The World Nutrition Summit is coming soon as a virtual meeting based in South Africa. The organisers kindly asked for a presentation based around the ROS hypothesis of obesity. I haven’t tried to make it particularly user friendly, Mike Eades has already done that, this is more of a brief whirlwind tour of the many technical papers which went to make up the Protons thread and spawned the idea.
Okay, this is the image I missed on my first read through the results/methods of Hall's latest offering:
I have to apologise for failing to pick this up possibly because I would never read the discussion or conclusion of a Hall paper. My interest in what Hall thinks is distinctly limited. The image shows a delay in the onset of fat mass loss with LC which then proceeds at a remarkably similar rate to the fat loss in the LF group.
But clearly, the changes in fat mass under LC trend upwards (ns) initially before trending downwards, (also ns) compared to fat mass at initiation of the diet.
As David Ludwig has pointed out, appetite suppression on LC can be delayed and shows most reliably from 2 weeks onwards. This might well be related to the rising levels of ketone bodies providing enhanced energy availability as ketosis develops using the concept outlined here:
Both diets were individualised to be weight maintaining, so we can say nothing about spontaneous food intake (ie appetite, ie weight/fat mass changes). The main point of interest is the very right hand end of Figure 1:
which shows calculated insulin sensitivity. Under high PUFA intake insulin sensitivity, after an overnight fast, is clearly enhanced.
We have no information about the FFA levels so assessing the "blood energy content" is impossible for either diet. However we can look at the relative changes in glucose vs beta hydroxybutyrate (BHB). They look to be reciprocal, but this is simply an artefact of percentage change.
Glucose drops by 10% from normal fasting levels in the PUFA group, measured as 79.2mg/dl, a sizeable number. Ketones in the PUFA group rise by 10% over the same period. But this is 10% of 1.34mg/dl, a vanishingly small amount, physiologically. There is no ability for this modest rise in BHB to offset the fall in glucose. The energy content of ketone bodies and glucose are approximately equal, per gram. Free fatty acids would be the unknown confounder.
That study looked at high (10% energy) vs very high (42% energy) from PUFA intakes, both under modest ketogenic conditions.
Hall's study looked at a mildly ketogenic diet with around 15% of energy from linoleic acid, a pathologically high intake, in comparison to a carbohydrate based diet deriving around 3% of calories from linoleic acid, ie a physiological linoleic acid intake.
Linoleic acid at 3% of calories, ie very low, will have no blunting effect on the ability to limit caloric ingress in to adipocytes ("loss") in the peak-absorptive period (so will not generate the need to eat more food to offset this loss in to adipocytes) and will not facilitate pathological insulin sensitivity to allow hypoglycaemia in the post-absorptive period (with subsequent hunger).
I think we can sum Hall's study up in the words of the abstract section:
"One participant withdrew due to hypoglycemia during the low-carbohydrate diet."
"One participant withdrew due to hypoglycemia during the high-polyunsaturated fatty acid diet."
The other participants had to (spontaneously) eat more to successfully avoid these problems during the first week.
The effect diminished through the second week of the study as ketones rose further to compensate for the hypoglycaemic effect of excess PUFA.
"The carbohydrate–insulin model of obesity posits that high-carbohydrate diets lead to excess insulin secretion, thereby promoting fat accumulation and increasing energy intake."
If the simplest CIM model of obesity was correct then eating a diet which raised insulin, like this:
to a peak of 120microU/ml and kept it elevated for three hours, should result in marked sequestration of lipid in to adipocytes with subsequent weight gain.
The question is how this occurs.
Over the years I have had a couple of attempts to understand "carbosis", which this low fat group appears to be demonstrating. None of my attempts have been particularly satisfying from the metabolic point of view. Let's try again.
In the immediate aftermath of the low fat test meal glycolysis is very active. This is facilitated by the elevated insulin translocating GLUT4s to the cell surface, which will facilitate the ingress of glucose.
Insulin will also facilitate the ingress of FFAs via CD36 translocation to the cell membrane but, at the same time, insulin will simultaneously lower FFAs:
With plasma FFAs at around 0.1mM it doesn't matter very much how many CD36 receptors are present on the cell surface, fatty acid oxidation will be a limited source of both NADH and FADH2 supply to the electron transport chain.
