Sunday, January 31, 2010

Who pays the piper: Disturbed earth

I hope this is going to be the last post on Bradley and Hunter. Once again Table 5, but now the HbA1c line.

An HbA1c of 5.2, 5.3, 5.4 or 5.5 is normal. Any variation within these limits is virtually irrelevant clinically. Moving a tenth of a percentage point in this region is utterly insignificant. The changes in HbA1c don't even get a mention in the discussion. But perhaps they should do.

The low fat group dropped their HbA1c from 5.4 to 5.3, shrug. The low carb group dropped their HbA1c from 5.3 to 5.2. Mega shrug. Sound just about identical, don't they?

There are two very strange things about this. The first is that the drop for the low fat group does not reach statistical significance. The p value is 0.4 for within LF group change.

But the drop from 5.3 to 5.2 in the low carbohydrate group is statistically very significant, p=0.01. How can this be?

The second strange thing is that, in the whole of Table 5, there is only one value expressed to three decimal places. That is the standard deviation of the HbA1c of the post diet low carbohydrate group. This sd has to use three decimal places because it is so small. Having zero would not look good.

What does an sd of +/- 0.003 mean? Because three standard deviations around a mean include 99% of the population we know that 99% of the values lie within the 5.2 +/- 0.009 range. But HbA1c is only reported to one decimal place and you can only ever deal with a whole number of people. So, by the end of the eight week study, 11/12 subjects on the low carbohydrate diet will have had an HbA1c of 5.2 and the 12th person will either have been 5.3 or 5.1, there is no other way of looking at the data.

This is what happens when you take a group of people eating anywhere between 190g/d and 275g/d of carbs and drop them all to close on 104g/d. Everyone's post prandial glucose drops and the range of HbA1c both drops and tightens up.

Not so the low fat diet people.

Before the study they started with carbohydrate intakes of about 120-240g/d and tightened in to a minimal spread round 226g/d on the low fat diet. So some individuals increased their carbs, some decreased them, some didn't change them that much.

The changes in glycaemia will have been all over the place following changes in individual carb intake. Some will have risen, some dropped. There was a spread to begin with and a spread at the end. The overall effect is a small and insignificant drop because the carb intake variation is blunted by a 500kcal energy deficit being made up of zero-carb butt fat!

Had this diet not included weight loss the low fat people would have really come out badly on an HbA1c basis as they would have been eating glucose instead of burning butt fat!

In terms of glycaemia, everyone benefited in the LC group. Some benefited and some were injured, ever so slightly, in the low fat group. This is hidden in the average but shows through the standard deviations!

Overall this aspect is more of interest for looking at how information gets buried, rather than any genuine clinical effect. Information gets buried all right. Luckily you can see the disturbed earth in the results tables.

That's it. I've had enough of Hunter and will give it a rest.


Saturday, January 30, 2010

Who pays the piper: Other lipids

About lipids, quote from the discussion in the Bradley and Hunter paper, related to Table 5:

"As expected, the low-fat diet decreased both LDL and HDL cholesterol. Although the low-carbohydrate diet did not decrease LDL cholesterol, it was not associated with a significant decline in HDL cholesterol. Given the established evidence that LDL lowering reduces the risk of coronary heart disease, the lack of a decrease may be of concern. In contrast to the lack of a change in LDL and HDL in response to the low-carbohydrate diet, there was a significant reduction in triglycerides within this group compared with no significant change within the low-fat group. This response has been consistently reported in other studies comparing a low-carbohydrate and low-fat weight reduction diet (12). It has been speculated that this result is due to a combination of a decrease in the VLDL production rate and an increase in triglyceride removal from the blood (31). Previous studies (32,33) indicate that increased triglycerides are an independent risk factor for cardiovascular disease, although it is impossible to predict the overall effect of the lipid changes with the lowcarbohydrate diet. Further examination of the lipid subfraction profile may help elucidate the effects of the dietary regimens on lipid metabolism. Indeed, previous studies have suggested that low-carbohydrate diets increase LDL particle size and decrease small dense LDL particles (34)"

