Jump to content

How Much CR is Enough? Or - What's Lost in (Rodent) Translation?

Dean Pomerleau

Recommended Posts


It has been thought that CR lifespan benefits are likely to linearly increase with degree of CR, up to very severe CR, based on rodent data like the famous Weindruch study [2], data from which is highlighted on the CR Society home page, and reproduced in this graph:


As you can see, compared with ad-lib fed mice, 25% CR provided some benefits. But severe CR, where mice ate 55% or even 65% less food than the AL group, lived much longer than the 25% restricted group. And note that this benefit was seen across the board. No sign of early mortality in the severely restricted mice - the survival curves for the 55% & 65% CR mice are completely to the right of the 25% CR mice survival curve, which in turn is completely right of the AL mice survival curve. From this it looks like the more CR the better.

But the recent NIA/Wisconsin monkey data doesn't look so cut and dried. In fact, one plausible interpretation of the NIA monkey results (as we've discussed elsewhere, e.g. here), is that if eating a healthy diet, restricting calories beyond that required to avoid obesity doesn't provide additional longevity benefits.

So how much CR is required to garner the benefits? Perhaps this new rodent study [1] sheds some light on the topic, as far as rodent studies can be extrapolated to humans (more on that below).

What they did was restrict the calories of a common strain of rats (F344 - all male) in two groups - 10% and 40% restriction relative to AL-fed controls. Then they let them all live out their natural lifespan to see how longevity and causes of death compared between the 3 groups (AL, CR25, CR40). Here are the two important graphs and the most important table from the paper, illustrating the main lifespan results, image originally from this analysis of the study on crvitality.com, red annotations are mine.

As you can see from the survival curves and Table 1, both the CR10 and CR40 groups had mean longevity that was significantly longer than the AL group, but they did not significantly differ from each other, at 796 days for AL, 918 days for CR10 and 947 days for CR40. One way to look at this result is that on average, restricting enough to avoid obesity (i.e. CR10), resulted in a 15% increase in mean lifespan. Bumping it up to severe restriction (CR40) bought the rats an additional (non-significant) 4% of mean lifespan relative to obesity-avoiding 10% CR (15% vs. 19% increased lifespan, respectively). That seems like a pretty modest (if any - given its non-significance) lifespan gain for a lot more severe restriction.

On the other hand, the longest lived 10% of the CR40 rats did indeed significantly outlive the CR10 rats, as can be seen from both the graph on the left above, and from the longevity table. Quantitatively, the longest lived 10% of the CR10 rats enjoyed a 37% longer life than the average AL rat, while the longest lived 10% of the CR40 rats enjoyed a 54% longer life than the average AL rat, a gain of 17% in lifespan from CR10 to CR40. That sounds pretty good, as long as you are one of the lucky, few longest-lived CR40 rats. From the survival graph, it looks like it was only the top ~20% of CR40 rats that enjoyed any longevity benefits relative to the CR10 rats, and it looks like this was almost (but perhaps not quite) fully counterbalanced by the bottom 20% of the CR40 rats dying before the CR10 rats (see red annotations on graph above for places where CR40 had an advantage and a disadvantage for lifespan relative to CR10). The Gompertz graph on the right above shows that rate of aging in the CR40 group appeared to be slowed relative to the other two groups, which didn't appear to differ from each other.

In short, if you're lucky (and the rat data in [1] translates to humans - see below), this data suggests that going for severe CR might gain you an extra ~14 years relative to obesity avoiding mild CR. But if you're unlucky, severe CR might kill you off early by almost an equivalent amount. If you are average (mean or median), you can expect to do about the same in terms of lifespan with either mild CR (i.e. enough to avoid obesity) or severe CR.

But wait a minute - one thing Michael always cautions about is the need to make sure the animal husbandry in a study is good, and the strain being used isn't one of those f**cked up ones that dies early, or is particularly fragile and therefore not a good candidate for CR research.