In addition to facilitating glucose ingress in to cells, insulin also drives the activation of the pyruvate dehydrogenase complex. This increases the translocation of pyruvate via its proton gradient coupled transporter in to mitochondria and ensures its metabolism to acetyl-CoA.
Glycolysis to pyruvate generates two molecules of cytoplasmic NADH for each molecule of glucose utilised. Much of this NADH will be reconverted to NAD+ by the malate-aspartate shuttle, passing the electrons to generate NADH within the mitochondria. If the rate of generation of NADH exceeds the capacity of the malate-aspartate shuttle, cytoplasmic NADH levels will rise and, secondary to this, the level of lactate will rise to resupply NAD+, using lactate dehydrogenase.
We can get an idea of how much the cytoplasmic NADH levels rise from the output of lactate in to the bloodstream of the study subjects after the low fat test meal. Like this:
I find a post prandial lactate of just under 3mM quite impressive. I would suggest that the NADH:NAD+ ratio is high.
So the next question is, what might an elevated cytoplasmic NADH level do to the glycerophosphate shuttle?
If we assume that high cytoplasmic NADH activates the glycerophosphate shuttle we will have the transfer of electrons from cytoplasmic NADH to the intra-enzymic FAD of the mitochondrial component of the glycerophosphate shuttle, mitochondrial glycerol-3-phosphate dehydrogenase.
From here the input is from the outer surface of the inner mitochondrial membrane directly on to the CoQ couple. From the mitochondrial point of view, that cytoplasmic NADH never arrives (it should have entered using the malate-aspartate shuttle). Instead it is "seen" as an FADH2 input, with all of the implications that has related to the FADH2:NADH ratio intrinsic to the Protons thread.
Under the influence of insulin there is clearly a great deal of cytoplasmic NADH generated from glycolysis and it is this action of insulin which must be limited to avoid excessive caloric ingress.
With glucose at 120mg/dl (sorry for the quaint units):
and insulin at 120microU/ml there are lots of calories entering cells. With FFAs at or below 0.1mM they will not be a significant source of either NADH or FADH2.
Fatty acids are out of the equation. So the signal for cellular satiety, driven by ROS, is going to come from the rising FADH2:NADH ratio generated by the glycerophosphate shuttle converting NADH to FADH2.
Ultimately the sensing of a cellular "satiety" level of substrate ingress will be signalled the generation of high-physiological levels of superoxide and hydrogen peroxide, facilitated by the glycerophosphate shuttle. I won't mention negative feedback from complex III but it will contribute too.
What will not be involved to any significant degree is beta oxidation. FFAs are low at the time of peak calorie availability/storage. And of those FFAs there will be very, very little linoleic acid.
A very low fat diet side steps the problems caused by linoleic acid failing to allow satiety-facilitating levels of ROS to be generated. Linoleic acid is simply out of the equation, what little there is of it from the diet being squirrelled away in adipocytes during the period of peak calorie availability. It is simply not there to interfere when cellular satiety is being successfully signalled.
Couple that with the fact that the low fat meal plans provided a linoleic acid supply limited to 3% of calories, it is quite easy to see how a group of 20 slightly chunky young Americans (BMI 27ish, 32% body fat, not the Arnie look for the slightly high BMI!) might be suffering from chronic linoleic acid toxicity. Dropping to 3% energy from linoleic acid is going to be markedly less fattening than the level which might be found in any version of the SAD.
Low fat is synonymous with low linoleic acid. High carbohydrate/high insulin has its own satiety mechanism, more dependent on the glycerophosphate shuttle, which is impervious to small amounts of linoleic acid. And with a diet of 2000kcal supplying LA at 3.1g/1000kcal, then just over 6g/day is not a lot to worry about.
I'll stop here to keep it simple. Obviously the tendency to normalise weight in the low carb period is directly related to low insulin, facilitating lipolysis. Once insulin is low enough, ie once carbohydrate intake is low enough, then linoleic acid, even at a total of around 40g/d (as was eaten during the LC phase), becomes unimportant.
EDIT, missed the lack of fat loss on LC! I guess 40g/d of linoleic acid with low but not basal insulin is too much! The is an update on this here. END EDIT.
