Monday, January 27, 2020

Rory Robertson and Protein Restricted Longevity

I looked at this paper when it did the rounds a fair while ago, saw that the only fat source used was soybean oil and decided that living on soybean oil, sucrose, maltodextrin and wheat starch was not a good idea and so I binned it as the junk it is:

The Ratio of Macronutrients, Not Caloric Intake, Dictates Cardiometabolic Health, Aging, and Longevity in Ad Libitum-Fed Mice

I missed the embedded problems which have since been brought to light by Rory Robertson, whose slightly over-the-top concerns are voiced here. I would perhaps disagree slightly with some of his opinions but, overall, he makes a rather good case. The first thing to note is that you have to go to the supplementary data to realise that a significant number of groups of mice were lost (and excluded) due to mortality problems. Table S1 describes all thirty of the diets which the study started out with. Five of these diets had to be discontinued because too many mice either died outright or (I suspect) were ordered to be euthanased on the authority of the supervising veterinary surgeon due to concerns about animal welfare. I'm assuming Oz has a Home Office much as the UK does which requires Named Veterinary Surgeons to be employed to supervise animal welfare in all laboratories.

We know this from the legend to supplementary table S1. Here is the last section of that legend (sorry that the small letter superscripting is lost, that's blogger for you):

"a Diets 2 low energy and 6 medium energy were discontinued within 23 weeks. b Diets 3 low energy, 3 medium energy and 6 low energy were discontinued within 10 weeks of treatment. These diets were discontinued due to weight loss (≥ 20%), rectal prolapse or failure to thrive".

Here is table S1 with the discontinued (and removed) groups outlined in red:

So, they started with 30 diets groups but five of them had such high early death rates that they were excluded from the study. This left 25 groups. Other than the legend to supplementary table S1 I am unable to find any reference to the loss of five diet groups anywhere in the main paper, which gives the distinct impression that 25 groups were all that were included from the start. So 17% of the mice died at under 23 weeks in to the study, many of those within less than 10 weeks, and you have to read the supplementary data to find out.

All of the high mortality groups were eating 5% of calories as protein.

Did you pick that up in the abstract? No, you didn't.

Is there any excuse for failing to discuss this crucial finding in the results and discussion sections of the paper? You can decide that. It's not exactly rocket science.

My feeling is that the authors could argue, if they were convinced that protein restriction was key to longevity (amino acids, cysteine, mTOR etc don'tchano), that studying early life mortality has no relevance to late life longevity. Why not leave early mortality to the paediatricians? That is a potentially arguable position and should, as it involved a huge chunk of the study mice, have been reported and been justified (if possible) in the results, discussion and especially in the abstract.

The other slight hiccup is this line from the main paper:

"Median lifespan was greatest for animals whose intakes were low in protein and high in carbohydrate... (Figure 2A)"

Figure 2A is not in a format which lends itself to simple interpretation and, obviously, excludes all of the mice which died or were euthanased at less than 23 weeks of age, all of which were in low protein groups. Anyway, you might want to see a simple table of median lifespan for each of the surviving groups. Like supplementary table S2. I've high-lighted the group which had the longest median lifespan in red:

Looks to me like the longest median lifespan group might have been eating 42% of it's calories as protein... Hmmmm. Worth repeating:

"Median lifespan was greatest for animals whose intakes were low in protein and high in carbohydrate... (Figure 2A)".

vs Table S2 giving 42% protein for longest median lifespan.


Let's make this crystal clear: The data demonstrating the actual outcomes are, absolutely, present in the supplementary data of the paper. It is also absolutely crystal clear that the paper itself, excluding supplementary information, does not accurately represent the the actual findings in the study.

If you had to summarise the paper in human terms you could say that applying severe protein restriction to your kids while topping up their calories with sugar and soyabean oil would hopefully result in them being taken away from you and placed in to care before they died.

Please don't try this at home.


My thanks to Rory Robertson for his attempts to have this paper retracted and more accurately rewritten and to Grant Schofield for tweeting about his efforts.

Saturday, January 25, 2020

Coronary Artery Calcium Score and Scleroderma

Dr Malcolm Kendrik has a very interesting post over on his blog relating to coronary artery calcium scoring. I think it is fair to say that he is not in favour of the test.

My ears pricked up (metaphorically) when he mentioned myositis ossificans, about which he comments "This does not end well".

I have spent some time in the past thinking about pathological arterial calcification, as applied to the aorta of of patients with familial hypercholesterolaemia. Bear in mind that the dietary advice for patients with FH is about the worst you could possibly imagine and, of course, has no evidence base. My thoughts and assorted links are in an old blog post here. At the time I had never heard of Sci-hub so was unable to access this rather neat diagram of the mechanism of action of insulin, Pi and pyrophosphate:

Back to pathological soft tissue calcification. Clearly the obvious question about myositis ossificans has to be to ask whether it is in part driven by hyperinsulinaemia/hyperglycaemia or both.

As far as I am aware this is not a question which had been asked. It is simply genetic and that's it.

However, a similar question has already been answered in relationship to a serious generalised soft tissue mineralisation condition described as "calcinosis and scleroderma", back in a publication from 1932 (apologies to the person who tweeted the link, I didn't note their name to acknowledge. And twitter is ephemeral). That is too long ago to be listed on Pubmed so if you would like to read it you can go and ask Elsevier how much they would like to charge you for a peek in to the past or you can go to that awful place that none of use ever use to download any paper for free.


