Bayesian Investor Blog

Book review: The Longevity Diet: Discover the New Science Behind Stem Cell Activation and Regeneration to Slow Aging, Fight Disease, and Optimize Weight, by Valter Longo.

Longo is a moderately competent researcher whose ideas about nutrition and fasting are mostly heading in the right general direction, but many of his details look suspicious.

He convinced me to become more serious about occasional, longer fasts, but I probably won’t use his products.

How to approach longevity

Longo tries, with partial success, to look at a broader range of evidence than many researchers, focusing on what he describes as 5 pillars:

• Basic research
• Epidemiology
• Clinical studies
• Centenarian studies
• Studies of complex systems

Longo advocates acting “in tune with evolution”. This is a welcome contrast to typical medical research, which often leaves me wondering whether the researchers believe in not-so-intelligent design.

Trying to extend your lifespan by increasing your intake of vitamin C is like trying to improve a Mozart symphony by increasing the number of cello players. … It took billions of years of evolution for it to reach the current state of near-perfection. We cannot expect a simple supplement to make something that’s almost perfect even better

Longo talks somewhat like advocates of a paleo diet, but instead of arguing that the last 10,000 years of evolution were unimportant for nutrition (or that evolution stopped 10,000 years ago, as some deluded people have), Longo errs in the other direction, by focusing more on what our grandparents ate, combined with what people eat in Blue Zones.

He imagines that 4 centuries is enough to adapt to adapt to foods that recently reached the West from the New World. If genetic changes had worked that fast, I think the selection process would have been pretty noticeable. Maybe Longo is thinking more in terms of the cultural evolution (i.e. something more like individual cultures taking a few centuries to figure out whether a new food can be made safe, using preparation techniques that that culture is willing and able to adopt). Whatever he’s thinking, I’m somewhat skeptical, but expect that he’s still doing better than those who ignore ancestral dietary patterns.

Fight evolution, or go along with it?

Longo talks about programmed aging for about 5 pages, then drops the subject, and writes as if that doesn’t affect his lifestyle advice [1]. I wondered for a while when “in tune with evolution” means returning to an environment in which we’re adapted to thrive, and when it means fighting nature’s plan to scrap us when we’ve outlived our usefulness.

I see several hints (the Mozart reference, and his description of the Blue Zones), which suggest that he sees his work as restoring the near-perfection that evolution produced. But he has co-authored papers on programmed aging with Joshua Mitteldorf, whose book hints that evolution may have programmed us to age faster in environments where food (especially protein-rich food) is abundant. Longo’s diet could be seen as fooling our bodies into thinking food is scarce, and that seems slightly out of tune with evolution. But Longo seems to have either decided that it’s currently more practical to restore nature’s near-perfection, or possibly that he needs to keep quiet about any hopes that are too far from the mainstream (he clearly doesn’t have much evidence that his diet can beat the Blue Zone results).

I’m a bit confused by this claim:

So for the great majority of human history, activating a program that reroutes so much energy away from reproduction to use for protection and repair would not have been advantageous.

It seems weird that any such trade-off could explain aging, when fasting and exercise slow aging.

It sure looks to me like any programmed aging is more plausibly due to the need for diversity, in order to make us less susceptible to infectious disease. I.e. in times of abundance, nature programmed our genes to focus more on experimenting with new combinations of genes that will keep up with newly evolved pathogens, whereas in times of famine, our ancestor’s genes couldn’t afford to throw away old bodies and build new ones.

Longo writes:

We think of poor nutrition, lack of exercise, and the genes we inherit from our parents as the major risk factors for diseases. But, by monitoring the age at which people are diagnosed with different diseases, we know that aging itself is the main risk factor for cancer, cardiovascular disease, Alzheimer’s, and many other diseases.

Longo is likely not the first to have said that, but I find it frustrating that his approach almost leads him to see a the other main risk factor: living in a wealthy society.

