Telomere Length and Mortality — Danish Study of 65,000 People

Last week, a Danish study was published that tracked 65,000 people over 15 years.  The bottom line was that telomere length robustly predicts longevity, even after factoring out the effect of age, smoking, exercise, blood cholesterol, BMI, and alcohol consumption.  This adds immensely to our knowledge of telomere length and its predictive power.  For perspective, the original [2003] study by Cawthon detected the relationship between telomere length and mortality based on fewer than 200 subjects.

The new data set is large enough to show trends over all of the health-related lifestyle variables.  Smoking, inactivity, weight (body mass index), and alcohol consumption all correlated negatively with telomere length.  So it should not be surprising that blood pressure and LDL choloesterol also correlated negatively with telomere length, and it is then a foregone conclusion that mortality must correlate negatively with telomere length.  This demonstrates without a doubt that unhealthy behaviors lead to shorter telomeres, as Epel and Blackburn have been telling us for a decade.  (They have emphasized the converse: that healthy life choices lead to longer, healthier life through the medium of longer telomeres [Ref, Ref, Ref, Ref]).

The bottom line of the new, large study is the extra predictive power of telomere length, even after all these other lifestyle and indicator variables are factored out.  Correcting for smoking, correcting for age, correcting for weight and cholesterol and exercise habits, there is still a powerful negative correlation between telomere length and mortality.  The shorter your telomeres, the greater your chance of dying.  The 10% of people with the shortest telomeres were dying at 1.4 the rate of the 10% with the longest telomeres, a result that was overwhelmingly statistically apparent (p<2×10-15).

Are Short Telomeres a Cause of Aging or Just a Marker of Aging?

There are many traits associated with aging that are mere markers.  For example, grey hair is associated with aging, but you don’t expect that coloring your hair will lead to longer life span.  Even if you found a treatment that restored your hair color by rejuvenating the pigment in your follicles, you wouldn’t expect to live longer as a result.

On the other hand, inflammation increases with age, and we know that it is not just a marker but a cause.  Reducing inflammation leads to longer life.

So are short telomeres like hair color or like inflammation?  Can we reasonably expect that lengthening telomeres will lengthen life?

Many lab scientists (and some gerontologists) think that it can’t be so easy to combat aging.  Theory says that if telomerase could increase life span, then evolution would have granted us more telomerase.  After all, the hTERT gene is already there in every cell, the metabolic cost of producing it is inconsequential.  Telomerase is free, and it can be released with the turn of a metabolic switch.

The theory continues: we know that diseases, lacerations, stresses all require more cell replication to repair them.  This must leave telomeres shorter than they would be otherwise. Smoking and inflammation are also known to shorten telomeres.  So (by this reasoning) people with shorter telomeres are expected to have shorter life expectancy because the shorter telomeres are telling a story that the person has suffered more stress.  Short telomeres are only a symptom of aging, and not a cause.

This new Danish study puts this theory to rest.  At last there is enough data that corrections can be made for smoking, obesity, exercise, and all major life style variables that could conceivably be have an impact on mortality comparable to the large effect we find associated with telomere length.

The correction is done using a statistical method called ANOVA, which can partition the causes of mortality into statistical bins and say how much is due to smoking, how much is due to blood cholesterol, how much is due to age, and how much is due to telomere shortening.

Results from the study:

  • Impact of telomere length on mortality, raw data:   3.38 (meaning that the 10% of people with the shortest telomeres were dying at a rate 3.38 as high as the 10% with the longest telomeres)
  • Same calculation, corrected for age:  1.54
  • Same calculation, corrected for age and all other hazard variables:  1.40

Conclusion: This demonstrates that age is the biggest factor in mortality, and telomere length has a strong effect, independent of age.  All the health variables together are a small factor compared to age and telomere length.

Short telomeres are not just a marker but a major cause of mortality.

Evolution has turned telomerase off such that short telomeres substantially affect our life span.  Turning telomerase on would not have cost anything, but that is not what evolution has done.

So the theory is wrong that says evolution has already made our life spans as long as possible.  Evolution has arranged for us to age and die “on purpose”.  Withholding telomerase is part of an evolved death program.


What does this say about the Cancer Hypothesis?

Suppose you believed that telomere length has been optimized by natural selection for a compromise between cancer prevention (short telomeres) and adequate capacity for tissue renewal (long telomeres).   Then you would predict that, since the length is at an optimal level, there is a smooth, flat top in the mortality curve.  People with slightly longer telomeres will have greater death rates from cancer, but lower from other causes; and people with shorter telomere length will have slightly greater death rates from other causes, but lower from cancer; and the sum (all-cause mortality) should be comparable for the two groups.

That would be the prediction.  But what the Danish group found instead (consistent with other studies in the last 10 years, but now unassailable because of the large sample) is that all-cause mortality decreases with telomere length.

We can only conclude that telomere length is not optimized for maximal life span.



The prediction is vindicated to the extent that telomere length is not as strongly associated with cancer mortality as it is with cardiovascular mortality.  (Cawthon, too, found this in his tiny data set.)  This shows that the theory is correct to the extent that short telomeres seem to offer some “protection” against cancer.  But this “protection” is relative only to mortality from other sources.  The net result of short telomeres is to increase cancer risk—just not as much of an increase as for heart disease. It is a safe bet that the reason for this increase is that short telomeres lead to more senescent cells (more inflammation) and less effective T-cells (less effective monitoring against early stages of cancer).  Hence, the net result is that longer telomeres offer cancer protection that more than compensate the increased risk.

