Caloric Restriction and Fasting - does it work?

Caloric Restriction and Fasting - does it work? At least in mice?

It is unfortunate that I have to ask this question in 2021, but that's life - full of surprises, some of them unpleasant. Caloric Restriction (CR) is known to extend lifespan of many mouse strains and of other species like spiders, butterflies and dogs (2). This assumption has been the cornerstone of modern biogerontology and has informed, if not driven, most of the research that has been performed in our field so far (4). However, more and more studies show that the CR state is fragile, easily disturbed, and not as robust as we once thought. In particular, there have been studies in wild-derived mice, the ILSXISS inbred panel, and now this work by Lamming, casting doubt on the CR hypothesis.

You may ask what is the CR hypothesis? When I say "CR works" or "does not work" I mean something specific. No one cares if CR is a lab curiosity that helps a couple dozen of strains live longer. We study CR because we hope it will be robust across species and because we're hoping CR in humans would either lead to non-zero lifespan (LS) extension or would pave the way for future CR-mimetic therapies (1). Perhaps we should call this the strong CR hypothesis. We know the weak CR hypothesis is true as something interesting is happening in these animals and life extension is real even if not robust across strains and species.

Today we will critically evaluate the "strong" CR hypothesis in light of a new publication by Heidi Pak from the Lamming lab (5). Without a long introduction, let me say that there is an ongoing debate about the ability of different "time-restricted" feeding protocols to extend mouse lifespan independently of CR. To make the confusion perfect different protocols and concepts go by easily distinguishable names like every other day feeding, intermittent fasting, time restricted feeding (let's call all of these fasting*). The major unresolved question is as follows. Are any, all or some benefits of CR due to fasting*?

For a long time I was a strong believer that only calories matter, but now I am not so sure. Right now I can imagine that some of the benefits of CR are indeed due to fasting.

A key observation, at least for me, that led to renewed interest (or reframing) of fasting* research was the eating behaviour of our little gluttonous furry friends. As pointed out by Pak et al: "CR animals rapidly consume their entire daily meal within ~2 h, and then fast for ~22 h until their next meal". It took me a while until I came to recognize this semi-trivial observation for its importance. In practical terms this also means we have two possibilities of studying fasting*: reduce calories while modifying the window during which the food is consumed, or keep calories constant while decreasing the same window (as was done more often in the past). The authors use both strategies, although the first approach takes center stage.

Abstract - TL;DR
Emerging data suggests that fasting* in mice contributes to the health benefits we see in animals that eat less. Inconsistent or flawed study designs, however, still limit the conclusions we can draw. While Pak et al., as discussed here, strengthens the fasting hypothesis, the use of dietary dilution - which I speculate is toxic - remains a major methodical limitation of this study. Nevertheless, it is intriguing to see that a calorie-restricted diet diluted with 50% cellulose fails to improve lifespan and mouse frailty. If true, this might weaken the CR hypothesis yet again because it shows that we do not even have a handle on the most basic aspects of a decade-old protocol, but other than that it is not a major attack on the strong CR hypothesis. CR could be real even if we have to rename it to "chronic intermittent fasting".

The Lamming paper: fasting* could mediate the benefits of CR

We find that fasting is required for CR-induced changes in insulin sensitivity and fuel selection. Additionally, fasting alone without reduced energy intake is sufficient to recapitulate the metabolic phenotypes and transcriptional signature of a CR diet. Finally, we show that fasting is required for CR-induced improvements in glucose metabolism, frailty and lifespan in C57BL/6J male mice. Our results overturn the long-held belief that the beneficial effects of a CR diet in mammals are mediated solely by the reduction of caloric intake, and highlight fasting as an important component of the metabolic and geroprotective effects of CR.

A  long-held belief overturned, or is it? (13)

In total five different dietary regimes were studied by Pak et al. 2021 (5). The authors used two special dietary approaches to reduce the effect of CR-induced fasting. One involves dilution of the standard (ad libitum) diet with 50% indigestible cellulose (diluted AL) and the other a contraption that dispenses food in regular intervals during the dark period when mice are active ( Both lead to 30% restriction, but should allow the mice to snack throughout the day. Then we also have standard CR and AL conditions. On top of that, in some experiments, we have time-restricted AL (, i.e., mice ate only during a 3h window (7).

All restricted diets led to improved glucose tolerance and reduced bodyweight in B6 mice but only CR proper (with the associated fasting periods) led to improved insulin tolerance. In contrast, in DBA/2J mice CR did not lead to changes in insulin tolerance and there were no differences between feeding groups. This is an interesting choice of secondary mouse model since this is one of the few classical strains where CR fails to extend lifespan.

