How much protein will I lose when I fast?
There is an excellent series of Videos online from this years Low Carb Vail conferences – but one of the interesting things about this conference was how significantly experts disagreed about the correct amount of protein people should eat, and in particular how much lean body mass we lose when fasting.
Dr. Jason Fung uses data from Kevin Hall published in “Quantitative Physiology of Human Starvation“. Dr Fung says in Therapeutic Fasting – Solving the two compartment problem –
Are we burning muscle? Hell no! It goes up slightly at the very beginning and then drops. Protein is not a storage form of energy.
Dr. Stephen Phinney in The Case for Nutritional Ketosis – uses data from George Cahills “Starvation in Man” study. Dr. Phinney says
In starvation your muscles are broken down. Even after a month of total starvation – no protein coming in, you are losing 4 oz of lean body mass every day.
What we have here is not just a difference of opinion … but two fundamentally different analyses of human starvation metabolism. Either one or both of them is wrong in general … or they are both right for a specific subset of humans.
So I looked around for any data that might give me a gut feeling for which might be closer to the truth. And found the story of Angus Barbieri (SHOW ME THE SCIENCE) who fasted for 382 days. He started at 456 lbs and ended around 180 lbs, and he could not have lost 4oz of lean body mass every day.
His plasma Urea, which is a marker of his consumption of amino acids (to either turn into glucose, or oxidise for energy) dropped to less than half of the normal rate for a man of his age. This example would appear to support Dr. Fung.
So I decided to do the math … and I apologise in advance for using a bastardization of metric, and imperial units – but nutritionists are most used to dealing with body mass in pounds (lbs), energy in kilocalories (kCal), and food mass in grams (g);
Let’s start with how much energy the Brain uses in a day;
Cahill et al., 1968 showed in “The consumption of fuels during prolonged starvation” that the adult Brain uses between 110g and 140g of glucose per day. Let’s say to be generous: 600 kCal
In the same study George Cahill’s team showed that once adapted to starvation, up to 2/3rd of the brain’s energy can be met from Ketones, leaving only 1/3 to be met from glucose.
So we need 400 kCal of ketones and 200 kCal of glucose to adequately fuel the brain. There are of course other glucose dependant tissue (red blood cells, renal tubules, etc). I don’t have data with which to include those, but when I do I shall.
Converting this to mass, this means we need roughly 50 g of glucose (200 kCal / 4 kCal/g) and 80g of ketones (400 kCal / 5 kCal/g ) – just to keep the brain alive.
I did the math once to work out the ratio of glucose that can be produced through Gluconeogenesis from the glycerol released from a model Triglyceride containing the 3 most common fatty acids in humans (Oleic: mw=282.46, Palmitic: mw=256.43, Linoleic: mw=280.45) and glycerol (mw=92.09) in order to work out the mass ratio of glycerol to that triglyceride … and it turns out that for every 10g of body fat burned there is very close to 1g of glucose that can be produced (10g Triglyceride -> 1.0104g glycerol ->0.988297g glucose).
To produce those 50 g of glucose it turns out that we need to burn 500 g of body fat to produce sufficient glycerol as a substrate for gluconeogenesis, or 1.1 lbs.
So to use ABSOLUTELY no protein to make glucose necessary to fuel his brain Angus Barbieri would have to be burning 1.1 lb/day of body fat. Using the rough figure of 3500 kcal/lb of body fat we can work out that would be 3850 kCal of body fat he would need to be drawing down every day – which is a pretty impressive metabolic rate but that is possible in a 456 lb man who is performing modest exercise.
We know (SHOW ME THE SCIENCE) that the maximum rate we can draw down body fat is around 31.5 kCal/day per lb of body fat. So he would have had to have had at least 122.22 lbs of body fat to be able to draw down 3850 kCal from it – which he probably had.
So I think we can say if he did enough exercise to keep his total daily energy expenditure up above 3850 kCal then, as long as he had more than 122.22 lbs of body fat, he could have got away with consuming NO protein to keep his brain adequately fueled for at least the first part of his fast until his availble body fat dropped below 122.22 lbs.
So a grossly obese subject such as Angus Barbieri would appear to support Dr. Fungs observation.