From [3] (and from memory), it is apparent that the F344 strain of rats is one that is commonly used in CR research. Below is the mortality curves for F344 rats fed AL or 40% CR diets from [3]. The mortality curve for the AL-M (ad lib fed male) group looks about the same in [3] as in study we've been looking at [1], with median lifespan around 800 days and max around 1000 days. But if anything the 40% CR group of F344 (male) rats looks better in [1] than in [3] below - where (eyeballing it) the median lifespan looks to be around 900 days (vs. nearly 1000 in [1]) and max is around 1150 days (vs. 1400 in [1]). Note also that unlike in the mortality curves above from [1], the mortality curve for the 40% CR rats from [3] spends time below the AL curve from around 400 to 600 days (see red annotations below), representing early mortality in the CR rats, likely due to the stress of severe CR.

So it doesn't look like there was any problem with strain or husbandry in [1] that might invalidate its results. In fact, based on the results of [3], it would seem that [1] may be overestimating the (max) longevity upside of severe CR, and underestimating the early mortality risk of severe CR, making severe CR in [1] seem like a better deal relative to a milder, obesity-avoiding level of CR that it actually may be. Unfortunately 10% CR wasn't tested in [3] - only AL and CR40.

The author's of [1] seemed surprised by the modest benefits of severe CR40 relative to mild, obesity-avoiding CR10, saying:

The data from this study clearly demonstrate that a 10% restriction of food significantly increases the lifespan of male F344 rats and, surprisingly, that the increase in lifespan is comparable to what was observed for rats that were restricted 40%. These data were surprising because of the general view that increasing the level of DR up to 40% would result in a continuous increase in lifespan, as reported by Weindruch et al.[2] for female C3B10RF1 mice in which a significant increase (over 20%) in the mean survival occurred between approximately 25% and 55% DR. ... These data in combination with the data from Duffy et al.,[4] which reported that feeding rats 10% and 25% DR was as effective as 40% DR in reducing the early mortality of male Sprague–Dawley rats, demonstrate that the lifespan of certain strains of rats and mice does not increase linearly up to 40% DR. Most of the extension of lifespan appears to be achieved by levels of DR much lower than 40% DR.

But what about translatability of such rodent data to humans? - a topic currently under discussion over on this thread, where just last night Michael said:

If you want to go beyond avoidance of specific diseases and premature incidence or accelerated progression of aging phenotypes and diseases, to test potential interventions in aging that might move the ball on one's "particular human longevity," you simply have to rely very substantially on cohorts of much shorter-lived model organisms. And for various reasons, the best model organisms available to us remain laboratory rodents [my emphasis - Dean]. There's a reason why they're still the go-to critters for all biomedical research.

So how good a model of human aging is the natural aging of these critters? Not very good it seems to me, if this study is typical of rodent causes of death (which I think it is), based on this table from the full text:


As you can see from the table above, like most rodents, a clear majority died from various forms of cancer, with leukemia being by far the most common single cause. As we all know (and which is graphically illustrated here) cardiovascular disease in its various forms is the #1 killer of humans (at least in the US and other developed nations) - cancer only accounts for 23% of human death. Further, leukemia is only the 6th most common cancer killer in people, accounts for only 5% of all cancer deaths and only ~1% of all causes of death in the US.

Based on this cause-of-death mismatch, these leukemia-prone rats, and cancer-prone rodents in general, don't seem like a very good model of the "diseases of aging" that kill people. As a corollary, the fact that CR40 significantly reduced leukemia deaths relative to CR10 in [1] seems unlikely to "move the bar" for human longevity, even if these results would translate directly to people.

So the takeaway message from this one seems to be that even CR rodent results, which have always been the best data we have to support large additional lifespan benefits of 'serious' CR relative to simply avoiding obesity, may be open to question both in terms of the magnitude of the benefit, and their translatability to humans.

Michael, if I'm missing something I'm sure I'll hear from you. :)xyz


[1] 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).

Free full text: http://onlinelibrary.wiley.com/doi/10.1111/nyas.12982/epdf

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
[2] Weindruch R, et al. (1986). "The retardation of aging in mice by dietary restriction: longevity, cancer, immunity and lifetime energy intake." Journal of Nutrition, April, 116(4), pages 641-54.

[3] J Gerontol A Biol Sci Med Sci. 1999 Nov;54(11):B492-501.

Growth curves and survival characteristics of the animals used in the Biomarkers
of Aging Program.

Turturro A(1), Witt WW, Lewis S, Hass BS, Lipman RD, Hart RW.

