Is Butter High in AGEs?

Visit Us
A butter-loving reader posed the following question to me:

Thanks for all the work you do. Any chance you could post something on the AGE content of butter? Does it make a difference if AGEs are endogenously or exogenously produced? Thank in advance.

Suffice it to say he liked my answer.  I'll repeat it here, with a little more detail.

The reader may have seen Dr. William Davis's recent post, “The Anti-AGEing Diet,” in which he wrote the following:

And minimize or avoid butter use, if we are to believe the data that suggest that it contains the highest exogenous AGE content of any known food.

Are we to believe this data?  I don't.

Don't get me wrong.  I love Dr. Davis's blog.  I'm just not feelin' the idea of measuring AGE contents with immunoassays instead of mass spectrometry.  Sorry for the big words.  I'll explain in a second.

Advanced glycation end products, or AGEs, are rogue chemical reaction products that result from the interactions between sugars or oxidized polyunsaturated fatty acids (PUFAs) and proteins.  The AGEs can then damage proteins, DNA, and possibly phospholipids, and thereby wreak havoc on our cells.

Dr. Davis links back to a previous post he made on butter, “Butter: Just because it's low-carb doesn't mean it's good.”  In that post he links to a 2004 analysis of the AGE content of 250 foods.  These authors recently produced a more extensive database of the AGE content of over 500 foods.  The results?  Butter, butter, butter, butter, butter.  The concentration of AGEs in butter was just massive.  In fact, they found that butter contains 5,000 times more AGE than whole milk!

Why don't I believe this?  For starters, it's not very realistic.   The specific AGE these authors were looking for is carboxymethyl-lysine.  CML is formed during the breakdown of glycated proteins — proteins that have been jumped and mugged by sugars such as glucose or, in the case of milk, lactose — or from the specific reaction of the amino acid lysine with a compound called glyoxal.  Glyoxal is formed from the spontaneous oxidation of glucose or as a byproduct of the oxidative destruction of PUFAs.  These processes are all accelerated by the presence of phosphate and trace metal ions in solution.

So where is all this CML coming from?  Milkfat is extremely low in PUFA to begin with.  Making butter from milk removes most of the protein and sugar and removes the majority of the phosphorus originally found in the milk.  On top of all of this, most of the water is removed, and removing water increases the stability of almost anything.

Is it possible for AGEs to form in butter?  Sure.  There's still lysine, and there's a little bit of PUFA, and there's still traces of carbohydrate.  But is it likely that butter has 5,000 times as much CML as whole milk?  Not very.  And someone who is going to claim this should have really good evidence to back it up.

So how good is the evidence?

The authors who composed these databases quantified CML with an immunoassay.  This means that they took an antibody that is supposedly specific for CML and quantified how much was in different foods by the relative amount of antibody-binding that occurred.  That's why you see the values reported in arbitrary “AGE units” instead of real units like “milligrams.”

There are a lot of problems with immunoassays.  Antibodies sometimes have spectacular specificitiy, but often have abysmal specificity.  We make a lot of antibodies in our intestines that are “polyreactive” to many different things so they can bind up bits of undigested food and other nasties.  In the lab, we try to make really specific antibodies, but it is often difficult.  In the case of AGEs, it's extremely difficult because AGEs can form on virtually any protein.  Some proteins may get modified with very many AGEs and others with very few.  Some proteins may be very big, some very small.  There is no one, specific shape of an AGE-modified protein and it is therefore impossible to make a highly specific, perfect antibody that precisely quantifies AGEs in foods.

By contrast, there are other ways to measure AGEs directly.  The gold standard is to use a technique called “liquid chromatography-tandem mass spectrometry with stable isotopic dilution.”  In short, this method digests the proteins in the food down to little itty bits so that all the individual AGEs are separated from their surrounding proteins.  The sample passes through a column that separates all the individual components.  As the AGE you are looking for plops out of the column, you determine exactly how much you have.  When you see reports using mass spectrometry, they report the AGE contents of foods in familiar terms like “milligrams.”  The drawback is that the method is more difficult, much more expensive, and that many labs are not equipped with the right machines.  The benefit is that the measurements actually reflect reality.

So what's the true CML content of butter as measured by mass spetrometry?  A study published last year showed that whole milk contains 40% more CML than butter.  Wow, big difference, huh?

Which do you believe?  The immunoassays that say butter has 5,000 times more CML than whole milk?  Or the mass spectrometry that says whole milk, which has more precursors, has 40% more CML than butter?

The same study found that evaporation of milk increased AGEs 10-fold, and that evaporated milk and various types of bread crust all had about 10 times the AGE content of butter and over five times the AGE content of beef.  Boiling beef increased the AGE content 7-fold, while frying the beef increased the AGE content 15-fold.

In addition to more posts on honey, fructose, and fatty liver, expect lots more information on AGEs in the coming months.

In the meantime, enjoy your butter!  Yum. 🙂

Visit Us

You may also like


  1. I know that this is an old post, but like Florent I'm really curious about ghee as well. I would really appreciate your weighing in on it!

