Insulin sensitivity makes you fat. Insulin resistance makes you fat. Discuss. (04) M scores

This is a study which Raphi threw out on Twitter several weeks ago. It can take me personally this long to understand some studies. It's interesting as it clearly shows that weight loss over 12 weeks causes a deterioration in insulin sensitivity in some people: 

Changes in adipose tissue lipolysis gene expression and insulin sensitivity after weight loss

They started with 52 obese people and ended up with 38 who completed this part of the study. On average insulin resistance dropped with weight loss.

On average, that is. It did for some. For others it didn't. Yes, in nine of the group of 38 subjects, insulin sensitivity deteriorated with weight loss. As the authors comment:

"Thus, it may seem paradoxical that some of the changes observed in the present study during weight loss prevented an increase in insulin sensitivity."

I love a paradox.

What is going on?

Here are the metrics from Table 2

The paradox comes from the M values.

These are the clamp values. Insulin was infused to maintain a plasma level of 80microIU/ml and hypoglycaemia was avoided by infusing glucose to maintain a plasma level of 5mmol/l. If you do this for 2h you can assume some sort of steady state is achieved and the M value is the rate of infusion of glucose needed to counteract insulin's hypoglycaemic effect over the final 40 minutes of the clamp. The higher the M value the better insulin is working and the less insulin resistant your subject is. This is the gold standard, HOMA-IR is the poor relative (though much easier to estimate and wouldn't have shown any paradox in this study).

So in nine subjects the M value deteriorated with weight loss.

The subjects were divided on the basis of the direction of this change in M score resulting from weight loss.

To cut a long story short the group which deteriorated their M score were obese but had normal insulin sensitivity before weight loss. The so-called metabolically healthy obese. The subjects whose M score actually improved were insulin resistant before weight loss.

From a previous paper by the same group the M score in a group of lean control subjects (not described here but part of the overall study) was 9.77mg/kg ffm/min and here, for the obese but insulin sensitive group M, it was 9.17mg/kg ffm/min, ie no difference. In the obese but insulin resistant subjects pre weight loss M values averaged 5.63mg/kg ffm/min, ie low, p<0.05.

The first question is: why should metabolically healthy, insulin sensitive obese people deteriorate their insulin sensitivity score as a result of weight loss?

That's the simple one. They're losing weight and have not stopped at the time of the hyperinsulinaemic clamp. Their diet is providing completely inadequate calories to maintain stable weight so they are losing some weight, mostly from fat. The process of weight loss involves releasing FFAs from adipocytes with an associated rise in plasma FFAs and subsequent oxidation of those FFAs. I did a ball park rough calculation and someone of my height (but obese) would have been on 1400kcal per day, ie they would be hungry and oxidising body fat to make it up to the well over over 2000kcal I'm considered to require to run my metabolism.

Aside: Obviously from the Protons perspective oxidising fatty acids will generate ROS irrespective of mitochondrial delta psi and the cell will resist insulin's signal in proportion to this fatty acid oxidation induced ROS signal. It is apparent that this cannot be suppressed by an hyperinsulinaemic clamp for two hours, see below. End aside.

There is a study which actually measured fat oxidation under an hyperinsulinaemic clamp under conditions of active weight loss. It's this one:

Prolonged Fasting Identifies Skeletal Muscle Mitochondrial Dysfunction as Consequence Rather Than Cause of Human Insulin Resistance

which combines a Dutch version of the M score with indirect calorimetry during the steady state conditions of the clamp. Neat hey? This time weight loss was not from an hypocaloric diet, it was from a zero caloric diet, ie extended fasting, which sources calories from a minimal carbohydrate, minimal protein, very high fat source, ie the subject's own body catabolism. Just for 60 hours. But by the end nearly 100% of calories from are fat oxidation.

The clamp data come from Fig 2, the rates of disposal of glucose in red come from Table 2 and I've clarified the labels to point out that "Fed" means "Overnight fasted", not post prandial:

The glucose disposal units don't include fat free mass as this doesn't noticeably change in 60 hours of fasting so this means the values can't be directly compared to M values. We can see that glucose disposal halves under hyperinsulinaemic clamp study conditions when 60 hours of fasting necessitates metabolism to be run on fatty acids.

Proof that fatty acid oxidation continues under hyperinsulinamia comes from measurements in a respiratory chamber while the clamp was on-going. From the bottom of Table 2:

After an overnight fast, on a balanced diet, lipid oxidation is 1.2μmol/kg/min and drops to 0.3μmol/kg/min under clamp. During body fat loss for 60 hours the clamp can only suppress lipid oxidation to 1.52μmol/kg/min, ie still higher than under basal conditions before the 60 hour fast. Oxidising lipids intrinsically generates mitochondrial ROS to limit ingress of excess calories by limiting insulin facilitated signalling. Even at clamp levels of insulin exposure.

So weight loss causes insulin resistance because it intrinsically involves fatty acid oxidation. Stopping the weight loss using a period of adequate feeding would allow the weight loss induced insulin resistance to fade away. The only "paradox" is that the changes which occur in the provision and oxidation of fatty acids cannot be immediately reversed by the external application of 80μIU/ml of insulin, at least not within two or three hours.

It's also worth mentioning that this insulin resistance during weight loss is utterly normal and pro-survival from an evolutionary perspective.

Okay. Much more interesting are the patients who do improve their insulin sensitivity with weight loss. Fasting FFAs before weight loss are around 0.71mmol/l, higher than the value of 0.52mmol/l found in the slim control group in the previous section of the study.

