Jump to content
Dean Pomerleau

Optimal Late-Life BMI for Longevity

Recommended Posts

Mike Lustgarten has penned an interesting blog post in which he looks at data from several sources, including these two meta-analyses [1][2]. Study [1] found the optimal BMI for adults in general (median age 58), was pretty flat and optimal between BMI of 19-25. Here is the graph:




But [2] found in older adults (65+) the optimal BMI was much higher:





As we've discussed here, this late-life "obesity paradox" might be a result of latent disease making people thin and more likely to die. Or it could simply be that heavier people have more metabolic reserves, which is important to enable the elderly to weather the "slings and arrows" of aging / decrepitude (e.g. falls & fractures, hospitalization, sarcopenia, loss of appetite, etc.)


But the most interesting graphic from Mike's post is this one, in which Mike looked through a bunch of references (see his blog post for the list of references) and apparently did his own meta-analysis of the average BMI of centenarians (thanks Mike!):




As you can see, most centenarians have a BMI between 19 and 24. He concludes:


Centenarians have a BMI between 19.3-24.4 kg/m2. Shouldn’t that be the BMI reference range for those interested in living past 100?


On the CR Society Facebook Group discussion of Mike's blog post, I question his rationale for this statement, saying:


To play devil's advocate, it seems like the only way to answer [the question of the optimal BMI for living past 100] is to see if [the centenarians] have maintained that BMI from a much younger age, or have only gotten that thin as a results of sarcopenia and other unintended weight loss. Or maybe they've gained weight relative to their younger selves. There just isn't enough information to know what is optimal based on late-life BMI in the extremely old. 


I further suggest something we've discussed before (in the thread mentioned above):


The optimal strategy may be to remain thin until one's elderly years to gain the benefits of CR, then put on weight to serve as a metabolic reserves when adverse events are likely to require them in old age.





[1] Berrington de Gonzalez A, Hartge P, Cerhan JR, Flint AJ, Hannan L, MacInnis RJ, Moore SC, Tobias GS, Anton-Culver H, Freeman LB, Beeson WL, Clipp SL, English DR, Folsom AR, Freedman DM, Giles G, Hakansson N, Henderson KD, Hoffman-Bolton J, Hoppin JA, Koenig KL, Lee IM, Linet MS, Park Y, Pocobelli G, Schatzkin A, Sesso HD, Weiderpass E, Willcox BJ, Wolk A, Zeleniuch-Jacquotte A, Willett WC, Thun MJ. Body-mass index and mortality among 1.46 million white adults. N Engl J Med. 2010 Dec 2;363(23):2211-9. doi: 10.1056/NEJMoa1000367. Erratum in: N Engl J Med. 2011 Sep 1;365(9):869.



[2] Winter JE, MacInnis RJ, Wattanapenpaiboon N, Nowson CA. BMI and all-cause mortality in older adults: a meta-analysisAm J Clin Nutr. 2014 Apr;99(4):875-90.


Share this post

Link to post
Share on other sites

Thanks, Dean, for the analysis. But I'd put the doubts more strongly. Found two old mailing list posts on my hard drive (I wish the archives were up!!) that summarize my view pretty well. (I'll post the header info also so people can look up the full discussion when the archives are finally back up; a few typos corrected).


I saw from at least one response to Mike L's post that people are considering trying to gain weight because of his post!! That represents a frightening failure of logic. Arg. Come on people, strap on your thinking caps. It could save your life.


- Brian


-------- Mensaje reenviado --------

Asunto: Re: [CR] Just what the doctor ordered - OVERWEIGHT !

Fecha: Thu, 03 Jan 2013 23:19:55 -0500

De: Brian M. Delaney

Para: The CR Society Main Discussion List


El 2013-01-02 08:33, Michael Rae escribió:

> All:




[someone wrote:]


>> Dr. Flegal et al. have done a lot of work on the subject of

>> overweight, obesity, BMI and mortality:

>> http://www.ncbi.nlm.nih.gov/pubmed?term=flegal%20km


>> but it seems to contradict almost everything we consider *essential*

>> as practitioners of calorie restriction.


> No, it doesn't. CR has nothing to do with your weight, and the people in

> population studies with low BMIs did not get there by being on CR.

> -Michael



It's astonishing that this notion just will not go away that BMI studies have any significant bearing on the relation between CR and longevity. Some people just don't get it, and maybe never will.




But it's even worse than the logical fallacy of "affirming the consequent", as I've pointed out a million times before. (Overweight people face social pressure to lose weight; some may do so by eating healthfully, may thus live longer, but still aren't skinny, just thinnER than they were.)









-------- Mensaje reenviado --------

Asunto: Re: [CR] Thin and Old? Leisure World BMI Study

Fecha: Tue, 23 May 2006 08:05:47 +0200

De: Brian M. Delaney

Para: The CR Society Main Discussion List


>> Wonderful report! I'd love to have the full copy of

>> the study.


>> It appears to me the take home message of the study is

>> there is a Goldilocks effect of calorie intake and

>> longevity. I.E. one can have too many or too little

>> calories over time and will live a shorter life as a

>> result.


> I think that we shouldn't make up too many conclusions based on this

> study.


Probably none.


This issue has been discussed ad infinitum. To think studies correlating body weight to mortality have something to do with CR is (among other things) to commit the fallacy known as "affirming the consequent" (for people who like fancy names). That is, it is to 1) take "If A then B," 2) find B, 3) conclude A.


From an older post to a different list:







But far more importantly, no conclusions about the benefits of a CR program can be drawn from studies showing a correlation between weight and mortality. People trying to draw such a conclusion are committing a serious logical error, one with potentially devastating health consequences for those who buy the error.


"CR leads to a reduction in weight." Yes.


"A reduction in weight is a sign of CR." No.


CR is not about being thin. Naturally obese mice (ob/ob) on severe CR are still chubby, but live much longer than naturally thin mice not on CR. Energy-restriction shifts resources away from growth and reproduction towards repair and maintenance. Doesn't matter what you weigh.



Indeed, the assumption (or false conclusion) that "accidental/unintentional CR" is more likely to be found among the underweight is not only wrong, it may even be backwards. People in the countries where these mortality studies tend to be conducted who are naturally thin have LESS reason to restrict their food intake (and note: food restriction is not the same as calorie restriction -- though that's a minor point), given societal pressures to be thin.


The way to determine whether or not CR reduces mortality is to look at people on CR and compare them to people not on CR. This is being done. Some initial results include those reported by Fontana [1].


It will take a long time before we can be certain whether CR dramatically reduces mortality, but it seems likely that it does so, and we can be certain that risks of diseases of aging (certainly, type 2 diabetes) are reduced significantly.


There are LOTS of sensible reasons not to be on CR. Believing that "it doesn't work" because of correlations between mortality and body weight isn't one of them.






[1] Fontana L, Meyer TE, Klein S, Holloszy JO.

"Long-term calorie restriction is highly effective in reducing the risk

for atherosclerosis in humans."

Proc Natl Acad Sci U S A. 2004 Apr 27;101(17):6659-63. Epub 2004 Apr 19.


Share this post

Link to post
Share on other sites



I agree with you when you say that:


"A reduction in weight is a sign of CR." No.




There are LOTS of sensible reasons not to be on CR. Believing that "it doesn't work" because of correlations between mortality and body weight isn't one of them.


There are many reasons why someone might be thin that have nothing to do with CR, and which aren't healthy (latent disease, smoking, heroin addiction, sarcopenia, digestive problems etc.). As a result, the observation that "thin people often die early" is not a good argument that CR doesn't work.


Nevertheless, I think to dismiss the apparent advantage of having "extra meat on one's bones" late in life as entirely irrelevant to CR folks may be being a bit too hasty. You may not be doing this - but just in case you or others are, I think its worth considering.


Frailty late in life has repeatedly been shown to predict mortality, as the study Al posted today [1] confirms. It can certainly be argued that thinness is not necessarily synonymous with frailty, and that CR may (hopefully) protect us from getting frail in our old age. But until aging can be halted, even CR folks will continue to fall victim to the negative effects of the aging process - perhaps a little more slowly, but we can't entirely escape things like sarcopenia, bone loss & fractures, loss of appetite, impaired nutrient digestion / absorption, impaired immunity, etc. 


So Brian, to make things concrete, I'd be interested in hearing what you would say to a hypothetical CR practitioner who is celebrating their 90th birthday today. Assume he/she has been practicing CR quite seriously for several decades, and have a thin physique to show for it (BMI ~18.5). For the last few years they've been feeling the effects of aging - no overt illnesses or disabilities yet, but they are starting to feel older and their biomarkers are starting to turn south. What would you tell them? Stay the course and continue hitting the CR hard? Or dial back and try to gain some weight?


Admittedly we can't know for certain, since we don't (and won't in our lifetimes) have clear evidence from a large, controlled group of people in their situation, who have been practicing CR for many years and then experienced old age. 


But I would opt to advise the latter - dial their CR back and gain some weight (i.e. targeting a BMI of 20-22), since any additional longevity gains from continued CR are likely to be small, and the benefits of having extra "meat on their bones" may help them deal with the inevitable challenges of getting old.


