1. Numerous online blog articles report ” 19 Gluten Cross-Reactive foods. “
2. Close inspection reveals questionable conclusions and significant overstatement and extrapolation from the original sources of information.
3. Alternate conclusions presented within this blog article would call into question existence of many blog-reported gluten cross-reactive foods.
4. Lessons about rigorous research and reporting of scientific data can be learned.
I have been reading the same article for more than two years now. Basically, it reads like the “Breaking News Alert” warnings about 19 Gluten Cross-Reacting Foods. Indeed, the same article (or only slightly altered versions of it) have been repeatedly featured on health diet and Paleo blogs alike:
• The Primal Docs: “19 Foods That Cross-React With Gluten”
• Mind Body Green: “Are You Not Healing Because Your Body Thinks Coffee, Chocolate & Cheese Are Gluten?”
• Dr. Perlmutter, author of Grain Brain: “Gluten Associated Cross Reactive Foods”
The basic gist of these above articles is that there are approximately 19 foods that “cross-react” with gluten—that these other foods can mimic gluten. This leads to your body reacting to these “cross-reactive foods” as if they were gluten, making these foods problematic for anyone who has a sensitivity to gluten (either Celiac disease or non-Celiac gluten sensitivity). That is, if you are making antibodies to gluten, those same antibodies recognize other food proteins.
However, just because it’s regurgitated doesn’t make it true.
The sheer number of articles that are circulating with the same information gives readers, naturopathic healthcare practitioners, patients, and the general public a false sense of informational credibility and reliability. It makes us think that because 207 internet bloggers have blogged “gluten cross-reactive foods”, it must have been independently verified 207 times, right? Wrong. Rather, it seems we have many examples of blatant regurgitation and reporting without serious scrutiny. That’s sensational journalism at the expense of analytical thinking.
So where did this information start? The list of “19 cross-reactive foods” is taken from a laboratory array made by a company called Cyrex (more details below). The array is reported to measure cross-reactivity of anti-gluten antibodies to other foods. It was developed in 2011, prior to the release of a paper entitled “Cross-Reaction between Gliadin and Different Foodand Tissue Antigens” in January 2013 which assesses this reported “cross-reactivity”. This paper will be referred to as Vojdani & Tarash 2013 throughout the rest of this article.
Cross reactivity in immunology has been defined as the ability for antibodies or self-reactive immune cells to recognize and bind different but similar antigens. So, if proteins in one substance are very similar to the proteins in another substance, there can be an immunological reaction to both. A good example of this is the Havein-like protein domains found in latex and bananas. People who develop sensitivity to latex are often also sensitive to bananas because they contain similar proteins. [Those interested in more detailed information can visit: here].
In the case of “gluten cross-reactive foods” the argument is that antibodies or cells recognizing wheat or gluten can also recognize other foods, such as eggs, chocolate, or coffee. Here’s a direct quote from one of the practitioners heralding the concept: “The [antibodies] are so busy attacking food every day, that it starts attacking foods that look similar.”
About 1 in 133 people have Celiac disease — an autoimmune disease that damages the intestine. The autoimmune reaction generates cells and antibodies which recognize gluten, or gluten complexed with tissue transglutaminase (tTG). The most common portion of the wheat gluten protein that is recognized is called α-gliadin. As a result of this, individuals with CD have cells which make anti-α-gliadin and anti-tTG antibodies (among others), and these are detectible using blood tests. CD is diagnosed with these blood tests along with a confirmatory intestinal biopsy.
Non-Celiac Gluten Sensitivity (NCGS) is different. Generally speaking, those with NCGS test negative for these autoantibodies — there is currently no biomarker or blood test to diagnose NCGS. The paper discussed here specifically deals with antibodies specific to α-gliadin peptide. Therefore, this paper can potentially only apply to individuals with CD and NOT those with NCGS. You can’t have “gluten cross-reactive” antibodies without anti-gluten or anti-tTG antibodies to start with. Though the authors use the term “gluten sensitivity” in the paper discussed below, they are not referring to those with NCGS, and instead are using the term somewhat synonymously with CD.
Take home message: if you do not have elevated blood autoantibodies indicating possible CD, these studies do not apply to you.
The “gluten-associated cross-reactive” foods list has been compiled from the Cyrex Labs Array 4. The Chief Scientific Officer of Cyrex, Aristo Vojdani, also used the specific foods listed in the array as antigens in a publication assessing their reactivity to α-gliadin antibodies (Vojdani & Tarash 2013). Most bloggers list all 18+ foods covered on the Cyrex panel and extrapolating to say that all are or can be cross-reactive to gluten. As you’ll see in the results below, not all of these foods are determined to be gluten cross-reactive in their studies. For example, hemp and potatoes were NEVER demonstrated to be — or even potentially be — cross-reactive. These bloggers have simply listed all of the foods that the original study addressed. Also note that the panel (as presented below) does not appear to assess reactivity of α-gliadin specific antibodies to other foods, but rather measures direct IgG antibody production against those foods.
