Quantcast

Intermittent and extended fasting

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
For those of you perhaps wondering in your head if CGM is right smack in crazy town:

The point of all this fasting and lowering carb business is to achieve lower glucose over time, and thus lower insulin over time. These are empirical methods of getting this done.

CGM is just the next step up in data to achieve this very same goal consistently, when used in this context for non-diabetics. It’s the continuous integration of the same data I strove to acquire with my carb challenge.

I think it’d be kind of amusing to show up to my PCP appointment on Dec 18 wearing a (self-prescribed) CGM monitor. At least then I could show hard data other than the two crappy data points of height and weight taken by their MA.
 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
This is a fantastic video re risk of high LDL-P if otherwise metabolically healthy

 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
Just called Dexcom to start the ball rolling on acquiring a G6 CGM. Turns out it’s a two step process: established an account with them on this phone call, answered some non-relevant questions (“what kind of diabetes do you have?” “None” “Can we get your insurance information?” “Not needed, as I’ll self pay as they won’t reimburse since I don’t have diabetes”), and provided them with my doctor’s info (me). Someone else with more domain-specific knowledge as opposed to the front line call lady will call back or email within 24 hours and continue this process.

Also ordered different lancets from a medical supply company. The tiny lancet used with the wife’s 0.6 microliter per test glucometer was entirely insufficient for the 40 microliters of blood that the Cholestech LDX needs. These new ones are much more serious business, 28 to 21 gauge, 1.8 to 2.0 mm depth. That’ll bleed, alright.
 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
They probably don't have anything on their bodies to ketoninate.

:p
That's totally not a word! (The process is of lipolysis, ketosis, then ultimately oxidation of ketones.)

Anyway, here's one of the activists from said group:



He's got a couple weeks' worth of fasting stores there.
 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
I picked up machine #2 from the post office today (machine #1 having been DOA). This one works!



Results from draw #2 (#1 having been the one just before I figured out that machine #1 was broken):

Context: Fasted 17 hours. Ate probably 150 grams carbs the day before, but also much of a lovely wagyu hybrid ribeye at dinner, so not a keto day for sure but not SAD, either.

Total cholesterol 263 mg/dL

HDL-C 25 <— I am guessing this is artifactually low as I’ve never been below 45 before

Triglycerides 53 (!!)

Calculated, corrected* LDL-C of 168
Relevant to others who may follow down my path, or who even get standard labs while on a similar diet:

Note the asterisk and “corrected” for that LDL-C. What the machine spit out was 227, but that’s because its assay (like most even used in real labs) doesn’t measure LDL-C directly. Instead it calculates it via the Friedewald Equation, which is:

LDL-C = [total cholesterol] - [HDL-C] - [non-HDL cholesterol] / N

Where N = 5. (This last term yields VLDL, for the curious.)
The problem with this equation is that the relationship between non-HDL cholesterol and VLDL falls apart in the setting of very low triglyceride values. How to correct it comes via a lovely JAMA paper that published a table of N (from the equation) values for different pairs of triglycerides and total cholesterol.



From this table and my values today N = 3.4, thus 168 mg/dL corrected calculated LDL-C. Quite a bit different, eh?
 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
Nothing super new here but a nice overview:

 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
Now for a Sunday mini science dump, most of these from diving down rabbit holes this morning before the kids awoke and I arose:

1) Eating a high-bacon diet is protective against colon cancer precursors, at least if you are a rat.

2) More on the whole “metabolic health changes how we should interpret individual biomarkers” bit with regard to lipid metabolism:

A) Small dense LDL (sdLDL, or “subclass B” in this paper’s terms) is what drives atherosclerosis, in part through screwing with NO/ONOO- ratios.

Conversely, large buoyant LDL (lbLDL, “subclass A” here) fixes those NO/ONOO- ratios so is protective… even though both are measured as part and parcel of LDL-C as a whole.

B) Who has bad ratios of sdLDL and lbLDL, with too much sdLDL? Did you guess people with metabolic syndrome or outright type 2 DM? Because you’d be right.

(See a pattern here? Insulin resistance really is the driver behind so many things!)

C) What’s worse than sdLDL for causing atherosclerosis? Oxidized sdLDL.

What drives oxidation? Elevated glucose levels. This promotion of LDL oxidation has been known for decades in diabetics, with this paper showing it promoted at 450 mg/dL glucose. But it’s much more than that admittedly very high level that causes oxidation: this group showed increased LDL oxidation in people with impaired glucose tolerance, defined as > 140 mg/dL on a 2 hour oral glucose tolerance test.