To support my contention, without human CR data, it seems that the best data we have to rely on is that of the general population, especially long-lived individuals from the general population, whose long lives suggest they've enjoyed the benefits of some combination of good genes and/or a healthy lifestyle that helped them avoid the typical diseases of aging that killed off most of their peers much earlier.


If we look at the meta-analysis of BMIs for centenarians that Mike Lustgarten put together in the table above, the average BMI of these 1550 very long-lived individuals was 21.4. Certainly on the slim side, but not rail thin either. In fact, given that weight loss is virtually inevitable at such an advanced age, it seems likely they spent much of their elder years at a BMI significantly higher than 21.4. And as the study Al posted [1] shows, thinness is correlated with frailty, and frailty with mortality in the oldest of the old.


The hypothetical 90-year-old in question might be one of the lucky few who escapes serious illness/injury as they get very old, and so not need the extra metabolic reserves.  But the odds would seem to be against them, and eventually the inevitable (for now anyway) process of aging is going to catch up to them, so the safer bet in my eyes would be to dial it back.


While obviously an argument from authority, I would also point to the fact that Luigi Fontana, perhaps the most knowledgeable researcher and proponent of human CR,  has advised several of the CR cohort he's been studying (including me) that if he were in our shoes, he'd dial back on the CR somewhat (i.e. a BMI < 18 is too low) - since being as thin has many of us are comes with serious risks and offers uncertain benefits. Needless to say all of us are a lot younger than 90.


I'm curious, what would you (or others) advise such a person?





[1] Understanding Risk in the Oldest Old: Frailty and the Metabolic Syndrome in a Chinese Community Sample Aged 90+ Years.

Hao Q, Song X, Yang M, Dong B, Rockwood K.
J Nutr Health Aging. 2016;20(1):82-8. doi: 10.1007/s12603-015-0553-5.
PMID: 26728938
To investigate the relationship between frailty and the metabolic syndrome and to evaluate how these contribute to mortality in very old people.
Secondary analysis of data from the Project of Longevity and Aging in Dujiangyan.
Community sample from Sichuan Province, China.
People aged 90+ years (n=767; baseline age=93.7±3.4 years; 68.0% women.
After a baseline health assessment, participants were followed for four years (54.0% died). A frailty index (FI) was calculated as the sum of deficits present, divided by the 35 health-related deficits considered. Relationships between the FI and the metabolic syndrome were tested; their effect on death was examined.
The mean FI was 0.26±0.11. Higher FI scores were associated with a greater risk of death, adjusted for age, sex, education, and metabolic syndrome items. The hazard ratio was 1.03 (95% confidence interval 1.02, 1.04) for each 1% percent increase of the FI. The mortality risk did not change with the metabolic syndrome (odds ratio=0.99; 0.71-1.36).
In the oldest old, frailty was a significant risk for near-term death, regardless of the metabolic syndrome. Even using age-adjusted models, the epidemiology of late life illness may need to account for frailty routinely.
[2] Ann N Y Acad Sci. 2015 Dec 22. doi: 10.1111/nyas.12982. [Epub ahead of print]
Significant life extension by ten percent dietary restriction.
Richardson A(1,)(2), Austad SN(3), Ikeno Y(4), Unnikrishnan A(1), McCarter RJ(5).
Although it is well documented that dietary restriction (DR) increases the life
span of rodents and other animals, this increase is observed at relatively high
levels of DR, in which rodents are typically fed 40% less than that consumed by
rodents fed ad libitum. It is generally assumed that lower levels of DR will have
a lesser impact on life span; however, there are very little published data on
the effect of low levels of DR on life span. In this study, we show that 10% DR
increased life span to almost the same extent as 40% DR. While both 10% and 40%
DR resulted in similar changes in non-neoplastic lesions, 10% DR had no
significant effect on the incidence of neoplasia (except for pituitary adenoma), 
and 40% DR resulted in a significant reduction (40%) in neoplasia. These data
clearly demonstrate that the life span of F344 rats does not increase linearly
with the level of DR; rather, even a low level of DR can substantially affect
life span. This rodent study has important translational implications because it 
suggests that a modest reduction in calories might have significant health
benefits for humans.
PMID: 26695614

Share this post

Link to post
Share on other sites

I forgot about this thread on Relationship Between BMI and Disease, and Longevity, started by Michael and with input from me which has a good discussion of the optimal mid-life BMI for longevity.


Overall, the data on that question seems to suggest that once other (detrimental) causes of low-weight are properly controlled for, being thin (BMI < 22.5, and even very thin, a BMI < 18.5), in middle age, is no worse, and may be better  for avoiding death, at least up to age 75 (when these studies ended) than having a BMI greater than 22.5.


Note - this data on optimal middle-age BMI for longevity is based on data from the general population, like the centenarian data above. While they controlled for things like latent disease, smoking, etc that could have cause "unhealthy thinness", it is likely that most of the thin people in these studies were eating a diet which was far from optimal. So the fact that their thinness didn't appear to do them harm, despite possibly poor nutrition, further supports the idea that CR, even possibly pretty serious CR, in mid-life might improve one's chances of living to a ripe old age, or at least through one's mid-70s.



Share this post

Link to post
Share on other sites

BMI doesn't differentiate between lean muscle mass and adipose tissue. For example I'm a 62 year old woman with a BMI of 19.2. My body fat is 18% (measured by hydrostatic displacement); yet it would be possible for me to have the exact same BMI of 19.2 but have 25% body fat, if I had less muscle and/or more adipose tissue. How does this ratio affect longevity? Is it factored in to any of these studies?

Share this post

Link to post
Share on other sites

BMI doesn't differentiate between lean muscle mass and adipose tissue. For example I'm a 62 year old woman with a BMI of 19.2. My body fat is 18% (measured by hydrostatic displacement); yet it would be possible for me to have the exact same BMI of 19.2 but have 25% body fat, if I had less muscle and/or more adipose tissue. How does this ratio affect longevity? Is it factored in to any of these studies?


Regarding the relative merits of lean body mass vs. fat mass for late-life survival in humans, from what I know/recall, there is pretty good evidence that having extra lean mass would be more beneficial than fat mass. But it may not be a causal relationship - i.e more lean mass means better physical condition / activity, more ability to take care of one's daily needs, less signs of aging (e.g. sarcopenia) and therefore less likelihood of dying.


This isn't exactly your question but the famous ob/ob mice results [1] found that mice genetically prone to obesity lived shorter lives than normal mice when fed ad lib. But when both types of mice were restricted to the same # of calories per day, the ob/ob mice lived just as long as the normal mice (and had longer max lifespan), gaining more relatively lifespan (as a percent of their normal ad-lib fed lifespan) than did the normal mice. Here is the lifespan table with the details from the full text, for current and future reference (the +/+ mice are the normal ones):



So having a natural tendency towards being overweight/obese, i.e. a so called "thrifty genotype", does not appear to be an obstacle to benefiting from CR. In fact some say it's an advantage, since (like the ob/ob mice), such people can retain more body fat for any given level of restriction, allowing for greater metabolic reserves in case of illness or injury. There is a whole thread on the genetics of obesity, on which some of us lament having a  'lean genotype" making CR harder to practice effectively, i.e. without losing too much weight.




Do you know this article by V. Longo and others http://www.sciencedirect.com/science/article/pii/S155041311400062X    , where they suggests that a higher protein intake may be protective for older adults (after 65 years old)?   Also here https://news.usc.edu/59199/meat-and-cheese-may-be-as-bad-for-you-as-smoking/


Yes - in fact I included reference to the Longo study you point to, PMID 24606898, in this post. Based on it and other data, in that post I recommended:

  • If possible, start CR in early or middle adulthood, when the body is more resilient.
  • Ease into CR, to avoid a big shock, especially if starting relatively later in adulthood.
  • Back off on the severity of one's CR practice (and perhaps add extra protein) when one's gets to be in the elderly range (65-70+), to avoid nutrient deficiencies and maintain some metabolic reserves so one can weather the 'slings and arrows' that will eventually come.

For your and everyone's information, it's easy to find out if an article like the one you asked about has been discussed before by searching the forums using its PMID number. That is also why those of us who reference specific studies try to always include the PMID number if available, so discussions of it will be easy to find in the future.





[1] Proc Natl Acad Sci U S A. 1984 Mar;81(6):1835-8.