This is a load of mealy mouthed beating around the bush. My translation (couched in terms of the lipid hypothesis, for believers such as Bardley and Hunter):

The low carbohydrate diet did no harm to LDL, this is really good. Low fat diet converted any bouyant LDL around from the pre study diet to sdLDL during the study. This is really bad, it just improved a lab number while actually injuring the participants. The low carbohydrate reduced triglycerides, this is really good. The low fat diet didn't. This is cr@p.

Executive summary: The low carbohydrate beat the low fat diet on lipids, hands down.


Who pays the piper. High fat diet drops HDL!

From the discussion of Bradley and Hunter's 2009 paper:

"A major concern associated with low-carbohydrate diets is that the reciprocal increase in dietary fat intake, particularly if this includes saturated and trans fat, may have detrimental effects on cardiovascular risk"

Who mentioned trans fats? To find out what the diets were actually like we need to look at Table 1:

Hard to say, but there are a few rather commercial sounding foods listed where we might find trans fat. How about the nutrient breakdown on the diet, that's here:

It looks like 60% of calories from fat in the LC group, 21% saturated, 21% monounsaturated and 13% polyunsaturated.

Now hang on, 21+21+13 does not equal 60. What was the other 5%? This was a study in which university nutritionists were providing the food. They should be able to make three numbers add up. It's not as if we are working with food frequency questionnaires or dietary recall from a year ago. As they say in the methods:

"Volunteers attended on alternate days throughout the study and were supplied with all appropriate foodstuffs (preweighed into daily portions) for their particular diet"

So 5% of the fat in the LC diet was not supplied in a form which could be listed comfortably in Table 4. Anyone for trans fats?

So lets go and look to see if there is any other evidence of trans fats as 5% of calories. We can try looking in Table 5.

Hands up who knows what saturated fat does to your HDL level? OK, go to the top of the class. The LC arm increased their saturated fat intake from 15% to 20% of a few less calories, with a similar increase in monounsats too, yet they DROPPED their HDL from 1.47mmol/l to 1.35mmol/l. Not catastrophic, just bad (if you believe in lipids being good or bad, all this means to me is that they fed trans fats as well as a little extra saturated fat). In just 12 people this is not STATISTICALLY significant. At a population level it would be biologically significant.

Of course the low fat folks dropped their HDL from 1.40mmol/l to 1.15mmol/l. Statistically (p=0.01) and biologically significant. Just what anyone would expect from a low fat diet, except that the drop seems even worse than you would expect from an ordinary university designed low fat diet! Table 4 suggests 4% unlistable fats for the low fat group too.

Trans fats shared equally? I don't know.

At this point I would just like to suggest that in NO WAY should ANYONE ever let a Belfast nutritionist design their diet. Just say NO.

Stick to eating Food.


Who pays the piper. Ignorant or bent?

Let's go back to 2006. Black and Hunter produced this paper with this as the final line of the abstract:

"In this study, a high-sucrose intake as part of an eucaloric, weight-maintaining diet had no detrimental effect on insulin sensitivity, glycemic profiles, or measures of vascular compliance in healthy nondiabetic subjects"

This statement is incorrect. The victims started with a fasting blood glucose of 4.8mmol/l, normal, and ended up as pre-diabetics with a FBG of 5.6mmol/l (discussed here). The glycaemic profile was NOT free of detrimental effect of either intervention diet. The only mystery was whether Hunter's group was ignorant of the concept of pre diabetes or bent. Now, if we go to the results section of the 2009 paper by Bradley and Hunter, currently under discussion, we can find this line:

"Although glucose tolerance tests were not performed, the mean fasting plasma glucose of 5.6 mmol/l was in the pre-diabetes range, consistent with an increased risk for development of diabetes."

