Insulin load index / most ketogenic foods

[tim2000s]

@martykendall I've a strong interest in this too and I and others have noticed a number of factors relating to protein ingestion on low carb diets, especially where high protein levels are eaten.

Recently I've been looking to understand the proportion of glucogenic amino acids versus the ketogenic amino acids. I just uploaded an updated version of the manifesto doc - https://www.dropbox.com/s/if9cs6u0achx4lj/Optimising nutrition, managing insulin.pdf?dl=0. It seems that up to about 80% of proteins could be converted to glucose if someone was very inactive and they were eating a lot of protein etc... So 54% is not a maximium it is more a typical average as shown by the FII data from The University of Sydney testing.

In respect to this, while I have limited study based evidence to work from, I also have my sample of one, and I think that you'll find there are two processes being undertaken that need to be accounted for when looking at the impact of protein on glucose and how it needs to be accounted for in the use of bolus insulin.

I have observed that when I eat a high protein intake, without high carbs, I get an immediate reaction that requires an approximately 50% of equivalent carbs bolus, plus any required bolus for carbs, much as your calculations show. I am certain that this is not related to GNG.

We all regularly discuss GNG as a mechanism by which protein is converted into glucose, and I have observed that this is a process that does take place, however it is best observed when eating a low carb meal with a significant protein "lump", such as a steak or chicken. With these sources, you see a relatively slow increase in blood glucose over the following three to four hours that requires a regular bolus to manage it. It is a proper case of sugar surfing.

The other process that is continually ignored, but I think has a greater part to play in the use of mealtime bolus insulin is the impact of amino acids on insulin production, specifically Leucine, and the non-gluconeogenesis reaction that takes place. Leucine is a well known stimulant of the beta cells, actively increasing the amount of insulin produced in a non-diabetic person (NIH paper here). There appears to also be an effect on the alpha cells, which has less documentation, which is that in the presence of Leucine, Glucagon is released and glycogen is converted to glucose to counter the Leucine insulin increase.

I've done a fair amount of reading on this following observations of what happens to my glucose levels when ingesting Whey Protein and BCAAs relating to working out. The carb content is practically nil (typically 2 or 3g) however the blood sugar reaction is such that the only food source that can be having the effect is the protein. Taking it a step further, I experimented with a protein bar to see what the impact was. My reaction is detailed here: http://www.diabetes.co.uk/forum/threads/how-does-protein-affect-blood-glucose.75156/

In order to take this experiment a step further, I was able to eat one of these protein bars as a treatment for a hypo. It was extremely effective. it contains practically no carbs. What I've also observed is that in the presence of a higher blood glucose level, when I would expect that the internal feedback loop would communicate that a reduced glucagon reaction is required to counter the diabetic non-existent insulin reaction to protein, blood glucose still rises. This leads me to the assertion that protein, and most likely leucine, has a direct impact on alpha as well as beta cells.


I think the crux of this is that GNG is not the only mechanism driving blood glucose levels up when ingesting protein. The insulin requirement is also driven by a reaction to amino acids and this is the primary driver of the mealtime bolus requirement for protein. GNG is a much slower effect and it is not the initial bolus that treats this but subsequent later boluses. I have also suggested that slower acting insulins might have a profile that better fit the GNG curve when compared to the rapid acting insulins currently used.

I hope this adds to your research.

 

[Spiker]

The protein allowance is subject to a bunch of variables. I could argue it's 54% based on the precendent of Wilders' formula, 60% based on the actual data, or 80% based on the glugenic protential of amino acids versus ketogenic (see http://en.wikipedia.org/wiki/Glucogenic_amino_acid).

The link you posted does not make the claim for 80% glucogenic potential, nor do the three most obvious links in the article you link to. Do you have a link for this 80% claim?

If it's a claim that one or two specific amino acids might approach an 80% GNG efficiency then as I said I don't think that's very important. Unless we are going to do an amino acid analysis on all the protein we eat, and do individual dose calculations for each of a dozen or more amino acids? It's not very practical. I'm not even sure it's practical to try to avoid / prefer protein sources with high / low content of more glucogenic amino acids. I really think that is just tinkering and will be ineffective, for T1s anyway.

 

[Spiker]

@tim2000s that is interesting and surprising in equal measure!

