Science journals: Anything for a headline

Well, this week's sensational result is reported in the Oct 5 Nature in a paper about limits to the human lifespan. The unsensational nature of this paper shows yet again how Nature and the other 'science' journals will take any paper that they can use for a cheap headline.  This paper claims that the human life span cannot exceed 115 (though the cover picture in a commentary in the same issue is a woman-- mentioned in the paper itself--who lived to be substantially older than that!).  The Nature issue has all the exciting details of this novel finding, which of course have been trumpeted by the story-hungry 'news' media.

In essence the authors argue that maximum longevity on a population basis has been increasing only very slowly or not at all over recent decades.  It is, one might say, approaching an asymptote of strong determination. They suggest that there is, as a result of many complex contributing factors-of-decline, essentially a limit to how long we can live, at least as a natural species without all sorts of genetic engineering.  In that sense, dreams of hugely extended life, even as a maximum (that is, if not for everyone), are just that: dreams.

This analysis raises several important issues, but largely ignores others.  First, however, it is important to note that virtually nothing in this paper, except some more recent data, is novel in any way.  The same issues were discussed at very great length long ago, as I know from my own experience.  I was involved in various aspects of the demography and genetics of aging, as far back as the 1970s.  There was a very active research community looking at issues such as species-specific 'maximum lifespan potential', with causal or correlated factors ranging from the effects of basic metabolism, or body or brain size.  Here's a figure from 1978 that I used in a 1989 paper




There was experimental research on this including life-extension studies (e.g., dietary restriction) as well as comparison of data over time, much as (for its time) the new paper.  The idea that there was an effective limit to human lifespan (and likewise for any species) was completely standard at that time, and how much this could be changed by modern technologies and health care etc. was debated. In 1975, for example (and that was over 40 years ago!), Richard Cutler argued in PNAS that various factors constrained maximum lifespan in a species-related way.  The idea, and one I also wrote a lot about in the long-ago past, is that longevity is related to surviving the plethora of biological decay processes, including mutation, and that would lead to a statistical asymptote in lifespan.  That is, that lifespan was largely a statistical result rather than a deterministically specified value.  The mortality results related to lifespan were not about 'lifespan' causation per se, but were just the array of diseases (diabetes, cancer, heart disease, etc.) that arose as a result of the various decays that led to risk increasing with duration of exposure, wear and tear, and so on, and hence were correlated with age.  Survival to a given age was the probability of not succumbing to any of these causes by that age.

This paper of mine (mentioned above) was about the nature of arguments for a causally rather that statistically determined lifespan limit.  If that were so, then all the known diseases, like heart disease, diabetes, cancer, and so on, were irrelevant to our supposed built-in lifespan limit!  That makes no evolutionary sense, since evolution would not be able to work on such a limit (nobody's still reproducing anywhere near that old).  It would make no other kind of sense, either.  What would determine such a limit and how could it have evolved?  On the other hand, if diseases--the real causes that end individual lives--were, together, responsible for the distribution of lifespan lengths, then a statistical rather than deterministic end is what's real.  The new paper doesn't deal with these, but by arguing that there is some sort of asymptotic limit, it implicitly invokes some sort of causal, evolutionarily determined value, and that seems implausible.

Indeed, evolutionary biologists have long argued that evolution would produce 'negative pleiotropy', in which genomes would confer greater survival at young ages, even if the result was at the expense of greater mortality later on.  That way, the species' members could live to reproduce (at least, if they survived developmentally-related infant mortality), and they were dispensable at older ages so that there was no evolutionary pressure to live longer.   But that would leave old-age longevity to statistical decay processes, not some built-in limit.

Of course, with very large data sets and mortality a multicausal statistical process, rare outliers would be seen, so that more data meant longer maximum survival 'potential' (assuming everyone in a species somehow had that potential, clearly a fiction given genetic diseases and the like that affect individuals differently).  There were many problems with these views, and many have since tried to find single-cause lifespan-determining factors (like telomere decay, in our chromosomes), an active area of research (more on that below).  We still hunger for the Fountain of Youth--the single cause or cure that will immortalize us!