Until you get just a little bit of carb creep of course... As carbohydrate and associated insulin rises then that approximately 13% of calories as LA is going to facilitate weight regain with a vengeance.
which sums up why I don't do twitter. To join this discussion would take weeks of careful thought and reference trail following. However this review (DO NOT download it from SciHub!!!!!! Or if you do, make a donation. I didn't say that) came out of it:
"In fact, they [FFAs] increase insulin precisely to the degree needed to compensate for the fatty acid–induced insulin resistance."
After that it's all about ROS and double bonds, a concept which Boden (who sounds a really interesting chap) lacked at the time. The basic CIM (Carbohydrate Insulin Model) of obesity is undoubtedly incomplete and, nowadays, is a straw man for people like Hall to use to facilitate career enhancement. As in
Note: There are buzz-words even in the title of this paper which make me uncomfortable!
As Raphi and I discussed, studies do not arrive out of the blue. Hall knew exactly how to set this latest CIM "destroying" study up. The rival camp (Ludwig's group are current torch bearers here) have pointed out that in short term studies that CICO applies quite well. In longer term studies (over two weeks, and yes, of course Hall knows this) low carb has a significant ameliorating effect on the fall in metabolic rate associated with caloric reduction/restriction. As in:
People may recall I've looked at Hall vs Ludwig in the past. Of the two camps I personally favour the one with the most honest approach to the data. My thanks to David Ludwig for updating me on this latest interchange within the on-going process.
Anyhoo, time to get the children's breakfasts and keep working on the house before the builder arrives on Monday, while working towards various deadlines off blog.
Figure 4, shown below, sums up why it interests me
Brief aside: The role of glucose in generating ROS from mitochondria is, to me, extremely dubious. It certainly can occur but it needs the activation of the glycerophosphate shuttle, which doesn't get a mention. But the GPS is how we convert cytoplasmic NADH in to mitochondrially inputted FADH2, with ROS generation resulting from the raised FADH2:NADH ratio intrinsic to this conversion acting on the CoQ couple. It will apply in the presence of insulin, not isolated hyperglycaemia. If you slog through the refs trail (not good) you need to realise FBS = insulin/IGF-1. Also few people even think of fatty acids in this respect, a serious omission. Enough of mitochondria, back to the cytoplasm. End aside.
It's worth pointing out that the insulin receptor is thought to act as a G-protein coupled receptor which signals to an NADPH oxidase (probably NOX4) using a specific G protein. The NADPH oxidase generates H2O2 extracellularly (this becomes important in future posts) which re-enters the cell (some papers suggest there is a specific transporter) to oxidise cysteines on protein tyrosine phosphatases, disabling these enzymes. Without PTPases maintaining the suppression of insulin signalling both the insulin receptor and insulin receptor substrates autophosphorylate and so signalling takes off.
That's all pretty straightforward and is verging on textbook.
Nothing happens without ROS generation.
This led me down a rabbit hole, thinking about how primordial insulin signalling might be and how primordial the ROS generation might be. Is insulin core, with ROS as a second messenger? People may have noticed that the most basic signalling is what interests me.
How far back does insulin go? If we have a look at this review
we can see that there is a recognisable insulin like receptor stemming from the common ancestor leading to both ourselves and sponges. That's pretty far back, marked by the blue lineage arrows in Figure 1 from the review:
Insulin signalling is thought to be present in most, but not quite all, metazoans (blue circle).
The review looks at the evidence for insulin signalling in yeasts, plants and a ciliated protozoan.
Sacchromyces has no suggestion of an insulin receptor. However it responds to exogenous human insulin with a response remarkably recognisable as the response of human cells to insulin.
Plants are more straightforward. They produce an insulin-like cysteine rich peptide which interacts with an insulin-like receptor to induce the effects typically seen in mammalian cells under insulin. In fact using this peptide on adipocytes produces exactly the same effects as human insulin.
Neither the "plant insulin" nor its receptor have anything in common with metazoan insulin/receptor protein amino acid sequences.
Except they have common "shape". They are immuno-related. They look similar enough (shape/charge distribution) that they can be recognised by the same binding antibody.