I can see how amino acid stimulation of insulin production in healthy individuals makes sense, since the influx of amino acids anticipates GNG. This is the same as insulin secretion starting when the saliva glands detect carbs in the mouth.

I'm not convinced of a direct leucine - > glucose conversion that isn't GNG. An alternative GNG would be that GNG has different timescales for different amino acids and while some part of GNG takes longer to finish than carb metabolism does, some other parts of GNG start almost immediately.

We shouldn't assume GNG has a delayed start. I don't think I have seen that cited. Definitely GNG takes longer to finish but that doesn't necessarily imply it has a delayed start. I would interpret your protein bar test (wherever did you find a carb-free protein bar?!?) as a clear demonstration that GNG begins immediately.

What is harder to explain is why you saw a BG rise when consuming protein but already in a high BG state. But I think I can explain it. Did you see the same degree of BG rise that you would have seen from eating the same protein in a low BG state? Higher or lower?

You'll have to remind me Tim if you are still producing insulin. GNG is down regulated by insulin, not by carbs, not by BG. So if you were not producing insulin, or only had Phase II response, I would expect to see the BG rise from the protein bar, whether your BG was high or low. In fact a rise would be MORE likely when your BG is high, since in that state your blood is likely to be low in insulin.

And during a hypo, when your blood is artificially high in insulin, I would expect diminished GNG but not necessarily nil. This is probably uncharted territory for biologists. I don't know how quickly the down regulation occurs, and maybe an artificially induced hypo moves faster than the down regulation response. And maybe the critical need for glucose triggers a separate counter regulatory response to promote GNG. It seems sensible that would happen but I can't cite evidence on whether glucagon or noradrenaline promotes 'emergency' up regulation of GNG.

 

[Spiker]

@tim2000s from your description of the leucine action directly on beta and alpha cells are you saying that in an IDD it would trigger a glucagon release and then glycogen dump, hence the need for extra insulin? Otherwise the impact on a (NIDD) diabetic with insulin response would be a drop in BG and a need to reduce insulin, not increase it, is that right?

Have you ruled out other reactions causing the immediate BG rise via a hormone response, such as allergy to the whey protein?

 

[tim2000s]

I'm not convinced of a direct leucine - > glucose conversion that isn't GNG. An alternative GNG would be that GNG has different timescales for different amino acids and while some part of GNG takes longer to finish than carb metabolism does, some other parts of GNG start almost immediately.

I'm not postulating this. Given that Leucine has a direct effect on the Beta cells increasing insulin production, I postulate that it has a similar action on the alpha cells stimulating glucagon release, causing glycogen to be converted to glucose, rather than Leucine.

We shouldn't assume GNG has a delayed start. I don't think I have seen that cited. Definitely GNG takes longer to finish but that doesn't necessarily imply it has a delayed start. I would interpret your protein bar test (wherever did you find a carb-free protein bar?!?) as a clear demonstration that GNG begins immediately.

I don't think that GNG is a delayed start, but I don't believe it has the intensity that the protein bar demonstrates. Additionally, the protein bar contains a 20g protein content that is not the purest form of protein (i.e. Whey) but is milk protein. Even by the greatest stipulated conversion of protein to glucose under GNG, 80%, that is an ingestion of 16g equivalent of Carbs. The glucose spike that was ongoing was considerably in excess of what I see from 16g glucose. Hence why I don't believe this can be considered to be purely a GNG reaction.

What is harder to explain is why you saw a BG rise when consuming protein but already in a high BG state. But I think I can explain it. Did you see the same degree of BG rise that you would have seen from eating the same protein in a low BG state? Higher or lower?

I consistently see a similar BG increase (and therefore treat with insulin) whether my BG level is higher or lower, when taking whey protein. I haven't done the comparison with the protein bars.

You'll have to remind me Tim if you are still producing insulin. GNG is down regulated by insulin, not by carbs, not by BG. So if you were not producing insulin, or only had Phase II response, I would expect to see the BG rise from the protein bar, whether your BG was high or low. In fact a rise would be MORE likely when your BG is high, since in that state your blood is likely to be low in insulin.