How about say a 158 lb lean weight lifter, with 10% body fat. He would also need to make 50 g/day of glucose assuming his brain was the same size. The problem for him is that he would only have 15.8 lbs of body fat and therefore he would only be able to generate 497.7 kCal/day from oxidizing body fat (15.8 lbs * 31.5 kCal/lb), which would consume 55.3g of body fat per day. Using the glycerol from fat to make glucose would produce 5.47g of glucose to contribute to his brain’s energy requirements, leaving the shortfall (44.53g) to be made up from other substrates. If that was all protein he would require 89.06g of protein from his body to be converted into glucose every day.
That is a lot closer to Dr. Phinney’s observation, and may be a better reflection of the subjects of George Cahills experiments in the 70s before the low fat revolution.
So yes … Dr. Phinney and Dr. Fung may both be correct, in context.
Previous Post
Next Post
Simply brilliant, enlightening and with a thorough explanation.
Thank you for putting this together as well as the podcasts!!
Still not sure I haven’t missed something important.
I’d pencil it in as a provisional hypothesis for now.
How does the glycogen store in the liver play into the extended fasting picture? Would it be used up first or in parallel?
My personal experience with my last extended fast (5 days) vs shorter fasts resonates with the implications of the math above. Especially if the glycogen were to be depleted first over the first few days.
Most of the data for fat transfer rates was from people who had been starving for over 40 days, in Ancel Keyes starvation experiments. They would have had no glycogen after the first 2 days.
The data that Dr Phinney was using was also after 2 weeks of fasting, so again there would have been negligible glycogen.
I suspect on a shorter fast, like a 1 day fast that liver glycogen could be used to supplement the obligate brain glucose requirements so it is possible that the scenario I propose would not come into effect until 2 days into the fast.
Hi Richard,
I really enjoyed this show but have a question about your calculator.
I’m super-low body fat at the moment. (We are talking around 6% according to my DXA scan result last week).
Similar to you my birthday is next week and I’m turning 50 yrs old. I made a decision a year ago to get in the best shape of my life for my 50th 🙂 I want a 6 pack abs by my birthday. I have been following a LCHF diet all year (no alcohol also) and feel amazing.
I’ve dropped 18.5 Kilos (I’m an Aussie)/40 Pounds since the beginning of the year and my % body fat has gone from 31.2% to as low as 5.7% last week as well has adding just over 9 kilo’s/19.8 pounds of Muscle Mass since the 1st Jan 2016. I have done “I.F” the whole time with a 15 hr fasting period each day. I have only done 8 days of “full” fasting since the beginning of the year but those have really been “Fat Fasting” days where I have exogenous ketones via BP coffee or Brain Octane/MCT oil & butter in Bone Broth (which I have called “upgraded bone broth”).
So here’s what I have noticed having just finished a 3 day fat fast…
I dropped my body fat an extra 2.5% over 3 days and gained muscle mass consuming approximately 90grams of fat each day but no carbs or protein.
I realise this is only an observation but shouldn’t this calculator account for the ability to consume fat as the source of extra fuel for someone who is low body fat rather than it have to come from protein?
Yes that is an excellent idea. It’s like a cross between the fat deficit calculator and this one.
I think we might end up doing a real keto calculator that takes all these things into account. But I’m keen for some real experts to weigh in on my speculations here. Until then, I’d have to call this provisional hypotheses.
Hi Phil,
Interested in what you are doing. When you are doing your 15hr IF, are you eating a ‘normal’ amount of calories in the 9hr window?
When Fat fasting, how much are your reducing your overall calories by?
The implications of Richard’s calculations for this (for me) is that, as a lean person, the body can provide relatively few calories daily from body fat, so any fasting should be fat supplemented.
Great explanation of how to reconcile these two views. I read Dr. Fung’s happened to come across this one for unrelated reasons, and I had some similar ideas but without such a good explanation.
Something that still seems like it might be a bit off though: for Dr. Fung’s objection, he mentions his own fasting habits and seeming lack of muscle loss as well, and he (as far as I can tell) is not overweight. I’ve heard from others that are likewise not overweight that rely heavily on longer term fasting than Dr. Fung typically does (as far as I know, thinking week long of 4 day fasts, though also 30 day). I’m not sure where his body fat is of course, so some things are hard to say, and we don’t have the data on him and the others I refer to of measurements before and after, and I doubt in any case we are dealing with sub 10% bodyfat individuals. All that being said, it also seems unlikely that they would have enough bodyfat to meet the needs you describe without resorting to use of protein somewhere during such fasts under the conditions described (if I understand correctly).