Exactly the same findings are duplicatable in the ciliate protozoan T pyriformis as for the Sacchromyces yeast.
So. Different insulin-like hormones, different receptors. Genetically completely unrelated, but causing the cell to respond in the same way.
The simplest answer is convergent evolution, as suggested in the review. I think this is correct. But there is a deeply insightful comment towards the end of the discussion. Almost insightful enough, but ever so slightly not quite there:
"The convergent evolution of ligand-receptor pairs alone cannot explain however the biochemical similarities in the intracellular response to insulin observed outside metazoans, as illustrated above. One way to overcome this seeming inconsistency is by considering that independently evolved upstream components of pathways devoted to processing environmental information may have been tied to evolutionary conserved core metabolic and cellular growth signaling networks".
The most obvious metabolic signalling molecules which adjust core metabolic function and cell growth are the ROS.
Metazoans, plants, yeasts, protozoa; all will use ROS signals to control metabolism and growth. This is the evolutionary conserved process on to which various environmentally responsive ligands and receptors have been co-opted to respond. On at least four separate occasions. My opinion.
A eukaryote is a derivative of a bacterium living inside an archaeon. Information about archaea is remarkably thin on the ground. I expect them to use ROS. Bacteria are more rewarding once you turn to Pubmed.
Perhaps bacteria are where we should be looking to find the origin of the primordial ROS signal.
This is Dr Mary Ramsay, head of immunisation for PHE (Public Health England, UK government). She is reiterating exactly the manifesto of the Great Barrington Declaration. I think it was April 2020 that I heard Prof Sunetra Gupta first advocate this concept. Now, in the early days of Lockdown 3, suddenly there is a voice of reason from a government department. It's echoing Vallance/Whitty from March 2020, before they both had all of their immunology knowledge, presumably with most of the rest of their brain function (sarcasm warning again) removed, sometime during Lockdown 1.
Here it is! (I can't see any way to embed and preserve Dr Ramsay's interview clip), it's in a Great Barrington Declaration tweet here
This is the text from the tweet because I always worry tweets may be ephemeral:
"Head of Immunisation for @PHE_uk -Dr Ramsey announced to the Science & Technology Committee that England may follow a focused protection strategy, where protection is given to the vulnerable and the disease is allowed to circulate among the young where its not causing much harm."
The text is true to the verbal narrative in the short clip.
While I'm on the subject of brain removals, people may be aware that daily COVID-19 deaths are currently exceeding those in the spring epidemic in the UK.
at a time when all-cause mortality is absolutely normal for the time of year, as mentioned in the last two posts. As in:
April was something exceptional. If anyone thinks December is in any way comparable to April you can head to the Funny Farm now. That's you, Prof Whitty.
Current COVID-19 death "data" are being used as an excuse for our government, particularly Matt Hancock (the UK Health Minister), to personally incite supermarket store managers to bully and harass people with mask exemptions into wearing masks in-store, if they want to buy food to eat. Apparently further measures are under consideration, god only knows what they will be.
Disgust is too mild a word.
I wonder whether Dr Ramsay will last at PHE? Perhaps the Guardian could run a hatchet job on her.
Here is the right hand end of the graph from the last post, showing the partially adjusted all cause mortality for the UK in week 52:
Clearly the down tick at the end is the result of PHE struggling to adjust for the incomplete data from a week with a bank holiday followed by a week with two bank holidays. Paperwork, such as registering deaths, tends not to get done at weekends or on bank holidays.
So, theoretically, we could have had any of the red dashed line corrections:
EDIT We now have the 2021 week 2 report with more delayed registrations included. Looks like all cause mortality is now on a par with 2017/18. Still watching...
What happened when the majority (most, though not all, comes through within a week) of data were added in? On the 7th of January an update was posted so here is the original graph directly alongside the new graph. Again, ignore the downstroke:
If anyone is struggling to see the difference I've circled the important areas in red:
To put the current situation in to perspective I've pulled the curve from the last years during which we had a significant winter flu epidemic, 2017/18:
and just to make it even clearer, here's a line to show peak all cause mortality from 2017/18 vs end of Dec 2020:
It's also worth noting, again on a terribly parochial UK basis, that the winter peak in all cause mortality is almost always at the end of the first week of January. As in now, though the numbers will take a week or two to fully come in. This year may be typical, it may not. So far it's typical.