PMID: 10619312

The collaborative Interagency Agreement between the National Center for
Toxicological Research (NCTR) and the National Institute on Aging (NIA) was aimed
at identifying and validating a panel of biomarkers of aging in rodents in order
to rapidly test the efficacy and safety of interventions designed to slow aging.
Another aim was to provide a basis for developing biomarkers of aging in humans,
using the assumption that biomarkers that were useful across different genotypes
and species were sensitive to fundamental processes that would extrapolate to
humans. Caloric restriction (CR), the only intervention that consistently extends
both mean and maximal life span in a variety of species, was used to provide a
model with extended life span. C57BI/6NNia, DBA/2JNia, B6D2F1, and B6C3F1 mice
and Brown Norway (BN/RijNia), Fischer (F344/NNia) and Fischer x Brown Norway
hybrid (F344 x BN F1) rats were bred and maintained on study. NCTR generated data
from over 60,000 individually housed animals of the seven different genotypes and
both sexes, approximately half ad libitum (AL) fed, the remainder CR.
Approximately half the animals were shipped to offsite NIA investigators
internationally, with the majority of the remainder maintained at NCTR until they
died. The collaboration supplied a choice of healthy, long-lived rodent models to
investigators, while allowing for the development of some of the most definitive
information on life span, food consumption, and growth characteristics in these
genotypes under diverse feeding paradigms.

[4] Duffy, P.H., J.E. Seng, S.M. Lewis, et al. 2001. The effects of different levels of dietary restriction on aging and survival in the Sprague–Dawley rat: implications for chronic studies. Aging Clin. Exp. Res. 13: 263–272.

Link to comment
Share on other sites

Since Michael hasn't (yet) taken the bait either here or on the related thread about the Ralston Purina dog CR study, I figure I'll goad him some more  :)xyz.


Recall, that in a nutshell, my contention is that the dog CR data, the monkey CR data, and now the rodent data from PMID 26695614 discussed in this thread, all support the idea that pushing CR beyond that degree needed to stay slim will only provide negligible longevity benefits to the average member of a mammal species, and comes with significant risk of early mortality, even when started at an early age, to say nothing of later in life.


In Michael's very thorough analysis of the monkey data, he acknowledges this as a pretty reasonable interpretation of the UWash & NIA primate CR results, saying:


A straightforward reading of the two nonhuman primate CR studies, then, is that in rhesus macaques, the relationship between energy intake, body weight, and lifespan is the commonsensical one, against which the rodent CR phenomenon stands as such a stark contrast: that overweight and excess adiposity are bad for one's health and prospects for long life, but that some normative "healthy" anthropometry is optimal, with diminishing returns at best as energy intake and body weight are progressively reduced beyond that juste milieu. Indeed, skeptics of the human translatability of CR have long argued that the weight loss that is associated with CR appears to only be salutary to health within a relatively narrow range. They argue instead that further limiting energy intake and adiposity will lead to progressively less marginal benefit, especially in light of the uniquely metabolically deranging effects of visceral adipose tissue, which is preferentially lost early in the process weight loss, whether achieved by diet or exercise.(69) Such skeptics also point to the importance of maintaining lean mass — both muscle and bone — for preserving health during aging, and to the large number of epidemiological findings (eg. (70,71)) suggesting a J-shaped or U-shaped relationship between body mass index (BMI) and mortality, although the relevance of these findings to the Calorie restriction phenomenon is dubious.*
In this interpretation, the slight restriction imposed on NIA control animals, leading to an energy intake and body weight that was intermediate between those of WNPRC's ad libitum and restricted groups, was sufficient to achieve or closely approach the point of diminishing health and longevity gains, and a further restriction from this point in the CR group therefore yielded no further extension of lifespan.
This explanation of the discrepancy in the effects of CR as compared to internal control animals in the WNPRC(5) and NIA(6) studies has much to offer. It is conceptually straightforward; it is consistent the major findings of the two studies, and with an important body of research in humans; it fits with some models of the postulated evolutionary basis for the slow-aging phenotype of CR; and is not exclusive of some of the other explanations we have explored. And, as we shall see in the next section, it can also provide a consistent explanation for several differences in health and metabolic outcomes between the two studies, and accommodate a broader body of research on diet and metabolism in nonhuman primates.
So Michael - given that PMID 26695614 shows that the "stark contrast" between rodents and primates might not be stark after all, but instead may amount to the same sort of "diminishing health and longevity gains" in rats beyond 10% CR that we saw in the primates, doesn't the "diminishing returns" hypothesis seem even more plausible today than when you wrote the primate study review in 2013?
Link to comment
Share on other sites

