Monday, May 26, 2014

HbA1c: Crack vs smack for a lower reading?

Having lived a great deal of my life in Norfolk, I’ve always rather liked the EPIC Norfolk publications. This was the first paper that I found which demonstrated, at the population level, that HbA1c is an excellent marker for your risk of both CVD and all cause mortality.

HbA1c is one of the most simple biomarkers to manipulate. If there was one single biomarker which you might want to modify, HbA1c is the one.

In Norfolk, people don’t.

Here people generally eat CIAB, more politely known as the UK version of the SAD or a mixed diet.

I doubt very much that anyone eats far enough away from “normal” or “unremarkable” macronutrient ratios that it has any significant effect on the distribution of HbA1c in the population. So, if you take a large group of people and set them free to eat whatever they want, HbA1c is probably a surrogate for mitochondrial health, particularly the mitochondria in adipocytes but more probably those throughout the body.

Nobody doubts that hyperglycaemia is damaging. Hyperinsulinaemia appears to problematic in its own right too. But my impression is that whether you succumb to a disease which is the result of hyperglycaemia directly or due to hyperinsulinaemia, this is just the label. The core problem is the underlying failure of mitochondrial function which signifies to a cell that it’s time to shut up shop. How this shows macroscopically, at the whole organism level, is probably up to a host of factors which I find of limited interest.

Sooooooo. Let’s be very specific: I consider, in the general population, that the linear positive relationship between HbA1c and all cause mortality is a reflection of failing mitochondrial health.

HbA1c is also the easiest of biomarkers to fudge, adjust or fake, call it what you will.

While diabetologists agonise over the inexorable progression of diabetes and wet their knickers at clinically insignificant drops in HbA1c on a calorie restricted mixed diet, those of us who eat LC are well aware that the 5% club is wide open to diabetics and the 4% club is not difficult for those of us who learned LC before mitochondrial damage became widespread. Most, although perhaps not all, people can drop their HbA1c to physiological levels with carbohydrate restriction, possibly with modest supplementation using metformin if they are significantly injured.

Dropping your HbA1c is easy. Very easy.

I have been thinking about this for a long time.

The really difficult question is whether an HbA1c of 4.4% achieved through the strict avoidance of post prandial hyperglycaemia is quite the same as an HbA1c of 4.4% achieved by a random chance combination of accidentally reasonable food choices, a functional pancreas, functional muscles, reasonable parental/intrauterine environment and good luck. Plus anything else you care to think of.

The LC eater is producing an effect on HbA1c which has never been investigated on a population basis, certainly not using modern, insulin resistant people who choose to normalise their blood glucose levels through diet. Whether it is as good for health and/or longevity as achieving it accidentally remains to be seen. No one need ask where my suspicions lie, but there are no hard data. There is absolutely no reason to doubt that being insulin resistant while eating A MIXED DIET indicates that you are in trouble, mitochondrially speaking. Sidestepping this with low carbohydrate eating will undoubtedly reduce your risk of those diseases which are a direct result of the hyperglycaemia and hyperinsulinaemia and which become inevitable on a mixed diet if you are insulin resistant. My own feeling is that this step is vital. I also suspect that some degree of reparation in mitochondrial health might be possible with long term near or frankly ketogenic eating.

OK, enough on the EPIC findings and my thoughts about adjusting HbA1c.

The next big step is to leave Norfolk’s EPIC beauty and wander over to the USA and look at the catastrophic association of a low HbA1c with death. As everyone undoubtedly knows, this link came to me from the exchange between Paul Jaminet and Ron Rosedale over safe starches. I was unaware of the study until Paul brought it up as a foil to the EPIC Norfolk findings.

It too, is a nice observational study.

I have to look at the data here in exactly the same way as I have looked at the data from EPIC. What might it mean? These are observational studies, their use is to generate hypotheses. What is the Hyperlipid view of an HbA1c of 2.8% (that is not a typo) on a mixed USA diet?

