by individual year of age up to the highest ages and by individual year of time back at least 4 decades are now available in the Odense Archive of Population Data on Ageing for most
developed countries .
The comprehensive data bases in this archive provide very strong support
for the ‘ mortality reduction paradigm ’ : evidence
from 27 countries show a dramatic decline in death
rates at older ages . In fact, the older the age
category the greater the absolute reduction in death
rate . The number of centenarians has roughly
doubled every 10 years in developed countries over
the past half century, and most of this rapid rise can
be attributed to the marked increase in survival from
age 80 to age 100 .
Rates of mortality im-
provement at older ages in countries with the longest
life expectancy are not, on average, slower than
rates of improvement in countries with shorter life
expectancies [44]. Rates of improvement for women,
who generally enjoy substantially longer life ex-
pectancy than men, are more rapid than for men.
These findings suggest that life expectancy in de-
veloped countries is not currently coming close to an
upper limit.
Why is life expectancy increasing ?
Genetic factors
have been shown to play an important role in
determining lifespans in various animals. For
example, a single mutation in the nematode Caenor-
habditis elegans doubles lifespan [49]. In humans,
gross genetic abnormalities like trisomy have proven
the influence of genetic factors in shortening lon-
gevity and, as discussed above, some genetic factors
appear to be associated with exceptional longevity.
However, the dramatic change in life expectancy
within a few generations cannot be due to change in
the genetic composition of the population.
Life expectancy is clearly influenced by environ-
mental and public-health factors. Up until 1950 or
so, most of the gain in life expectancy was because of
reductions in infant, childhood, and early adult
mortality, as the result of rising standards of living,
public-health interventions and medical develop-
ments that reduced deaths from infectious diseases.
Since 1950, however, the increase in life expectancy
in developed countries is largely attributable to
reductions in mortality at older ages [50]. Thus, the
key question is : why are death rates declining, even
at and, indeed, especially at advanced ages ?
It is well known that socio-economic status,
education and occupation influence mortality pat-
terns throughout life .
However, the causal
pathways by which social status, wealth, education
and occupation influence mortality are not well
understood. Part of the problem is that it is difficult to
unravel the intricate web of causality linking socio-
economic factors and longevity. For instance, both
wealth and health at older ages are influenced by
both wealth and health at younger ages and, through
both genetic and environmental influences, by the
wealth and health of one ’s parents.
A person who is
healthy when young may get a better education and
a better first job. The health in youth may carry over
to health (and longevity) in old age, which may also
be influenced by education and occupation.
In today’s western world, cigarette smoking is the
most prominently known environmental factor that
influences lifespan. In a large study of English doctors
it was found that the age at which half of cigarette
smokers had died was 8 years less than the age for
nonsmokers, and for heavy cigarette smokers
10 years less .
The influence of cigarette smoking on longevity is probably mainly mediated through an
increased risk for cardiovascular and lung diseases, as well as for cancer. Declining cigarette smoking as
well as changes in diet and physical activity may be important factors in rising life expectancy in at least
some countries.
Recent research has confirmed the increased risk associated with excessive alcohol consumption, but has also showed reduced mortality amongst individuals with moderate daily wine consumption compared to teetotallers . Comprehensive population studies of height and weight have shown that
high and low body-mass-index (weight divided by
height squared) is associated with increased mortality.
A long-standing debate concerns the extent to
which intrauterine and perinatal exposures influence
late life morbidity and hence lifespan.
In particular, nutrition and infection early in life have been claimed
to be a strong determinant of late life mortality. In a
series of papers Barker and co-workers have reported
a follow-up of the individuals found in a large UK source of detailed birth and infant records from the
beginning of the century . These studies have indicated that low growth rates in utero and during infancy are associated with adverse health outcomes at adult ages.
In particular, a strong
relationship between reduced growth early in life
and high death rates for cardiovascular diseases has
been reported. Based on these observations, Barker
and co-workers have proposed ‘ the fetal-origins
hypothesis ’, which asserts that a baby’s nourishment
before birth and during infancy programme its
susceptibility to diseases and mortality later in life. A
number of studies, however, have failed to support
this hypothesis.
For example, Swedish and Danish studies have shown no increased mortality in adult-
hood amongst large samples of twins, although
twins experience severe growth retardation in utero
during the third trimester .
A significant proportion of the increase in life
expectancy might be attributable to medical progress.