Apparently, and perhaps not surprisingly, PMID 26695614 is not the first study to show a mild level of CR (10% restriction relative to ad lib) in rodents provides a large fraction of the survival benefits of more severe CR. I came across this older (2001) study [1] of another common rat strain (Sprague Dawley rats), which tested various levels of CR. They found:


The average lifespan of AL rats was 115 <sic> months [they mean weeks].

At 104 weeks on study (110 weeks of age), the

survival rate for the AL and 10%, 25%, and 40% DR groups was 63.4, 87.5, 87.5,
and 97.5%, respectively. The largest increase in survival (24.1%) occurred
between AL and 10% DR, indicating that very low levels of DR have a significant
effect on survival.


This further supports the idea articulated above, that most of the benefits of CR, at least for the average individual, can be had via modest, 10% CR. Unfortunately all the rats in this study were sacrificed early to study their organs, rather than allowing them to live out their natural lives, so there isn't data on total lifespan, just survival to 104 weeks.





[1] Aging (Milano). 2001 Aug;13(4):263-72.

The effects of different levels of dietary restriction on aging and survival in
the Sprague-Dawley rat: implications for chronic studies.
Duffy PH(1), Seng JE, Lewis SM, Mayhugh MA, Aidoo A, Hattan DG, Casciano DA,
Feuers RJ.
Author information: 
(1)Division of Genetic and Reproductive Toxicology, National Center for
Toxicological Research, FDA, Jefferson, AR 72079, USA. pduffy@nctr.fda.gov
Comment in
    Aging (Milano). 2001 Aug;13(4):261-2.
    Aging Clin Exp Res. 2002 Apr;14(2):152-4.
A study was undertaken to determine the effects of incremental levels of dietary 
restriction (DR) in rats. Survival, growth, reproductive, and dietary intake (DI)
variables were monitored in a chronic study in which male Sprague Dawley (SD)
rats (NCTR colony) were fed their ration ad libitum (AL), or DR. The main
objectives were to determine if low levels of DR could be used to increase the
survival rate of SD rats in the chronic bioassay, and to identify the survival
characteristics of a long-lived SD rat strain (NCTR colony). The average life
span of AL rats was 115 months. At 104 weeks on study (110 weeks of age), the
survival rate for the AL and 10%, 25%, and 40% DR groups was 63.4, 87.5, 87.5,
and 97.5%, respectively. The largest increase in survival (24.1%) occurred
between AL and 10% DR, indicating that very low levels of DR have a significant
effect on survival. Whole-body, liver, prostate, and epididymis weights and body 
length were decreased by DR, whereas brain weight, testicular weight, and skull
length were not altered by DR. Rats from the NCTR colony were found to be ideal
for chronic studies because they are much longer-lived than other SD stocks.
Although the 104-week survival rate for these SD, non-obese AL rats exceeds the
FDA's "Redbook" survival guideline (> 50%) for chronic bioassays, the use of DR
is advocated because it reduces individual variability in body weight.
PMID: 11695495
Link to comment
Share on other sites

  • 2 months later...

A possibly interesting parallel is wrt. exercise vs all-cause mortality. There is of course no consensus on the amount/intensity of exercise needed to get optimal effects vs all-cause mortality, but there does appear to be something of an emerging consensus that the vast majority of the benefit of exercise is going from completely sedentary to modest amounts of exercise, and doing more the benefits start flattening pretty drastically, and for higher levels may even reverse. That first fraction of exercisers can be called "avoidance of sedentary behavior" where you get most of the benefits parallel to "avoidance of obesity/overweight" wrt. optimum caloric intake vs all-cause mortality. The question becomes, perhaps there is no such thing as slowing down aging or pausing aging, and most life span benefits are in "morbidity avoidance" (with suitably broad definition of what constitutes morbidity).

Link to comment
Share on other sites


This topic is now archived and is closed to further replies.

  • Create New...