Which seems to nicely answer that question. I particularly like both studies having exactly 5.6mmol/l as the value cited. A pure fluke, but so informative....

Ignorant or bent? You decide!


Friday, January 29, 2010

Who pays the piper for arterial stiffness? Now for Table 7...

First the problem: TABLE 7

OK, first the easy bits. Both diets dropped systolic and diastolic blood pressures by the same amount. I'll come back to the pulse pressure (systolic - diastolic) later, but it didn't change by any significant amount either.

The last line, brachial pulse wave velocity, didn't change and was the same on both diets. This is a marker of how stiff the brachial artery is, a muscular artery which doesn't store blood in the same way as the elastic arteries such as the aorta. You can say muscular arteries didn't change with diet or weight loss.

There are then a set of numbers derived from the aortic arterial waveform, this waveform, deep in the aorta, is estimated from a peripheral pulse (illustration stolen from some random Nature paper where folks probably knew what they were talking about):

The time-to-wave reflection is the time from the start of contraction to a very early reflected wave from the periphery. It comes much sooner than the wave described in the previous post and just shows as a "bumping up" of the pressure in the aorta.

The sooner this wave arrives the more rigid the aorta. To find it you look for an inflection point on the upswing of the pulse pressure. It's marked as Tr in the diagram. It's probably similar in the information it carries as aortic pulse wave velocity in the last post but requires much more sophisticated software to extract.

This marker of aortic health improved by twice as much in the low fat diet as the low carbohydrate diet. That's interesting, but note that both groups improved.

The amount that the blood pressure gets "bumped up" by using this wave is the augmentation pressure, recorded as "augmentation" in the table. To get the augmentation index you should just divide the augmentation pressure by the pulse pressure and multiply by 100 to get a percentage. ie how much of your pulse pressure is from augmentation due to early reflection... The numbers don't add up exactly as the machine seems to be doing something slightly different and describing the end result as aortic augmentation index as compared to the usual augmentation index... That's just inside the box of tricks used and I can't tell exactly what they've done.

So, because pulse pressure didn't really change, augmentation and augmenation index should change together. They do, but only approximately. The difference has, again, to be within the machine and how it looks at the waveforms and does its calculations.

What's bad is that, as reported by Hunter, AI got worse in the low carb group and got better in the low fat group. Oh no! So Hunter can say in the discussion:

"It is possible that the high fat content of a low-carbohydrate diet exerts detrimental effects on endothelial function, which raises concerns regarding the long-term safety and efficacy of low-carbohydrate diets"

But when you look at the numbers the low carbohydrate group started with an augmentation of 7.4 mmHg and this increased to 8.4 mmHg.

The low fat group started with a higher value of 9.2 mmHg and this dropped to 8.3 mmHg.

So the low carbohydrate group started 1 mmHg of augmentation below the end result of 8.4 mmHg. The low fat group started at 1mmHg of augmentation above the end result of 8.3mmHg.

I defy anyone to claim 8.4 is different from 8.3 with SDs of 5.4 and 7.8 respectively.

The end values were the same.

So there are two insoluble mysteries. First is how the "aortic" augmentation index, the only value which made p<0.05, is different for the simply calculated augmentation index percentage, which probably didn't reach p=0.04.

The second is how come a group of 12 people can improve one measure of aortic stiffness (time-to-wave reflection) and worsen another (augmentation index), when both parameters are derived from the same 12 cardiovascular systems! Dunno on either of those.

But I would also add that augmentation should never be zero! You're pushing blood down a pipe. Some augmentation is essential and normal and a quick look around pubmed has some poor folks with augmentation indices of >40%, to put things in perspective... It's not like cholesterol (jk), the lower the better blah blah blah !

This comes down to clinical vs statistical significance. Both groups ended up with the same augmentation, neither change was significant, biologically or statistically, but a machine derived variant made p=0.04 in favour of low fat diet, but with dubious biological significance.