We don’t have evidence that’s quite rigorous enough to say that living in the developed world is a larger risk factor for common diseases than is age, but that’s where the evidence points. Cardiovascular disease, diabetes, and dementia are pretty clearly absent from at least some hunter-gatherer societies. It looks like experts have searched for those diseases in at least a thousand New Guinea hunter-gatherers over 60 years old, and reached a consensus that there are no signs of those diseases.

For cancer, the evidence is more confusing, and aging seems to be the biggest risk factor there. Longo points to places that have 1/4 the cancer rate of the US, and that’s the clearest evidence I’ve seen on the subject.

For cardiovascular disease, the evidence is stronger: Lindeberg’s table 4.1 compares the U.S. to Uganda in 1951-1956, showing how many deceased men had signs of previous myocardial infarction at autopsy:

Age	USA	Uganda
40-49	31 of 178 (17%)	0 of 178 (0%)
50-59	51 of 199 (26%)	1 of 199 (0.5%)
60-69	32 of 98 (33%)	0 of 98 (0%)
70-79	8 of 24 (33%)	0 of 33 (0%)
80+	2 of 9 (22%)	0 of 9 (0%)

These types of evidence add up to strong reasons to say that modern lifestyles are the biggest single cause of disease in the developed world, although age is intertwined in a way that makes it hard to say which is the “main” risk factor.

The concept of Western disease fits pretty well with what Longo is trying to fix, even if Longo doesn’t think of it that way.

Insulin resistance

This book caused me to update further in the direction of believing that insulin resistance causes a large fraction of age-related suffering and death in wealthy societies.

I’ve been reading various sources that hint at this for the past few years, but Longo focuses a bit more directly on how most Western disease can be avoided by lifestyle changes that overlap a good deal with the kind of changes that reduce insulin resistance.

Yet Longo isn’t very focused on insulin resistance – that’s just one of several pathways that are influenced by fasting. And contrary to many authors who care about insulin resistance, he recommends getting 55-60% of calories from carbs.

What diet does Longo promote?

It’s quite plausible that our ancestors evolved in conditions under which hunger was somewhat common. That could create adaptations that depend on something like fasting.

This basic idea is also supported by the success of calorie restriction at extending lifespans in animals, and by some evidence from fasting for religious reasons.

Longo also brings in a theory about fighting cancer: normal cells respond to a “starve signal” by going into protective mode, whereas cancer cells became dangerous in the first place by ignoring that type of signal. Longo has a nice analogy: normal cells act like Russians who know how to hunker down for winter, cancer cells are vulnerable since they continue advancing like Napoleon did.

Longo advocates a decent and not too remarkable diet for a typical day, with occasional 5 day periods of his special formula Prolon, for a Fasting-Mimicking Diet (FMD).

He did studies showing an increase in the average (but probably not the maximum) lifespan of mice, and human trials which show good changes in biomarkers.

Getting better biomarkers than what the standard American diet produces is only hard in that it’s hard to get people to follow a better diet. And mice may be less well adapted to dietary protein than humans, or maybe evolution selected for rapid aging in mice in a way that makes them not indicative of human aging. Most diseases for which mouse studies seem to translate well to humans, seem to be diseases that are not currently major problems in the developed world. So his research only adds a modest amount to my favorable opinion of occasionally restricting calories.

Where’s the version of Prolon that looks like what my grandparents ate? Prolon has foods that my grandparents likely didn’t eat, such as quinoa, refined inulin, turmeric, and potassium sorbate.

Longo includes lists of foods that are high in some vitamins and minerals for which he’s concerned about deficiencies. Why does he list calcium, but not zinc? Calcium consumption only looks low because people are looking at an RDA from a country that has had high dairy consumption (RDAs range from 500mg to 1200mg, depending on what authority you ask). Okinawans got 505.3 mg/day in 1949, on their way to becoming the culture with the world’s longest-lived people. I don’t think evolution is fast enough to have made some humans dependent on getting twice that. At most, high calcium diets make up for deficiencies in other nutrients (vitamin D? potassium?).