Note: “Although testing for the presence of food-specific IgGs has been regarded as a potential tool for the diagnosis of food allergy/intolerance, it’s [sic] the accuracy and clinical utility of such testing remain unclear.” Source: Zeng et al. 2013
In addition to this blatant reporting error, some bloggers have added foods to this “gluten cross-reactive” list that don’t even show up in the conclusions of the original paper and have no additional citations, including sugar, “chemicals,” and GMO’s.
Now, let’s examine the main highlights and conclusions of Vojdani & Tarash 2013 which deals with the “gluten cross-reactive foods” found on the Cyrex Labs Array 4.
1. Some CD patients (~30%) continue to experience symptoms despite adopting a gluten free diet (GFD). [Note: they did not study those with NCGS]
2. The authors hypothesized if continued symptoms despite adoption of a GFD could be caused by cross-contamination with gluten-containing foods or with cross-reactivity between α-gliadin (the main immunogenic epitope of wheat in celiac disease).
Purchased foods from the supermarket and processed them for study. The left side of Table 1 shows foods that were used for antigen testing. Note that the foods used are identical to the foods contained in the Cyrex Array 4 which then show up on every “19 gluten cross-reactive foods list.”
Note: because extra-intestinal symptoms of CD have been reviewed in the scientific discourse, I will not be commenting on the portions of this paper discussing reactivity to tissue antigens.
2. Polyclonal rabbit anti-α-gliadin antibodies were made from rabbits injected with α-gliadin.
3. Highly specific monoclonal mouse anti-α-gliadin antibodies were purchased from a commercial source.
4. Used indirect ELISA and dot blot analysis to assess cross-reactivity of polyclonal and monoclonal anti-α-gliadin antibodies to food proteins.
In Fig 1 and 2 (using polyclonal and monoclonal antibodies respectively), the authors asked whether certain food proteins would be recognized from antibodies specific for anti-α-gliadin. They found high binding against α-gliadin (positive control). They also found binding “…against α- + β-casein, followed by yeast, casomorphin, oat cultivar #2, fresh corn, milk, millet, milk chocolate, instant coffee, rice, milk butyrophilin, and whey protein.” They did not find binding of oat cultivar #1, sesame, buckwheat, sorghum, hemp, amaranth, quinoa, tapioca, teff, soy, egg, and potatoes — foods commonly cited as “gluten cross-reactors.“ Note that only the mean for the assays (stated to be performed in quadruplicate) are reported, that “error bars” are not included, nor is there comment on whether the results were significantly different from the control. By this standard, I could flip a coin, get tails 3/4 times, and conclude tails is 3 times as common as heads. Scientific conclusions can only be trusted when tested to be true with statistical significance (typically p≤0.05).
The authors asked: can food proteins displace binding of α-gliadin-specific antibody to α-gliadin? They used high (10mg/ml) and low (150 ug/ml) concentrations of the proteins. At the high concentration, α-gliadin inhibited anti-α-gliadin antibody binding to plate bound α-gliadin at 79% (this is the positive control). At the low concentration no displacement was observed. At higher concentrations, yeast, instant coffee, and casein appear to promote displacement of anti-α-gliadin antibodies to α-gliadin. They also report no inhibition from soy at any concentration.
From Table 2: “The absorption with corn, yeast, millet, α- + β-casein, instant coffee and rice antigens inhibited the binding of α-gliadin 33-mer peptide to anti-α-gliadin 33-mer peptide by 27%, 26%, 23%, 21%, 16%, and 15%, respectively.” While the authors report differences in the means, are they actually significant differences? First, we need to evaluate what the specific hypotheses for this particular experiment are. Let’s take a look:
1. Null hypothesis: (Food) antigen X does not inhibit α-gliadin antibody binding relative to control.
2. Alternate hypothesis: (Food) antigen X inhibits α-gliadin antibody binding relative to control.
Let’s also take a moment to talk about statistical significance:
Biologists typically use a threshold value of p≤0.05, which is considered to be statistically significant, and p≥0.05 as statistically not significant (this threshold value may change depending on the application and appropriateness for a given assay). A p≤0.05 means that the chance of the null hypothesis being true (for a given data set) is equal to (or less than) 5%. Therefore, we can confidently reject the null hypothesis, and accept the alternate hypothesis.