There is a huge goldmine of information in the results which need talking about but I think that's enough for now. It's all pro LC and anti LF, as you would expect.


BTW I've just calculated the AI% from changes in pulse-pressure and reported augmentation and, using the simple percentage technique [of AP/PP X 100], the end of study AI% for LC (17.14%) actually ends up slightly lower, compared to the end value for the LF group (17.29%). Hunter no doubt would argue the direction of change being awful despite LC end result beating LF etc etc but still, it's pleasant pedantry to look at the numbers...

Thursday, January 28, 2010

Who pays the piper for arterial stiffness?

General anaesthesia for horses is a remarkably high risk procedure. So high risk that most clued up anaesthetists will routine place an arterial catheter for direct measurement of arterial blood pressure, usually using an oscilloscope type display of the wave form, even for "short" procedures. Back in my training days I had the pleasure of many such events and, in those where there was not too much to worry us, my extremely well informed supervisor and I would discuss the applied physiology of keeping 500kg of unspeakably valuable equine alive, asleep and immobile. I remember just one brief comment about the double nature of the systolic pulse peak. Polly's comment was that it was something to do with a pressure wave reflection from the hind limbs (we usually had the catheter in the facial artery, up at the front end, as far away from the surgeons as we could get. If you've never been there...

Anyway. The heart contracts, there is a rise of pressure in the peripheral arteries to the systolic peak followed by a decay down to the diastolic plateau before the next beat. Except it's not smooth. A few milliseconds after the systolic peak, part way down the fall to diastole, there is a second peak. Who ordered that?

It turns out that this second peak really is due to a reflected pressure wave. The systolic pressure wave travels directly from the heart to the facial or the radial artery, wherever you have managed to get an arterial line in. The same pressure wave travels off down the aorta until it hits the resistance of the hind legs. Some of the pressure wave carries on, some is reflected back. The reflected portion goes back up the aorta in the wrong direction before spreading out through the arteries at the front end of the patient, those few milliseconds later.

If all horses where exactly the same size the time between the systolic pulse and the second peak would always be the same if the speed of travel of the pressure wave was always the same.

It's not. The speed of the wave is influenced by the stiffness of the aorta. The stiffer the aorta, the faster the wave travels and the shorter the time to that second peak. Also, the stiffer the aorta in a human the more likely that person is to die of cardiovascular disease. The two things are associated.

So, in a human, if you correct for height, the time difference between first and second peaks gives you an index of aortic stiffness. This usually worsens with age so you can then give a crude cardiovascular "age" to a person.

There are several methods for getting aortic pulse wave velocity and, luckily, most of them don't involve you having a catheter placed in an artery!

The easiest and most expensive way is to buy one of these machines, which is a commercialisation of the system described here. This is from the paper and the caption is self explanatory:

By an absolute quirk of fate I was put in touch with someone who owned one of these commercial machines. I was invited to play with it and came out with a cardiovascular age of 32 years. I think I was 51 or 52 at the time, already several years in to LC/high saturated fat eating. The girl who owned the machine was in to "healthy" eating plus arginine supplementation for her own CV issues. The CardioCheck gave her a CV age of about 70 years and she was having frequent palpitations. I was polite enough not to ask her chronological age but it was a sight younger than mine!

I lent her Life Without Bread. The palpitations stopped immediately on LC but it took six months to get her CV age down to her chronological age. I guess it's well below that now.

I don't have the equivalent of $3,999 in pounds sterling for a CardioCheck machine. But I do have access to a doppler ultrasound used for the determination of blood pressure in cats and dogs.

You can HEAR the double peak in the doppler arterial flow. You can also sit your computer next to the loud speaker of the doppler, record the sound output, smooth it electronically and actually measure the time between the peak systolic flow and the smaller second peak. Factor in your height and you have an aortic stiffness index.