One obvious way to quantify a fast is to measure the calorie deficit during the duration of the fast.

It seems like Longo started researching 3-day fasts that would have produced 6000 calorie deficits, and when he had trouble getting participants to do that, he switched to a 5-day period of 800-1100 calories per day, producing a 5700 calorie deficit. He also hints that the 5-day version is safer [2].

Longo doesn’t explicitly talk in terms of calorie deficits, but I infer he used something like that in designing his diet.

Why did Longo study a fixed-length, fixed calorie diet, rather than a diet that goes until some biomarkers, such as ketones, reach a good threshold? My intuition says that the biomarker-based approach would produce healthier results in those who followed it. I presume part of his reason is that his model would urge use of harder-to-frequently-test biomarkers such as IGF-1 and IGFBP1. Maybe some of it is that he knows that fewer people would follow such a protocol.

I fear that another part of his reason has something to do with the way that science is usually done. I.e. focus first on the hypothesis that’s simplest to describe, or maybe simplest to test.

His Mozart reference suggests he knows at some level that the simplest hypotheses are unlikely to be right. Yet his actual research only reflects that a little. I wanted to see him asking more questions of the form “if it’s that simple, why hasn’t at least one of {evolution, the medical establishment, or the diet book industry} done something to fix it?” (see the protein discussion below for an example where I want him to ask more about evolution).

How hard is the diet to implement?

Longo’s human RCT says that only 6 of the 95 participants who were told to fast were excluded for noncompliance, 18 withdrew for other reasons, and 71 reportedly complied.

That’s with a group that was presumably selected for their willingness to try a new diet. Don’t expect average people to do that well.

I wouldn’t be shocked if the actual compliance was worse than reported (if so, that would mean that the study underestimates the biomarker improvement among those who do comply).

I expect this diet requires a good deal of willpower, although the willpower needs will tend to decline over time more than with most diets. Also, it likely causes several days per year of very low energy (the amount likely varies unpredictably between people).

My experience says there are lots of subtle things that influence how bad the hunger feels.

The full 5-days of calorie restriction would be a bit unpleasant for me now, and would have felt nearly prohibitively hard if I had started it back before I did any form of intermittent fasting.

Some of my negative reaction seems due to my mind deciding that novel patterns of calorie deficits are scarier than ones that look like calorie deficits that have been previously followed by feasts. So I’m inclined to gradually build my way up to bigger calorie deficits at a time.

The book convinced me to try a 3-day diet that combined a goal of achieving ketosis with the goal of getting a bit closer to the FMD diet, summarized here:

Day	Calories	Protein (g)	Net Carbs (g)	Fiber (g)	Ketones in urine (mg/dL)
1	0	0	0	0	7
2	990	13	44	20	6 to 30
3	1835	21	28	36	40 to 60 (then 75, dropping to 8, the next day)

I had tried to achieve ketosis before, using a low-carb diet, but not restricting calories or protein much, and gave up because I felt lousy for days, and didn’t have a good idea of how much longer it would require (I didn’t do urine tests until this time). This time my mood deteriorated until about 44 hours into the diet, then recovered more than halfway, and stabilized.

I lost about 6 pounds, and have been slowly regaining that weight (I’d prefer to keep my weight near the lower end of that range, but it doesn’t seem worth much willpower).

I had some problems with muscle cramps the night of day 3, and my muscles still felt weak during a hike on day 6, but recovered soon after that.

My guess is that it would be a bit easier for me to gradually work my way up to a 3-day fast with zero calories than it would be to do the full 5-day FMD version, although I expect either one is more drastic than I really need.

My experience with meditation suggests that it can influence how much suffering is associated with hunger – I’ll likely expand on that in a meditation-related post soon.

Is protein harmful?

Longo wants a low protein diet (about 10% of calories).

He has decent centenarian studies and observational studies to support that advice.