The easiest question to answer is whether this might be a surrogate for exceptional mitochondrial health. If you assume that the formula for conversion of HbA1c to average blood glucose levels holds true at the 2.8% level of HbA1c (which it obviously doesn’t), it would represent an average blood glucose of 1.0mmol/l. The lady would be dead. So, before we make idiotic assumptions about average blood glucose levels and mitochondrial health here, we had better think very carefully about what, exactly, a very low HbA1c might really mean. We can say for a start that it is not a simple reflection of glycaemia.

By far the most plausible explanation is not that the red blood cells have been exposed to 1.0mmol/l for the last three months, it’s more likely that these red cells have been exposed to a higher level of glucose, possibly somewhat physiological, for a significantly shorter period of time i.e. the red blood cells are young. Always young. They don’t hang around for long enough to become old and glycated before they are lost. Obviously the study group was well aware of this possibility and corrected for it as best they could.

The lowest HbA1c group had the lowest haemoglobin count, the highest red cell width variability (suggestive of regenerative anaemia) and the highest red blood cell volume (each RBC having relatively little haemoglobin). If you don’t think this group contained some seriously anaemic people you’re probably mistaken. Correcting for red blood cell variables in model 2d of Table 2 gave the bottom limit of the 95% confidence intervals closest to 1.0 of all of the adjusted low HbA1c models.

A quick glance at the ferritin levels, paradoxically, shows that these folks also appear to have iron overload. This may or may not be real as the group also contains 11% people sero positive for hepatitis C virus. As ferritin is an acute phase protein its elevation in a hepatitis C patient may be related to inflammation rather than to iron overload. Correcting for iron based parameters (model 2e) resulted in a worse "higher lowest" limit of confidence intervals than the basic correction for age, ethnicity and sex. You have to wonder if this might be due to an incorrect assumption about ferritin levels.

Of course there were four times as many hep C carriers in the low HbA1c group as the reference group. Model 2f tried to correct for this but I doubt they were able to control for those people using iv heroin (glycaemia neutral) vs those on iv crack vs those on oral stimulants (glycaemia lowering). Anorexia and hypoglycaemia are not uncommon under extended periods of recreational stimulant use. This information seems unlikely to be available to allow adjustment in the NHANES III models.

So I see two studies and each generates a different hypothesis. EPIC suggests that progressive mitochondrial failure on a mixed diet shows as raised HbA1c. The second, from NHANES III, is that sub-physiological HbA1c readings correlate to changes other than physiological glycaemia. Many of these changes appear to bad news, and HbA1c here is the messenger, just as it is in Norfolk.

Using this sort of observational data to convince yourself to increase your starch intake may be a mistake. Avoiding iv drug use and the metabolic consequences there-of might be a better approach.

If anyone can achieve an HbA1c of 2.8% by avoiding starch consumption I'll plaster it all over this post as an edit.

Above all of this, in the realms of sensible living, is the fascinating question of whether an HbA1c of 4.4% achieved by VLC eating has the same (positive) health implications as an HbA1c of 4.4% on a mixed diet.


BTW the EPIC researchers had a look at low HbA1c in Norfolk. Looks like speed is not the preferred recreational drug here. This fits with my experience dealing with the general population where I live.


Unknown said...


" whether an HbA1c of 4.4% achieved by VLC eating has the same (positive) health implications as an HbA1c of 4.4% on a mixed diet."

So - what do you think?

Scott Russell said...

Variability in RBC turnover always made me a bit wary of the usefulness of HbA1c. I seem to recall most infections and parasites drastically increasing RBC turnover. Or maybe its just that very sick people tend to give a lot of blood for bloodwork.

And re: VLC 4.4%, if memory serves RBC turnover is reduced with low carb eating, resulting in paradoxically high HbA1c readings. Obviously this wasn't from postpranial hyperglycaemia, but can easily scare an unaware doctor/patient.

If they found a way to normalize the test for RBC turnover I imagine a lot of these artifacts would disappear.

Peter said...

At a given level of mitochondrial injury the VLC will utterly knock spots off of the mixed diet. That is simple. Reversing mitochondrial pathology may not be possible, but if it is then ketones and fatty acids will do a much better job than running a cell on glycolysis. This is a no brainier. Crabtree rules.


Puddleg said...