In addition the authors performed metabolomics which is hard to interpret because we know little about the expected changes in contrast to transcriptomics. As far as I can tell, looking at these PCA plots in the supplementary data diluted AL is generally different from AL and again different from CR. To me they look distinct even in the liver, although the authors say the following: "CR-fed mice had a distinct metabolomic signature from AL-fed mice [in muscle]; however, unlike in the liver, Diluted AL-fed mice had a distinct signature from AL-fed mice". animals were not studied for some reason. Not clear this strengthens the main point.

Then following that, the authors switch to studying time-restricted AL feeding ( They perform transcriptomics and state that "Fasting and calorie restriction result in highly similar transcriptomic signatures in liver and white adipose tissue" without providing any clear quantification. But assuming that it is true, what does it tell us? Shouldn't this analysis include diluted AL or to show that it produces a response similar to AL? I am very confused by this choice. The authors also noted that led to a small caloric deficit. This issue has plagued most of the published fasting* research and was often used by critics as an argument to discount the effects of fasting.

Most importantly, the fact that fasting drives some of the metabolic changes during CR is not a sufficiently strong argument to imply that it also drives the lifespan benefits of CR. As we have seen, many of the candidate (metabolic) changes during CR make no contribution to lifespan or a very minor one. For example, the quantitative role of insulin sensitivity during CR remains questionable, among other reasons, because low insulin by itself fails to robustly extend lifespan (6). Similarly, in the ILSXISS replication effort, there was no relationship between glucose tolerance under CR and lifespan extension.

Dilution induced CR: impact on lifespan and healthspan

Knowing this, the authors move on to hard outcomes like healthspan and lifespan, but this is where it gets interesting. "[the] regimen is similar to a regimen previously shown to extend the lifespan of mice", yet the authors refrain from using it in this study. Sure, cost is the obvious limiting factor because the setup appears technically challenging at first, but either way this is an important shortcoming (as the authors acknowledge).

When someone explains dietary dilution to me, my immediate gut reaction is "this must be toxic". I mean, come on, have you ever tried eating Walden Farms zero-calorie sauces? I experimented with these once and quickly found out how hard it is to suppress your gag reflex confronted with something so otherworldly. Even though some people swear by it, I just couldn't. Now imagine half of your diet, almost 1kg, per day is made up of these sauces! Jokes aside, it is a safe intuition that (3) massive dietary dilution could be unhealthy and this is something the authors need to actively disprove.

The authors (or reviewers), to their credit, do realize this. They successfully rule out some forms of toxicity, for example, "when accounting for the amount of cellulose consumed, Diluted AL-fed mice absorbed digestible macronutrients similarly to AL-fed mice [but micronutrients were not investigated]" and gut barrier integrity measured using fluorescent dextran was also comparable between the groups.

Other signs however point in the direction of dietary-dilution-is-killing mice. The frailty profiles between AL and diluted AL are not similar although they should be if the null hypothesis is true: "the specific deficits of AL-fed and Diluted AL-fed mice that contributed to their equivalent frailty varied, with Diluted AL-fed mice developing kyphosis, and AL mice having declining grip strength and body composition". To be fair that assay is probably too noisy for such a strong assumption.

Perhaps more importantly, if you ask me, the diet seems to be outright killing the mice. Diluted AL does not merely offset the benefits of CR, it actually shortens the mean lifespan of mice. As if this was not problematic enough, the difference emerges at around 250 days, which is way too early to have anything to do with aging. Finally, "the incidence of cancer in Diluted AL-fed mice was low, perhaps due to their shorter lifespan" - which is also consistent with a benefit of diluted AL, and induced CR, on cancer incidence that is overridden by acute toxicity.

Another indication that something is off with the diluted AL diet is the large behavioral change of the mice, who appeared rather unhappy with the diet and ended up "shredding" their pellets (looking for some other food source? [8]) These mice also lost even more weight (including lean and fat mass) than mice subjected to 30% CR.

While this is undoubtedly a great paper I remain a bit stumped by the many "convenient" choices. Diluted AL was used instead of which would be the superior protocol that still reduces the period of fasting. Several experiments with restricted AL mice were performed that would have benefitted from an / diluted AL control. Overall while the authors advance an interesting hypothesis, I do think the paper fails to upset the established consensus in the way the ILSXISS inbred panel did. If the CR benefits were only or predominantly due to fasting this would be indeed a huge sensation. Right now we do not know.

I would have preferred if the authors looked at other markers of the CR state like IGF-1 and also hunger signaling. The non-physiological nature of diluted AL raises a lot of intriguing issues in relation to hunger. One far fetched hypothesis is that calorie sensing relies more heavily on sensing the "volume" or "mass" of the consumed food rather than the calories itself (9). If this were true, it could explain why diluted AL mice performed so poorly, since the volume of the consumed food would be quite high. If the food was diluted by 50% then the mice need to raise their intake by roughly 40% to reach a caloric intake of 70% relative to ad libitum. That is a tremendous increase.