The point here is that the new paper is at most a capable but modest update of what was already known long ago.  It doesn't really address the more substantive issues, like those I mention above.  It is not a major finding, and its claims are also in a sense naive, since future improvements in health and lifestyles that we don't have now but that applied to our whole population could extend life expectancy--the average age at death--and hence the maximum to which anyone would survive. After all, when we had huge infectious disease loads, hardly anybody lived to 115, and in the old days of research, to which the authors seem oblivious, something like 90-100 was assumed to be our deadline.

The new paper has been criticized by a few investigators, as seen in reports in the news media coverage.  But the paper's authors probably are right that nothing foreseeable will make a truly huge change in maximum survival, nor will many survive to such an extended age.  Nor--importantly--does this mean that those who do luck out are actually very lucky: the last few years or decades of decrepitude may not be worth it to most who last to the purported limit. To think of this as more than a statistical result is a mistake.  Not everyone can live to any particular age, obviously.

The main fault in the paper in my view is the claim in essence to portray the result as a new finding, and the publication in a purportedly major journal, with the typical media ballyhoo suggesting that.

On the other hand....
On the other hand, investigators who were interviewed about this study (to give it 'balance'!) denigrated it, saying that novel medical or other (genetic?) interventions could make major changes in human longevity.  This has of course happened in the past century or two.  More medical intervention, antibiotics and vaccines and so on have greatly increased average lifespan and, in so doing in large populations, increased the maximum survival that we observe.  This latter is a statistical result of the probabilistic nature of degenerative processes like accumulating wear and tear or mutations, as I mentioned earlier.  There is no automatic reason that major changes in life-extending technologies are in the offing, but of course it can't be denied as a possibility either. Similarly, if, say, antibiotic resistance becomes so widespread that infectious diseases are once again a major cause of death in rich countries, our 'maximum lifespan' will start to look younger.

Those who argue against this paper's assertions of a limit must be viewed just as critically as they judged the new paper.  The US National Institute on Aging, among other agencies, spends quite a lot of your money on aging, including decades (I know because I had some of it) on lifespan determination.  If someone quoted as dissing the new 'finding' is heavily engaged in the funding from NIA and elsewhere, one must ask whether s/he is defending a funding trough: if it's hopeless to think we'll make major longevity differences, why not close down their labs and instead spend the funding on something that's actually useful for society?

There are still many curious aspects of lifespan distributions, such as why rodents have small bodies that should be less vulnerable per-year to cancer or telomere degradation etc. that relate to the number of at-risk cells, yet only live a few years.  Why hasn't evolution led us to be in prime health for decades longer than we are?  There are potential answers to such questions, but mechanisms are not well understood, and the whole concept of a fixed lifespan (rather than a statistical one) is poorly constructed.

Still, everything suggests that, without major new interventions that probably will, at best, be for the rich only, there are rough limits to how long anyone can statistically avoid the range of independent risk our various organ systems face, not to  mention surviving in a sea of decrepitude.

One thing that does seem to be getting rather old, is the relentless hyperbole of the media including pop-culture journals like Nature and Science, selling non-stories as revolutionary new findings.  If we want to make life better for everyone, not just researchers and journals, we could spend our resources more equitably on quality of life, and our research resources on devastating diseases that strike early in the lives we already are fortunate to have.

Cancer moonshot and slow-learners

Motivated by Vice President Biden's son's death at an early age from cancer, President Obama recently announced a new health initiative which he's calling the cancer 'moonshot'.  This is like a second Nixonian 'war' on cancer but using a seemingly more benign metaphor (though cancer is so awful that treating it as a 'war' seems apt in that sense). Last week the NYTimes printed an op-ed piece that pointed out one of the major issues and illusions belied by the rhetoric of the new attack on cancer, as with the old:  Curing one cancer may extend a person's life, but it also increases his or her chances of a second cancer, since risks of cancer rise with age.

Cancers 'compete' with each other for our lives
The op-ed's main point is that the more earlier onset cancers we cure, the more late onset, less tractable tumors we'll see.  In that sense, cancers 'compete' with each other for our lives.  The first occurrence would get us unless the medical establishment stops it, thus opening the door for some subsequent Rogue Cell to generate a new tumor at some later time in the person's life.  It is entirely right and appropriate in every way to point this out, but the issues are subtle (though not at all secret).