But what about the evidence from hunter-gatherers, who get 19 to 35% of their calories from protein? And why do the correlations suggest that only a subset of protein is associated with bad health (proteins from plants and seafood seem fine)?

The centenarian evidence seems like good evidence that a low-protein diet can be safe, but rather weak evidence about the alternatives (the centenarian evidence is at least partly explained by them being too poor to afford meat).

The hunter-gatherer evidence might be somewhat exaggerated by researchers looking more at the dry-season diet than the mud-season diet, and by Inuit, who maybe aren’t so healthy, and live in an environment that’s rather different from our ancestors. But the evidence still seems like enough to say that a 20% protein diet is insufficient to cause Western diseases.

Studies suggest that people with Laron syndrome have a significantly reduced risk of cancer and type 2 diabetes, and Longo attributes that to effects of low growth hormone levels and low IGF-1, similar to what protein restriction produces. That suggests that protein can cause problems, although it says little about the importance of the effect.

Methionine restriction is possibly the primary factor in the effectiveness of calorie restriction, and that presumably applies to fasting. This seems true for many species, probably including primates [3].

Why would we have evolved so that methionine is harmful? Methionine doesn’t seem to have been particularly rare in ancestral environments. Maybe high methionine is bad only when combined with a glycine deficiency? A glycine deficiency seems like a plausible consequence of our recent practice of discarding animal skin and eating only the muscle.

Longo relies a fair amount on evidence from observational studies, but I only see weak support there for his claim (see detailed discussion in footnote [4]).

Here are some hypotheses that might explain the observational evidence:

• protein affects insulin/IGF-1/growth hormones (why doesn’t this show up in observational studies for plant, fish protein?)
• methionine creates excess homocysteine (why doesn’t this show up in observational studies for fish protein?)
• average protein levels aren’t important, and we should focus more on periods of low protein (or maybe periods of no animal protein?) (Josh Whiton)
• some proteins that are new to adult human diet in the past 10,000 years or so cause problems (e.g. Gundry’s rants against lectins)
• modern habits of eating animal protein create unnatural amino acid ratios (glycine deficiency – why doesn’t this show up in observational studies for fish protein?)
• protein is fine, it just happens to be correlated with other bad features of food. Here are some speculations about those bad features:
  • fat cooked at excessive temperatures
  • heavily processed meat (e.g. smoked, or whatever it is they do to make sausages)
  • animals that were fed a really crappy diet
  • saturated fat

Lots of hypotheses, and I can’t find compelling reasons for picking one over the rest.

Evolutionary considerations lead me to doubt that average protein consumption is an important variable to look at, so if Longo’s analysis is partly right, then I’d expect the optimal analysis would combine his ideas with some model of fluctuating protein levels.

The paper Metabolic effects of milk protein intake strongly depend on pre-existing metabolic and exercise status suggests that milk proteins have some signaling effects, which tend to be appropriate for a rapidly growing body. But those signals can be undesirable in a body that isn’t evolved to grow any further, and can be diabetes-inducing in a body that’s sedentary enough to only grow by accumulating fat (and close to neutral in a body that gets enough exercise that it wants to regrow muscle?).

That fits nicely with Longo’s discussion of protein’s effects on growth hormones (which he studied in people with Laron’s syndrome, where lack of growth hormone receptors mimics a low-protein diet).

But observational studies tend to show that dairy consumption is associated with good outcomes. That’s not strong evidence, but it’s inconsistent with milk protein causing major problems.

There’s likely something more complex going on than just milk protein causing diseases, but that hypothesis seems more plausible than Longo’s position that protein in general is bad.

I’m reluctant to cut my protein consumption, for several reasons:

• protein increases satiety (e.g. see the Fullness Factor formula)
• taste
• my diet already has enough restrictions that I avoid a majority of restaurants; I don’t want to limit my variety further

Still, there’s enough confusion about protein that I should assign at least a 10% chance that high protein levels will hurt my health. So I’ll move a little bit in the direction of more protein-restriction days, remain cautious about dairy, and be sure to consume a bit of collagen to make up for the low glycine content in the meat that I eat.