This reminds me of a story: I once got a rather terse mail when I suggested to a researcher that their data was interesting but strangely at odds with my intuition and thus felt incomplete, scoffing "the data is the data", but this could not be farther from the truth in biology. All our data is subject to implicit and explicit biases, and we test so few conditions that there is a lot of room for genuine disagreement and professional intuition. This is exactly such a case, this study seems "wrong", not in the sense of being incorrect, but rather like a piece of a puzzle that is annoyingly at odds with the other pieces. Reconciling such inconvenient studies can be very worthwhile.

The fact that Solon-Biet (10) arrived at a similar conclusion also using dietary dilution, while the only known study of showed lifespan extension (11), may be part of the answer. The fact that time restricted feeding leads to smaller health benefits than CR is consistent with the idea that only some of the CR benefits are due to fasting (12) and inconsistent with the data shown by Lamming and Solon-Biet using dilution approaches.

Ultimately, we may be dealing with more than one puzzle here, with pieces yet to be found and assembled.

References and notes

*) fasting in this article is an umbrella term and shorthand, where appropiate the exact protocol is mentioned

1. I am starting to wonder if CR-mimetics and anti-anabolic pathways could work in humans even if CR is not robust. It's certainly possible but a failure of CR would constitute a major downgrade for the Bayesian prior here.

2. not all data is as experimentally rigorous as one would hope, though.
Nakagawa, Shinichi, et al. "Comparative and meta‐analytic insights into life extension via dietary restriction." Aging cell 11.3 (2012): 401-409.

3. part of the intution goes as follows we know that (ultra-)processing of foods is generally a problem as are other extraordinary deviations from the diet an organism evolved to deal with. We also know that not all types of fibre are beneficial at all doses, e.g.: 

Bonithon-Kopp, Claire, et al. "Calcium and fibre supplementation in prevention of colorectal adenoma recurrence: a randomised intervention trial." The Lancet 356.9238 (2000): 1300-1306.

4. these days, luckily, we are getting more and more crazy out of the left field ideas.

5. Pak, Heidi H., et al. "Fasting drives the metabolic, molecular and geroprotective effects of a calorie-restricted diet in mice." Nature Metabolism (2021): 1-15.

6. Nelson, James F., et al. "Probing the relationship between insulin sensitivity and longevity using genetically modified mice." Journals of Gerontology Series A: Biomedical Sciences and Medical Sciences 67.12 (2012): 1332-1338.

7. "Mice in the group were entrained to eat comparably to the AL-fed group within a 3-h period during the first 2 weeks, and food was always removed 3 h after the start of feeding. "

8. "Due to shredding behaviour of the Diluted AL group, we performed a comprehensive food consumption where we measured average shredded food and food left in the hopper. We found an average of 23% of the food was shredded on the bed of the cage. We accounted for this value to calculate for Diluted AL food consumption presented in this paper."

9. Non-caloric effects via food sensing would not be totally unexpected for sure:
Libert, Sergiy, et al. "Regulation of Drosophila life span by olfaction and food-derived odors." Science 315.5815 (2007): 1133-1137.

10. would have to re-read to check if they had a non-dilution approach
Solon-Biet, S. M. et al. The ratio of macronutrients, not caloric intake, dictates cardiometabolic health, aging and longevity in ad libitum-fed mice. Cell Metab. 19, 418–430 (2014).
Also note this beautiful study that Lamming dug up totally at odds:
Kokkonen, G. C. & Barrows, C. H. The effect of dietary cellulose on life span and biochemical variables of male mice. Age 11, 7–9 (1988).

11. unfortunately the control mice are a bit shortlived at around 770d mean LS
Nelson, W. & Halberg, F. Meal-timing, circadian rhythms and life span of mice. J. Nutr. 116, 2244–2253 (1986).

12. Mitchell, Sarah J., et al. "Daily fasting improves health and survival in male mice independent of diet composition and calories." Cell metabolism 29.1 (2019): 221-228.

12a. Mice are not very bright but you can train them using food: "The MF mice learned quickly that food would not be continuously available, and thus, tended to gorge, spending hours each day without food. "

Note that the eating window of ML (fasted) mice was half that of controls but still higher than CR. Thus this study cannot disprove the hypothesis that an isocaloric reduction of the eating window to CR levels would lead to further LS extension and that most CR benefits are due to fasting.

Another strength of that study: "The bodyweight gain trajectory of NIA-MF mice was significantly lower compared to NIA-AL controls (p=3.48E-05), while no significant difference was observed between WIS-MF and WIS-AL mice (p=0.55) (Fig. S1, upper panels)"

Given that both WIS and NIA groups show similar LS extension after MF (fasting) argues against so called crpto CR effects.

13.  Actually I like the conservative phrasing of the abstract. However, this is not why this paper is so controversial and interesting. It is the cellulose dilution lifespan study that is key. The cellulose data, if true, would suggest 100% of the CR effect is due to fasting.