This group has metabolic syndrome, ie they do not have the ability to suppress lipolysis using insulin because enough of their adipocytes are "leaking" FFAs via basal lipolysis and so are exposing their peripheral tissues to an on-going supply of FFAs, which are being oxidised continuously whether insulin is present or not. When insulin is present then this unstoppable supply of FFAs necessitates enough resistance to insulin's signal to balance caloric ingress. Glucose disposal under hyperinsulinaemic clamp (M value) stays low at 5.63mg/kg ffm/min because fatty acids are being oxidised, so glucose is significantly excluded. We don't know the FFA concentration under clamp or fatty acid oxidation in this study but my view is this is what is happening.

With weight loss adipocytes shrink. Two things happen. FFA levels rise because weight loss requires fat release. And, as adipocytes shrink, they decrease their size-determined basal lipolysis rate and so restore insulin's ability to suppress FFA release. These opposing factors, in this case, balance out in a trivially raised level of fasting FFAs to 0.76mmol/l, but the release of these FFAs can be suppressed by insulin, so M rises from 5.63mg/kg ffm/min to a respectable 8.55mg/kg ffm/min under the hyperinsulinaemic clamp after weight loss.

Fat oxidation obligates a degree of resistance to insulin's signal to maintain physiological caloric ingress. Protons also suggests some fatty acids do this better than others.

I was going to stop here but I may as well throw this in. People may recall this graph:

Now we can get rid of most of the labelling and insert the concepts discussed in this post. These arrows represent what I consider the situation to be under weight stability, so the situation is not complicated by the level of insulin resistance superimposed by weight loss.

It's easiest to think of the arrows representing the study subjects before 12 weeks on a severely hypocaloric diet. During weight loss the metabolically healthy group will temporarily move their location to the right due to increased fatty acid oxidation resisting insulin. Those with metabolic syndrome will try to do the same but this is offset by shrinking adipocytes decreasing their basal lipolysis so allowing insulin to work and actually improve the M score overall.

I could have stopped here too but I still have a final arrow to add. This comes from Arne Astrup's lab in Denmark. There are no clamp data but under weight stability the M score and HOMA-IR would correlate perfectly well, assuming everyone was on a similar mixed diet. Astrup looked at formerly obese (FO) women who had maintained, through sheer willpower, a normal body weight using conventional dieting. Rare as hen's teeth, but they do exist.

Fat metabolism in formerly obese women

No clamp data but just look at the insulin and glucose values. C is control, FO is formerly obese. Formerly obese ladies are very, very insulin sensitive. Pathologically so:

If we wanted to add these ladies to the graph they would be here, at the blue arrow:

There is a body of opinion on t'internet that holds that insulin resistance causes obesity. They have some explaining to do.

Now I really will call it a day.

Peter

Insulin sensitivity makes you fat. Insulin resistance makes you fat. Discuss. (03) Visceral vs subcutaneous

Visceral adipocytes are exquisitely insulin sensitive. Visceral adipocytes are insulin resistant. Let's go.

This is another of the anchor studies which has informed my world view over the last decade or so:

Insulin signaling in human visceral and subcutaneous adipose tissue in vivo

I like it because it used real live intact human beings, it used a physiological dose of insulin and it looked at the signalling pathways within cells. It then compared the effects within visceral adipocytes (omental in this case) with those in subcutaneous adipocytes. I'm not sure that I had much of an opinion about this before I read the study but it locked me in to my still current opinion that visceral adipocytes are particularly insulin sensitive.

This has been reinforced by the PET scan data from this next study. The mouse section is poor as it compares SC adipose tissue to epididymal adipose, the effect would be more marked if they had used omental/mesenteric as the visceral adipose source but hey, the PET data in real live humans are nice...

Increased glucose uptake in visceral versus subcutaneous adipose tissue revealed by PET imaging

This property of visceral adipocytes appears to be intrinsic to their stem cells. If you differentiate stem cells from SC and visceral tissues in to adipocytes those adipocytes, generated in-vitro, retain this exquisite insulin sensitivity.

Fat depot-related differences in gene expression, adiponectin secretion, and insulin action and signalling in human adipocytes differentiated in vitro from precursor stromal cells

So visceral adipocytes are the most insulin sensitive in the body and this is programmed in to their stem cells, which makes me accept that it is evolutionarily conserved and so essential their function. Which, surprisingly, is neither to kill us nor give us metabolic syndrome.

But we also have this:

Subcutaneous and visceral adipose tissue: structural and functional differences

All of the below quotes are taken from the same screenshot.

This was 2010. I doubt that there is any less confusion nowadays.

"Adipocytes from VAT are more insulin-resistant than SCAT adipocytes (39,40)."

"Visceral adipose tissue has higher rate of insulin-stimulated glucose uptake compared with SCAT adipocytes."

"Visceral adipocytes are more metabolically active and have a greater lipolytic activity than SCAT adipocytes (44,45)."

For my sins I started with reference 44, Arner's 1995 paper

Differences in lipolysis between human subcutaneous and omental adipose tissues

which had slightly weird methods for investigating adipose tissue function. This referred me to the same group's 1983 paper

Differences at the receptor and postreceptor levels between human omental and subcutaneous adipose tissue in the action of insulin on lipolysis

and finally back to Arner 1981

The antilipolytic effect of insulin in human adipose tissue in obesity, diabetes mellitus, hyperinsulinemia, and starvation

which gave me this image:

The lines represent paired adipose samples, a control sample labelled "Basal" (open circles) and a sample to which a fixed concentration of noradrenaline (NA) had been added at all insulin exposures.

The first red flag is that basal lipolysis can be suppressed by insulin. So what this paper means by basal lipolysis and what other papers mean by basal lipolysis appear to be quite different. The other really strange finding is that under noradrenaline induced lipolysis insulin is suppressive (as you would expect) up to 250μU/ml at which point insulin enhances lipolysis, markedly. Hmmmmmm. I would suggest insulin induced insulin resistance but that seems like a long shot. Ultimately the model doesn't seem to be a very good one.