Effects of food restriction on aging: separation of food intake and adiposity.
Harrison DE, Archer JR, Astle CM.
Restricted feeding of rodents increases longevity, but its mechanism of action is
not understood. We studied the effects of life-long food restriction in
genetically obese and normal mice of the same inbred strain in order to
distinguish whether the reduction in food intake or the reduction in adiposity
(percentage of fatty tissue) was the critical component in retarding the aging
process. This was possible because food-restricted obese (ob/ob) mice maintained 
a high degree of adiposity. In addition to determining longevities, changes with 
age were measured in collagen, immune responses, and renal function. Genetically 
obese female mice highly congenic with the C57BL/6J inbred strain had
substantially reduced longevities and increased rates of aging in tail tendon
collagen and thymus-dependent immune responses, but not in urine-concentrating
abilities. When their weight was held in a normal range by feeding restricted
amounts, longevities were extended almost 50%, although these food-restricted
ob/ob mice still had high levels of adiposity, with fat composing about half of
their body weights. Their maximum longevities exceeded those of normal C57BL/6J
mice and were similar to longevities of equally food-restricted normal mice that 
were much leaner. Food restricted ob/ob mice had greatly retarded rates of
collagen aging, but the rapid losses with age in splenic immune responses were
not mitigated. Thus, the extension of life-span by food restriction was inversely
related to food consumption and corresponded to the aging rate of collagen. These
results suggest that aging is a combination of independent processes; they show
that reduced food consumption, not reduced adiposity, is the important component 
in extending longevity of genetically obese mice.
PMCID: PMC345016
PMID: 6608731

Share this post

Link to post
Share on other sites

A few posts above in this thread, I lamented the fact that all we've really got to go on when it comes to the optimal strategy for late-life CR & BMI is human data from the general population. This isn't quite true, and I owe thanks to Al Pater for pointing out that there is rodent CR data to help answer (or at least hint at answering) this question.


In particular, this study [1] tested 3 strains of male mice (C57BL/6, DBA/2, and B6D2F1) in order to answer several interesting questions.


First, they did a standard study of AL vs CR40 started at 4 months (the human equivalent of early adulthood - 20-30 year-old) to test the effects of CR on lifespan in these three strains. Interestingly, the C57BL/6 and B6D2F1 strains benefited from CR, but the CRed DBA/2 mice lived no longer than AL-fed DBA/2 mice, and had higher earlier mortality than AL-fed DBA/2 mice. Here are the survival curves for the three strains:



DBA/2 mice are naturally thinner than the other two strains, and the authors postulated that this natural thinness may or may not be why they didn't benefit from CR in this or other studies. But whatever the reason, the fact that some strains (particularly naturally thin ones) don't benefit from CR is a bit troubling for the prospects for human CR, as the author's observe (my emphasis):


These findings do not support the popular conception that CR might be a practical tool for the
retardation of the aging process in all of the various animal species, including humans. The
nature of the mechanisms by which CR affects life span of rodents, or why different strains of
laboratory mice and rats respond differently to CR, is presently unknown.


But the second part of this study was at least as interesting and more germane to the question of what a CR practitioner should do late in life to maximize their (note - using 'they' as a singular pronoun :)xyz) longevity. What the researchers did with other groups of mice from all three strains was to switch them from AL to CR feeding (AL->CR) or from CR to AL feeding (CR->AL) at various ages (7, 17 or 23 months) for 11 weeks, to see what impact such changes had on longevity/survivorship. 


So what happened?


Switching from AL to CR at 17 months (the human equivalent of starting CR at about age 55) or 24 months (human equivalent of about age 69) was generally bad news:


 Switching the old AL mice to the CR regimen (AL→CR) tended
to decrease their probability of survival relative to the AL→AL groups. This effect was
pronounced for the DBA/2 mice, but the same trend was evident in the C57BL/6 and
B6D2F1 mice. The decrease in probability of survival was evident by 17 months in the DBA/
2 and C57BL/6 mice but not until 24 months in B6D2F1. 


In contrast, switching from CR to AL late in life was not detrimental relative to continued CR:


Transfer of CR mice to the AL regimen (CR→AL) had little or no effect on
survival (relative to CR→CR groups) in the three genotypes.
In fact, looking at the survival plots of the two strains that benefit from CR (C57BL/6 and B6D2F1 in the top two rows below), we see that the CR->AL mice (solid circles) had a slight (non-significant) survival advantage relative to the CR->CR condition (hollow circles) when the transition was done at 24 months, the human equivalent of ~69 years old. And look at the poor late-onset CR mice (AL->CR) from all three strains (solid squares) - they are dropping like flies in the 11 weeks after CR onset.
In short, this rodent data suggests that the advice to "back off" serious CR when one reaches one's elder years is sound. It appears that backing off won't reduce any CR longevity benefits you've accrued, and it could be beneficial in the "real world" for recovering from injury/illness, something that rodents housed in a safe & sterile environment don't have to worry about. And this data clearly suggests starting serious CR in one's elder years is a really bad idea, particularly if done abruptly.


[1] Genotype and age influence the effect of caloric intake on mortality in mice.

Forster MJ, Morris P, Sohal RS.
FASEB J. 2003 Apr;17(6):690-2. Epub 2003 Feb 5.
PMID: 12586746 Free PMC Article
Long-term caloric restriction (CR) has been repeatedly shown to increase life span and delay the onset of age-associated pathologies in laboratory mice and rats. The purpose of the current study was to determine whether the CR-associated increase in life span occurs in all strains of mice or only in some genotypes and whether the effects of CR and ad libitum (AL) feeding on mortality accrue gradually or are rapidly inducible and reversible. In one experiment, groups of male C57BL/6, DBA/2, and B6D2F1 mice were fed AL or CR (60% of AL) diets beginning at 4 months of age until death. In the companion study, separate groups of mice were maintained chronically on AL or CR regimens until 7, 17, or 22–24 months of age, after which, half of each AL and CR group was switched to the opposite regimen for 11 wk. This procedure yielded four experimental groups for each genotype, namely AL==>AL, AL==>CR, CR==>CR, and CR==>AL, designated according to long-term and short-term caloric regimen, respectively. Long-term CR resulted in increased median and maximum life span in C57BL/6 and B6D2F1 mice but failed to affect either parameter in the DBA/2 mice. The shift from AL==>CR increased mortality in 17- and 24-month-old mice, whereas the shift from CR==>AL did not significantly affect mortality of any age group. Such increased risk of mortality following implementation of CR at older ages was evident in all three strains but was most dramatic in DBA/2 mice. Results of this study indicate that CR does not have beneficial effects in all strains of mice, and it increases rather than decreases mortality if initiated in advanced age.
Keywords: caloric restriction, aging, C57BL/6, DBA/2, B6D2F1

Share this post

Link to post
Share on other sites

I think to dismiss the apparent advantage of having "extra meat on one's bones" late in life as entirely irrelevant to CR folks may be being a bit too hasty. You may not be doing this - but just in case you or others are, I think its worth considering.


I am not doing that, but I would say: we just don't know.



But I would opt to advise [for an elderly CRer who is starting to feel the effects of aging] the latter - dial their CR back and gain some weight (i.e. targeting a BMI of 20-22), since any additional longevity gains from continued CR are likely to be small, and the benefits of having extra "meat on their bones" may help them deal with the inevitable challenges of getting old.


I would probably opt for that as well, but not for reasons based on data from the general population. (CRers are not the general population!). But my first bit of advice would be to test bone strength, immune markers, etc., -- and also do tests of nutritional status (at least mineral status) -- and do a few-month trial of increased eating, and see how those markers change.



About Sohal: he's very strongly anti-CR (but that doesn't mean we shouldn't take his arguments seriously!), so I'm a little -- just a little: he's also a competent researcher, from what I see -- concerned that bias might come into play in his work, but the results of his paper are certainly relevant to this discussion. But we all must remember something: CR rodents are likely much more restricted than most of us are. So the 40% restriction used in the study likely doesn't correspond to how much we restrict.


Far more importantly: the switch to CR in adulthood was made all at once! We've known for years that switching non-gradually to CR in adulthood doesn't work in rodents (and likely not in any species). This was a key finding that Walford and Weindruch made a couple decades ago. It was thought since the 1930s that CR only worked when initiated extremely early in life. But all that was needed to make it work when initiated in adulthood was do the shift gradually, for example, go down by 10% in three or so phases.


I'm surprised Sohal et al weren't aware of this fundamental finding, or chose not to take it into account in the design of the study.


- Brian

Share this post

Link to post
Share on other sites



Sohal may have an anti-CR agenda, and it is unfortunate he chose to use a study design with an abrupt onset of CR in the adult / aged mice. But I was less interested in the strongly detrimental effects of abrupt onset CR (which I think we all know is bad news, at least in middle to late adulthood), than I was in the opposite, the effects of transitioning from CR to AL in one's later years - which from this mice data shows little negative, and perhaps a small positive, effect, at least for mice mortality in the 11 weeks after the switch.


And while he may be anti-CR, as you suggest, the evidence and arguments in the paper you point to [1], are well worth considering and quite germane for discussion of optimal weight which is the topic of this thread. Thanks for drawing my attention to it!


In it, Sohal is (as you would predict) pretty negative about CR.


His most convincing argument (to me) from the data he presents is that CR in mammals works because of, and to the degree that, it prevents the animals from becoming unnaturally obese as many laboratory strains of mice/rats are prone to when given ad lib access to food.


More quantitatively, he shows that across a range of rat & mice strains, the lifespan extension resulting from lifelong 40% CR is a linear function of the area under the curve of weight gain that the CR rodents avoid as a result of not getting fat eating AL. The concept is a little complicated, so allow me to illustrate. Here is a graph of lifelong body weight trajectory for AL or CR40 mice from one common strain (C57BL/6), fed either a normal diet (left) or a high-fat/high-protein diet (right). 