Now, take a look at the “p values” reported in Table 2:
Using the p≤0.05 threshold standard, we can reject the null hypothesis for α-gliadin (outlined in blue, positive control). If we’re generous and we set our threshold to p≤0.1 (meaning the chance of the null hypothesis being true is less than or equal to 10%) we can include yeast, millet, and corn.
p=0.5 means that there is a 50% chance that we will erroneously reject the null hypothesis and accept the alternate hypothesis as true. This calls into question the interpretation of the results. While the means of the assays (reported as a % inhibition) may have been reduced indicating binding, there was apparently such high assay variability that it was not a significant difference from the control.
Dot blot analyses (Fig 8 and Fig 9 of the paper, using polyclonal and monoclonal antibodies respectively) showed slight staining (possibly indicating antibody reactivity) against instant coffee, corn, yeast, milk, α- + β-casein, millet, rice, milk butyrophilin and oats.
No staining was observed with casomorphin, egg, pure coffee bean, cocoa and sorghum.
A note on the dot blot experiment: polyclonal antibodies (used in Fig. 1 and 8) are notorious for yielding high background (i.e. not real) signal.
The direct conclusion from Vojdani & Tarash 2013 is that “milk, casein and milk-containing products such as milk chocolate should be thought of as containing gluten-like peptides, at least in individuals whose symptoms fail to improve significantly on a GFD.” My thoughts are that milk is probably not good for people with CD, especially if they fail to get better on a GFD. Secondary lactose (milk sugar) intolerance is common in people with CD. It has also been demonstrated that about 50% of CD patients (who do not get better on a GFD alone) can have a specific mucosal response to cows milk, with casein being the likely culprit. Antibodies to milk proteins can be elevated for some people; these responses have not been demonstrated to be due to mimicry or cross-reactivity. As Vojdani & Tarash 2013 cite, there is some computational data to suggest that these antibodies may cross-react, and some of their own data might also suggest that is the case. Milk tends to be problematic in a variety of other ways, and people who adhere to a paleo diet are probably not consuming it.
Instant coffee often contains gluten. Be aware of labels which state “may contain traces of gluten.” Pure coffee (including Starbucks) is probably fine.
Some oats contain gluten. We already knew this, since, though they are different plants, and oats do not contain gluten, they are processed together, and therefore are often contaminated with wheat/gluten/α-gliadin. If you do eat oats, you should make sure they are tested to be gluten free; even being processed in an oat-only facility may not be enough. But what about the avenins, you say? People with Celiac disease may also have T-cell specific reactions to the oat prolamine avenin. Monoclonal antibodies against γ-gliadin (not α-gliadin antibodies like are used in this study) can react to oat avenin. However, avenin is not proline rich like gliadin and therefore is able to be better degraded by GI tract enzymes, and this lessens potential immune reactions. Additionally, studies looking at immune activation in the small intestines of Celiac disease patients (following a GFD) failed to show histological consequences of immune activation following oat consumption. However, some studies do demonstrate oat-specific responses particularly in patients with complicated Celiac disease, although these are typically not large or randomized studies. Moreover, the variety of the oat can make a difference, and this was reported in Vojdani & Tarash 2013.
Vojdani & Tarash 2013 conclude that they “don’t know” whether the reaction they observed is a true cross-reaction or if the yeast used is contaminated with gluten. In these experiments, Brewer’s and Baker’s yeast appear to be combined when cross-reactivity was assessed. Brewer’s yeast is often a byproduct of beer, and not considered gluten free. Sometimes it can be made from beet sugar, but typically practitioners recommend avoiding Brewer’s yeast for those following a GFD. Why didn’t they look at the two types of yeast separately? They can make no direct comments regarding cross-reactivity and yeast in this particular study.
Reactivity was noticed with milk chocolate (see results from milk section). There was no reaction with pure cocoa, or milk-free dark chocolate. Note that many milk chocolate products can be cross contaminated with wheat, so it’s important to purchase chocolate that has been certified gluten free. Sadly nothing Godiva makes can be considered gluten free.
As you can see from Figure 3, there was no inhibition of anti-α-gliadin antibody binding when soy was added. However, the dot blots and other data are potentially suggestive. The inconsistency of reported results (i.e. which foods are “gluten cross-reactive?”) is concerning. Note that milled soy flour may be contaminated by gluten (the linked pilot study found more than 2,900 ppm of gluten in one product). There may be other arguable issues with soy that are not appropriate for this article, and will not be discussed here.