The only difference I can see between this group's model and the rest of the world is that they use tiny cubes of adipose tissue rather than isolated adipocytes. If I had to guess I would suggest that there are the end terminals of noradrenergic neurons present in their adipose samples and these are modifying the response to insulin and noradrenalin. It could also be that having structure around adipocytes might alter their behaviour ie this might be a better model than isolated adipocytes. Hard to compare when almost all of the surrounding work is done with isolated adipocytes. Generally I would be very cautious about looking for understanding in such a model as the results are counterintuitive, but I could be wrong. That's a few days of my free time I'll never get back! Next I checked the other three refs.

Ref 45

Metabolic complications of visceral obesity: contribution to the etiology of life of type 2 diabetes and implications for prevention and treatment.

is a review and not available on Pubmed or Sci-Hub and I doubt it it's worth chasing.

Ref 39:

Relationships of Generalized and Regional Adiposity to Insulin Sensitivity in Men

does not, as far as I can make out, in any way support the statement it is purported to. The study wasn't designed to do so and I can only hope it was a typo when it was cited.

Ref 40 is gold dust

Visceral fat and insulin resistance – causative or correlative?

because it says this:

"In one study in vivo , subjects were given isotopically labelled fatty acids, and biopsies of different depots were taken at abdominal surgery 24 h later (Marin et al. 1992); accumulation of label was most marked (per g TG) in the omental and retroperitoneal depots."

which allows us to find Marin et al's paper 

The Morphology and Metabolism of Intraabdominal Adipose Tissue in Men

which used real live humans in a tracer study coupled with adipose tissue biopsies during abdominal surgery. My sort of study. Just from the abstract we have, in-vivo:

"Adipose tissue lipid uptake, measured after oral administration of labeled oleic acid in triglyceride, was approximately 50% higher in omental than in subcutaneous adipose tissues."

and in vitro

"Adipocytes from omental fat also showed a higher lipolytic sensitivity and responsiveness to catecholamines"

So, easy in via insulin, easy out via sympathetic nervous system stimulation.

But also:

"Furthermore, these adipocytes were less sensitive to the antilipolytic effects of insulin."

which is illustrated like this

Here we have visceral adipocytes as black circles and SC adipocytes as open circles. It's true that visceral adipocytes exposed to noradrenalin at 10⁻⁴ mmol/l can suppress this lipolysis less effectively than SC adipocytes do but the difference is relatively small compared to the very marked extra lipolysis under noradrenalin exposure in visceral adipocytes. Visceral adipocytes *should* stay small by balancing insulin signalling against sympathetic lipolysis.

So let's summarise:

Visceral/omental adipocytes are exquisitely insulin sensitive.

They perform markedly augmented lipolysis under sympathetic stimulation.

They have mildly impaired insulin suppression of sympathetic lipolysis.

I think that sets us up to think in sensible terms about what visceral fat does and why it is associated with metabolic syndrome.

Peter

Insulin sensitivity makes you fat. Insulin resistance makes you fat. Discuss. (02)

This is just a longevity study but it provides insight following on from my last post.

An isocaloric moderately high-fat diet extends lifespan in male rats and Drosophila

Let's just recap:

If we start in "control" state and switch to an insulin sensitising diet we have an initial sequestration of fat in to adipocytes and rapid weight gain, red bracket. As adipocytes distend they increase their basal lipolysis, which cannot be suppressed by insulin, and both fasting insulin and post prandial insulin levels must rise. If we fail to observe the initial acute insulin sensitivity phase and limit our observations to the rising insulin resistance phase (blue bracket) we could easily observe that elevated insulin levels are *associated* with future weight gain and assume, incorrectly, that there is causality.

On to the study.

The fat in the control diet is 15% soybean oil, so approx 7% of energy intake as LA. This is not a particularly low LA diet and, as you would expect, fasting insulin levels slowly rise throughout life.

All of the following graphs are extracted from Figure 2 section J, which looks like this

giving this as the control insulin levels:

The high fat (aka insulin sensitising, aka increased LA) diet substituted lard for some of the carbohydrate and was fed ad lib. We don't know how much LA was in the lard, but probably at least 10% of the lard calories were LA. We miss the insulin sensitising hypoinsulinaemic phase as it's long gone by 28 weeks of age:

If we take this graph at face value, at any given age the insulin value is higher in the rats which are fattest, so the logical assumption is that this is causal. QED.

EDIT: I have used my imagination to butcher the above graph back to week eight. Accept or reject at your own discretion!

END EDIT.

But my contention is that these insulin levels are only elevated as a consequence of adipocyte size, ie the hyperinsulinaemia is a secondary phenomenon and could be eliminated by suppressing basal lipolysis, usually using good old acipimox. It works.

But acipimox cannot maintain suppressed lipolysis from distended adipocytes for more than a few hours.

Undoubtedly the most effective way of limiting basal lipolysis is to limit absolute adipocyte size. If we could maintain the adipocytes of the rats fed the insulin sensitising diet at the same size as those of the control fed rats there would be no "masking" effect of elevated basal lipolysis to hide the on going life long insulin sensitising effect of the obesogenic diet.

So they pair fed, by calories, exactly the same "calories" to a third group of rats so that they received exactly the same calories of insulin sensitising food as the chow fed mice eating ad lib.

These rats dis not become fat (I know, I know, heavy calories vs light calories, forgive me) and their weights exactly matched those of the control rats. So now we have rats with normal levels of basal lipolysis for age but no excess through eliminating "over eating" (stop sniggering) the amount of food which hunger, driven by excess insulin sensitivity, demanded in the "green-line" rats.