As you can see, the mice fed the high-fat diet in ad lib quantities (solid line in the right graph) gained a lot more weight than the AL-fed mice on the normal-diet (solid line in the left graph). So the shaded area representing weight gain area under the curve (Wt. Gain AUC) is a lot larger for the high-fat AL mice than the normal diet AL mice. And note that this weight gain AUC is primarily a function of how fat the AL mice are getting, not how much weight the CR mice are losing - in fact the CR mice are  fairly stably maintaining their early adult weight until they get pretty old and start to slowly lose weight.


And as you can see along the bottom, the survival increase of the CR mice relative to their corresponding AL peers was higher in the high fat condition relative to the normal diet condition. This illustrates his thesis - that CR lifespan benefits are proportional to weight gain AUC - i.e. the more obesity avoided, the larger the lifespan gains from CR.


But what is really striking is how accurately this proportionality appears to predict CR benefits across different rodent strains, as illustrated in these two graph:



What this shows is Weight gain AUC (the shaded area we saw in the graph above) along the X-axis vs. lifespan benefits along the Y-axis. The data is quite linear, especially for the various mice strains/diets.


What this results seems to strongly suggest is a stronger version of the thesis that I've been advancing recently (e.g. here, here and here), that most of the benefits of CR can be achieved by simply avoiding obesity. He goes a step further, contending quantitatively that the more "summed obesity burden" (i.e. lbs overweight * years spent in that overweight condition. The equivalent of cigarette pack-years for smoking burden) you avoid by maintaining a constant, low body weight, the greater will be your lifespan benefit relative to AL.


He even uses this model to explain the ob/ob mice results. Rather than presenting Sohal's explanation for the ob/ob mice results and its implications in my own words, I'll quote the paper, which does a really good job I think:


While the hypothesis that CR increases longevity by preventing excessive body mass and adiposity has received some attention [82–85], it has historically been downplayed in the rodent biogerontology literature [19, 86–88], partly based on the results of a study by Harrison et al. [87], in which they compared the effect of CR on longevity in genetically obese (ob/ob) and normal C57BL/6 mice. The obese mice were fed AL or placed on a CR regimen that maintained their weight at a level equivalent to the non-obese [AL-fed] mice. An additional group of normal mice was fed approximately 33% less food than their AL fed counterparts. Compared to the AL fed normal mice, the food- restricted ob/ob mice had a higher percentage of body fat, yet the life span of the two groups was nearly equal. This outcome was widely interpreted to mean that a reduction in body weight/adiposity should be ruled out as a mechanism for the CR-related increase in longevity. However, this explanation ignores a rather crucial observation in the study, namely that the cumulative weight gain in ob/ob mice under the AL regimen was ~ 4-fold greater than in the normal mice (Fig. 5 - [reproduced below]). Thus, an alternative interpretation of the results of the study by Harrison et al. would be that CR was more effective in lengthening the life span of ob/ob mice than the normal mice because it attenuated their energy imbalance and the consequent deleterious effects of fat accumulation to a relatively much greater extent [66, 82, 89–98]. This explanation accords with the data shown below:




What you can see in these graphs is that the CRed ob/ob mice (dashed line on the right graph), weighed about the same amount as the AL fed normal mice (solid line on the left) - around 30g. What isn't shown is that the CRed ob/ob mice maintained much more fat mass (higher percent body fat) than the AL fed normal mice despite having the same weight. The fat ob/ob mice fed AL died early, at about 18 months. But the CRed ob/ob mice lived almost as long as the normal mice and gained a heck of a lot of lifespan benefit relative to AL-fed ob/ob mice. This has been used as evidence by some to show that weight doesn't really matter, you can maintain quite a bit of fat mass while practicing CR and still garner lifespan benefits - i.e. its "calories, calories, calories" that matter.  


But what Sohal argues, and this graph illustrates, is his theory that what matters for lifespan benefits is the amount of obesity burden avoided. The CRed ob/ob mice avoided a lot of obesity relative to AL-fed ob/ob mice, and so lived a lot longer than AL-fed ob/ob mice. The CRed normal mice didn't avoid nearly as much obesity, since the AL-fed normal mice didn't get nearly as fat as the AL-fed ob/ob mice, so the Wt. Gain AUC was much smaller. And so the CRed normal mice didn't gain nearly as much lifespan advantage over their AL-fed normal mice peers.


Sohal points out that this model matches the Wisconsin/NIA CR primate results as well, where there were significant health and lifespan benefits of CR relative to controls in the Wisconsin monkeys, largely as a result of the control monkeys getting seriously obese since they were fed a crappy diet in literally ad lib amounts. In contrast, the CR monkeys in the NIA study didn't live longer than the NIA control, which Sohal suggests was probably because the NIA control monkeys were prevented from getting obese. Michael seems to concur, both overtly in his blog post about the monkey trials where he too acknowledges the strong possibility of this hypothesis, and tacitly here by his silence on the topic, despite my repeated goading of him  :)xyz.


In short, Sohal says the rodent and monkey data (and I'd add to that the dog CR data I discussed here) shows that the fatter a particular species or strain gets when fed AL, the greater the lifespan benefits of CR will be relative to AL feeding, likely because of the greater obesity burden the CR animal is avoiding by remaining thin.


Or put more succinctly - Sohal appears to be arguing that it's not so much that CR is good for health/longevity as it is that getting and staying obese is actively bad for health and longevity. In his own words in the last sentence of the paper he puts it this way:


In a nutshell, CR increases life span when it counteracts a significant energy imbalance.


 My two (tentative) corollaries to this theory are that most of the CR benefits can be achieved simply by maintaining a reasonably low (but not necessarily rail-thin) body weight, and that one can remain thin and thereby gain most of these benefits by several means, including reduced calories, increased exercise, or cold exposure.





[1] Free Radic Biol Med. 2014 Aug;73:366-82. doi:

10.1016/j.freeradbiomed.2014.05.015. Epub 2014 Jun 2.
Caloric restriction and the aging process: a critique.
Sohal RS(1), Forster MJ(2).
The main objective of this review is to provide an appraisal of the current
status of the relationship between energy intake and the life span of animals.
The concept that a reduction in food intake, or caloric restriction (CR), retards
the aging process, delays the age-associated decline in physiological fitness,
and extends the life span of organisms of diverse phylogenetic groups is one of
the leading paradigms in gerontology. However, emerging evidence disputes some of
the primary tenets of this conception. One disparity is that the CR-related
increase in longevity is not universal and may not even be shared among different
strains of the same species. A further misgiving is that the control animals, fed
ad libitum (AL), become overweight and prone to early onset of diseases and
death, and thus may not be the ideal control animals for studies concerned with
comparisons of longevity. Reexamination of body weight and longevity data from a 
study involving over 60,000 mice and rats, conducted by a National Institute on
Aging-sponsored project, suggests that CR-related increase in life span of
specific genotypes is directly related to the gain in body weight under the AL
feeding regimen. Additionally, CR in mammals and "dietary restriction" in
organisms such as Drosophila are dissimilar phenomena, albeit they are often
presented to be the very same. The latter involves a reduction in yeast rather
than caloric intake, which is inconsistent with the notion of a common, conserved
mechanism of CR action in different species. Although specific mechanisms by
which CR affects longevity are not well understood, existing evidence supports
the view that CR increases the life span of those particular genotypes that
develop energy imbalance owing to AL feeding. In such groups, CR lowers body
temperature, rate of metabolism, and oxidant production and retards the
age-related pro-oxidizing shift in the redox state.
Copyright © 2014 The Authors. Published by Elsevier Inc. All rights reserved.
PMCID: PMC4111977
PMID: 24941891

Share this post

Link to post
Share on other sites



In the post immediately above this one, I briefly alluded to the analogy between pack-years of cigarette smoking and what I'll call "lifetime obesity burden", which is approximately (# lbs overweight/obese) * (# of years spent overweight/obese), and which Sohal quantified as "weight gain area-under-the-curve".


At the time I didn't realize how apt this parallel was, but this new study [1] I stumbled across today seems to reinforce the analogy. It found that a person's "maximum attained weight" over their entire past life history was a much better predictor of future mortality than was their weight at the time of the survey. In particular, people who were formerly overweight or obese but who had lost weight to regain "normal weight" status when they filled out the questionnaire, were at an elevated risk of dying. From the full text, they found that the death rate for people who were previously overweight, but reported normal weight at the time of survey was 27 percent higher than the rate for people whose weight never strayed from the normal range.


People have speculated that this might be due to latent disease causing subjects to have lost weight at the time of the survey, and which later would end up killing them.


But the authors seem to lean towards Sohal's explanation instead. Namely that like smoking, the burden of overweight/obesity has a long tail, so having been overweight/obese earlier in life can come back to bite you later. This is directly analogous to smoking. Former heavy smokers often die early from cardiovascular disease, respiratory disease or lung cancer, despite having quit smoking many years earlier - they had too many "pack years" under their belt to avoid the inevitable negative consequences of their unhealthy habit. 


This "long tail of obesity" hypothesis makes quite a bit of sense, given the fact that our two biggest killers, cardiovascular disease and cancer, are both exacerbated by being overweight/obese, and result from the slow accumulation of damage over many years, to artery walls on the one hand and DNA/cellular machinery on the other.