First, the risk of some cancers slows with age.  Under normal environmental conditions, cancers increase in frequency with age because they are generally due to the accumulation of multiple mutations of various sorts, so that the more cell-years of exposure the more mutations that will arise.  At some point, one of our billions of cells acquires a set of mutational changes that lead it to stop obeying the rules of restraint in form and cell-division that are appropriate for the normal function of its particular tissue. A tumor is a combination of exposure to mutagens and mutations that occur simply by DNA replication errors--totally chance events--when cells divide.  As the tumor grows it acquires further mutations that lead it to spread or resist chemotherapy etc.

This is important but the reasons are subtle.  The attack on cells by lifestyle-related mutagens like radiation or chemicals in the environment becomes reduced in intensity as people age and simplify their lives, slowing down a lot of exposures to these risk factors. However, cell division rates, the times when mutations arise, themselves slow down, so the rate of accumulation of new mutations, whether they be by chance or by exposures, slows.  This decrease in the increase of risk with age at least tempers the caution that curing cancers in adults will leave them alive for many years and hence at risk for at least some many more cancers (though surely it will make them vulnerable to some!)

Apollo 11, first rocket to land humans on the moon; Wikipedia

Competing causes: more to the story, but nothing at all new
There's an important issue not mentioned in the article, but that is much more important in an indirect way.  This is an issue the authors of the op-ed didn't think about or for some reason didn't mention or perhaps because they are specialists they just weren't aware of.  But it's not at all secret, and indeed is something we ourselves studied for many years, and we've blogged about here before: anything that reduces early onset diseases increases the number of late onset diseases.  So, curing cancer early on (which is what the op-ed was about) increases risk for every later-onset disease, not just cancer.  In the same way as we've noted before, reducing heart disease or auto accident rates or snake bite deaths will increase dementia, heart disease, diabetes, and cancer--all other later-onset diseases--simply because more people will live to be at risk.  This is the Catch-22 of biomedical intervention.

In this sense all the marketing rhetoric about 'precision' genomic medicine is playing a game with the public, and the game is for money--research money among other things.  There's no cure for mortality or the reality of aging.  Whether due to genetic variants or lifestyle, we are at increasing risk for the panoply of diseases as we age, simply because exposure durations increase.  And every victory of medicine at earlier ages is a defeat for late-age experience.  Even were we to suppose that massive CRISPRization could cure every disease as it arose, and people's functions didn't diminish with age, the world would be so massively overpopulated as to make ghastly science fiction movies seem like Bugs Bunny cartoons.

But the conundrum is that because of the obvious and understandable fact that nobody wants major early onset diseases, it seems wholly reasonable to attack them with all the research and therapeutic vigor at our disposal. The earlier and more severe, the greater the gain in satisfactory life-years that will be made.  But the huge investment that NIH and their universities clients make in genomics and you-name-it related to late-age diseases is almost sure to backfire in these ways.  Cancer is but one example.

People should be aware of these things.  The statistical aspects of competing causes have long been part of demographic and public health theory.  Even early in the computer era many leading demographers were working on the quantitative implications of competing causes of death and disease, and similar points were very clear at the time.  The relevance to cancer, as outlined above, was also obvious.  I know this first-hand, because I was involved in this myself early in my career.  It was an important part of theorizing, superficial as well as thoughtful, about the nature of aging and species-specific lifespan, and much else.  The hard realities of competing causes have been part of the actuarial field since, well, more or less since the actuarial field began.  It is a sober lesson that apparently nobody wants to hear.  So it should not be written about as if it were a surprise, or a new discovery or realization.  Instead, the question--and it is in every way a fair question--should be why we cannot digest this lesson.  Is it because of our normal human frailty wishful thinking about death and disease, or because it is not convenient for the biomedical industries to recognize this sober reality front and center?