Sweet. With this post and the last, I do like the way you've gotten philosophical in your old age (so to speak).
Re: ferretin - see "anaemia of chronic disease vs. iron-deficiency anaemia".

HCV is also a condition where higher LDL is protective.

Ash Simmonds said...

"Using .. observational data to convince yourself to increase your starch intake"

That right there covers 99.99% of diet ideologies.

Stuart said...

Low HbA1c is as bad or worse than high HbA1c in non-diabetics
Low Hemoglobin A1c and Risk of All-Cause Mortality Among US Adults Without Diabetes

Made me think of this and maybe it is related.
I'm eating lower carb <100g and have 4.5 which would be low but not ketogenic.

Peter said...

@Charles Grashow, after a little consideration; having mitochondria which do not require LC to maintain low HbA1c is the best choice. In their absence (you have no choice in this), LC is essential. If you have decent mitochondria to begin with then running them uncoupled, using elevated FFAs and chronic normoglycaemia, looks to be most likely ploy to keep them that way. You can work this out for yourself.

Ta George. You have to keep thinking about what stuff means. Nice article BTW.

heartlifetalk, I’ve never read anything of Evelyn’s where she didn’t perfectly fail to understand what is going on. She must have a PhD in it. I’m grateful to her for introducing me to Axen and Axen, who’s work she completely fails to understand of course. Disclaimer: I never, ever read what she says since Axen and Axen (I was pointed to this subject by Melchior, reading carbophilia is uncomfortable to me), so my above statement carries limited weight!


Unknown said...


SO - in you opinion - what would ideal HbA1C be? IS there an optimal range for one to be in?

Peter said...

If you have good enough mitochondria you might run in the low 4% region. But who knows what is ideal? And HbA1c is quite, quite separate from insulin signalling, another ball game.


Stuart said...

Peter the study she quotes seem to say that at low H1ABC one increases mortality never mind that she fails completely to understand the study but the study is claiming a sweet spot which might shed light into your idea here.

anyways the pathways to carb and fat burning switching is just being learned about

altavista said...

+1 on the philosophical bent.
"recreational stimulant use" LOL

Off on a tangent here, I find dr. Rosedale talks a good talk, but when you go to the specifics and see his dietary guidelines, it's all over the place. No way anyone can live on that and stay LC and under 1g protein/lean body mass/day.

OldTech said...

Hmm... it may be more complex. The idea that raising HDL would be beneficial failed for the niacin trial. And after listing to Dr Dayspring it may be that HDL simply performs so many different functions it is difficult to generalize. So you can not simply say raising HDL-C is good. It depends upon what the HDL is doing.

However, from a practical perspective, it still looks like increasing HDL may be positive on a very low carb diet, depending upon what else you eat, just because of the role that carbs play in inflammation.

My open question is the what else? In particular what dietary fats? To answer that, I need to know what is the role of dietary fats in oxidation of phospholipids?

Ash Simmonds said...

Again, HDL-C is just a vague biomarker.

Nobody ever said having more of it cures heart disease, just that people with higher levels *tend* to die from other stuff.

Don't focus on eating what supposedly increases biomarkers, focus on avoiding what we know actively damages you.

OldTech said...


I agree that we should not focus too much on vague bio markers, but my questions on dietary fats is related to what looks like a fundamental biochemical process - the oxidation of phospholipids which is involved in atherosclerosis.

Zorica Vuletic said...

Ya, I only care about insulin signalling tbh. I really don't think I have any hyperglycemic issues so outside of that only insulin signalling matters. Like you, I too like to 'resist insulin'. :)

karl said...

I've always been a little leery of HbA1c. It involves too many assumptions and my experience is that it can mislead.

It is supposed to measure the area under the curve - thus assumes that damage of high BG is linear with dose ( doubtful).

If I want to spur my mitochondria - it seems that resistive high intensity training (Rippetoe etc) blows away the standard advice - and does it using much less time for the effect. (You also get a much bigger change in muscle insulin sensitivity than with treadmills etc. )

I don't like to post links to videos - but this is pretty good - (download it and watch it with VLC at 1.6x real time).

Meat-heads seem, once again, to be on to something.