  2. I eat all animal products raw and only cook vegetables/starches(and perhaps bones). Boiling and steaming are the only cooking methods I use.

  3. Hi Chris,
    I had to check out your blog after your excellent talk on Saturated fats and PUFAs on the cruise. Great technical analysis on what I hope can be refiled as a non issue for butter lovers. It's good to know your informative blog is available for us microbiologists who know just enough analytical chemistry to be dangerous.

  4. Ed, certainly the butter and milk were both pasteurized. There is a much newer paper I cited by the same authors that has similar data but many more foods. I do not think they fried the butter. That would be extremely disingenuous, and I think they make it clear that they fried the thigns that are commonly consumed fried. No one repeatedly fries things in butter and then consumes the butter.

    I think it is very much missing the point to believe that the result is an effect of temperature. The result is most likely an effect of the assay having no specificity for CML.

  5. I've read the paper that lists butter as a high-AGE food: "Advanced Glycoxidation End Products in Commonly Consumed Foods" (2004, Journal of the American Dietetic Association, via Google Scholar cache).

    Here are some numbers from Table 1:

    Milk, cow, whole …. 0.05 kU/mL
    Butter ………….. 265 kU/g

    The table caption refers to "foods prepared by standard cooking methods" (these include frying). The table doesn't say that the butter was used in cooking, but I'd be willing to bet that it wasn't raw!

    Expecting high AGEs in uncooked butter — over 5000 times the level in milk! — would make little sense. There's every reason to think that this butter had been exposed to high temperatures.

  6. Hi Chris,

    Awesome. I was puzzled by the butter result too. Why would butter contain way more CML than you'd expect from taking the fatty fraction of whole pasteurized milk? Butter isn't heated after extraction, so it makes no sense. Also, according to their data, extra virgin olive oil has nearly as much CML as butter if I recall correctly. Now that makes no sense whatsoever. Dr. Davis neglected to mention that in his post.

  7. Hi Chris,

    Excellent work. I was not devastated by Dr. Davis' article. I knew someone would put the smack down on this. Your points make great sense. Anyways, if butter really did have 5000 times the AGEs of whole milk, we would be seeing many of the well known whole health gurus who eat butter as a staple develop significant health issues after years and years of consumption. But, alas, that is not the case. Sometimes, common sense can lead the way. Here, you have bridged common sense with a detailed explanation. Well played.

    Jack Kronk

  8. Austin,

    The mass spectrometry study looked at raw and pasteurized whole milk and found about 70% higher CML in the pasteurized milk. The comparison I made between butter and whole milk was based on the pasteurized milk. In the antiobody study, the only milk they looked at was commercial whole milk.


  9. Chris,
    Thanks for the post! Do you know if the milk in these trials was raw or pasteurized? It seems that would make a big difference in the quantity of AGEs in the whole milk.

  10. James,

    There is some evidence that about 10% of AGEs enter the bloodstream in both humans and rats. The rat evidence is of good quality whereas the human evidence is a much more poorly designed experiment by the same group. In humans without kidney disease, these AGEs are rapidly excreted.

    There is pretty good evidence that high-heat cooking has a variety of negative effects on clinical parameters in people who are not in the best of health, but very little indication it is due specifically to the exogenous AGE content of those foods.

    So, there are two reasons why it is ridiculous to make dietary conclusions from an AGE database: first, the evidence is incomplete to say the least that dietary AGEs are a problem; second, AGE databases with immunoassays should be more or less ignored until the immunoassay is validated with mass spectrometry for use across a broad spectrum of food groups and processing types, which is not the case currently.


  11. Dr Davis has been hating on butter for a while. Someone made a comment on his blog along the lines of whether exogenous AGE's even enter the blood stream after digestion. Does a high consumption of exogenous AGE's promote higher levels of endogenous AGE's?

  12. Donny and Stephen,

    Yes, I saw that it was whipped. I find it much more likely that, as Donny suggested, the whipping distorted the stucture, perhaps shredding it apart, to make the existing CML better targets for the antibodies, than that whipping butter is 500 times more effective at producing CML than evaporating milk.

    That said, this doesn't explain why, on the whole, all their fats and oils were very high. So, I think part of the issue is that these antibodies just aren't accurate. In fact I have more certain reasons to believe that AGE antibodies have abysmal specificity and I'll write about those reasons in the future. Suffice it to say for now, I take anything on AGEs that uses immunoassay quantification with a grain of salt and if I am forced to use these antibodies for my dissertation work (which, thankfully, it appears I will not be), I would be embarassed to publish it.

    Nathaniel, AGEs are present even in raw foods. Yes, most of us are equipped to deal with them. Whether we deal with them perfectly is an issue worth discussing. In any case, that's another matter for a future post.

    Yes, endogenous AGEs are of much greater concern. More posts on that coming in the next few months too.


  13. With dietary AGEs being present in virtually all cooked food to some degree, should we even care? I have to think that we're equipped to deal with such things.

    It is endogenously-produced AGEs that I worry about.

  14. So maybe some of the more AGE-resistant foods (like butter) resulted in less complex structures, better targets for the antibodies?

Leave a Reply

Your email address will not be published. Required fields are marked *