I'm consistently equivalent to the low 30's in CV age. I've played with other people of my age who also eat LC and they too have a "young" CV system. Normal diet people come out all over the place, some scarily high.

So what?

Well, it's time to look at Bradley and Hunter's latest sugary offering.


EDIT added in response to comments

OK, the formula is simply height in metres divided by time between peaks in seconds. I can't find the graph anywhere on my hard drive so this jpeg is the closest I can come to putting values in to ball park areas.

One 50+ female LC friend has a time between peaks of 320ms and a height of about 1.65m giving an SI of 5.12. That puts her CV age somewhere in her 20s...

I can't convert the Audacity file to anything visual to show you. The estimations are pretty crude but if you check various pulses you get a reasonable idea of a typical waveform.

Relax before you try, be sitting, no crossed legs!


Friday, January 22, 2010


Flying south for the weekend for my wife's graduation, I'll try and catch up with comments and emails next week.

Have a great weekend


Sunday, January 17, 2010

Liver and onions at 100 after Keep Fit

Just tripped over this one while checking emails:

"But for Miss Easter, who was born when Edward VII was still king and the Titanic was only being built, the keep-fit classes are just a bit of fun. She said: "What's my secret? Keep away from junk food. I also do a bit of gardening, I cook every day and I am going to have liver and onions ... when I get home."

Well, I can think of worse things to have for supper!


Fruit and Vegetables link

This just fits in the fruit 'n' fibre and glycaemic load aspects that run through Hyperlipid: Elizabeth sent me the link to a nice post by Dr A here.

What can you add to that?


Friday, January 15, 2010

Paleo and fructose

I've had a couple of queries which tie together neatly and involved some detailed paper reading, so I thought I'd throw them out for general discussion.

A friend is struggling to keep adequate carb intake on a starch free diet. Anyone who has tried this will soon run in to kilos of cabbage. The easiest answer is fruit and honey but this then means about half of your carbohydrate calories are going to come from fructose. Does this matter?

The other query was from David. He looked through Stephan's discussion of this paper and noted the huge intake of fructose from just about every modern fruit or fruit juice that anyone can imagine a hunter gatherer gathering. You know how it goes, gathering was good that day! This is the same quandary as the first query but worse. For a LC paleo person to eat a little fructose is one thing, to eat nearly quarter of a kilo (dry weight!) of fruit and juice to stay politically correct is pushing it a bit. Yet the results of the study are good. How come? Here's the menu.

The column on the right gives the bit we want. Ignore diets called ramp 1,2 and 3. You can see "Paleo diet" is quite fructose rich. Before we can get on to discuss this, a problem needs to be addressed. I'm either going senile or there is a double typo. It's here

Aside: Look at the HDL. Zilch increase. I'm amazed it didn't drop with all those PUFA!

Fasting glucose pre study is 18mmol/l and on the paleo diet it's 17mmol/l. These are wrong. Certainly for people described as healthy. The HOMA scores allow back calculation, first converting the insulin from the nice pmol/l quoted to grubby microIU/ml needed (HOMA seems to mix IU and SI units...) but even this gives a fasting BG of 7.2mmol/l... Still not exactly normal! And it didn't normalise on the paleo fructose diet either. So either it's a typo, my math is useless (that WOULD be embarrassing!) or these folks had fasting hyperglycaemia. Take your pick!

OK, back to working out what is going on. Here is the key:

If you look at the HOMA-change graph Figure 2 you can see that 4 people (upper left) were normal (HOMA <2), two people were almost normal (HOMA<4) and two were frankly pre diabetic (HOMA 6 and 7.5). In my terms rather than ADA classification. None of the normal or near normal people improved HOMA much on the diet. Why? You can't drop HOMA below normal as it's already normal.