Start at 8 min - but pay attention to the whole thing.

Puddleg said...

I am pretty sure that 1) there is a "sweet spot" for HbA1c, given that it's a measure of both glycemia and RBC life-span.
2) the "sweet spot" range can go lower if you are VLC without this indicating poor RBC life-span
3) if you had hemolysis or poor RBC production, HbA1c could be normal and you might still be in trouble from hyperglycemia.
Still, nothing good about elevated HbA1c.

karl said...

@George Henderson -
Yes nothing good about a high HbA1c - but my take is it is less useful information than teaching folks to take postprandial - then they learn how much and which foods get them in trouble.

If I see my number over 110, I figure I ate to much of something or the wrong something. (One thing it taught me is that Salsa can be spiked with sugar that you can't taste due to the spices!).

My beef is MD's that hand out pills instead of explaining how they can reduce BG with diet. (Insurance won't pay for that 20 mins - but will pay for CABAG)..

Beyond HbA1c there are the fancy lipid tests - the patterns they detect are best modulated by reducing carb intake - In short they are really mostly a very expensive way to measure BG.

NY said...

Hi Peter and all,

I would really appreciate your take on raised HbA1c which some people experience after being low carb or ketogenic. There seems to be a discordance between glucose levels and HbA1c levels for example as BJJ caveman has shared his experience here
If one experiences higher HbA1c levels (from 5% to 5.7%) on a low carb diet how should they take it? Is it something they need to be concerned about?


BigWhiskey said...

@George Henderson


Peter said...

NY, You can always check post prandials but for a 24/7 estimate fructosamine is your best bet. Always assuming no active protein loss issues. Fructosamine looks at the last 2 weeks rather than the last 2-3 months.


steve said...

So what does an A1c reading of 5.3% mean when the BG consistently measures in the low 80's?

NY said...

Hi Peter,
Thanks for your response. So do you think there exists a discordance between HbA1c and BG levels due to the longer life span of red blood cells in a low carb environment?
Post prandial in a low carb environment would be usually high if eaten rice or potatoes maybe upto 8-9 mmol/l which is a poor glucose tolerance according to the prponents of safe starches but this post prandial drops if one keeps eating carbs for a few days which again is claimed the role of safe starches in improving glucose tolerance but my question here is what about insulin? Is it worth it if better glucose numbers come at the cost of higher insulin levels?

Also, Peter have you looked into resistant starch's role in making one insulin sensitive? Or is it just like any other fibre that will improve insulin sensitivity by slowing the BG responce?


Unknown said...

Hi Peter, just discovered your blog, pretty interesting stuff, i'm pretty new to LC, 1 month, less with carbs <30gr and proteins <60gr, i'am female 5'3, 100 lbs ( little fluctuation in 30, since highschool) have been running most days for 5 years, and there is absolute not know history of diabetes  in my family.  After beginning low carbs, mostly trying to solve some stomach issues, i got a glucometer and began monitoring my Fg that was usually between 95-105 with peaks of 109 and 111, after searcing a little i realized it could be a common effect of LC, but then I got my HbA1c at it is 6.3 that seems to be prediabetics, my posmeal BG in LC are not high, and I haven't tested with carbo loading, since I'm just adapting and don't want to mess with it.   I understand that this could be the result of having long living Hb, but don'believe there is any way to know. So my question is, should i do anything at all about it, and if so what could it be, since i already seems to meet all common recommendations to lower BG? And do you think that if this result be valid, could be caused more by impaired insulin production instead of insulince resistance, since this last seem to lead almost always to overweight.  Thanks in advance for your advise

Peter said...

Hi Liz P,

Fructosamine should check your more recent glucose levels and is independent of RBC turnover, though affected by over rapid protein turnover. At your low bodyweight it is certainly possible that you have a relatively low insulin production compared to average. If you do test this you can't draw any logical conclusions unless you are well carb loaded as glucokinase in your pancreas down regulates in LC eating and so pancreatic insulin secretion is normally is blunted under these circumstances.

If you do carry a gene for reduced insulin secretion then LC eating in the long term would seem the preferred option to reduce glycaemic damage...