It's hard to accept mortality and that life is finite.  Some people want to live as long as possible, no matter the state of their health, and will reach for any life-raft at any age when we're ill.  But a growing number are signing Do Not Resuscitate documents, and the hospice movement, to aid those with terminal conditions who want to die in peace rather than wired to a hospital bed, continues to grow.  None of us wants a society like that in Anthony Trollope's 1881 dystopic novel The Fixed Period, where at age 67 everyone is given a nice comfortable exit--at least that was the policy until it hit too close to home for those who legislated it.  But we don't want uncomforable, slow deaths, either.

The problem of competing causes is a serious but subtle one, but health policy should reflect the realities of life, and of death.  I wouldn't bet on it, however, because there is nothing to suggest that humans as a collective electorate are ready or able to face up to the facts, when golden promises are being made by legislators, bureaucrats, pharmas, and so on.  But, science and scientists should be devoted to truth, even when truth isn't convenient to their interests or for the public to hear.

The Elephant (not) in the Cancer Ward

Recently, Tomasetti and Vogelstein (the latter a senior and highly regarded cancer geneticist) suggested that most cancer is due just to bad luck.  We discussed that study here.  When cells divide, DNA is copied, but that is a molecular process that isn't perfect (see discussion of Wednesday's Nobel Prize in Chemistry, e.g., for the discovery of DNA repair mechanisms and their association with cancer).  There are mutation detection mechanisms of various sorts (the BRCA1 gene whose mutations are associated with breast and some other cancers, is one with that sort of function).  The more at-risk cell divisions, the more mutations, and the higher the likelihood that one cell will experience a combination of mutations that (along with inherited variation) transforms the cell into the founder of a cancer.  T and V's assertion based on statistical analysis of numbers of cells at risk, their division rate for given tissues, and age of onset patterns, was that random mutation was a major contributor to cancer, rather than inherited genotype or environmental exposures, which they argue would account for this substantial fraction of cases.

Naturally, those whose grant fortunes depending on the idea that cancer is 'genetic' and/or 'environmental' roared in opposition to an idea that could threaten their perspective (and empires). Some of the T and V paper's statistical methods were questioned, and perhaps their paper was over-stated or less definitive than claimed.  Nobody can doubt that genetic variation and environmental exposures that could cause cells to be more likely to experience mutations, play a role in cancer.  But in any practical sense, it is hard to deny that luck plays a role (even with environmental exposures, because if they cause mutations, they basically strew them randomly across the genome, rather than causing them in any particular gene, etc.).

But we mentioned an important issue then that had been raised 40 years ago by epidemiologist Richard Peto.  Essentially it is that other mammals, like mice, experience a similar array of cancer types, with similarly increasing risk with age....but that increase is roughly calibrated with their life span. In fact, mice have far fewer stem cells in, say, their intestine or blood than humans, but their risk of cancer in these tissues increases far more rapidly (in years) than does human risk, though we have orders of magnitude more at risk cells and cell divisions.  This became known as Peto's Paradox.  It has not really been answered though there are some attempts to determine how it is that different species, of different sizes, calibrate their cancer risk in relation to their observed typical lifespan.

"Elephas maximus (Bandipur)" by Yathin S Krishnappa - Own work. Licensed under CC BY-SA 3.0 via Commons - 

For example a 2014 paper in Nature Reviews Genetics by Gorbunova et al. documents the very different typical lifespans of rodent species, and suggests some plausible genetic mechanisms that may protect the longer-lived species from cancer.  There must be some such mechanism, or else we misunderstand something very important in the carcinogenesis process.

Now a new commentary has been discussed in the NY Times of a JAMA paper, that makes similar genetic arguments for the very out-of-line cancer-free longevity of elephants.  Based on their numbers of at-risk cells, elephants should drop over with cancer at a very young age, but instead they typically live for a very long time.  How can this be?

The JAMA authors, Abegglen et al., found that a gene, called TP53, that is clearly related (when mutated) to cancer susceptibility in humans and in experimental assays, at least in part because it detects and effectively kills misbehaving mutated cells.  The study included humans with Li Fraumeni syndrome (LFS), a genetic disorder that greatly increases the risk of developing cancer, susceptibility to which has long been known to be associated with variants in TP53, and blood samples from Asian and African elephants.  