I don't produce insulin, and my comparison is based more on whey protein than the bars particularly. There is still some experimentation to be done with the bars, so I can't confirm when using the bars where the effect is strongest. Based on the whey protein and the duration of the insulins I use, plus the timing of exercise and whey protein ingestion though, the rises, as said before, are consistent.

 

[tim2000s]

@tim2000s from your description of the leucine action directly on beta and alpha cells are you saying that in an IDD it would trigger a glucagon release and then glycogen dump, hence the need for extra insulin?

Correct.

Otherwise the impact on a (NIDD) diabetic with insulin response would be a drop in BG and a need to reduce insulin, not increase it, is that right?

In a non-diabetic I would expect the reaction to be a flat lined BG level as I would expect the glucagon and insulin to offset each other.

Have you ruled out other reactions causing the immediate BG rise via a hormone response, such as allergy to the whey protein?

Yes. I see the same reaction with scrambled eggs and breakfast meats, a meat and spinach omelette, non-whey protein and to a lesser extent chicken breast.

 

[tim2000s]

wherever did you find a carb-free protein bar?!?

It's amazing what you can find on bodybuilding websites!

 

[Spiker]

I responded on the thread @tim2000s posted. http://www.diabetes.co.uk/forum/threads/how-does-protein-affect-blood-glucose.75156/

You have convinced me that there is a prompt glucagon->glycogen response from certain free amino acids, particularly found in pure or purified protein foods such as eggs, whey protein, etc. A response that is additional to GNG. So thanks for that!

In turn I can see how someone might correctly have a 100% protein to carb ratio, if they were eating these kinds of foods. As per Tim's experience in fact.

Coming back to ILI and the manifesto, I think what would be much more useful than assuming a 54% protein ratio across the board, is to use the ILI to hunt for protein foods that have particularly high or low insulin response. These may well correspond to the eggs, whey protein etc. We then know to dose differently for these foods, and/or avoid them.

Funnily enough over the last few years my response to "the problem with protein" in low carb diabetic dieting is to minimise the reliance on protein, and to consistently eat the same protein foods. In a way this protein minimising and standardisation is a throwback to the minimising and standardisation of carbs in the diabetic diet prior to QA insulins and the resulting freedom of basal-bolus. So potentially here there is a "basal-bolus" revolution for protein that offers the same freedom while still low carbing.

I'm actually slightly depressed that this may mean there is no such thing as a constant protein ratio even for one individual at one point in time - that it depends on the specific food. :-(

I may have to start carrying around ILI tables.

 

[tim2000s]

I'm actually slightly depressed that this may mean there is no such thing as a constant protein ratio even for one individual at one point in time - that it depends on the specific food. :-(

I may have to start carrying around ILI tables.

I'm sure there will be an app for that

 

[martykendall]

Outside of the US and some other countries this can be ignored as fibre is already excluded from the carb count in the nutrition information. Are you sure that the Sydney data hadn't already excluded the fibre from their data?

My understanding is that in Australia and the US fibre is not subtracted while in the UK and some other European countries fibre is already removed, so it's a null argument. The nutrition guidelines have already decided it is irrelevant.

Unfortunately many people count total calories (including fibre) which I think is dangerous because it discourages vegetables which is not ideal.

 

[martykendall]

Me too, and in my experience your suggested ratio of 54% is much too high to be conservative. I will hypo when low carbing if I use a ratio above 33%. I suggest a conservative number to start with is 25%, test and titrate from there.

54% is also implying a degree of precision that simply doesn't exist given the level of variation. Given all the uncertainties, both theoretical and practical, I think you should use a round number, say 50% or 60% - though as I say I think those are too high.

Whether the maximum value is 54% or 58%, there's no way you can say it's "conservative" to use a maximum value that would never be reached even if a person ate nothing but pure protein (since even then not all the protein would enter GNG).

I've never heard this 80% number quoted. I would dismiss it as an outlier, rather than using it to justify well referenced experimental reports and well analysed metabolic sequences of the theoretical maximum as being "middle" values.

[/QUOTE]

I agree with the concept of being conservative initially titrating up from there for type 1s.

I couldn't find any data on the proportion of aminos that were glucogenic versus ketogenic so I ran some numbers on the USDA foods database myself.