So to bastardize Mr. Spock: stay thin to live long and prosper...





[1] Proc Natl Acad Sci U S A. 2016 Jan 4. pii: 201515472. [Epub ahead of print]

Revealing the burden of obesity using weight histories.
Stokes A(1), Preston SH(2).
Analyses of the relation between obesity and mortality typically evaluate risk
with respect to weight recorded at a single point in time. As a consequence,
there is generally no distinction made between nonobese individuals who were
never obese and nonobese individuals who were formerly obese and lost weight. We 
introduce additional data on an individual's maximum attained weight and
investigate four models that represent different combinations of weight at survey
and maximum weight. We use data from the 1988-2010 National Health and Nutrition 
Examination Survey, linked to death records through 2011, to estimate parameters 
of these models. We find that the most successful models use data on maximum
weight, and the worst-performing model uses only data on weight at survey. We
show that the disparity in predictive power between these models is related to
exceptionally high mortality among those who have lost weight, with the
normal-weight category being particularly susceptible to distortions arising from
weight loss. These distortions make overweight and obesity appear less harmful by
obscuring the benefits of remaining never obese. Because most previous studies
are based on body mass index at survey, it is likely that the effects of excess
weight on US mortality have been consistently underestimated.
PMID: 26729881

Share this post

Link to post
Share on other sites



In my previous post, I talked about cardiovascular disease (CVD) as having a very long tail - i.e. taking many years to develop, and having a long-term impact that isn't easily reversed once it has developed.


I was shocked today by the amazing evidence in support of this in Dr. Greger's latest video. It talks about a study [1] showing the presence of fatty streaks, the precursor of CVD, in the heart and arteries of human fetuses!


In fact, babies in the wombs of mothers with high serum cholesterol levels had 10x the fatty streaks of babies inside moms with normal cholesterol levels:


These results demonstrate that LDL oxidation and formation of fatty streaks occurs already during fetal
development, and that both phenomena are greatly enhanced by maternal hypercholesterolemia.


One more way that the sins of the mother (but thankfully, probably not the father in this case) are visited upon the child...





[1] J Clin Invest. 1997 Dec 1;100(11):2680-90.

Fatty streak formation occurs in human fetal aortas and is greatly enhanced by
maternal hypercholesterolemia. Intimal accumulation of low density lipoprotein
and its oxidation precede monocyte recruitment into early atherosclerotic
Napoli C(1), D'Armiento FP, Mancini FP, Postiglione A, Witztum JL, Palumbo G,
Palinski W.
Author information: 
(1)Department of Clinical and Experimental Medicine, Federico II University of
Naples, 80131 Naples, Italy.
To determine whether oxidized LDL enhances atherogenesis by promoting monocyte
recruitment into the vascular intima, we investigated whether LDL accumulation
and oxidation precede intimal accumulation of monocytes in human fetal aortas
(from spontaneous abortions and premature newborns who died within 12 h; fetal
age 6.2+/-1.3 mo). For this purpose, a systematic assessment of fatty streak
formation was carried out in fetal aortas from normocholesterolemic mothers (n = 
22), hypercholesterolemic mothers (n = 33), and mothers who were
hypercholesterolemic only during pregnancy (n = 27). Fetal plasma cholesterol
levels showed a strong inverse correlation with fetal age (R = -0.88, P <
0.0001). In fetuses younger than 6 mo, fetal plasma cholesterol levels correlated
with maternal ones (R = 0.86, P = 0.001), whereas in older fetuses no such
correlation existed. Fetal aortas from hypercholesterolemic mothers and mothers
with temporary hypercholesterolemia contained significantly more and larger
lesions (758,651+/-87,449 and 451,255+/-37,448 micron2 per section, respectively;
mean+/-SD) than aortas from normocholesterolemic mothers (61,862+/-9,555 micron2;
P < 0.00005). Serial sections of the arch, thoracic, and abdominal aortas were
immunostained for recognized markers of atherosclerosis: macrophages, apo B, and 
two different oxidation-specific epitopes (malondialdehyde- and
4-hydroxynonenal-lysine). Of the atherogenic sites that showed positive
immunostaining for at least one of these markers, 58.6% were established lesions 
containing both macrophage/foam cells and oxidized LDL (OxLDL). 17.3% of all
sites contained only native LDL, and 13.3% contained only OxLDL without monocyte/
macrophages. In contrast, only 4.3% of sites contained isolated monocytes in the 
absence of native or oxidized LDL. In addition, 6.3% of sites contained LDL and
macrophages but few oxidation-specific epitopes. These results demonstrate that
LDL oxidation and formation of fatty streaks occurs already during fetal
development, and that both phenomena are greatly enhanced by maternal
hypercholesterolemia. The fact that in very early lesions LDL and OxLDL are
frequently found in the absence of monocyte/macrophages, whereas the opposite is 
rare, suggests that intimal LDL accumulation and oxidation contributes to
monocyte recruitment in vivo.
PMCID: PMC508471
PMID: 9389731

Share this post

Link to post
Share on other sites

Not a chance on low BMI not signaling CR compared to higher BMI on a population basis.  You can get low BMI in one of two ways:  CR or physical activity (PA).  There's a third, rarer mechanism:  certain disease processes, like some cancers and hyperthyroidism.  But again, that's rare.  Most diseases associated with low BMI achieve this through lower food intake...or CR.


PA doesn't actually differ a huge amount across cultures, especially when you exclude the extraordinarily, freakishly sedentary.  Increasing intentional activity beyond a certain level decreases non-exercise activity thermogenesis to compensate in free-living people.  The fairly tight cluster of body fat percentage with BMI testifies to this, too.


So that leaves people mostly eating fewer calories, which is "non CR" that's really CR, though still not CRON.  In cases of drug addiction, low BMI is typically a result of either loss of appetite, which causes CR, or loss of prioritization for food, which causes...CR.  Making CR something other than restricting calories is beyond silly.  It doesn't matter WHY you restrict--restricting is restricting, period.  Intention doesn't matter.  ON does, of course.  And the disease itself matters, because if it's a respiratory disease or anorexia or weight maintenance or religious reasons, you'll get different outcomes, for sure.  But not being one of a specially privileged group believing themselves to be part of some grand experiment doesn't magically make their restriction no longer CR.  (Seriously, distinguishing between "food restriction" and "calorie restriction"????  Anorexics are practicing calorie restriction, guaranteed, and usually also food-type restriction.)


If you want involuntary CR, there's been plenty of that historically, too.  In certain countries, like Bangladesh and parts of India, girls are routinely underfed from childhood because they are considered to be worth less compared to sons.


Recognizing this means you have to admit that CR's not going to get you past the age of 100 and change, even if practiced perfectly.  And guess what?  It's not.  But the people on the opposite end are dying in their 60s (well, the super obese in their 50s), so I'm thrilled with the expectation of a high likelihood of a healthy 90s.


Low weight resulting in high death in old age has been shown again and again to be related to a loss of lean mass.  When the only exercise you get is from moving your own body weight around a little bit, as is common in old age in places with very sedentary retirement like the US, weight is protective because it increases your lean mass just a little.  There are a number of studies, including tracking BMI over time and controlling for PA in the elderly (which is freakishly depressed in some places...like the US) that show no risk from leanness that isn't associated with lower lean mass.  Keep a lean mass like a younger person through the healthiest diet, calorie quantity, and appropriate exercise that you can, and then you'll stay at the lowest risk possible in old age.


There's nothing special about CR here except that we can be more methodical about the ON half.  And we've already seen the maximum longevity that CR can get you.  It's well within the 90-110 year maximum lifespan.


Comparing ob/ob mice to humans makes no sense.  We don't have appreciable numbers of people walking around with untreated leptin deficiency or any similar condition.  Leptin deficiency inhibits puberty in people, and that tends to get noticed and treated.  We can also see VERY clearly from hundreds of studies that in humans there's a pretty tight correlation between body fat and BMI.  There are studies with over a thousand people, for example, that don't find a single man with less than 15% body fat or woman with less than 25% with a BMI above the overweight range.  We have samples of 6000 people measured by DEXA scan in the NHANES that don't find even one extreme outlier--hardcore weightlifters are super-rare.  There are no ob/ob equivalent clusters where at the same BMI one group of people is walking around "naturally" with twice...or half...the body fat percentage of other people.  Even a 33% difference is incredibly rare and always, in the absence of severe, life-altering mutations, the result of deliberate and very high intensity bodybuilding.  To get ANYTHING like that variation through genetics, you would have to isolate groups like those with Down or Prader Willi syndromes.  You can't have something like that and just not know.  


The genetic contribution to weight among ostensibly normal people is no more than 10lbs, MAYBE 15lbs, with all known genetic variations apart from a severe syndrome stacked together.  Our apparent genetic heterogeneity actually reveals an even greater homogeneity--there is a lack of breeding of groups of people to a specific purpose or mutation and a powerful regression toward a norm.  The environmental pressures haven't been tremendously different for most groups of people across time.  We might, MIGHT see some unusual results in genetically isolated Saami and Inuit populations, and maybe Australian Aborigines, who had extreme environmental pressures measurably different than most populations. But so far, even things like the Inuit's high levels of brown fat turned out to be, surprise surprise, entirely environmental.  We're just not going to see wide variations, apart from genetic disease, in typical Afro-Eurasian populations.  That means that BMI is an incredibly powerful proxy for caloric intake and physical activity at a given height and that these studies ARE meaningful, and to assume a special privilege because you're doing CR to live longer is very, very wrong-headed.