So the red line is the true insulin sensitivity of rats fed the obesogenic diet when not allowed to superimpose the increased basal lipolysis generated by becoming fat per se:

I hope this post complements the previous post and makes it completely clear that an insulin sensitising diet causes rapid sequestration of fat in to adipocytes. This lost fat will a) make you hungry, b) distend your adipocytes, c) increase their basal lipolysis and d) this latter will cause systemic insulin resistance.

This is a very basic and very simple to understand concept.

Aside: d) is core to limiting caloric ingress to cells when FFAs are available in the presence of insulin/glucose, which should not occur. You know, as in

Insulin resistance is a cellular antioxidant defense mechanism

which is a fundamental paper for understanding hyperinsulinaemia. Ignore at your peril. End aside.

What happens to make adipocytes insulin resistant (a late change in the words of Bill Lagakos, and I agree) is much, much trickier and then we're slipping in to Nick Lane territory.

Peter

Insulin sensitivity makes you fat. Insulin resistance makes you fat. Discuss.

My apologies for disappearing off of the blog, after the trip to Alicante and subsequent half term I got out of the habit of blogging and let the fructose thread go cold. I'll get back to it at some stage but this is what is currently making me think. Here we go.

I have a certain specific outlook on life which suggests that insulin sensitivity facilitates fat storage and insulin resistance limits fat storage.

The two anchor points to this world view are here

Insulin sensitivity is increased and fat oxidation after a high-fat meal is reduced in normal-weight healthy men with strong familial predisposition to overweight

and here

Insulin resistance associated with lower rates of weight gain in Pima Indians

I think it is fairly reasonable to suggest that this is not a particularly popular point of view and that it may simply reflect my personal biases and my very selective view of the literature. While there are rodent data to support it, the overwhelming evidence from human studies is that insulin resistance/fasting hyperinsulinaemia are strongly associated with future weight gain. I'll cite the below study (let's call it the Odeleye study after the first author) in particular as it too is derived from the Pima Indians and actually mentions the second of my two anchor studies, link courtesy of Gabor Erdosi via Twitter

Fasting Hyperinsulinemia Is a Predictor of Increased Body Weight Gain and Obesity in Pima Indian Children

This is merely the tip of the iceberg and Gabor has many, many studies which show exactly the same finding. I'll stick a list of them down at the end of this post but we only need one for the purpose of my current discussion, the others give a flavour of how universal the finding is. I've not read all of Gabor's links but the titles alone give you the basic gist of their findings. I chose the Odeleye example as it's a Pima study contradicting my own favourite Pima study. It also leads us on to ventromedial hypothalamic lesions in rats, which I like. And the authors are a bit dodgy, which I also like. But there is a vast literature supporting their findings.

So here is the paradox:

Protons says:

Excessive insulin sensitivity facilitates excessive fat storage

and conventional wisdom says:

Insulin resistance makes you hyperinsulinaemic and this hyperinsulinaemia overcomes insulin resistance to facilitate excessive fat storage.

If we set this up as a simple dichotomy and assess evidence by the kilogram (once upon a time studies were published in *paper* journals. Getting a copy involved going to a library and paying 10p a page to photocopy them. You could assess likelihood of Truth by the weight of the paper used to publish. The "weight" of the evidence... Showing my age here) then clearly insulin resistance precedes weight gain. All one needs is insulin hypersecretion to overcome insulin resistance and you have obesity.

How does one tease out the explanation, especially when both directly contradictory concepts have the potential to be correct?

Let's go to the Odeleye Pima study and see what we can find. Here is this line in their discussion:

"However, our results are in contradiction to those obtained in adult Pima Indians in whom insulin resistance and high insulin secretion were associated with a lower weight gain (7,8). Reasons for the discrepant results among studies in children and adults remain to be clarified."

which is as true today as it was in 1997.

In rodent derived support of their findings in Pima children they mention

"These results [in Pima children] are consistent with those in animal studies in which hyperinsulinemia precedes an increase in body weight. In rats, lesions of the ventromedial hypothalamus result in hyperinsulinemia before excess weight accumulation (26)."

giving us, as ref 26, this study:

In vivo metabolic changes as studied longitudinally after ventromedial hypothalamic lesions

Unfortunately this study says nothing of the sort. From the abstract 

"One week after the lesions, total glucose metabolism was more sensitive and responsive to insulin than in age-matched controls."

and from the discussion

"Thus, for a similar increase in plasma insulin levels (i.e., +100μU/ml over basal insulinemia) VMH-lesioned rats actually showed a 92% reduction on hepatic glucose production, while such percentage inhibition was only 46% in control rats (Fig. 1)."

Aside: Also, by six weeks this enhanced ability to suppress hepatic glucose output of week one was severely compromised in the VHM rats. Secondary to adipocyte distension and increased FFA release in my opinion. It is essential to distinguish primary from secondary effects. End aside.

and from Table 1 we have

This is the best rodent study that Odeleye's group could find. I'm afraid their contention that hyperinsulinaemia precedes weight gain is simply not supported by the study they provide. Fasting insulin of 111μU/ml is neither biologically nor statistically different from 94μU/ml.

Of course they could have cited

Molecular and metabolic changes in white adipose tissue of the rat during development of ventromedial hypothalamic obesity

which was available at the time and is even less supportive of their statement

but they didn't. I said they were a bit dodgy! Made me giggle anyway.

There are a number of other fascinating snippets in the rodent studies but I'm wandering and so will try to get back on topic.