Vojdani & Tarash 2013 conclude (based on Figures 1 & 2 and 4 & 5) that “millet may not be a good substitute for gluten-containing grains for some individuals” suggesting that this is due to “gluten cross-reactivity.” It is important to note that milled millet often contains contaminating gluten – even on those products listed as gluten free, or “naturally gluten free;” this makes millet often unsuitable due to contaminating gluten, and not necessarily due to cross-reactivity. The authors note a reference stating “amylase inhibitor from barley had a significant homology with millet” however this does not appear to have any direct bearing on the antibodies specific for anti-α-gliadin used in this study.
Corn and rice:
These are briefly discussed in the conclusions of Vojdani & Tarash 2013, only to state that “We also obtained fresh corn on the cob and various rice grains and observed significant immune reactivity…” and make mention of a lipid transfer protein with ” high cross-reactivity rate with various grains and vegetables.” The cited studies appear to be specific for allergic (IgE-mediated) type reactions. Other studies have suggested that corn might be a problem for some people with Celiac disease, and some zein moities are computationally predicted to have sequence similarity to α-gliadin. However, CD-specific antibodies don’t appear to cross-react to corn zein. The rice/gluten cross-reactive study cited by Vojdani & Tarash 2013 is specific for IgE mediated responses, whose dominant epitopes are different than the epitopes recoginzed by anti-α-gliadin antibodies presented in this study. There may be other arguable issues with corn and rice that are not appropriate for this article, and will not be discussed here.
It is astonishing that more people in our community have not voiced concerns about the issues raised in this post.
Some have raised these concerns— notably Chris Kresser and Jane Anderson. Erica Dermer and Rebecca Black also wrote short synopses of a “Gluten Myth” panel presented at ICDS13 that discussed the lack of scientific and clinical basis for gluten cross-reactive food avoidance and testing. Even The Paleo Mom (in her updated post on the issue) has stated more than once that she questioned the validity of the cross-reactivity findings published in her original post.
“I could not find any published studies confirming the results from Cyrex Labs.”
When she contacted Cyrex labs, they were able to provide her with some research articles (though they couldn’t divulge “proprietary information”); however,
“…the paper did not show up on my PubMed searches.”
Yet, despite reading (and rereading) the text, she apparently came to the same conclusion – all 19 foods are potential gluten cross-reactive foods.
Note: The Celiac Disease Center does not recognize Enterolabs or Cyrex stool tests for cross-reactivity (or for CD for that matter). Simply, they are “not sensitive or specific enough” and just haven’t held water (yet) in the scientific arena. Maybe they will one day, but not today.
I would like to plead with the audience to take this post for what it is: a critical review of an article which has been misinterpreted and has spread throughout the blog-o-sphere like wildfire; it is intended to start a conversation. It is not a personal attack on the authors of the study, Cyrex labs, practitioners who use Cyrex labs, nor is it an attack on The Paleo Mom, whose work I very much appreciate and admire. I am very much open to continued discussion on the methods, conclusions, and results of the particular paper discussed here, as well as links to any scientific and peer-reviewed articles.
I also don’t want this post to detract from legitimate arguments concerning avoidance of some of the foods on the Cyrex Array 4 list. I believe it is true that “gluten free isn’t always enough” and this is true for those with CD as well as NCGS. I myself am diagnosed with NCGS, and have found symptom relief from several autoimmune and autoinflammatory diseases with a gluten free diet and also with the avoidance of sugar, non-gluten containing grains, alcohol and other foods – some of which are on the Cyrex Array 4.
There are many reasons to avoid other foods than gluten, and it is relatively easy (and cheap) to determine other foods you are sensitive to using a food journal and food rotation diet.
My only purpose is to highlight the importance of being rigorous with our own research and reporting within the ancestral health community, and to highlight that the rationale for avoidance of some foods may have arrived through inflated interpretation of inconclusive results.
Scientific research is only beginning to reveal the intricacies of “leaky gut” and how it may promote development of or exacerbate autoimmunity — and only time, clever questions, and intelligently designed experiments will yield fruitful results to push forward clinical and scientific dogma.
“The important thing is to not stop questioning.” – A. Einstein
Christina Graves is a doctoral candidate pursuing her degree in interdisciplinary biomedical sciences - her research focuses on the innate sensing mechanisms of intestinal epithelial cells and how this sensing may go awry in autoimmune disease. In addition to her graduate studies, she also writes for the AutoimmunePatient.com blog to communicate recently published scientific research to the public at large. She is a passionate patient advocate for individuals with narcolepsy and works to enhance public understanding of the complex nature of the disease. You can learn more about her professional career at www.clgraves.com.