I'd guess the people at the upper left had a moderate American fructose intake pre paleo diet and probably didn't change this much on the honey and carrot juice in the study diet. They started normal, they stayed normal and we can forget about them. The two people with IR on the lower right of the graph are the two who did all of the improving. This certainly applies to HOMA scores but, because we don't have the data broken down by individuals for anything else, we cannot see if that was the case for all improvements in all of the other parameters looked at. I think it might be. Is there anything to support this idea?

The thing to look at is the food intake pre and during diet.

People entered the study on a self selected diet averaging 254g/d carbohydrate. But the standard deviation is 128g/d. The rule of thumb is that three standard deviations include about 99% of the population. In the USA I doubt many people with a BMI of 27 were eating below 200g/d of carbohydrate, so a few individuals will have to have been eating much more than 254g to get that big SD. The easiest way to get your carbohydrate intake up towards three SDs above the mean of 254g [ie 254+(3X128)= 638g/d] is to drink Big Gulps. Big Gulps provide 800kcal per serving of HFCS calories. This has to be a guess as the results do not break carbs down by fructose. Sigh.

On the paleo diet food and drink intake was tightly controlled and everyone will have been consuming near identical fructose. The fall in mean carb intake from SAD to the paleo diet is only 5g/d, but the SD of the drop is 126, essentially the same as the 128 of the pre study carb intake's SD. Taking people from a wide range of macronutrient intakes and putting them all on the same diet will do this. The total carb intake during the paleo diet is 254 minus 5, ie 249g/d but this will have a SD of near zero, because all of the subjects were on university prepared meals of precisely controlled calories and composition for the study.

Anyone who is used obtaining large numbers of calories by drinking Big Gulps will have had a marked reduction in fructose intake on this paleo diet.

From hepatically crippling fructose to liveable with, if not ideal, fructose.

So, in answer to the question, yes, the diet is high in fructose by my standards. But my guess is that it is markedly reduced in fructose intake for those two people who did all of the improving in terms of insulin sensitivity. Most of the other changes will follow on from this specific change. We are very lucky to have Figure 3 to allow us to tease this out! Few papers include individual data nowadays, though it was quite common in early research (1950s etc).

That's how I see this paleo diet. Good, but could do better.

Two asides:

The second aspect is fatty acids. The PUFA increased, which should have worsened liver damage, but omega 3s were added in modest amounts, which should have ameliorated liver damage. Fatty liver from omega threes is unheard of in remotely recognisable human diets. End result of fatty acid manipulation; probably neutral. They didn't hurt anyone. Good.

Finally; endotoxin. Endotoxin is needed to convert the relatively benign condition of NAFLD to the much more serious and cirrhosing NASH. Endotoxin uptake is markedly facilitated by gut damage. Gluten is the primary opener of intestinal tight junctions. There were no grains in this diet. That should reduce liver insult. Liver insulin resistance is primary to metabolic syndrome.

Back to the first query: For those wanting to avoid ketosis when unable to tolerate starch. Some fruit seems relatively benign. You can get away with moderate amounts of fructose for years before it gets you, probably half a lifetime. Unless you make Big Gulps your only significant source of calories.

It really does seem possible to break your liver in childhood if you really work at the Big Gulps.


Thursday, January 14, 2010

Saturated fat meta analysis: Krauss

From Elizabeth: A new meta-analysis.

Generally I hate meta-analyses. It's usually very difficult to go through all of the references that a group has assembled to support their hypothesis (usually that saturated fat is bad). Even harder to find out what studies they have excluded. I tend to re-title them as "A meta-analysis of all statin/PUFA/low fat studies supportive of the lipid hypothesis shows marked support for the lipid hypothesis". At least someone has done the footwork for the opposing view of saturated fats this time.

It reminds me of this commentary, unfortunately associated with Nestle (the food giant), which sets out the basic concept (six years ago) that there is no evidence of harm from saturated fat and that efforts to reduce it throughout the food chain might be mistaken. Very mistaken.