Fact checks

I checked some of his claims.

These subjects consumed a Mediterranean diet supplemented with either one liter of extra-virgin olive oil per week …

Someone misread the PREDIMED study (which says “Participants in the group assigned to a Mediterranean diet with extra-virgin olive oil received 1 liter of the oil per week per household, with the recommendation to consume at least 4 tablespoons per day of extra-virgin olive oil per person”), and that misreading seems to have spread widely. I would think that someone whose profession involves nutrition would notice something suspicious about a claim that seems to imply people getting more than half their calories from olive oil.

The book’s table 4.5 (on the Okinawan diet) has a strange category of food: dairy/seaweed.

I checked the paper he lists under that table, and instead of explaining that category, I ended up wondering where Longo got the data? Not from that reference.

The paper says Okinawans got 1% of calories from fish (in 1949), but Longo says 11% of calories from “omega-3 rich foods (fish, etc)”. Some legumes are somewhat rich in omega-3, and the paper lists those as 6% of calories. But Longo has a separate listing for “soy and similar foods” (12%), so I’m pretty sure he didn’t include legumes in “omega-3 rich foods (fish, etc)”.

I’m confused!! My best guess is that Longo took data from some other source, reflecting more a recent shift in their diet. Wikipedia says::

The traditional Okinawan diet as described above was widely practiced on the islands until about the 1960s. Since then, dietary practices have been shifting towards Western and Japanese patterns, with fat intake rising from about 6% to 27% of total caloric intake and the sweet potato being supplanted with rice and bread.

Also:

Recently, there is a decreasing trend in the life expectancy in Okinawa.

The paper says Okinawans got 69% of calories from sweet potatoes. I’m a little confused about the exact numbers [5], but this clearly indicates a high sugar diet.

That’s important because it conflicts somewhat strongly with Longo’s advice to limit sugar (I don’t see any explicit numbers for his usual diet; on page 174 he advises a 10g/day limit on sugar for days when we’re doing FMD for cardiovascular reasons).

Longo does better on the references that I discuss in footnote [4], but overall he doesn’t seem careful enough to inspire much trust.

While writing this review, I noticed a Red Pen Review that checked random references, and those checked out better than the ones I checked, so maybe my intuition led me to his weakest references.

Conclusion

I learned a modest amount by reading this book, and more by checking some of its references.

The book is mostly about diet. It has a chapter about exercise, but it’s easy to forget that he says exercise is important, since he doesn’t say anything novel or controversial there. Please don’t let your interests in novelty and controversy distract you from the importance of exercise.

Longo has persuaded me to slightly increase the lengths of my fasts and/or calorie restriction periods.

Longo has convinced me to aim for better insulin-related blood tests results than Bredesen recommends.

This review has focused on problems with the book, but Longo’s advice seems around 90% right, and his mistakes don’t look particularly harmful. Following the advice might add a month or a decade to your lifespan, and additionally may reduce the time you spend in hospitals or nursing homes.

The quality of life issues deserve more emphasis than Longo or I have gotten around to discussing – dementia and dialysis machines have some hard-to-quantify tendency to make sensible people suicidal, and chemotherapy is about as bad as the disease that it treats. Hunter-gatherers seem to have pleasanter ways of dying of old age. Let’s aim to do at least as well as they do.