Just before I move on to what is actually happening I'd like to throw in the insulin section of Figure 1 from my first anchor study. Pre-obese people are particularly insulin sensitive. I've always cited Table 2 which has HOMA-IR scores to show this. It is quite possible to argue that these fasting data do not represent the dynamic effect on insulin levels in the period after a meal and that post prandial hyperinsulinaemia could easily negate the effect of the fasting values. Luckily for my position the study also fed a mixed meal and tracked insulin levels for six hours following the meal. Like this:

The pre-obese, excessively insulin sensitive people are consistently and significantly hypoinsulinaemic cf controls for at least six hours after a mixed meal. Nice.

Summary:

People at significant risk of obesity are consistently hypoinsulinaemic cf controls, in both fasting and post prandial states.

You can model this by inducing an hypothalamic lesion in rats.

You can model it by high level neonatal MSG exposure in rats

You can model it by the feeding of linoleic acid in rats.

What's happening?

This is my summary:

The time scale might be in weeks for a rodent model changed from chow to a high linoleic acid diet or years for an human neonate given his first bottle of soybean oil based infant formula soon after birth. The intervention might be even further back if his mother has been heeding a cardiologist's advice to limit saturated fats and substitute with polyunsaturates.

Okay, that's really the end of the post but here are those human observational studies. My assumption is that they are *all* observing the blue bracketed section of the graph labelled "Slowing weight gain".

I have absolutely no doubt that their data are accurately reported. I have serious doubts that any of the authors could explain the left hand end of my graphic. Which is also correctly describing reality. Any hypothesis must explain all the data. Both aspects are correct.

All links via Gabor, much appreciated.

Insulin resistance as a modifier of the relationship between dietary fat intake and weight gain _ International Journal of Obesity

Long-Term Change in both Dietary Insulinemic and Inflammatory Potential Is Associated with Weight Gain in Adult Women and Men

A longitudinal study of serum insulin and insulin resistance as predictors of weight and body fat gain in African American and Caucasian children

Ten-year weight gain is associated with elevated fasting insulin levels and precedes glucose elevation

Pre-teen insulin resistance predicts weight gain, impaired fasting glucose, and type 2 diabetes at age 18–19 y: a 10-y prospective study of black and white girls

Acute Postchallenge Hyperinsulinemia Predicts Weight Gain: A Prospective Study

Long-Term Change in both Dietary Insulinemic and Inflammatory Potential Is Associated with Weight Gain in Adult Women and Men

Insulin resistance as a modifier of the relationship between dietary fat intake and weight gain

Peter

Chicken fillets are not meat. If you're a cat.

I've just had a very near miss with feeding one of my cats.

Mini came to us as a kitten around 2012. Back in those days I fed raw meaty bones and the only food available with bone sizes appropriate for a cat easily available in the UK was chicken wings. During bone growth I wanted to ensure adequate calcium and phosphate and nutritional grade bone meal had gone from routine use to virtually unavailable. I'd not, at that time, delved deep in to the problems of PUFA which are plentiful in the chicken wings.

Over the years she had low grade on-going gingivitis problems, with hindsight probably related to the high linoleic acid content of her diet. As I  came to realise that the omega 6 PUFA are a problem I transitioned most of our other cats to 12% fat beef mince and that's what they all eat nowadays.

Not Mini. She routinely threw up on beef mince but was fine on the chicken wings, so there she stayed.

Then she broke two teeth (on a chicken wing!) and when I did her dentistry it was obvious that her low grade gingivitis was not low grade and that all of her teeth were in a bad way, so I took them all out.

No cat is going to eat raw chicken wings with her gums. I tried her on beef mince again, she threw up again. Eventually I decided to try her on chicken breast meat, well cut up and relatively low in fat so low in PUFA. She liked it, maintained weight and had no suggestion of ammonia toxicity from an almost all protein diet. She's a cat after all.

Then last November she hid up under one of the children's beds and when I managed to get her out she was very distressed and had laboured breathing. A quick trip to the practice and 30 seconds of ultrasound by someone better at it than me confirmed heart failure.

Heart failure in cats is common. It's usually hypertrophic cardiomyopathy and is essentially untreatable. I assumed that if HCM was present then high protein + high glucose (cats really do convert protein to glucose continuously) -> high GH -> IGF-1 -> cardiac hypertrophy. Probably wrong but who knows? My colleague of the ultrasound machine was uncertain if she could see hypertrophic or dilated cardiomyopathy so I booked a scan with a cardiologist. A few frusemide tablets stopped her drowning and she had no suggestion of a thrombus or pre-thrombus in either atrium.

It was dilated cardiomyopathy. Or end stage hypertrophic cardiomyopathy where replacement of cardiac myocytes with fibrous tissue looks, on ultrasound, just like the dilated form.

DCM is genetic or caused by taurine deficiency. So I PubMed-ed:

The taurine content of common foodstuffs

and found this:

Poultry breast meat is special when compared to poultry leg/wing meat. It has the lowest taurine content of any meat. If you feed your cat on chicken fillets she will develop dilated cardiomyopathy. If the cardiomyopathy allows the generation of an atrial thrombus which breaks off the syndrome produced (iliac thrombosis) is terminal, whatever hope your vet might cautiously suggest re "treatment".

On the plus side DCM from taurine deficiency is completely reversible. Mini was re-scanned recently by the same cardiologist and now has a normal heart. She has been off of frusemide for months.

Taurine comes by the kilo as a sports supplement. I put a pinch of it on her food each night.

Due to my initial thoughts before getting the diagnosis I'd changed her on to beef mince yet again and was going to put up with paddling in cat vomit on the kitchen floor occasionally. Didn't happen. No more vomiting on beef. Huh?

So now all of our cats are on beef mince.

Moral of the story:

Don't feed you cat on chicken fillets.