I believe Dr. Fung suggested the protein can be taken from sources other than muscle, such as from the breakdown of skin tissue, but that seemed even in his blog to primarily apply to those loosing much weight (already overweight), so I don’t know if that would play into much here.
On Dr. Phinney’s side, beyond the external data he looked at, I’ve been under the impression he and Dr. Volek have done a lot of their research and experiments on pretty high level athletes anyway, which would favor your explanation of working with people both already lean and very active, who may really require more protein as a result.
In any case, do you think the calculations you show and explanation can account for someone like Dr. Fung and others of normal-lean (but not super lean athletes) being able to fast frequently without muscle loss? From your calculator, it looks like you’d have to be 75% body fat at 200 lbs, or 50% at 300lbs, to not require any intake of protein. Even in the case of Angus Barbieri, it seems unlikely (to me) that he would maintain enough bodyfat throughout his fasting time as he continuously lost weight (granted, I don’t know off hand if he lost muscle towards the end).
Yeah towards the end of Angus Barbieri’s fast he would have been gradually using more protein to make glucose for his brain – but until he had 124lbs-ish of body fat he could have needed to use none.
I agree with Dr. Fung that there are many sources of protein that are not muscles. We know that HGH is increased during fasting so it could well be that muscle turnover is low on the totem pole. I would expect the glucagon mediated autophagy during fasts would be in a position to liberate protein from cells in line to be recycled. And I would hope that would include empty adipocytes, skin cells.
Also there are sources of protein like the collagen matrix that now depleted fat cells are supported by, lipoproteins no longer needed to transport the now much lower amount of dietary fatty acids and liver repackaged free fatty acids.
As for why Dr Fung doesn’t appear to lose protein – You make a good point. I don’t know that he doesn’t lose protein. We’d probably have to know how much urea he is producing to really know. I’m sure he has the ability to run those diagnostics easily.
Hi Richard,
Thanks for this. This is is very interesting. But I am puzzled by some of the implications. As a relatively lean runner, this calculator says I should only be able to access ~650Kcal per day, but I know from experience and events that my body can provide far an excess of this. Obviously, some initial energy will be coming from muscle glycogen, but once this is gone, does that mean that I have only 650Kcal available before muscle starts to be broken down?
If true, I need to do less fasted exercise and ensure that my fat stores are topped up with dietary fat during long exercise.
I believe that is exactly the case. You will draw down the energy that your body fat is willing to provide, you will additionally have a little energy in circulation in lipoproteins, you will also have some muscle glycogen, and then you will be into burning protein for energy. That doesn’t necessarily imply muscles.
You have maybe 1% of your protein at any time that is labile, and then you have structures that are excess to requirements like excess skin, or the extracellular matrix that used to support now unused fat cells. But keep going and you will need to break down muscles.
The alternative is to supplement with a little dietary fat. I Have myself experimented with doing fasted hill sprints on my bike until I feel cold in my extremities, and then eaten a teaspoon of butter and within minutes the symptoms disappeared.
I repeated the experiment without the butter and the symptom persisted for about an hour. So I think that cold extremities might be a sign of some budgetary reconfiguration and maybe the heating bills are no longer being paid.
I want to follow up with Andy’s comment. I ride a bicycle and am training for a 150 mile ride for MS which is 75 miles each day. I have approximately 35 lbs of fat so for the calculations I can access (35x3500x.10)or 1225 calories for fat and the balance should come from glycogen storage or supplementation. Storage for the entire ride sounds impossible. I am trying to avoid the sugars and fruit and cereal because I understand from Dr. Naiman that the insulin will prevent lipolisis. If I burn 1500 calories an hour for 5 hours at 15 mph I need 7500 calories. Would I have to gain fat of 150 lbs to make the ride on storage 7500-1220-1500(approx glycogen storage)/31? I do not know how to supplement for that much fat. Perhaps the extrapolation of the formula may not be correct.