The study needs close scrutiny for methodological issues, but the authors make what they feel, reasonably, is a relevant finding.  There is only one copy of the TP53 gene in humans, but in elephants there are 20.  In blood cell assays this gene's activity was higher than in humans.  The inference is that elephants' longevity relative to cancer is due to this gene. If that is indeed the (or at least, an) explanation for the elephants' cancer-related longevity, it raises some other important questions, which should at least raise eyebrows and the need for ever-present skepticism.

Questions raised by the results
As in the rodent paper cited above, single-gene mechanisms for complex traits are appealing and publication-worthy, but in a sense such claims raise questions about themselves.  Elephants live long lives relative to other diseases that essentially have little if anything to do with cancer.  One can think of heart disease, dementia, stroke, kidney failure, liver disease, neuromuscular and joint disease, and waning immune systems.  Are these traits all due to having more TP53?  That seems unlikely.  

Alternatively, apparently whales are known not to have multiple TP53 duplicates, and I don't know about other very large animals like rhinos, giraffes, and so on.  A standard argument would be that in ecological circumstances when natural selection favors longer lives for some species, it uses whatever mechanism happens to be available--that is, selection has no foresight and can't just choose genes to duplicate.  Each species will have experienced the longevity advantage in its own local time, place, and ecosystem.  Just as the genes whose mutation yields resistance to malaria in humans vary from continent to continent, so will longevity-related genes favored by selection

So, Peto's Paradox remains curious.  If each species has its own protective mechanism (and perhaps several for its different organ and physiological systems), then we can account in a reasonable way for longevity patterns.  There is no need to find, or even to expect the same thing in all species' evolution: variation in response to selection can vary by organ system, species, and location even among species.  This is exactly the sort of thing that we should expect when we think of the complexity of genomic mechanisms--and what has consistently been found by genome mapping studies (GWAS) of late onset traits (and, for that matter, even early onset ones).

In turn, that means that each paper that claims subtly or overtly to have found 'the' or even a widespread important mechanism related to aging needs to be taken circumspectly.  Aging and lifespans are complex phenomena.  We will learn from each example we document, as with GWAS results, that a simple anti-aging strategy can't be inferred.  It's not likely to be a single magic bullet.

On Being Mortal

Ken and I have written a lot about disease causation, prediction and prevention but we haven't written much about the other side, when prediction, prevention and treatment aren't enough, when disease becomes fatal.  We have just read Dr Atul Gawande's book Being Mortal, a beautifully written heartfelt exploration of the end of life.  Dr Gawande is a surgeon at a major teaching hospital, and a professor of health policy, and his job is to save lives, and to teach medical students how to do the same.  He is presumably very good at this.  In the book, though, he writes about the process of learning how to be a doctor when there is no cure and he can't save a patient's life, something he didn't learn in school, and that has taken him decades to learn.  Presumably these are lessons he now teaches to his students, to the great benefit of us all.

Dr Gawande tells his story through many case histories, including that of his father, as he made decisions about how to live, and die, with an untreatable cancer.  He told some of these same stories in the BBC Reith Lectures last fall.  He writes about the tremendous regret he now has about instances in which he just was not able to have the kind of conversation with a dying patient that he now knows he should have had.

We are used to two common medical models, he says, the 'paternalistic' model of the 1960's, when a patient could be treated with a blue pill or a red pill, and the doctor made the choice. "Take the red pill.  It will do you good."  Then, the 'information' model took over -- the doctor supplied information, telling the patient that his/her disease could be treated with a blue pill or a red pill, explaining the pros and cons of each and then asking the patient to choose.  But, we don't face the end of our lives statistically, and weighing the pros and cons of different treatments is not what helps us make decisions about how to proceed, which is what Gawande finally realized after too many painful conversations with his very sick patients.

When terminally ill, a patient is overwhelmed with fears and concerns, and recognizing and acknowledging these is the truly important role a doctor can play when a patient is facing life-threatening illness. After much thought, hundreds of conversations with gerontologists, palliative care physicians, managers of the best assisted living facilities, and with patients, Gawande has come to see that there's another model, the 'interpretive' or 'shared decision making' model.

Gawande now asks his patients, What are your priorities if your time is limited?  What are your goals for treatment?  What are your fears?  And what trade-offs are you willing to accept as a result of your care?  And, he, the patient, and the patient's family choose the course of treatment with the patient's answers in mind.