Check out this spreadsheet https://www.dropbox.com/s/k1j2vabv9112y0t/food ranking.xlsx?dl=0 . Column KJ sums the total glucogenic aminos (74% by weight), column KL is the ketogenic aminos (12%) and the remainder is the aminos that fit into the "both" category (14%). Using the correlation analysis it appears that between 78 and 89% of protein not used by the body could be turned into glucose and stored in the liver and muscles for energy with the excess being converted to fat.

It appears to me that people haven't contemplated this much because protein has a slow impact on glucose levels, however I think this means that we underestimate the effect of protein containing foods on insulin, particularly on a low carb, hight protein diet.

 

[martykendall]

Yes as I said, he is unfortunately pretty vague about it, and no one else is speaking with any authority about it. So like you have done, I had to go to the underlying research papers. I then (as Dr Bernstein has always done) had to use myself as the experiment. So I have empirical data from a real human. There are a u other humans in this small data set, enough to convince me that effective protein ratios vary in T1 LCHF people, from below 25% to as high as 60%.

I'm on the TYPEONEGRIT Facebook group and some people are using 100% of protein as carbs. This is not advocated by Berstein and I would think on the high side for most people. But it illustrates the variability in protein responses.

My question would be how you would accurately test / titrate for the protein factor when protein takes a very long time to digest and show up in the blood stream (i.e. three to ten hours).

I think a lot of type 1s just probably already have their basal insulin set to cover the protein in their diet. This was one thing I noticed from reading the TAG page on the TuDiabetes. Lots of people needed to cut back on their basal insulin once they started to dose for dietary protein.

In theory it would be more accurate to set your basal with fasting and then have the meal bolus based on the carbs and protein. This would enable you to be more flexible with your meals.

 

[tim2000s]

It appears to me that people haven't contemplated this much because protein has a slow impact on glucose levels, however I think this means that we underestimate the effect of protein containing foods on insulin, particularly on a low carb, hight protein diet.[/QUOTE]
@martykendall, whilst I agree with this statement relating to the speed at which protein affects glucose levels and the effects on insulin requirements, especially on a low carb diet, high protein diet, I still think you are missing the non-gng component.

Looking at your ILI, it seems to mostly account for the early push up of the glucose level that I have observed in my experiments, that I am sure are not conversion of protein to glucose, without helping to offset the later highs caused by the gng conversion. This can be handled with longer term boluses on a pump.

One of the things I note is that my basal is set without gng being taken into account so after a meal with high protein (lunch today had 80g) I have to bolus small amounts at roughly 1.5 and 2.5 hours to stop the inexorable rise that is taking place. These amounts have been titrated based on bg increase rate rather than protein eaten earlier. Stephen Ponder' Sugar Surfing, if you like.

 

[martykendall]

I responded on the thread @tim2000s posted. http://www.diabetes.co.uk/forum/threads/how-does-protein-affect-blood-glucose.75156/

You have convinced me that there is a prompt glucagon->glycogen response from certain free amino acids, particularly found in pure or purified protein foods such as eggs, whey protein, etc. A response that is additional to GNG. So thanks for that!

In turn I can see how someone might correctly have a 100% protein to carb ratio, if they were eating these kinds of foods. As per Tim's experience in fact.

Coming back to ILI and the manifesto, I think what would be much more useful than assuming a 54% protein ratio across the board, is to use the ILI to hunt for protein foods that have particularly high or low insulin response. These may well correspond to the eggs, whey protein etc. We then know to dose differently for these foods, and/or avoid them.

Funnily enough over the last few years my response to "the problem with protein" in low carb diabetic dieting is to minimise the reliance on protein, and to consistently eat the same protein foods. In a way this protein minimising and standardisation is a throwback to the minimising and standardisation of carbs in the diabetic diet prior to QA insulins and the resulting freedom of basal-bolus. So potentially here there is a "basal-bolus" revolution for protein that offers the same freedom while still low carbing.

I'm actually slightly depressed that this may mean there is no such thing as a constant protein ratio even for one individual at one point in time - that it depends on the specific food. :-(

I may have to start carrying around ILI tables.

I think there is a subtle but CRITICAL differentiation here:

1. Some foods will digest quickly and raise your blood sugars. This is generally thought if in terms of glycemic index for carbohydrate containing foods, but I think there is a similar effect for proteins.