The one finding in the article that could most easily be mistaken for vindication of the cancer hypothesis is that there are three genetic markers of telomere length that were also tracked in these 65,000 subjects, and these markers are also correlated with higher cancer risk.  The three genetic markers correlate with higher cancer risk and also with longer telomeres.

The way in which this result is reported is biased, a bit misleadingly, toward the standard cancer hypothesis.  The authors write,

“We found that genetically short telomeres were associated with low cancer mortality but not low cardiovascular mortality, death from other causes, or all-cause mortality.  This implies that genetically long telomeres are associated with higher cancer mortality.”

The misleading thing is the use of the words “genetically short telomeres”.  You might read this and think it was referring to telomere length that the subjects were born with.  But in fact, this was not measured.  Without being able to go back in time, we have no way to know what was the subjects’ telomere length at birth.  The fine print in the article tells us what they mean by “genetically long telomeres” is the variant of these three genetic polymorphisms that is statistically associated with longer telomeres late in life.

(A bit of background: a SNP is a “single-nucleotide polymorphism”.  This refers to the smallest possible genetic difference.  Our DNA is made of units labeled A, C, T and G, and the vast majority of your DNA is absolutely identical to mine and every other human beings’ DNA.  What makes us unique is these small differences, and the lowest-level, smallest differences are these SNPs.  A SNP is a place in the DNA where some people have an A while others have a C, for example, amidst a sea of letters before and after that are identical.

There are three SNPs in the region around the telomerase gene that presumably help to determine when and how much telomerase is transcribed.  Some people make more and others less, based on these tiny differences.  What the new study shows is that the version of these three SNPS associated with longer telomeres is also associated with higher cancer rates.)

This result is expected, and could hardly be otherwise.  If the standard selfish-gene version of evolutionary theory works anywhere at all, it must work for the tinest variations.  Indeed, that is where the theory logically must apply, and that is the only place it has been tested.

I have written extensively (in this blog here and here) that this same standard population genetic theory, the selfish gene, does not explain the big picture in evolution.  By the big picture I mean sex, aging, cooperation, speciation, evolvability, the structure of the genome.  But I would have to be radical indeed to deny that standard selfish gene theory can explain the small picture.  For the record: I think that selfish gene theory offers a good account of SNPs.

Withholding telomerase, allowing cells and whole animals to senesce, is a “big picture” adaptation, built deeply into the structure of the genome.  One small part of this picture is ruled by just three SNPs, and this small part appears to be guided by a tradeoff between cancer and other forms of mortality.

What else can be concluded from this huge new data sample?

This is far and away the world’s largest record of individual telomere lengths.  Here is their scatter plot of telomere length vs age:

Telomere length vs age, new data

This is amazingly detailed compared to what had previously been available.  Here’s an example of what we had been working with before the Danish study:

Old data, telomere length vs age

The first thing we notice is how many more data points there are in the new study.  The second thing is that the scales are so different.  In the older samle, between ages 20 and 80, telomere length decreases from 7,500 to 6,000 base.  In the newer sample, telomere length decreases from 5,000 to 4,000 over a similar age range.  This must reflect a difference in methodology for measuring average telomere length.  Both studies characterize their target variable as the average telomere length for chromosomes in white blood cells from a blood sample.  (Curiously, red blood cells have no cell nucleus, no chromosomes, no DNA.  They are meant to do one thing only, to live a short time, and then be replaced.).

Both the new and old plots show a substantial scatter around the trend line.  Standard deviation looks to be almost 1,000 bp, and there are especially many outliers thousands of bp to the north of the trendline.  Although the downward trend is obvious with so many data points, it is also clear that it is not strong enough to account for the powerful mortality trend with age.  For example, eyeballing from the plot, I would guess that ¼ of 60-year-olds have the telomere length of the median 90-year-old.  But we know that the mortality risk of a 60-year-old is 17 times lower than for a 90-year-old (if Americans and Danes are similar), and if ¼ of all 60-year-olds fared as poorly as the average 90-year-old, then the difference between 60 and 90 could not be any larger than a factor of 4.  This kind of reasoning reminds us that age is a larger factor in mortality risk, and independent of telomere length.

This suggests a strategy for seeking in the new data an answer to an important question for life extension science:  How great are the potential benefits of telomerase activation?  In the next few years, we expect telomere-lengthening treatments to be available, so that telomere length is no longer a factor in mortality.  How many years can we expect this to add?

We might look for an answer in the factor 1.54 quoted above.  People with the shortest telomeres have 1.54 times the mortality risk of people with the longest telomeres.  Referring to the same life table I cited above, that factor of 1.54 represents five years of aging.  From middle age onward, mortality is rising exponentially with a slope of about 0.038 per year.

Log of the probability of death as a function of age

Only 5 years?  From one perspective, 5 years is a huge benefit–about what we might get if there were a universal cure for cancer.  But 5 years is a disappointing prospect for people (I am one) who have said that telomerase therapies are the most promising near-term technology for life extension.  The calculation may be misleading for a number of reasons.  For example, it may be that the average leukocyte telomere length is the most convenient thing to measure, but not the most salient.  It may be that the shortest telomeres are more important than the average, and telomeres in stem cells are more important than telomeres in leukocytes.  It may be that telomerase itself has benefits above and beyond the lengthening of telomeres [ref, ref].  Or it may be, as Mike Fossel has emphasized, that absolute telomere length is not as important as relative telomere length, separately calibrated not just for each species but each individual.  For readers who know more than I about the biology of telomeres, I invite your comments.


This article originally appeared in Josh’s blog Aging Matters here. Republished with permission.