Footnotes

[1] – I was disappointed with how the otherwise good Red Pen Review handled this issue:

The theory of programmed aging unravels when you consider that, for most species, extrinsic factors (e.g., predation, infection, starvation, cold, etc.), and not intrinsic factors from aging, are the primary causes of death. Most species simply do not survive long enough to even experience aging, more accurately termed “senescence.” Hence, senescence has very little impact on the biological fitness of most species in their natural environment, severely limiting the degree to which the forces of natural selection could have selected for a so-called programmed aging trait. Moreover, rarely does the benefit of the overall group (and group selection) override the strength of individual-level selective forces. The very suggestion of this book that individuals can “cheat” the system and live longer is an example of how quickly the whole notion of “programmed” aging would fall apart. If the system can be cheated, there will always be “cheaters” who will game the system; within any group those cheaters will be represented in higher numbers in future generations, even if it means not doing as well as an entire group.

The 2002 paper which the review references does seem to represent a quasi-consensus of experts at the time that few wild animals die of senescence, but more recent research appears to strongly dispute that.

Longo’s diet won’t do much to cheat any programmed aging. The most the diet can hope to achieve is a Blue Zone style lifespan where we remain active until 100 or so, and avoid western diseases. Longo has published a paper in Nature Reviews Genetics (available here) which outlines why the cheater objection isn’t conclusive.

The idea that programmed aging could resist cheaters seems strange, but no stranger than the idea that an adaptation requiring two parents for reproduction can resist being overtaken by individuals who reproduce parthenogenetically.

[2] – The book includes too many generic disclaimers of the form “don’t try this without consulting a doctor”, but also includes some much better warnings that cover a wide variety of specific risks associated with fasting. Yet he left me unclear whether he knows of important safety reasons for preferring the 5-day FMD to a 3-day total fast.

[3] – Longo points out that the NIA calorie restriction study on rhesus monkeys show no benefit compared to controls that got 17.3% of calories from (mostly plant) protein, whereas the mostly similar University of Wisconsin study showed a benefit from calorie restriction compared to controls that ate a more Western diet with unlimited protein, mostly from milk. The latter study would have had a much larger methionine difference between controls and calorie restricted monkeys.

[4] – From Low-carbohydrate diets and all-cause and cause-specific mortality: two cohort studies:

Consistent with our results, higher intake of vegetable protein, but not animal protein, was associated with a lower risk of coronary heart disease mortality in the Iowa Women’s Health Study(30). The lower CVD mortality observed with higher vegetable low-carbohydrate score in our study is likely in part due to the established benefit of unsaturated fats, dietary fiber, and micronutrients such as magnesium and potassium, as well as other bioactive compounds including vitamins, minerals, and phytochemical (31).

I.e. they saw a correlation that has many possible causes. I’d expect that correlation simply based on the differences in fruit and veggie consumption.

From Dietary protein and risk of ischemic heart disease in middle-aged men:

CONCLUSIONS: We observed no association between dietary protein and risk of total IHD in this group of men aged 40-75 y. However, higher intake of animal protein may be associated with an increased risk of IHD in “healthy” men.

That conclusion sounds weaker than Longo’s description of it. But that’s mostly just the authors being cautious about statistical significance. The data seem consistent with Longo’s interpretation.

There’s a clear positive correlation between protein and IHD. The correlation with animal protein is stronger than with total protein. The authors control for a bunch of variables, and that weakens the correlation enough to drop below statistical significance. I’m unclear how they got from there to “no association”, when I would conclude that there seems to be an association, but that the sample size is small enough to leave doubts.

Who knows whether it was appropriate to control for those variables? I’m fairly sure some are confounders that ought to be controlled for (e.g. fiber). But I’m guessing that BMI and “History of diabetes” are more mediators than confounders. The authors show some awareness of this problem, and tried to tried to use a statistics approach to infer causality, but there’s no substitute for the wisdom of Pearl here.

[5] – I’m confused, because the paper gives sweet potato data in both calories and grams. I calculate anywhere from 48.7 to 88.7 grams of sugar per day, depending on whether I use the paper’s weight or calorie data, and also depending on which way the sweet potato was prepared, and probably some other source of variation that I don’t understand. The paper must use nutritional data that say sweet potatoes have around 1.45 calories per gram, whereas the sources I use to get sugar levels believe it’s more like 0.9 calories per gram.