That was a very, very near miss and (another) lesson to me that I do not have all of the answers! Occasionally reality bites you.

Peter

Addendum. There is a similar known problem with feeding whole ground rabbit carcasses to cats (of which I was aware, but had missed the selective taurine deficiency in white poultry meat):

Rabbit Carcasses for Use in Feline Diets: Amino Acid Concentrations in Fresh and Frozen Carcasses With and Without Gastrointestinal Tracts

Fructose (10) A can of cola

At some stage we have to transition from "fructose is good, it increases glycogen storage in the liver" through "if fructose is storing glycogen in the liver you can be damned sure it's storing lipid" through to "it causes hepatic insulin resistance" and the follow-ons from there.

This next study is pretty well supported by an established base of in-vivo and end-of-vivo studies which I think I can safely pass by and look at the induction in changes to insulin signalling, which might be interesting

Fructose Selectively Modulates c-jun N-Terminal Kinase Activity and Insulin Signaling in Rat Primary Hepatocytes

We're looking at this:

Aside: c-jun N-terminal kinase is what we expect to kill cells severely injured by being cultured in "fasting" levels of unadulterated palmitate plus glucose at 25mmol/l for 24 hours. You know the studies. Assume intolerable ROS. End aside.

The basic message from the paper is this:

The interesting parts of the paper allow us to ask hepatocytes very, very carefully, about the level of fructose exposure which inhibits insulin signalling. We can accept JNK activation as a crucial messaging step in converting fructose exposure to insulin resistance. The concept that this might be ROS driven is my own rather than anything in the paper per se.

This is the bar chart:

This model used steady state fructose exposure over four hours and is quite convincing that there is a "switch" somewhere between 0.4mmol/l and 0.6mmol/l. Our previous steady state study was that of dogs using fructose at 2.22 micromol/kg/min and this achieved a portal vein fructose of 0.1mmol/l:

Inclusion of low amounts of fructose with an intraduodenal glucose load markedly reduces postprandial hyperglycemia and hyperinsulinemia in the conscious dog

We can combine the data from both studies and suggest that, in summary, exposing hepatocytes to fructose a 0.1mmol/l is insulin signal augmenting and exposing them to fructose at 0.6mmol/l induces insulin resistance.

It is also perfectly reasonable to assume that the level of ROS which indicate that is a necessary time to induce insulin resistance are converted to a signal to be carried by the JNK pathway, exactly the one which carries the mitochondrial ROS indicator that it is necessary to induce insulin resistance.

Whether the pre-emptive signal generated by palmitic acid even at low delta psi is the same one as is generated by simple caloric overload (ie failure to resist insulin in time, ie linoleic acid) remains to be seen (by me at least, so far).

We can summarise that nibbling an apple might augment storage of a bowl of porridge as hepatic glycogen but downing two cans of fructose sweetened soft drink might do other things, not least of which is to induce hepatic insulin resistance.

Looking at things fundamentally, this is a story told by ROS signalling. All the signals downstream are certainly interesting and complex but tell us little about the underlying essential process, the information derived from which they are carrying and refining.

Peter

Fructose (09) The Surwit hepatocyte

I hope everyone has forgotten this diagram

which I simplified to this:

Well, now it's time to butcher it further, to an even simpler diagram:

And now I can get rid of the background faint image and shift things around a little to make some more space. I've also converted all "carbohydrate" pathways to blue.

which leaves room to add in mitochondrial, saturated fat derived ROS, the physiological antagonist to the ROS signal which we name as the "insulin" response, though insulin is but a partial contributor to the genuine ROS signal. We can show the blockade like this:

All very simple. Now lets look at the Surwit diet, 59% fat calories, mostly coconut oil, very low PUFA (~2% LA) and modest fructose (~6% of calories).

Aside. Oooh, look, the Surwit diet provides something very close to the 5% of calories as fructose which was used to augment glycogen formation in dogs. Neat. End aside.

I was going to get in to a deep morass at this point about why MCTs fail to generate an ROS signal in proportion to their chain length and degree of saturation. The aside became progressively larger and, not unexpectedly, more theoretical. For the sake of the discussion of the Surwit diet we just have to accept that coconut oil contains fatty acids which are dealt with differently to longer chain fatty acids and which generate a limited ROS signal.

So let's go back to the Surwit diet, 59% fat, mostly coconut, 13% sucrose, some maltodextrin and some casein.

First, at 6% fructose this will produce an insulin-augmenting level of ROS generation, solid blue arrow below. Next is the glucose from the sucrose and maltodextrin, generating a fairly low level of ROS, mostly via the insulin receptor, shown as a thin arrow because this diet is almost a low carbohydrate diet. Both the above generate ROS which signal "insulin" activation "downstream".

Next let's add in octanoate. For whatever reason this is a poor generator of ROS under physiological levels of exposure. It will do nothing to inhibit the "insulin" effector actions of the carbohydrate generated ROS:

I suppose you could even make a case that the relatively minor production of ROS from octanoate might actually produce an activation signal for the "insulin" effect. A possible explanation for this paper. Then you're really in trouble.

Or at least your liver is.

During the earlier posts on this thread about the actions of fructose I've cited papers which suggest that fructose derived ROS augment the formation of glycogen within hepatocytes. Very clever people are very welcome to look at which substrates change in which direction activating which enzyme pathways to generate this glycogen. It's complex.

To me it's much simpler. Augmented low level ROS -> "do what insulin does". One effect being glycogen accumulation.

In a Surwit type diet the directly supplied fatty acids are heavily slanted towards medium chain fatty acids. Mammals do not use MCTs for bulk caloric storage, the preference is toward a mix of palmitate +/- oleate. The liver is quite capable of converting caprylate to palmitate +/- oleate. In fact MCTs are segregated away from chylomicrons by the enterocytes in the gut and are diverted, as free fatty acids, to the portal vein and so directly to the liver for this to happen.