Krauss has been more and more open as a supporter of saturated fat after his very cautious and mixed beginnings. This sort of publication is useful when confronted with the garbage from the Food Standards Agency condemning saturated fat. Cordain might be going the same route.

There is a general feeling in THINCS and the Nutrition and Metabolism Society that 2010 could be a good year for saturophiles and the rest of mankind too of course, should they care to listen.


Saturday, January 09, 2010

Gluten: Does coeliac disease require an infection?

OK, time to post again. This one is a bit of a stand alone for those of us with food phobias which include gluten.

A zymogram is a very interesting tool, one I had never heard of until I received the pdf of this article via a friend in Scandinavia. It's from the same research group as discussed in this post. It's just a letter rather than a full research paper as the study is small, but possibly very important.

Anyway, a zymogram is an electrophoresis technique where an electric field is used to move a charged protein through a gel. It's used for hunting enzymes and what is special about the gel is that it contains the substrate for the enzyme you are hunting. If you bind gliadin in to the gel you can go hunting gliadinases in the gut contents of a human who does or does not have coeliac disease. Use your electric field to separate out the proteins and in the location where a hole gets burned in your gel, that's where a gliadinase is active. You can then work at characterising the gliadinase.

Here is a zymogram. The two ladders on the left are from normal people, there are no white bars. The next ladders have faint white bands where the gliadinases have eaten in to the gel. These are untreated coeliacs. The last lane, lane 5, has that massive burn out at 33kDa. This is a fully treated coeliac with no symptoms yet who has plenty of gliadinase in their gut.

Proline is an amino acid with a bent back. It's side chain is attached to it's amino group, putting a rigid bend in to a protein's structure. Apart from putting the essential tight twists in to collagen, proline also puts kinks in to gliadin. Lots of kinks. The sorts of kinks which stop gliadin fitting in to normal mammalian protein digesting enzymes. Particularly difficult to deal with are pairs of adjacent prolines. In fact there are only a handful of mammalian enzymes known which can act at this point. None of them fit the size/charge of the gliadinase found by Bernardo et al in their zymograms, at least one of which can do this. They think the enzyme is of bacterial origin. They only find it in the gut contents of people with coeliac disease.

It doesn't matter if you have active or diet controlled coeliac disease, the enzyme, and presumably the bacterium producing it is (pretty well) always and only present in coeliacs. So is coeliac disease an infection? Or are people with coeliac disease exquisitely good hosts for the gliadinase producing bacteria which do not establish in normal people?

This is fascinating. It's probably not the enzyme itself which is the trigger for coeliac disease (as pullulonase is for IBS) because it is detectable in controlled coeliacs as well as those with active disease. When there is no gliadin present to promote production of the enzyme it has to have other uses beyond splitting double prolines in gluten and it simply goes in to production overdrive when gliadin arrives. This seems likley.

If it's not the bacterium or the protease which trigger coeliac the other logical explanation is that the gliadinase is particularly good at producing immunogenic peptides from gliadin. This might be to do with the ability to split paired prolines, not something pancreatic lipase or brush border peptidases can do.

At the moment the only information available is observational. An intervention trial would eliminate the bacterium and its enzyme, check this on a zymogram, then re challenge with gliadin. We are probably years away from being able to try this but, if it turns out that this is a cure for coeliac disease, just think of the implications...

A genetic coeliac could eat real artisan breads, drink pints of Nelson's Revenge, freely eat pizza until having to buy a longer belt, develop hypothyroidism, get a wheelchair for gluten ataxia or multiple sclerosis... In fact all of the gifts of direct gluten consumption and toxicity (which probably don't need enzymic digestion to be received) could be shared by people who might have had to avoid gluten for digestive reasons! Share and share alike. We all need wheat.

Oh, except those of us who have contracted food phobias through dabbling in the scientific literature!

Hee hee, makes me think of "Three Men in a Boat" with pubmed substituting for the British Museum.