One patient said that as long as he could watch football and eat chocolate ice-cream he wanted to keep living, so treatment continued for this man longer than it did for, for example, Gawande's father who said that he wanted not to suffer, did not want to be paralyzed, and if he couldn't enjoy seeing friends and family, he wanted no more treatment.  So, he refused further chemotherapy when the trade-offs were no longer acceptable to him.

In modern medicine, it's important to recognize that 'no cure' is not the same as 'no treatment'.  There is almost always something else that can be tried, some heroic measure, some experimental surgery or medicine that can be used to give a patient hope, or even a little more time, even when the illness can't be cured.  Doctors are very good at plugging ahead with all of this, without stopping to ask their patients the kinds of things that Gawande now asks.  The proper goal of the medical system, Gawande now believes, is not to stave off death as long as possible, or even to make a good death, but instead to assist in assuring "a good life to the very end."

According to Dr Gawande, modern medicine is very good at a lot of things, but preventing and treating aging and death are not among them.  Until the 1950's, people in the developed world most often died at home.  Then, increasingly, as it became more and more possible for medicine to intervene in the process, people began to die in hospital -- indeed, at ever increasing expense.  Now, however, people are beginning to choose to die at home again, and the hospice movement is largely responsible for making this work as well as it can.

The primary role of nursing homes (an industry which, according to Gawande, began to grow when the number of hospital beds for the elderly wasn't sufficient once aging and dying were medicalized) is to keep the elderly safe, but at the cost of lost privacy, dignity and control over one's own life. Nursing homes are run for the convenience of the system, not the residents. Fortunately, there are increasingly alternatives that allow people to 'age in place,' in their own homes, or if that's not possible, in an assisted living alternative, with as much or as little aid as they want or need.  

If Gawande's book is an indicator that we are wresting aging and dying back from a system that appropriated it, at great cost in money and suffering, it is reminiscent of the movement to demedicalize pregnancy and childbirth, with the increasing popularity of birthing centers and home births, or of menopause, which once meant hormone replacement therapy for all but no longer does.  There are many things modern medicine does very well, of course.  But there are things it can't and will never do well, including preventing aging and death.  

Still, many people do opt for heroic measures at the end of life.  This is in a sense because of the hope that they can be cured, and perhaps a deeper yearning for immortality.  Is this because medicine has over-promised?  Surely in part.  As Gawande says, patients are usually thinking in terms of 10 or 15 additional years when they hear that yet another treatment can be tried, not weeks or months, but it's more like weeks or months that these heroic measures have to offer.  

But this over-promising is nothing new.  Genetics has been doing it for decades, and the new commitment to precision medicine, genetics and so much more, is more of the same.  Some of this is because of snake oil salesmen, certainly, but not entirely.  Just as we have to blame Trump's popularity not just on Trump, but on the people buying his 'message' as well, it is the age of genetics because the people have bought the message being sold.  This isn't so different from the promise of miraculous cures by some religions (or mountebanks).

Surely there will come a time when we recognize that all that has been promised just can't be done, we won't be able to foretell our medical, academic, economic, or romantic futures from our genomes at birth,  and we'll understand that geneticizing our lives is as much over-promising as is the idea that one more experimental chemotherapy is going to finally cure our incurable disease.    

We put our faith in medicine when we are most vulnerable, hoping against hope that it will save us.  Perhaps it was the miracle drugs of the mid 19th century that encouraged this faith -- antibiotics really did save lives.  And then technology -- kidney dialysis and heart transplants, hip replacements and triple bypass surgeries.  We're very good at technology.  But, we still don't really understand cancer, or mental illnesses, or the cause of so many diseases.  And we won't be able to predict complex disease from our genes (which we've written about many times before on the MT), and we certainly can't prevent aging or death.  Despite the promises.  

Atul Gawande's message is sane and oddly reassuring, but as such it's a radical one as he aims to return control of a patient's present and future back to the patient.  This is a challenge to vested interests, yes, as well as a challenge to the usual way medicine is done in the industrialized world.  But it's a welcome and important one, because it's something we all will face.