Given augmented hepatocyte insulin signalling combined with augmented access to free fatty acids, what is the likely effect of augmented "insulin cascade activating" ROS levels?

Could that be the accumulation of lipid in the cells subject to this combination of circumstances?

We call this fatty liver.

That's the first step of several.

Peter

Fructose (08) Acipimox and FFAs

Just a tidy up post. I got bored with acipimox. This is why inhibiting lipolysis using acipimox doesn't make you fat.

Different acute and chronic effects of acipimox treatment on glucose and lipid metabolism in patients with type 2 diabetes

I've taken this from the top left panel of Fig 1 and removed the "day three" line so we just have the pre acipimox treatment solid line and the 28 day dashed line. I've added in red arrows to mark the acipimox dose times. Across the bottom are clock times for 24 hours:

It should come as no surprise that the AUC for FFAs on day 28 is virtually the same as that for pre treatment.

Whether this is a simple drug withdrawal effect or an effect mediated via increased size of already distended adipocytes ramping up basal lipolysis in the aftermath of being stretched by acipimox is hard to say, but the bottom line is it won't make you fat.

The beauty of linoleic acid as an obesogen is that it's there all the time from diet or released from lipid stores. All it needs is a decent level of carbohydrate induced insulin signalling for it to accentuate and you're away.

Under hypoinsulinaemia linoleic acid becomes (almost) powerless to augment insulin signalling because there's not much insulin there. Hence the efficacy of the low carbohydrate diet.

Should get back to fructose next.

Peter

Fructose (07) Acipimox tangent II

Acipimox is a drug which keeps on giving. Before I get back to fructose thoughts there's this post and maybe another given over to acipimox.

This is the current paper of interest:

Overnight Lowering of Free Fatty Acids With Acipimox Improves Insulin Resistance and Glucose Tolerance in Obese Diabetic and Nondiabetic Subjects

It has good points and bad points. The worst is the fact that the figures are of such low resolution as to be illegible in places, especially numerical scales to graphs.

On the plus side the tables are fine and the data extraordinarily confirmatory to my biases. Today I'm just looking at the "normal" slim people.

Here is the effect of acipimox on HOMA-IR from an on-line calculator using numbers from the slim, non-diabetic group:

Despite these subjects not supposedly having impaired glucose tolerance the HOMA-IR score derived from their mean fasting values is >2.0 suggesting some degree of insulin resistance. It's not easy to be healthy even as a slim 40 year old in Brazil.

This is easily corrected by acipimox giving an HOMA-IR score of 1.13, well under the cut off for IR.

It does this by locking fatty acids in to adipocyte triglyceride droplets and keeping them there. This, fundamentally, is what acipimox does. It locks lipids in to adipocytes.

From the ROS point of view there is then minimal fatty acid oxidation, minimal superoxide produced by reverse electron transfer through complex I, minimal inhibition of the insulin cascade at the level of insulin receptor substrate so maximal insulin signalling. The end result, in the basal state, shows as increased glucose oxidation and decreased fat oxidation:

These changes are absolutely secondary to the suppression of lipolysis by acipimox at the level of the adipocytes.

I hope all of this is starting to sound familiar.

I can't get this table out of my head

taken from here:

Insulin sensitivity is increased and fat oxidation after a high-fat meal is reduced in normal-weight healthy men with strong familial predisposition to overweight

Acipimox is reproducing both the low HOMA-IR score and reduced lipid oxidation seen in slim people who are destined to become obese. Ignore the comment about genetics at the end of the abstract. People with obese parents are going to become obese themselves via the action of linoleic acid locking triglycerides in to adipocytes.

Before obesity develops they have "excellent" insulin sensitivity because they are metabolically hypocaloric due to concurrently starting becoming obese via lipid *loss/sequestration* in to adipocytes. They will have to eat more [carbohydrate] to make up their metabolic needs by however much lipid is sequestered in to their adipocytes. They will only become insulin resistant once those adipocytes become large enough that lipolysis cannot be adequately suppressed by insulin.

Acipimox recapitulates linoleic acid's insulin sensitising and obesogenic effects, but only for 6 hours at a time.

Linoleic acid accumulates in your adipose stores (and deep fat fryer) and is available continuously for years making you hypocaloric [ie hungry], especially when you avoid saturated fat. As cardiologists have advised for decades.

Of course, if you took acipimox every 6 hours for the rest of your life you would get fat, wouldn't you? I only discovered acipimox in pre-Protons days when I worked from the simplistic, partially correct idea that insulin inhibits lipolysis to make you fat. Pubmed "lipolysis" "inhibitor" "obesity" and acipimox pops out.

BTW It doesn't make you fat. Basal lipolysis was another gift of acipimox.

Peter

Fructose (06) Acipimox tangent

I've glibly stated that I consider insulin resistance to be a simple but imperfect adaptation to the inability of insulin to suppress basal lipolysis, that is the rate of FFA release from isolated adipocytes in an ex vivo situation.

One of the best suppressors of basal lipolysis is the nicotinic acid derivative acipimox.

There's a review of our knowledge in 2007 here:

Nicotinic Acid Receptor Subtypes and Their Ligands

which has a nice diagram like this, slightly edited:

The process is quite simple. Acipimox, a nicotinic acid derivative (NA), interacts with a cell surface receptor (here HM74A but it has many, many names) which is coupled to a signaling G protein (G) which inhibits adenyl cyclase (AC), lowering cellular cyclic AMP so deactivating a cAMP dependent protein kinase (PKA) which does something magical to all lipases to shut down lipolysis.