Did I make that clear enough? Glucose levels seen with pre-diabetes—which is insulin resistance by definition—increases LDL oxidation and thus cardiovascular disease risk.

D) What decreases LDL oxidation? We can theorize that sustained low glucose levels would do so, thus the whole continuous glucose monitoring idea.

This rather crappy little paper did show that age and BMI-matched patients with not substantially different LDL densities (e.g. not sdLDL-prevalent in one group per points A and B above) showed an association with athletic training and LDL oxidation resistance. They didn’t show why this happens.



Cliffs Notes: For good colon health eat more bacon—attn @SkaredShtles . For heart health: Exercise. Become more insulin sensitive.
 
Last edited:

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
Two more to round it out, re utility of low carb high fat in improving lipid markers: Not talking about LDL-C here—pay attention! But rather reducing sdLDL, reducing Lp(a), reducing oxidized LDL and oxidized sdLDL in particular.

This nice 4 week per arm crossover study showed that one month on each diet could demonstrate that low carb high fat diets lead to lower Lp(a) and lower oxidized LDL.

This paper sums up the evidence more broadly, noting that I haven’t read each cited paper in turn:

The assumption that a low-fat diet reduces the ‘bad’ cholesterol (ie, LDL) is an imprecise notion. While total LDL may be lowered with a reduced intake of dietary fat, if replaced with carbohydrate, this may increase sdLDL particles (ie, pattern B),10 ,11 which are more atherogenic than large buoyant LDL particles (ie, pattern A).12 Additionally, data indicate that a high saturated fat intake lowers sdLDL particles and raises large buoyant LDL particles.13 Thus, replacing carbohydrate with fat may improve the LDL particle size distribution (eg, pattern B shifted to pattern A). Lastly, if fat is replaced with carbohydrate, this may worsen the overall lipid profile (decrease in HDL-C, increase in triglycerides and increase in sdLDL particles).10 ,11
 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
More re continuous glucose monitoring, and whether there's evidence to support Peter Attia's and my working theory, that "flatlining" as much as possible ~90 mg/dL is preferential if one's goal is reducing cardiovascular disease:



That's from https://www.sciencedirect.com/science/article/abs/pii/S0002870312003134 . These data seem to show that lower is better, even within the nominally normal range. This is evidence for the 90 figure itself, which includes fasted and post-prandial periods.

Then we have this study: https://onlinelibrary.wiley.com/doi/full/10.1111/j.1464-5491.2010.03059.x . This one shows that endothelial function is impaired when glycemic variability is higher, even in the subgroup of people who are fat but don't have DM2 or metabolic syndrome. (Although 30% in this subgroup smoked and 27% were hypertensive, so not quite ideal there. A1C of 5.3% +/- 0.2 in this group.) This is evidence for the flatlining bit, of minimizing the variation.
 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
I’m planning on fasting from after dinner tonight through Thursday morning or noon, assuming I feel good.
Still on my fast. 72 hours in at this point. I think I’ll resume eating tomorrow morning. Got some things I’d like to eat: shoku pan with delicious Kerrygold butter, some unagi sushi, and who knows what else before afternoon Thanksgiving foods at my sister’s place.

I skied fasted today. 13,000 vert at Copper. Felt fine. Easy pace most of the day as I was skiing with my wife.
 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549

Fascinating. T cells mediate “obesity memory” in rats. Unclear if true in humans to same extent or what to do about this, but food for thought indeed.
 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
Mini science dump:

1) Vegetable oil leads to obesity. Counteract via adding omega-3s (and avoiding omega-6 vegetable oils in the first place! ubiquitous in packaged foods).


2) Vegetable oil reduces LDL-C... but in reanalysis of the original data it does NOT reduce coronary heart disease death or all cause mortality. Indeed there's a non-significant trend towards higher death, again with lower LDL-C.


3) Half life of linoleic acid (read: high omega-6 vegetable oil) in human fat about 680 days. So if one reduces vegetable oils today then it'll take about 9 years before 97% of it is out of one's fat:

 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549

Interesting. Stearic acid enriched foods are difficult to come by. References this:

 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
More or less final data from my n=1 experiment with the goal of illustrating the effect of fasting duration on triglycerides.

Screen Shot 2019-12-08 at 11.14.12 AM.png


These data were acquired over multiple fasts, not just multiply-repeated samples at each time point from one fast. I included the data points from glucose as well to show that the elevation in day 2 triglyceride was independent vs. some cortisol spike-mediated mechanism, for instance. I didn't include HDL-C or LDL-C as I don't trust the data for the former, and both seem more or less independent of fasting duration.

Moral of this short tale is that one should fast for at least 14 hours before a lipid panel (but before maybe 27 hours and not on day 2 of an extended fast!), at least if one eats and metabolizes things similarly to me.