2. Some foods will digest slowly and will not raise your blood sugars significantly because they digest slowly but will still require a significant amount of insulin to metabolise.

3. Some protein containing foods have more ketogenic versus glucgenic potential based on their amino acid profile.

Most people focus on #1 because it's what they can see on their metre or CGM.

I would argue that #3 is largely irrelevant because most protein containing foods have a high potential to covert to glucose rather than ketones. I have looked into this and I can't see that there's anything useful or applicable. See section 2.10 of the updated manifesto at https://www.dropbox.com/s/if9cs6u0achx4lj/Optimising nutrition, managing insulin.pdf?dl=0

As someone who is interested in getting lean I am personally more interested in #2 - that is, the total insulin load of my diet - in order to not store excess fat and keep insulin levels low which will enable my body to use fat for fuel.

The recent ACCORD studies showed that dosing with meds or insulin to keep blood sugars down wasn't good for health. So based on this I would argue that keeping insulin down (rather than just blood sugar) is a priority for diabetics as well for long term health.

Rather than using insulin load to calculate insulin doses I think the primary power is to use it to select for low insulin load foods and then the margin for error is lower (whether you elect to allow for 0 or 80% for protein).

 

[martykendall]

It appears to me that people haven't contemplated this much because protein has a slow impact on glucose levels, however I think this means that we underestimate the effect of protein containing foods on insulin, particularly on a low carb, hight protein diet.

@martykendall, whilst I agree with this statement relating to the speed at which protein affects glucose levels and the effects on insulin requirements, especially on a low carb diet, high protein diet, I still think you are missing the non-gng component.


@tim2000s, can you clarify what ILI is? Sorry. Feeling naive.

I see lots of people on the TYPEONEGRIT group use the small doses to combat GNG. A nice luxury for those on a CGM (they're prohibitively expensive in Australia).

The TAGers will split their carb and protein doses with the protein dose being given as a square wave over a number of hours. This overcomes the issue with the insulin acting before the protein has time to digest.

Another possible option could be to dose for the insulin load (based on the carb and protein calc) with the meal and rely on the fact that insulin takes a long time to act anyway so your chance of the insulin driving you low before the protein digests is low, particularly if you start out with a high blood sugar at the meal. Probably safer to split the doses if you test low at the meal?

 

[martykendall]

@martykendall I've a strong interest in this too and I and others have noticed a number of factors relating to protein ingestion on low carb diets, especially where high protein levels are eaten.


In respect to this, while I have limited study based evidence to work from, I also have my sample of one, and I think that you'll find there are two processes being undertaken that need to be accounted for when looking at the impact of protein on glucose and how it needs to be accounted for in the use of bolus insulin.

I have observed that when I eat a high protein intake, without high carbs, I get an immediate reaction that requires an approximately 50% of equivalent carbs bolus, plus any required bolus for carbs, much as your calculations show. I am certain that this is not related to GNG.

We all regularly discuss GNG as a mechanism by which protein is converted into glucose, and I have observed that this is a process that does take place, however it is best observed when eating a low carb meal with a significant protein "lump", such as a steak or chicken. With these sources, you see a relatively slow increase in blood glucose over the following three to four hours that requires a regular bolus to manage it. It is a proper case of sugar surfing.

The other process that is continually ignored, but I think has a greater part to play in the use of mealtime bolus insulin is the impact of amino acids on insulin production, specifically Leucine, and the non-gluconeogenesis reaction that takes place. Leucine is a well known stimulant of the beta cells, actively increasing the amount of insulin produced in a non-diabetic person (NIH paper here). There appears to also be an effect on the alpha cells, which has less documentation, which is that in the presence of Leucine, Glucagon is released and glycogen is converted to glucose to counter the Leucine insulin increase.