Despite the fact that acipimox can reverse diabetes overnight it has been a clinical complete flop and is nowadays only used to improved a few lab numbers of people eating a diet of complete crap.

I woke up this morning thinking about why "Magic happens" in the above diagram. Not how, but why.

Niacin is a drug acting on the ketone body receptor of adipocytes. The endogenous activator is shown as beta-hydroxy butyrate:

The obvious explanation for ketone bodies inhibiting lipolysis is that this is a negative feedback loop. The original diagram also had FFAs inhibiting lipolysis too.

There are upper limits to the body's tolerance of elevated FFAs and ketosis is a marker that decreasing the FFA supply to the liver might be a good idea.

The most obvious inhibitor of lipolysis via hormone sensitive lipase is insulin. Under ketosis (ignoring MCT exposure) insulin is at absolutely rock bottom levels and if you want something to limit lipolysis under hypoinsulinaemia you had better develop a system which does so independent of insulin's action.

Ketones do that (pax hyperglucagonaemia and ketoacidosis, there are limits).

So if a cell has elevated basal lipolysis (which cannot be shut down by insulin) choosing an alternative mechanism, insulin independent, for suppressing lipolysis is effective stratagem.

Acipimox activates this process in the absence of the normal physiological ligand, ie ketones.

ATGL is key to mediating lipolysis triggered by increased adipocyte lipid droplet size where as elevated insulin exposure cannot do this. You can shut down this elevated basal lipolysis with a "ketone mimetic".

Which reverses type two diabetes. Until the drug wears off.

Peter

Oh, another thought: Is this how exogenous ketones improve insulin sensitivity? Never mind all those complicated intracellular switches (I told you I was lazy). The "problem" is that adipocytes are leaking FFAs. Acipimox fixes this problem. Acipimox is a ketone mimetic. Do exogenous ketones "mimic" acipimox? Amusing thought for the day!

Fructose (05) Four out of five diabetics

This follow on study looking at fructose in people with type 2 diabetes was a bit of a disappointment. 

TLDR: Fructose works pretty well in four out of five people with diabetes just as it does in "normal" people with a relatively poor OGTT result.

Acute Fructose Administration Improves Oral Glucose Tolerance in Adults With Type 2 Diabetes

This is what the title of the study is describing:

It's clear that adding 7.5g of fructose to a 75g OGTT load improved both glucose and insulin curves. There is nothing exciting about this. What I thought might be interesting was that, in a study of five subjects, two of them were outliers of sorts.

I'd hoped these outliers might give more insight in to the state of ROS signalling within hepatocytes of people with severe diabetes. Not really, much of the data you need is not in the paper. For completeness here are the anomalous results:

Subject 2 failed to drop their systemic glucose in response to added fructose, like this

Not unsurprisingly they also failed to drop their insulin level, clearly insulin level follows exposure of the pancreas to systemic glucose, which didn't change, so neither did insulin.

Can we guess what might have been happening in hepatocytes to nullify the "beneficial" effects of fructose? Let's assume that fructose is being absorbed, entering hepatocytes and generating ROS via a NOX enzyme in proportion to the rate of fructose ingress. So the question becomes

"Under what circumstances does a modest increase in ROS fail to activate the insulin cascade?"

We have no idea what the fasting insulin level was, nor the fasting glucose, so this is a little difficult. We are told that subject 2 had an HbA1c of 9.0 or 10.1, one of the highest in the study, which implies the worst average glucose excursion over the last three months or so. We could choose extreme insulin resistance, failing beta cell function or a combination of both to explain this. The fact that they produced a minimal increase in AUC for insulin when presented with 75g of glucose with or without fructose suggests that beta cell function was limited and underlying hyperglycaemia might both be present. I favour this explanation.

Both hyperinsulinaemia -> NOX4 activation via G protein coupled signalling and hyperglycaemia -> NOX2 activation via calmodulin kinase signalling have the potential to spill ROS over from activating via phosphatase inhibition to being inhibitory via insulin receptor substrate inhibition. Adding a small increase in ROS from fructose to a maximally stimulated system may affect both aspects, producing no net change. That's my guess.

The second unusual result was from subject 5. They produced a perfectly reasonable drop in systemic glucose in response to fructose addition but had an unusual increased insulin response despite reduced systemic glucose.

I struggle to explain this "paradoxical" rise in insulin despite a successful fructose induced fall in pancreatic exposure to systemic glucose. You could argue that the increase in insulin came first and this lowered the systemic glucose level but this would make quite an exception and need either a pancreatic response to a very small systemic fructose rise or a pancreatic effect derived from gut hormone signals, none of which were measured and none of which are needed for an explanation of the fructose effect in all other subjects.

If I had to guess I would suggest the fasting insulin level was very high in this person. We know from this 2011 study that OGTTs are reproducible on repeat testing (in "normal" people) over 3 days unless you have to be hyperinsulinaemic in order to be "normal":

Reproducibility of multiple repeated oral glucose tolerance tests

"However, our cases of individuals who exhibited hyperinsulinaemia in order to maintain glucose homeostasis, suggest that repeated OGTT’s may not produce a reliable estimation of insulin sensitivity for people with pre-diabetes and diabetes."

Subject 5 was also an individual with very high HbA1c (9.0 or 10.1) suggesting poor control and compatible with high insulin levels. So the insulin anomaly may just be random finding in one person secondary to chronic hyperinsulinaemia... We'll never know.

Overall the fructose effect on hepatic glucose output seems to be genuinely maintained in patients with DMT2 unless they are approaching seriously poor levels of control.

Should individuals with more "mild" diabetes add a little fructose to each bowl of porridge they consume in real life?

Rhetorical question.

Peter