So what next now that I've done this, and earlier the carb challenge? (The latter with the wife's glucometer that reads 13 points higher on average from the same fingerstick as compared to my Cholestech LDX's lipid + glucose cassettes…) I'm thinking that I'll do a fat challenge.

What's this? The short version is that it's an oral triglyceride tolerance test, analogous to the carb challenge being an oral glucose tolerance test. One can do the challenge either with pure fat (e.g. 1 g heavy cream per kg body weight) or with a mixed input (e.g. 50 g fat and 50 g carb to mimic a fast food meal).

But why do this, one might ask? This article explains the background and why this test's results might be relevant. My summary of these findings are:

a) The predictive value of fasting triglyceride levels with regard to vascular disease largely goes away in most trials once one accounts for HDL-C levels. (Why? Likely because low HDL-C and high fasting triglycerides comprise metabolic syndrome/insulin resistance, so both are largely due to the same underlying cause.)

b) Non-fasting triglyceride levels seem to be equally predictive of vascular disease. I don't think anyone is claiming that an individual non-fasting value has added incremental predictive value over a single fasting value, mind you, but it's plausible to me that measures such as triglyceride area under the curve from 0-6 hours since eating may add information.

c) Interpretation of a single non-fasting triglyceride level is very crude, which isn't surprising given heterogeneity in what people eat and at what interval from eating (1-8 hours?) the data were acquired. Thus the challenge, with a standardized meal.
I'm guessing that this will prove to be yet another measurement that correlates with overall insulin sensitivity, but I'm always game apparently to extract more blood from myself in the name of science. (And, at the moment, I have 6 remaining Cholestech LDX lipid + glucose cassettes.)

So I'm thinking that I'll do one last series of tests on December 18, at 0, 2, 3, 4, 5, and 6 hours after eating. The 0 will serve as calibration with the official fasting lipid draw from earlier that same day, and the others as part of the oral triglyceride tolerance test.

References for when I interpret my data:

https://www.omnicalculator.com/health/cholesterol-units for units relative to other papers (TG and cholesterol mmol/l to mg/dL is not the same!)
 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549

Randomized controlled trial with the intervention adding n-6 (omega 6) fatty acids. Group was men with prior heart attack.

Adding in the n-6 fatty acids (read: vegetable oil, safflower here since nearly all n-6) increased mortality, cardiovascular disease, and coronary heart disease over their followup period.

Cliffs Notes: in the low carb high fat community it’s more or less part of the canon that one should avoid vegetable oils. Here’s some solid evidence with hard outcomes and in humans for why this is likely a good idea (as opposed to evidence from, say, rats or endothelial cells in a petri dish).
 

Cerberus75

Monkey
Feb 18, 2017
520
194
Yeah, people should really look into what fats should be in their diet. Nuts, olives, coconut and Avacodos are good for you. You should have a higher amount of saturated fats to protect the monounsaturated from oxidation.
 
Last edited:

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
I just completed and returned the Certificate of Medical Necessity... for myself, for a continuous glucose monitor. (In my defense it appears my dad was a type 2 diabetic for a few years without telling us! and also I get nominally hypoglycemic, albeit when extended fasting. :lol: )

The process of obtaining this Dexcom G6 has been a pain in the ass and a half.
 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
I just ordered a test from omegaquant.com . I went for the Complete option. I’m curious to see what it shows—I’ve upped my fish oil game to ~4 g/day EPA+DHA recently after reading about the recent trials, such as REDUCE-IT.

https://www.nejm.org/doi/full/10.1056/NEJMoa1812792

Admittedly the population tested in REDUCE-IT isn’t me (on statins, TG 135+ mg/dL), but its effect was quite impressive.
 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
I just ordered up some stearic acid, food grade, from Amazon. Time to experiment with adding it to my diet!

Why?

Theory, noting increased beta-oxidation (i.e. use of fatty acids for energy), and increased mitochondrial function:


Practice, by a semi-crazed chef who has been cooking himself up stearic-acid-rich croissants:

 

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
On the primary care doc’s office scale with similar if not slightly heavier clothing (probably was in shorts last time) down about 41 lbs just now as compared to July.

So begins my day of tests.


Edit: Other results that are back so far:

41 or so lbs down since July, scale at PCP vs scale at PCP. 30.5 lbs down since August, DEXA derived weight vs the same (as in measuring fat and non-fat mass and summing). Body fat % was essentially unchanged, though, as I proportionally lost as much lean as fat mass.

LFTs nearly normal now. ALT still slightly high. Ferritin normal now. (AST/ALT/ferritin was high in August, ferritin higher yet in July.)