However, it is hard to disentangle the effect of the
simultaneous improvement in living conditions and medical treatment. In the industrialized world, the classical example is tuberculosis incidence and mor- ntality, which has declined steadily through the first 8 decades of this century with no increased decline after the introduction of antibiotics after World War II .
Because many new treatments are being intro-
duced following evidence of increased survival in a
clinically controlled trial, it seems likely that medical
treatment has, at least in recent decades, reduced
death rates at older ages and that such improvements
will continue. An example of a promising devel-
opment is that large randomized trials have demon-
strated that fibrinolytic therapy can reduce mortality
in patients with suspected acute myocardial infarction (AMI) .
Improved treatment of frequent and potentially fatal diseases, such as AMI and
serious infections, can be expected to increase life
expectancy. Furthermore the emergence of geriatric
medicine, with emphasis on both prevention and
intervention initiatives amongst elderly persons, is
likely to reduce the number of premature deaths
amongst elderly persons .
# 1996 Blackwell Science Ltd Journal of Internal Medicine 240 : 333–341
Table 3 Determinants of mean lifespan in the industrialized
world
Genes
Environment
Medicine
Probably multiple
Year of birth
Socio-economic status
Education
Occupation
Smoking
Alcohol
Body-mass index
Diet (?)
Physical activity (?)
Intra-uterine conditions (?)
Probably multiple
In the USA, an activist approach towards medical treatment of elderly persons seems to be more prevalent than in Europe and Japan. This is a possible explanation of the recent finding that survival after age 80 in the USA is considerably greater than in Europe and Japan . If this explanation is verified,
then more intensive medical treatment of the elderlym may lead to substantial increases in mean lifespans.
The discussed determinants of extreme lifespan are
listed in Table 3.
Determinants of variation in lifespan
Whatever mean lifespans are, some individuals live
longer than others. Why ? The correlation in lifespan
between different kinds of relatives can shed light on
the importance of genetic and environmental factors
in variation of lifespan. Folk wisdom suggests that if
you want to live long you should choose two long-
lived parents. Family studies, however, have gen-
erally found only small correlations of between 0.01
and 0.15 in lifespan between parents and offspring
whilst the correlation between siblings has been
found to be considerably higher, in the range of
0.15–0.35 . Correlation within families
can be due either to genetic factors or to shared
environment. The finding of a higher correlation
between siblings compared to the parent–offspring
correlation could either be due to a higher degree of
shared environment amongst siblings (perhaps be-
cause they belong to the same generation) or to
nonadditive genetic factors.
The concept of nonadditive genetic factors can be
described by an example. Assume that a gene is
associated with longevity ; for simplicity, suppose
two different alleles, A and A , exist for this gene.
depends on the other allele, that is to say the effect of
same genotype as either parent. In such a situation,
nonadditive genetic factors are not transmitted from
one generation to another but they will generate
some similarity between siblings. More generally,
the interaction between genes is not necessarily
limited to one locus but can also be between different
gene loci.
Nonadditive genetic factors are of special interest
in studies of longevity. In 1930, Fisher described
what he called the fundamental theory of natural
selection for ‘ fitness traits ’ . Fisher pointed out
that reproduction is the key feature in evolution and
hence there has been a strong selection pressure on
fitness traits such as survival to reproduction. Such
selection pressure tends to minimize variance in
additive genetic factors.
If a specific additive genotype is very important to survival to reproduction, there
will be a tendency for only those who have this genotype to pass on their genes to the next generation. After many generations of selection, all
individuals in the population will have the genotype
(except for new mutations). Therefore, there will be
little or no variation in additive genetic factors for
fitness traits but there will still be variation because
of nonadditive genetic factors. Longevity might be
such a trait if it is highly correlated with the chance
of surviving to reproduction.
The correlation in lifespan described above – with
higher correlation between siblings compared with
parent offspring correlations – suggests the influence
of nonadditive genetic factors. To disentangle the
effect of shared genetic and environmental factors,
twin and adoption studies are required.
Both twin and adoption studies have demonstrated a strong genetic component of premature death. The only adoption study published shows a correlation between Danish adoptees and their biological parents, especially for death due to infection and vascular causes. In contrast, death from cancer
appears to be influenced by the family environment
. A Swedish twin study showed that death from
coronary heart diseases was influenced by genetic
factors both in women and in men, but that the
influence of genetic effects on cardiovascular mor-
tality decreased at older ages .