Share this post

Link to post
Share on other sites

Sarcopenia Best Predictor of Late Life Mortality Risk


Two of the biggest problems with maintaining a low BMI into one's senior years are sarcopenia (muscle loss) and osteoporosis (bone loss). This new study [1] posted by Al (thanks Al!) illustrates how damaging the former can be for longevity. 


It followed 350 community-dwelling elderly natives of Italy ranging in age from 80-102 for 10 years. It correlated a bunch of measures, including diagnosis with sarcopenia, with all-cause mortality risk.  It found that the group with low muscle mass had 3.67 times the 10 year mortality risk of the people with sufficient muscle mass, after adjusting for age and gender. Even after adjusting for a bunch of other potential confounders (including adjusted for age, gender, ADL, IADL impairment, cognitive impairment, BMI, CRP and IL-6), sarcopenia was associated with more than a doubling of 10-year mortality risk. Here are the 10-year survival curves for people with and without sarcopenia:




Shockingly, the ill-effects of sarcopenia were virtually independent of whether or not subjects also had 2 or more conditions/diseases typical of old age, including: "obesity, coronary heart disease, cerebrovascular disease, congestive heart failure, peripheral artery disease, hypertension, lung disease (chronic obstructive pulmonary disease, emphysema or asthma), osteoarthritis, diabetes, dementia (Alzheimer's disease and other forms of dementia), Parkinson's disease, renal failure and cancer (non-melanoma skin cancer excluded)."


In other words, have sarcopenia, die early, regardless of what other ailments you do or do not have.


The authors conclude:


Our findings show that sarcopaenia ... is associated with higher mortality rates in older adults living in the community, independent of age, gender and several clinical and biochemical parameters.


Results from this study also show that physical function impairment, not multimorbidity, intervenes in the relationship between sarcopaenia and mortality. Specifically, sarcopaenic participants with poor physical performance, as indicated by lower scores at the SPPB, showed higher mortality rates relative to their well-functioning peers. In contrast, the presence of two or more disease conditions did not impact 10-year mortality of older community-dwellers with sarcopaenia. 


Which group would you rather be in? The frail sarcopenia folks or the long-lived muscular folks? And in fact, the sarcopenia folks weren't even that thin, with an average BMI of 23 vs. nearly 27 for the long-lived folks without sarcopenia.


Once again we see having meat (especially muscle) on your bones when you get old is associated with a greatly reduced risk of dying.






[1] BMJ Open. 2016 Jul 25;6(7):e008281. doi: 10.1136/bmjopen-2015-008281.


Impact of physical function impairment and multimorbidity on mortality among community-living older persons with sarcopaenia: results from the ilSIRENTE prospective cohort study.


Landi F, Calvani R, Tosato M, Martone AM, Bernabei R, Onder G, Marzetti E.

Free full text: http://bmjopen.bmj.com/content/6/7/e008281.long




Sarcopaenia and physical function impairment may have a greater effect on survival than other clinical characteristics, including multimorbidity. In this study, we evaluated the impact of sarcopaenia on all-cause mortality and the interaction among muscle loss, physical function impairment and multimorbidity on mortality risk over 10 years in older community-dwellers.


Prospective cohort study.


Population-based study.


All persons aged 80+ years living in the community in the Sirente geographic area (L'Aquila, Italy) (n=364). Participants were categorised in the sarcopaenic or non-sarcopaenic group based on the European Working Group on Sarcopenia in Older People criteria.


(1) All-cause mortality over 10 years according to the presence of sarcopaenia and (2) impact of physical function impairment, assessed using the Short Physical Performance Battery (SPPB), and multimorbidity on 10-year mortality risk in persons with sarcopaenia.


Sarcopaenia was identified in 103 participants (29.1%). A total of 253 deaths were recorded over 10 years: 90 among sarcopaenic participants (87.4%) and 162 among non-sarcopaenic persons (65.1%; p<0.001). Participants with sarcopaenia had a higher risk of death than those without sarcopaenia (HR=2.15; 95% CI 1.02 to 4.54). When examining the effect of sarcopaenia and physical function impairment on mortality, participants with low physical performance levels showed greater mortality. Conversely, the mortality risk was unaffected by multimorbidity.


Our findings show that physical function impairment, but not multimorbidity, is predictive of mortality in older community-dwellers with sarcopaenia. Hence, in sarcopaenic older persons, interventions against functional decline may be more effective at preventing or postponing negative health outcomes than those targeting multimorbidity.


PMID: 27456324

Share this post

Link to post
Share on other sites

Study of Older Twins Shows Higher BMI Associated with Fewer Heart Attacks & Lower Mortality


The evidence just seems to keep piling up that being overweight/obese in late middle-age and older does not seem to increase one's risk of dying from CVD or all-causes.


In this new study [1] from JAMA, researchers followed ~4000 Swedish identical twins with discordant BMIs (i.e. one was thin and one was chubby) for 12 years starting at an average age of 58 (late middle age on average in my book, but with a range from 42 to 92). They compared the rate of heart attacks, all-cause mortality, and diabetes between the two twins to see how much BMI matters against a background of identical genetics, to focus on the impact of lifestyle-induced overweight/obesity.


The average BMI of the heavy twins was 25.9, which is close to the long-lived sarcopenia-free group (avg BMI ~27) I discussed in the post immediately above, and the optimal BMI of 25 as determined across the meta-analysis of 4 million never-smokers discussed in this post, and the optimal BMI of 26 discussed in this post.  The skinnier twins had an average BMI of 23.9, which obviously isn't especially skinny compared with most of us. 


What did they find? Not surprisingly given the previous results, they found that the skinnier twins had no advantage in terms of risk of heart attacks or all-cause mortality. In fact, the chubby twins had a lower rate of heart attacks and mortality:


During a mean follow-up period of 12.4 (2.5) years, 203 MIs (5.0%) and 550 deaths (13.6%)
occurred among heavier twins ... compared with 209 MIs (5.2%)  and 633 deaths (15.6%)
among leaner twins...


Here are the survival curves for the leaner and heavier twins, showing the modest mortality advantage the heavier twins enjoyed, which appeared to be growing as the follow-up period increased:




It turns out in the fully adjusted model, the chubby twins had a statistically significant advantage when it came to heart attacks (MIs) and all-cause mortality:


In the fully adjusted model, the odds ratio (OR) for the combined primary end point of MI or death
was 0.75 (95% CI, 0.63-0.91) in heavier compared with leaner twins.


And yes, they did correct for smoking and other potential confounders. In fact here is the cool sensitivity analysis plot for different subpopulations, which several interesting things highlighted:




First, you'll notice that the odds ratio of all the subpopulations are less than 1.0, meaning they favor a lower risk of MIs or death for the heavier twin.


Interestingly, the advantage of being heavy was much more pronounced in twins who were younger at baseline (top highlight), as well as in twin pairs where the baseline weight of the heavier twin was less than 25 and so the skinnier twin was especially skinny, and had a BMI less than 23 (second highlight)


Also paradoxically, the advantage of being heavy was most pronounced in the subpopulation of twins who were both sedentary (which chubbies being 50% less like to have MI or die!), but even among those pairs of twins who both characterized their level of physical activity as "rather much or very much", the heavier twin was on average 24% less likely to have a heart attack or die during follow-up (third highlight).


Perhaps not as surprising, the chubbiness advantage was greater in pairs of twins who both smoked vs. both never-smokers, but even in pairs where both were never-smokers, the heavier twin was on average 14% less likely to have a heart attack or die during follow-up (bottom highlight).


Where things really get crazy and surreal is when the authors stratified by the difference in BMI between the twins. Even as someone who has become skeptical of the ill-effects of being chubby, I figured that at some large BMI difference between the twins the ill-effects of being really fat would kick in and cause the skinny twin to have an advantage over their obese sibling. Boy was I wrong. Here is the graph:




As you can see the advantage in favor of the heavy twin pretty much monotonically increased with the difference in BMI between the two siblings, all the way up to to point when the difference between the BMI was ≥ 7 points, and the average BMI of the heavy twin was 32.4 and the thin twin was 23.6. Incredibly, the very obese twin in these extremely lopsided pairings was on average 58% less likely to having a heart attack or die during follow-up:


In this [bMI difference > 7] subgroup, 10 MIs or deaths occurred during follow-up in heavier
twins compared with 17 such events in leaner twins (unadjusted OR, 0.42; 95% CI, 0.15-1.18)


Notice from the "Lesser Twin Mean BMI" column, it wasn't that in these lopsided pairs the thin twin was emaciated and close to death (perhaps as a result of smoking or latent disease). Nope - the average BMI of the thin twin in the lopsided pairs remained virtually constant (at ~23.7) and it was the heavier twins being fatter and fatter that created the within-twin BMI discrepancy. 