CRP 1.11.

Normal CBC, with 48.8 hematocrit, which is where I usually run.

Normal chem panel.

At time of labs 100 mg/dL glucose, 13 uIU/mL insulin. That’s a HOMA-IR of 3.2, so I’m still insulin resistant.

52 HDL-C, 251 TC, 63 TG (yes!), 186 calculated LDL-C via the Friedewald equation.

A1C of 5.1, as compared to 5.2 before.
 
Last edited:

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
A potentially interesting thing that resulted today (as opposed to the Lp(a), directly measured LDL-C, and the NMR LipoProfile pending) is my resting metabolic rate test.

August:

2,118 kcal/day resting metabolic rate. 2,000 of those calories via fat oxidation, 118 via carbohydrate.

Decembre:

2,051 kcal/day resting metabolic rate. 2,115 of those calories via fat oxidation, -62 (!) via carbohydrate.

Negative calories, you say? This is a math/conversion artifact caused by measured respiratory quotient values below 0.7, wherein 0.70 on the nose is 100% fat oxidation.



Note the string of values at and slightly below 0.7, thus the artifact. This has been described before and appears to be related to a ketogenic diet:



Anyway, if one ignores the negative value's contribution then I didn't drop RMR at all. If one includes it then I dropped 3.2% RMR. This is in contrast to dropping 12.8% non-fat mass over the same interval.
 

rideit

Bob the Builder
Aug 24, 2004
23,055
11,298
In the cleavage of the Tetons

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
49 hours into a fast. Glucose holding steady, 88 mg/dL at the moment. Feeling ok, so I'll push on through at least tonight. I'm interested to see what my glucose levels will do on day 3.

Edit: Ate this morning. 60 hours. Time to see what my glucose versus time looks like on my monitor!
 
Last edited:

Toshi

Harbinger of Doom
Oct 23, 2001
38,029
7,549
982EDA03-EF2E-4159-99C9-5622F70B153E.jpeg


So I have discovered something interesting based off of my continuous glucose monitor usage. The above 24 plot of blood glucose versus time is precise as compared to the wife’s AccuChek Nano (although that, in turn, reads 13 mg/dL higher on average than my Cholestech LDX). Let’s assume it’s accurate, though.

The back story for what I’ve been up to the past few days matters in interpreting these glucose data. I ate on Thursday, probably < 50 g carbs, quite clean. Then I fasted for 43 hours through 1:30 PM on Saturday. During that fasted time I bike commuted to and from work on Friday and then skied fairly intensely for 4 hours on Saturday morning—see my scribble on the screenshot. I ate over the period of 1:30-6:30 PM on Saturday, probably 100 g carbs if not a bit more.

As one can observe I didn’t have any sort of immediate glucose spike but was persistently high through this morning. The grey bar’s range is from 60-140 mg/dL, so I was flirting with 140 mg/dL for a good portion of the night!

So am I diabetic now? Should I be worried about this?

The answers are no (A1C 5.1% as of the latest) and no, I don’t think so, respectively. That I exercised while fasted and usually eat low-ish to low carb is significant because the high effectively fasted glucose levels probably are due to adaptive glucose sparing/physiologic insulin resistance. In this state one’s muscles are used to not using glucose—again, during my bike commute and skiing they were most likely operating nearly exclusively via beta oxidation of fat or, alternately, via ketones. Therefore they don’t respond to insulin signaling like “normal” and one may see higher glucose than one might other expect given a normal A1C.

(My family also arrived back from Seattle yesterday afternoon so some element of stress also is likely at play. House was much quieter without them!)

The related question would be of whether glucose hanging out in the 120-140 mg/dL range but without spikes is dangerous. If it is then I’d either eat more carbs more regularly (as in fewer/no +24 hour fasts) so as to reverse the adaptive glucose sparing effect, or I’d consistently eat fewer carbs yet a la @stoney . But I haven’t found evidence that this is actually harmful.

With true diabetics then complications of diabetes scale with changes in A1C, which reflects an average of glucose levels over ~3 months. Said complications can be seen scaling all the way down to A1Cs in the non-diabetic range. However, the key point is that for diabetics on a normal-carb-load diet they experience large post-meal spikes in glucose. The amount of time above 10 mmol/l glucose, which is 180 mg/dL, is supposed to be what’s bad, because that level leads to LDL oxidation and other advanced glycation end-products. Until I see evidence to the contrary I’m going to keep on doing what I’ve been doing.

Re 10 mmol/l threshold see this, for which I don’t know the exact source:

BB1078DD-C8C2-42E7-860F-6AE227CCA570.jpeg