I believe your energy contribution from 35 lbs of body fat would max out at
31.5 kCal/lb/day x 35 lbs = 1102.50 kCal/dayFor the event you will also have whatever triglycerides are circulating in lipoproteins, so chylomicrons carrying fat from a recent meal, and LDL particles carrying fat from the liver. You’ll also have glycogen stored in muscle and your liver.
Dr. Naiman is correct that eating glucose or starch will prevent lipolysis during the event. That will raise insulin which will inhibit your fat cells releasing energy, and it’ll inhibit fatty acids getting into your energy organelles to be oxidized. So even if you have circulating lipids you’ll not be able to use them until your insulin drops.
I’m neither a physician, nor an exercise physiologist, but I suspect the algebra will look like this;
7500 kCal = 1102.5 kCal (body fat) + 1500 kCal (glycogen) + 4897.5 kCal (circulating triglycerides)That last figure works out to be around 1.2 lbs of fat in circulating triglycerides which would likely be from a recent meal. You might be able to also carry some fat for the road like one might a starch gel, but I’m sorry I don’t know if there is a limit on fat digestion during exertion as there is for starch.
It does sound like a fun ride, and an interesting exercise to adequately fuel it – best of luck with the ride.
This article https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1274154/pdf/westjmed00207-0055.pdf mentions that protein loss in obese persons is 3-6 g/day (no timing provided).
3-6g / day Nitrogen loss equates to 19g to 37g per day of protein intake to offset. This subject was a single 41 year old non-obese religious person. His rates were 10-12 in the first week, 5-7 after the 3rd week.
This shows the contextual difference between someone with adequate body fat to supply their caloric needs, which also contributes substrate for GNG which will also defray the loss of protein. Compared to someone who may not have sufficient body fat so they may need to recruit protein as both a substrate for energy for survival as well as the primary substrate for glucose for the brain.
As my math showed, it is possible if you have enough fat and run your metabolism high enough, that you are generating sufficient glycerol that you may not need to use protein at all for GNG.
The other thing I would point out is that the difference between the first week and the 3rd week is ketoadaptation, the liver making more ketones, the body protein sparing in response to circulating ketones, and the brain needing less glucose because it’s running 75% on ketones. Most of us who are keto and fast can probably expect to start the fast in a protein sparing state.
This is very interesting!
But I believe you need 2 molecules of glycerol to make 1 molecule of glucose, since glycerol has 3 carbon atoms and glucose has 6. (Molar mass 92 g/mol for glycerol and 180 g/mol for glucose)
This would imply that you only get 5g of glucose per 100g of triglycerides via GNG, which in turn would mean that you need DOUBLE the body fat than what Richard has calculated to keep from not using any protein at all to fuel the brain.
But I can most certainly be wrong…
It is true you can make 1 glucose from 2 glycerols, but the glycerol molecule (92.09382 g/mol) has roughly half the molecular weight of glucose (180.156 g/mol).
By my calcs if you have 230g of fat being burned, you have roughly 23.23g of glycerol, which would be 0.2523 moles of glycerol, doubling that to make glucose and you’d end up with 0.1262 moles of glucose, which would be roughly 22.73g of glucose.
Very interesting. But does it apply to short term fasts or just long term fasts? ie If less than 24 hours, do you have to worry about this?
I also wonder if you ever combined this with the max amount of fat loss calculator. That one tells me I need to eat about 750 calories of fat per day, since I can’t get it all from my body fat. But this one says I need 336 calories of protein a day to fuel my brain. If you add these together I can only do a 300 calorie deficit per day… hardly a fast. So not sure how to combine these. Is it really that I need 750 total calories (even though it says fat calories) and that 336 of them should be protein? Thanks!
Interesting post.
How does one factor into this, the seemingly adopted fact that higher protein intake helps one conserve lean body mass ( https://www.reddit.com/r/ketogains/comments/8qwnzb/one_of_the_myths_busted_by_the_faq_aint_a_myth/ )?
Some of this studies compare groups with equal calorie intake, with a calorie deficit and normal and high protein groups. The high protein group tends to conserve or gain more lean body mass. If what is presented in this post is true, the case where calorie deficit is below what’s available to process/day, the protein intake should not make a difference.