"Calcinosis and scleroderm" looks to be one of a family of soft tissue calcification diseases. The case report from 1932 describes the complete remission of this extremely unpleasant condition in a child following a period of time on ketogenic diet of the type used at the start of the last century, before dieticians were invented/summoned from Hades.

Did the ketogenic diet resolve this child's pathological calcification by suppressing insulin levels, glucose levels or both? Does it work by lowering alkaline phosphatase production by cells in/around inflammatory lesions? Or by some other mechanism?

Would it do the same for pathological arterial calcification? Given a tool like the ketogenic diet, perhaps there is some logic to CAC testing?

Unless you feel that tissue calcification is an appropriate part of healing until it gets to scleroderma levels...


Tuesday, January 21, 2020

Barja, an aside

I quite enjoyed Barja's review

The Cell Ageing Regulatory System (CARS)

but found this section a little uncomfortable:

Hmmmmmm. Plant based, healthy fruit and vegetables, bad animal fats. Not my sort of outlook really.

In another of his publications here

Highly resistant macromolecular components and low rate of
generation of endogenous damage: Two key traits of longevity

there is this comment

"It was also found that 6–7 weeks of dietary restriction are enough to decrease MitROS production and 8-oxodG in mtDNA and nDNA in rat liver (Gredilla et al., 2001a )".

Gredialla et al (incl Barja) 2001a is

Effect of short-term caloric restriction on H2O2 production and oxidative DNA damage in rat liver mitochondria and location of the free radical source

Here they found, by eyeball, an approximately 50% reduction of in 8-oxodG in mitochondrial DNA after those six weeks of quite severe caloric restriction:

Now let's compare this with the degree of damage reduction (this time using the term 8-OHdG as the marker rather than oxo-8dG, which appears to be the same thing).

Here's the change in mtDNA damage marker in brain mitochondria using F3666, one of the worst ketogenic diets around:

Just by eyeball I make the drop in mtDNA damage out to be greater than 50% by two days and something like 75% by three weeks. On ad-lib food consumption. No hunger.

Considering that F3666 does not extend longevity in mice (it doesn't shorten lifespan either, despite causing liver damage and it does actually improve health during ageing in rodents) this does, for me, slightly knock some of Barjas core ideas.

Sad but true.


Wednesday, January 15, 2020

Stearic acid again

Better post this one while I have a few minutes. I picked it up while looking for refs for Gustavo Barja's epic The Cell Ageing Regulatory System (CARS) in which longevity is tied to the Double Bond Index of the mitochondrial inner membrane (Thanks Bob!). BTW it is possible to modify the DBI but, with current data, it looks like you cannot alter the saturated or MUFA percentages, it is replacing omega 3s with omega 6s which mimics the mitochondria of long lived mammals!

Anyway, here is the cocoa butter paper:

Differential effects of saturated versus unsaturated dietary fatty acids on weight gain and myocellular lipid profiles in mice

Here are the diet compositions:

The line in red is the total percent of calories from linoleic acid in each diet. Here are the body weight changes:

The bottom two lines are the low fat high carbohydrate diet which happens to come in at just 1% linoleic acid and the cocoa butter diet which comes in at 1.4% of calories as linoleic acid. The high palmitic acid gives the most weight gain as it delivers 4.5% of calories as PUFA. Olive oil is a close second, also with 4.5% linoleic acid. The oddity is the safflower oil diet which is very high in PUFA but only gives intermediate obesity. Quite what is going on here is difficult to say but you have to wonder at what level of omega 6 PUFA that "next level up" signalling (lipid peroxide based) kicks in. No data on that, just a guess/excuse from the Protons perspective. There are a number of other studies showing this phenomenon of limited weight gain with safflower oil.

Still, stearic acid as cocoa butter is still looking pretty good. All of the high fat diets were based around different fat sources placed in to the D1245 background so are equally high in sucrose and starch too, comparable amounts across all of the higher fat diets.


Thursday, January 02, 2020

Protons (53) a formula

A couple of things came up in emails recently. First is that I never mention that I had a chat with Ally Houston on the Paleocanteen podcast. It was fun. I think I sound like me. It's here

Second is that karl asked if there was a general formula for working out the F:N ratio for assorted fatty acids.

Edit: cavenewt pointed out that for people unfamiliar with the FADH2:NADH ratio concept there is a reasonable introduction at Protons: FADH2:NADH ratios and MUFA. PubMed-ing Dave Speijer and CoQ makes good reading too. End edit.

There wasn't but given a few minutes and some algebra it works out like this for even-numbered, fully saturated fatty acids of carbon skeleton length n:

F/N   =   (n-1)/(2n-1)

So stearate (C18) is 0.486

Palmitate (C16) is 0.484

Caprylate (C8) is 0.467

For MUFA/PUFA you just subtract one FADH2 per double bond (db). This doesn't affect the NADH term.

F/N  =  (n-1-db)/(2n-1)

Oleate (db = 1) is 0.457

Oleate is the MUFA of stearate. Saturated fats allow us to resist insulin, MUFA allow insulin to act.

Linoleic acid, also C18 but with two double bonds, gives 0.429

This is lower than stearate or oleate. The switch for ROS generation occurs between roughly 0.486 (high physiological ROS) and 0.457 (low physiological ROS). LA is lower than oleic acid.

Glucose has an F/N ratio, from memory, of 0.2 so LA is the "glucose-like" of the common fatty acids, in Mike Eades' terminology, and so will fail to generate fatty acid appropriate ROS. Which will allow continued insulin action when it should be resisted. That will make you fat, and the loss of calories in to adipocytes will make you hungry. The exact opposite of stearic acid...

Happy New Year all.