Most studies of lifespan in twins have been based
on populations in which many twins were still alive,
in which there was only a short period of observation,
and in which no information was available about the
zygosity (fraternal or identical) of the twins . A recent study by Herskind et al. followed more than 2800 twin-pairs with known zygosity, born between 1870 and
1900, from birth to death.
This study showed that
approximately one-quarter of the variation in lifespan
in this population could be attributed to genetic
factors, whilst the remaining three-quarters was
due to nonshared environmental factors. As expected
from the fundamental theory of natural selection,
the genetic factors were found to be nonadditive.
Twin and adoption studies do not shed light on
which environmental and genetic factors influence
the variation in lifespan. It seems plausible that the
candidates are the same as those described above as
determinants of mean lifespan (for example smoking,
nutrition, medical treatment, socio-economic condi-
tions and various candidate genes).
Probably, a large number of environmental and genetic factors interact
to determine lifespan. A recent analysis by our group suggests that only 1 % of the variation in lifespan can
be explained by the variation in the genetic factor apolipoprotein E, the most potent ‘ longevity gene ’
yet uncovered in humans.
It should be noted that the one-quarter or so of the
variation in lifespan due to genetic factors is not
necessarily beyond intervention. Insight into the
function of genes can immediately open up possi-
bilities for modifying adverse genetic effects if gene–
environment interaction plays a part in the aetiology.
A classic example of such an interaction is phenyl-
ketonuria, an autosomal recessive trait with elevated
blood and urine phenylalanine, caused by a defect in
a locus, on chromosome 12, coding for phenylalanine
hydroxylase. Normal development is obtained in
homozygous individuals with a diet low in phenyl-
alanine and supplemented with tyrosine. Mental
retardation appears if the individuals consume an
ordinary diet .
Another example is the Smith–Lemli–Opitz syn-
drome (SLOS), first described in 1964. SLOS is an
autosomal recessive disorder with variable expres-
sivity. Craniofacial anomalies are predominant, in
addition to limb and genital anomalies, failure to
A newborn homozygous for the autosomal recessive trait phenylketonuria
thrive and mental retardation. SLOS is thought to be
the second most common autosomal recessive dis-
order amongst white North Americans after cystic
fibrosis, with a carrier frequency of 1–2 %. Recent
research has shown that SLOS results in an error in
cholesterol synthesis. Insight into the aetiology of
this genetically determined disease has induced a
medical treatment that may reduce some of the
postnatal symptoms. The treatment is a diet high in
cholesterol .
Conclusion
Knowledge of the determinants of human longevity
is still sparse. On the population level, a number of
factors associated with mortality can be identified.
On the individual level, however, these associations
are too weak to be reliable predictors of a person ’s
lifespan. This can be illustrated by a case history. A
Danish woman, who was born in 1890 and who
grew up in a poor family, was sent away from home
when she was 17 because she had severe tuberculosis
that was a potential hazard to the rest of the family.
The infection was treated with an operation. At the
age of 48 she got breast cancer, first in one breast
and 6 years later in the other, and had both breasts
removed. In her 80s she had three minor strokes.
The woman herself told this story, which was verified,
in 1996 shortly after her 105th birthday.
The prospects for deeper understanding of the
determinants of longevity may, however, be good.
Rapid progress in genetics may add considerably to our understanding of survival. Because processes of
ageing in such species as yeast, worms, insects, and rats are similar in some ways to processes of ageing
for humans, advances in experimental gerontology may prove to be informative
Demographic and epidemiological studies of human populations may also play an important role
as more reliable and more extensive data are collected and analysed. Mortality is changing in different
regions and countries : life expectancy is declining in Russia and parts of Eastern Europe, and rapidly
increasing in Japan and some other developed countries. Mortality change in Denmark is following
a different pattern than in the other Nordic countries
. In many countries, the changes in mortality rates vary substantially at different ages.
The variety and speed of mortality change provides excellent opportunities for identifying underlying mechanisms and causal factors if appropriate data are gathered, especially data on the specific charac- teristics of individuals. The development, particularly in the Nordic countries, of large registers of health- related information about individuals – including such special populations as twins, adoptees, and centenarians, as well as the general population– may permit significant breakthroughs in knowl-
edge about the determinants of the duration of life.
Comentários
Postar um comentário