Obviously the number of twins in each BMI-difference subgroup is pretty small, so the confidence intervals are large in the above table, but as I said, the results seem pretty crazy.  Crazy to the point of wondering if these guys might have flipped a sign somewhere in their analysis. But the fact that the study is in JAMA means the peer-review process must have been pretty rigorous and would have caught any obvious mistakes.


What makes it even odder was the fact that all wasn't peaches and cream for the fatter twins. They were nearly twice as likely to report having diabetes during the follow-up:


During follow-up, 345 incident cases of diabetes occurred in
the heavier twins compared with 224 cases in the leaner twins.
After adjustment for physical activity and smoking, the OR of
incident diabetes was 1.94 (95% CI, 1.51- 2.48) in the heavier
compared with leaner twins (Table 2). The risk of incident diabetes
was increased in the heavier twins in most subgroups
(eFigure 1 in the Supplement) and rose with increasing BMI dissimilarity
within each pair (eFigure 2 in the Supplement). Accordingly,
in pairs with BMI discordance of 3.0 or more, the
ORs of incident diabetes in heavier twins compared with the
ORs in leaner twins was greater than 5.0.


So it appears there is a substantial penalty to be paid for being fatter (in terms of increased diabetes risk), which makes it all the more strange that the chubby twins had fewer heart attacks and deaths during follow-up, since having diabetes is a risk factor for both of these outcomes.


Interestingly, and more in line with my intuition, when they ignored the twin relationship and simply analyzed the 8000 subjects individually, they apparently (I can't access the supplemental material to verify) saw a nadir in death and MIs for a BMI around 24, with a j-shaped risk curve for BMIs above or below 24:


In an ordinary nonpairwise cohort analysis of the total twin
cohort (n = 8092), the association between BMI and the primary
end point (death or MI) was j-shaped, with greater risks
of death for individuals with low and high BMIs (eFigure 4 in
the Supplement). In individuals with BMIs of 24.0 or higher
(n = 4626), the risk of MI or death rose by approximately 5%
per 1-unit increase in BMI (hazard ratio (HR), 1.05; 95% CI, 1.02-
1.08); the rate of MI or death during follow-up was approximately
37% higher in obese individuals (BMI, ≥30.0: HR, 1.37;
95% CI, 1.12-1.69) than in normal-weight individuals (BMI, 18.5-
24.9) after adjustment for age, current smoking, and physical


So that sounds a lot more like what you would expect, with increasing mortality above normal weight, and with a modest mortality penalty for being too thin, even after adjusting for age, smoking and physical activity.


So what the heck is going on with the twin analysis? How do the authors interpret their results, and more importantly, explain their results? Here is what they said (my emphasis):


Our results show that the risk of death or MI was
lower in heavier than in leaner twins overall after taking all
available covariates into account. Subgroup analyses indicated
that this lower risk was attributable to effects in the subgroup
with baseline BMIs of 24.9 or lower (ie, in those with normal
BMIs according to the World Health Organization’s
definition). Explanations could include residual confounding
from unknown diseases or general frailty not captured
at baseline in leaner twins. Such influences could also result
in a decrease in BMI before the baseline assessments and is one
explanation put forward to explain the obesity paradox.


So the authors speculate it could be a residual confounder, like a latent disease or previous smoking habit that made the skinnier twin skinnier and also more likely to die or have a heart attack. But even among the never smokers it was better to be heavier, so at least a history of smoking doesn't explain it all. The latent disease explanation also seems suspect, give the following:


The risk of death or MI in heavier compared with leaner twins was constant during follow-up (eFigure 3 in the Supplement). Thus, the exclusion of twins with less than 1 year of follow-up time did not change the odds of the primary end point (unadjusted OR, 0.80; 95% CI, 0.69-0.92).


If it were a latent disease that was making the skinny folks more likely to Die / have a MI, then you'd expect the fatness advantage to dwindle as the duration of the follow-up increased, as the sick skinny folks kicked off early leaving the robust skinny folks to duke it out with their fat siblings. But that wasn't the case. Further:


[W]hen twins with low or normal BMI at baseline were excluded, the odds of MI or death were not higher in subgroups in which the heavier twin was overweight or obese. The results also remained consistent with increasing BMI dissimilarity in twin pairs, including a subgroup in which all high-BMI twins were obese (mean BMI, 35.0).


So even when they threw out twin-pairs where one of the twins is on the skinny side, the fatness advantage remained, i.e. even when both members of a twin pair were pretty hefty or even when both were obese.


Here is the kicker of the whole study (my emphasis):


The results of the present study suggest that lifestyle-obtained higher BMI with no genetic contribution, including direct genetic effects and gene-environment interactions, is not causally related to an increased risk of CVD or mortality.


In other words, if you're fat because of your behavior (i.e. eating a lot and/or not being very physically active), rather than a result of your genetics, you don't have to worry. If anything, if you are fat you'll be at less of a risk of having a heart attack or dying than if you were thinner.


How crazy is that!?


Despite the authors reasonable argument to the contrary, I still lean towards the idea that there must have been some residual confounder that explains the paradoxical results (i.e. for a given set of genes, being fatter almost always means less likely to die).


But then again, this finding is consistent with the "fragility" hypothesis I've mentioned elsewhere (i.e. the flask vs. test tube analogy, and also here), and which the authors also suggest as a possible explanation in the quote I highlighted above. In short, the twins in this study weren't all that old on average, 58 at baseline and 70 at the end of the follow-up. So perhaps the fragility disadvantage of being the thinner twin outweighed the disadvantages of being the fatter twin (i.e. increased risk of diabetes, CVD, etc.) which hadn't quite had time to catch up with (and kill) the fatter twin yet.


Whatever the explanation though, once again we see no advantage, and quite possibly, a significant disadvantage of being thinner when it comes to risk of heart attacks and all-cause mortality during late middle-age.





[1] JAMA Intern Med. 2016 Aug 1. doi: 10.1001/jamainternmed.2016.4104. [Epub ahead of


Risks of Myocardial Infarction, Death, and Diabetes in Identical Twin Pairs With
Different Body Mass Indexes.

Nordström P(1), Pedersen NL(2), Gustafson Y(1), Michaëlsson K(3), Nordström A(4).

Full text: http://sci-hub.cc/10.1001/jamainternmed.2016.4104

Importance: Observational studies have shown that obesity is a major risk factor
for cardiovascular disease and death. The extent of genetic confounding in these
associations is unclear.
Objective: To compare the risk of myocardial infarction (MI), type 2 diabetes,
and death in monozygotic (MZ) twin pairs discordant for body mass index (BMI).
Design, Setting, and Participants: A cohort of 4046 MZ twin pairs with discordant
BMIs (difference >0.01) was identified using the nationwide Swedish twin
registry. The study was conducted from March 17, 1998, to January 16, 2003, with
follow-up regarding incident outcomes until December 31, 2013.
Main Outcomes and Measures: The combined primary end point of death or MI and the
secondary end point of incident diabetes were evaluated in heavier compared with
leaner twins in a co-twin control analysis using multivariable conditional
logistic regression.
Results: Mean (SD) baseline age for both cohorts was 57.6 (9.5) years (range,
41.9-91.8 years). During a mean follow-up period of 12.4 (2.5) years, 203 MIs
(5.0%) and 550 deaths (13.6%) occurred among heavier twins (mean [sD] BMI, 25.9
[3.6] [calculated as weight in kilograms divided by height in meters squared])
compared with 209 MIs (5.2%) and 633 deaths (15.6%) among leaner twins (mean [sD]
BMI, 23.9 [3.1]; combined multivariable adjusted odds ratio [OR], 0.75; 95% CI,
0.63-0.91). Even in twin pairs with BMI discordance of 7.0 or more (mean [sE],
9.3 [0.7]), where the heavier twin had a BMI of 30.0 or more (n = 65 pairs), the
risk of MI or death was not greater in heavier twins (OR, 0.42; 95% CI,
0.15-1.18). In contrast, in the total cohort of twins, the risk of incident
diabetes was greater in heavier twins (OR, 2.14; 95% CI, 1.61-2.84). Finally,
increases in BMI since 30 years before baseline were not associated with the
later risk of MI or death (OR, 0.97; 95% CI, 0.89-1.05) but were associated with
the risk of incident diabetes (OR, 1.13; 95% CI, 1.01-1.26).
Conclusions and Relevance: In MZ twin pairs, higher BMI was not associated with
an increased risk of MI or death but was associated with the onset of diabetes.
These results may suggest that lifestyle interventions to reduce obesity are more
effective in decreasing the risk of diabetes than the risk of cardiovascular
disease or death.

DOI: 10.1001/jamainternmed.2016.4104
PMID: 27479111

Share this post

Link to post
Share on other sites

So how does this jive with the paper your paper refers to, Dean, a randomized control trial of weight loss?