I have absolutely no doubt whatsoever that work in the ITUs (which are busy but not full, on a national basis) is hell at the moment. In hot-spots, worse. Given the level of barrier nursing needed, staff shortages because people are not allowed back to work until they eventually become PCR-ve (which can take months), social distancing within wards etc life must be awful and exhausting.
I absolutely stand ready to accept that there is a catastrophe ongoing at the moment. It is quite possible that this winter's peak deaths will exceed that of 2017/18.
So far it hasn't.
I would suggest that the Guardian and the BBC may not be the most reliable sources of information about the current situation in the UK.
Another (as so often!) addendum:
Here are the ITU admissions for COVID-19 from ICNARC (which I consider to be unbiased reporting) up until just before Christmas:
which looks pretty grim, despite UK ITUs being at under 100% occupancy. But these are just COVID-19 patients. I don't have a similar curve for non-COVID-19 patients, but I do have this graph which includes both groups (and the rest of England), should they go on to die:
The last column is too low in both death categories due to delayed reporting, as always.
It's quite clear that for every death from COVID-19 there is nearly one less death from anything else. It could be that the virus is so lethal it is impossible to die quickly enough from anything else before it gets you.
Or it could be that managing to get a SARS-CoV-2 PCR-ve death certificate is becoming harder and harder to achieve in a hospital setting when PCRs are still being run to 45 cycles during a winter of endemic COVID-19.
Please, please, please be aware that the week 52 data are only ESTIMATED.
The real value might be higher. Equally, it might be lower.
Hat tip to Mike Yeadon.
An afterthought: How do you explain the 2020/21 curve?
I don’t know. However: most “COVID-19 on the death certificate” deaths occur in hospitals and are reputed to be registered very promptly compared to community deaths. There is a massive need for the numbers of COVID-19 fatalities in the current situation, the government needs (and demands) these numbers fast to drive policy. This will inflate the uncorrected data through early December. However non-COVID-19 deaths are currently massively and exceptionally below normal for the time of year and these will only be incorporated in to the data more slowly than the rapidly registered COVID-19 deaths. So through early December the overall value has been correctly reported as being higher than normal because the low number of COVID-19 negative fatalities from the community (or even from hospitals) aren’t yet included. Now they are now coming in. I doubt PHE are remotely interested in assimilating this anomaly in to their estimating process (which will be based on previous years normal delay patterns of registration) but eventually the EuroMoMo absolute death data will have the truth out. Not that it will make any difference.
Why are non-COVID-19 deaths so low? People, sadly, die in some excess during the Winter. If you are hospitalised for anything at all leading to your death, the chances of you reaching your end without achieving a +vePCR test are very low. You will be a COVID-19 death and so you will be missing from other datasets.
***Broken links to some Pubmed papers. See the post "A aargh BROKEN LINKS" at the top of the top of the Labels list to see how to get round them***
I am Petro Dobromylskyj, always known as Peter. I'm a vet, trained at the RVC, London University. I was fortunate enough to intercalate a BSc degree in physiology in to my veterinary degree. I was even more fortunate to study under Patrick Wall at UCH, who set me on course to become a veterinary anaesthetist, mostly working on acute pain control. That led to the Certificate then Diploma in Veterinary Anaesthesia and enough publications to allow me to enter the European College of Veterinary Anaesthesia and Analgesia as a de facto founding member. Anaesthesia teaches you a lot. Basic science is combined with the occasional need to act rapidly. Wrong decisions can reward you with catastrophe in seconds. Thinking is mandatory.
I stumbled on to nutrition completely by accident. Once you have been taught to think, it's hard to stop. I think about lots of things. These are some of them.
The "labels" function on this blog has been used to function as an index and I've tended to group similar subjects together by using labels starting with identical text. If they're numbered within a similar label, start with (1). The archive is predominantly to show the posts I've put up in the last month, if people want to keep track of recent goings on. I might change it to the previous week if I ever get to time to put up enough posts in a week to justify it. That seems to be the best I can do within the limits of this blogging software!