I've done a fair amount of reading on this following observations of what happens to my glucose levels when ingesting Whey Protein and BCAAs relating to working out. The carb content is practically nil (typically 2 or 3g) however the blood sugar reaction is such that the only food source that can be having the effect is the protein. Taking it a step further, I experimented with a protein bar to see what the impact was. My reaction is detailed here: http://www.diabetes.co.uk/forum/threads/how-does-protein-affect-blood-glucose.75156/

In order to take this experiment a step further, I was able to eat one of these protein bars as a treatment for a hypo. It was extremely effective. it contains practically no carbs. What I've also observed is that in the presence of a higher blood glucose level, when I would expect that the internal feedback loop would communicate that a reduced glucagon reaction is required to counter the diabetic non-existent insulin reaction to protein, blood glucose still rises. This leads me to the assertion that protein, and most likely leucine, has a direct impact on alpha as well as beta cells.


I think the crux of this is that GNG is not the only mechanism driving blood glucose levels up when ingesting protein. The insulin requirement is also driven by a reaction to amino acids and this is the primary driver of the mealtime bolus requirement for protein. GNG is a much slower effect and it is not the initial bolus that treats this but subsequent later boluses. I have also suggested that slower acting insulins might have a profile that better fit the GNG curve when compared to the rapid acting insulins currently used.

I hope this adds to your research.

Wow. Thanks @tim2000s.

These are questions I'm trying to get my head around.

Leucine and lysine are the two amino acids that are exclusively ketogenic. See http://en.wikipedia.org/wiki/Ketogenic_amino_acid. I'll have to read the paper, but I wonder if the ketones produced might look like an insulin reaction. Did they measure insulin?

Again, I think there are two issues. One is that eating protein seems to stimulate release of glucagon from the liver. Hence increase in BG is not directly related to ingested protein.

But at the same time the liver act as a reservoir for glucose that that reservoir can be filled by GNG from ingested protein. If you get to a point where the liver is full then the excess spills over into the blood. Perhaps this is why you see a BG rise some hours later, once the liver reservoir is refilled.

My point though is that I think regardless of when it's used the glucose that's produced from GNG (from protein in excess of the body's requirements for muscle repair) will require insulin at some point whether it is stored temporarily in the liver or not.

Moreover, reducing the insulin load from carbs and protein will reduce the size of the waves to a more manageable size when it comes to your "sugar surfing".

What do you think?

 

[Spiker]

total glucogenic aminos (74% by weight), column KL is the ketogenic aminos (12%) and the remainder is the aminos that fit into the "both" category (14%). Using the correlation analysis it appears that between 78 and 89% of protein not used by the body could be turned into glucose and stored in the liver and muscles for energy with the excess being converted to fat.

OK but, 100% of the weight of glucogenic (and multipurpose) amino acids entering GNG do not turn into 100% of that amount of glucose. 100% of amino acids entering GNG turn into 58% of that weight of glucose. The rest are lost due to overheads and inefficiencies of the process. Partly this is just thermodynamics - you always lose energy (as waste) when you change the state of something from one state to another.

 

[Spiker]

It appears to me that people haven't contemplated this much because protein has a slow impact on glucose levels, however I think this means that we underestimate the effect of protein containing foods on insulin, particularly on a low carb, hight protein diet.

There is no doubt in my mind that LCHP diets fail, and the reason they fail is that people (including me) just replace carbs with GNG-generated carbs from protein. Nothing really changes.

 

[Spiker]

My question would be how you would accurately test / titrate for the protein factor when protein takes a very long time to digest and show up in the blood stream (i.e. three to ten hours).

Same as anything, frequent testing. Lots of people on here routinely do +2 +4 hr tests after food. That would cover most protein action.

It doesn't require inventing a new dosing regime to fix this problem. It just requires people to apply existing rules of correct basal testing and correct bolus testing. The only really tricky part is matching the insulin to the protein without a pump.

Well, ok, and the calculation for the protein itself is not trivial, as we've discussed. But finding the correct value by experimentation is not difficult and not that much harder than finding the correct value for carbs.

 

[Spiker]

I think a lot of type 1s just probably already have their basal insulin set to cover the protein in their diet. This was one thing I noticed from reading the TAG page on the TuDiabetes. Lots of people needed to cut back on their basal insulin once they started to dose for dietary protein.

Yes, and the same is true for carbs - lots of type 1s have their basal covering some of their standard daily carbs. But the right fix for that is to do correct basal testing, which requires it be fasting testing.

But yes I get your point, the ability to obscure real variations with a blanket of basal does tend to cover up imprecise bolus dosing, whether carb or protein, or fat even.