Cardiovascular effects of intensive lifestyle intervention in type 2 diabetes.
Look AHEAD Research Group, Wing RR, Bolin P, Brancati FL, Bray GA, Clark JM, Coday M, Crow RS, Curtis JM, Egan CM, Espeland MA, Evans M, Foreyt JP, Ghazarian S, Gregg EW, Harrison B, Hazuda HP, Hill JO, Horton ES, Hubbard VS, Jakicic JM, Jeffery RW, Johnson KC, Kahn SE, Kitabchi AE, Knowler WC, Lewis CE, Maschak-Carey BJ, Montez MG, Murillo A, Nathan DM, Patricio J, Peters A, Pi-Sunyer X, Pownall H, Reboussin D, Regensteiner JG, Rickman AD, Ryan DH, Safford M, Wadden TA, Wagenknecht LE, West DS, Williamson DF, Yanovski SZ.
N Engl J Med. 2013 Jul 11;369(2):145-54. doi: 10.1056/NEJMoa1212914. Epub 2013 Jun 24. Erratum in: N Engl J Med. 2014 May 8;370(19):1866.
PMID: 23796131
Free PMC Article



Share this post

Link to post
Share on other sites

Al, from the discussion section of the paper you linked to showing no effect of weight loss intervention on cardiac events for overweight people with diabetes...


Another possibility is that sustained weight losses larger than those achieved in Look AHEAD may be required to reduce the risk of cardiovascular disease. In this regard, it is noteworthy that the differential weight loss between the two trial groups averaged 4% over the course of the study but only 2.5% at the end.


There are several limitations to these findings. We used a specific lifestyle intervention that focused on achieving weight loss through caloric restriction and increased physical activity. It is unclear whether an intervention focused on changes in dietary composition, for example the Mediterranean diet,18 might yield different outcomes.


Both of those discussion points are likely relevant, but I think the second is undoubtedly more critical.  The people in the study lost a small amount of weight but they were likely still eating the same crap that made them sick in the first place.  Eating a healthier diet obese people ought to be able to lose a much higher percentage of bodyweight than they accomplished in this study.

Share this post

Link to post
Share on other sites



You terse message once again leaves it very unclear what you are suggesting when you say:

So how does this jive with the paper your paper refers to, Dean, a randomized control trial of weight loss?


Is this some kind of joke or sarcasm Al?  In fact, it seems to me, as Todd points out, the paper you point to jibes almost perfectly with the twins study I referencs. 


In the twins study, being heavy was not associated with an increased risk of heart attacks or overall mortality, despite the fact that being heavy was associated with an increased risk of diabetes.


In the study you cite, being obese and having diabetes was not associated with an increased risk of heart attack or CVD-related mortality, relative to being thinner (i.e. after intensive intervention to lose weight the subjects didn't experience fewer heart attacks or CVD death).


In both studies, being heavier, and even having diabetes as a result, was not associated with a worse outcome when it came to heart attacks or mortality (at least CVD mortality).


Al, do you see a contradiction between these two paper that I'm missing? If so, please elaborate rather than employing your usual cryptic style and forcing us to guess what you mean.



Share this post

Link to post
Share on other sites

In both studies, being [my emphasis] heavier, and even having diabetes as a result, was not associated with a worse outcome when it came to heart attacks or mortality (at least CVD mortality).



Hey Dean!


Fortunately, being is not becoming, to put it tersely (though quite correctly): none of this bears more than highly speculatively on the decision to go on CR (or to go off CR), for all the reasons discussed a zillion times previously.


- Brian

Share this post

Link to post
Share on other sites



Great to hear from you! I hope all is well, and you are settling in to your new home - Boston area IIRC? My wife's family lives in Worcester (actually Auburn) not too far from Boston. Very nice part of the country, except in the winter when the snow becomes really heavy (literally and psychologically!). Speaking of becoming heavy, you wrote:


Fortunately, being [heavier? - DP] is not becoming [heavier? - DP], to put it tersely (though quite correctly): none of this bears more than highly speculatively on the decision to go on CR (or to go off CR), for all the reasons discussed a zillion times previously.


That is a bit terse and cryptic. Who do you think you are, Al? ☺ Is it too much to ask for you to elucidate on what exactly you meant? Maybe we have discussed it a zillion times, and if so perhaps you can point to some of those discussions in support of CR and contra to the evidence discussed in the Will Serious CR Beat a Healthy, Obesity-Avoiding Diet and Lifestyle? thread, and the posts that discussion links to. To me the evidence pretty clearly suggests the answer to that question is "almost certainly not". But I take it you still think otherwise? I'd love for you (or anyone) to argue the opposing position on that thread.


Regarding being heavy vs. becoming heavy, the study under discussion (PMID: 27479111) does address the issue, and contrary to (my interpretation of) your speculation that the two aren't the same, at least in terms of MI and all-cause mortality risk, it didn't seem to matter whether the heavier twin was fat(ter) from childhood or grew fatter during adulthood. Being fatter meant less risk of death, period, regardless of when they packed on the pounds.


More specifically, they started tracking the 4000 twins between 1998 and 2003, and finished the study in 2013 (avg follow-up 12.5 years). But they also had data on the BMI of the twins from a questionnaire they filled out in 1973 (those Swedes do a lot of tracking of their citizens - lucky for us!). Interestingly, in 1972 (28 years before the study began, when the average twin in study was only 30), the average BMI of the twins who would go on to be fatter was virtually identical to the twin that would stay thin (21.7 vs. 21.3, respectively). In other words, when they were younger, most of the twins who would eventually become fat(ter) 28 years later were actually quite thin.


So if you're terse statement "being ≠ becoming" when it comes to obesity/overweight, in response to the conclusion I drew from these study, namely that:


"being [brian's emphasis] heavier, and even having diabetes as a result, was not associated with a worse outcome when it came to heart attacks or mortality (at least CVD mortality).


was meant to suggest that becoming chubbing is bad for mortality while simply being heavier isn't bad, it seems the data from this study contradicts you. Why? Because on average, the chubbier twins weren't chubby when young, but instead became fat(ter) after age 30, and nevertheless enjoyed lower rates of heart attacks and mortality many years later. So this study suggest becoming chubby in middle age conveys survival benefits later in life, between the average ages of 58 and 70 (an important qualifier, I realize). In fact, the authors state exactly this point about becoming fat being no worse than being fat for all of one's life:


Increases in BMI during approximately 30 years before baseline were not associated with the outcome in this model (OR, 0.97; 95% CI, 0.89-1.05).


Does this analysis address the point you were trying to make Brian, or am I misinterpreting your distinction between being and becoming?



Share this post

Link to post
Share on other sites

Maybe Brian is saying it could be the question of whether weight status changed intentionally versus unintentionally, Dean.

Share this post

Link to post
Share on other sites


Maybe Brian is saying it could be the question of whether weight status changed intentionally versus unintentionally, Dean.


Perhaps that is what Brian meant, but it would seem irrelevant to this discussion if that is the case.


It's almost certain that the weight gain by each of the heavier twins in this study was unintentional - very few people want to become chubby as they age. Nonetheless the twins who (likely unintentionally) gained weight and became chubby lived longer. So how exactly would the relatively1 intentional nature of the weight loss associated with CR have any bearing on the results of this study, suggesting that gaining weight is beneficial for mortality?


The reason for weight change would certainly be relevant if we were talking about the shortening of life associated with weight loss - where the reason for weight loss (i.e. intentional vs. unintentional) might indeed make a difference for health outcomes. But as I described above, neither of the twins in the study was losing weight, at least on average. That is, even the thin twins drifted from a BMI of ~21 at age 30 to ~24 at age 58 when the study began. It's just that the BMI of the fatter twins drifted higher faster, and (perhaps as a result) they lived longer.





1 'Relatively' because weight loss is usually considered, esp by Brian, as an unintended and unfortunate side effect of CR...

Share this post

Link to post
Share on other sites

Thanks, Dean,


I guess that the way I was looking at it, generally healthy people have a healthy appetite.  Hence the obesity epidemic.  Heck, I would weigh a lot more if I did not restrict mine.  Not having a healthy appetite seems to me to go against our nature.  Having an unhealthy lack of appetite and hence less weight gain might be exactly that -- unhealthy due to lack of health.

Share this post

Link to post
Share on other sites


I guess that the way I was looking at it, generally healthy people have a healthy appetite.  Hence the obesity epidemic.  Heck, I would weigh a lot more if I did not restrict mine.  Not having a healthy appetite seems to me to go against our nature.  Having an unhealthy lack of appetite and hence less weight gain might be exactly that -- unhealthy due to lack of health.


Who said the thinner twins didn't have a healthy appetite? They didn't lose weight, in fact they gained weight over the years. Just not as much as their chubbier twin. I think it is a pretty big stretch to suggest a group of people (i.e. the thin twins) whose BMI drifted up from 21 to 24 as they grew older are candidates for "having an unhealthy lack of appetite" suggesting a "lack of health".


It seems more likely that they simply showed more dietary restraint and/or had a more active lifestyle than their chubbier twins. Paradoxically, the thinnies seem to have paid a price for what would normally consider healthier habits & weight between the ages of 58 and 70, in the form of more heart attacks and higher mortality.



Share this post

Link to post
Share on other sites



Don't forget that physical age is not necessarily the same as cronological age --  and perhaps those on CRON are at a younger physical age than our cronological age.


Perhaps CRONnies differ from ad-libbers wrt optimal level of thinness at a given cronological age.


  -- Saul 

Share this post

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now