Diagnostic Testing Algorithms for Celiac Disease
Published: June 2010
Hot Topic Q&A is an opportunity for viewers to ask questions of the Hot Topic presenter.
For this Hot Topic Q&A, a conference call was held June 22, 2010 at 3:00pm CST. The questions presented below are unedited.
Good afternoon and welcome to Mayo Medical Laboratories’ Hot Topic Q&A. Our program today is a follow-up to the Hot Topic “Diagnostic Testing Algorithms for Celiac Disease” and the Clinical Insight Hot Topic “Diagnostic Challenges of Celiac Disease.” Our presenters are Dr. Melissa Snyder and Dr. Joseph Murray. Dr. Snyder is Director of the Antibody Immunology Laboratory, in the Division of Clinical Biochemistry and Immunology at Mayo Clinic. Dr. Murray is a practicing gastroenterologist and Professor of Medicine and Immunology at Mayo Clinic, as well as Director of Mayo’s Celiac Disease Program.
Sometimes our technologists see a borderline positive tTG and EMA neg and wonder which one is more accurate as a screen test. As you know the EMA-IFA is technologist-driven by expertise and technique, and may suffer as such in terms of inter-tech reproducibility. Whereas the tTG is an automated assay that potentially has better reproducibility. What are your thoughts on this?
(Dr. Snyder) So getting back to the question, I think the reproducibility for tTG and EMA are very similar. Again, those assays in the upper range probably have a little bit more issues with precision. I mention that because I think it’s relevant because these antibodies can be used to monitor patients with celiac disease so we want to keep in mind that especially at the higher levels that precision can be somewhat of an issue. Dr. Murray, do you have anything to add to that?
(Dr. Murray) Frequently, we’ll see patients with celiac disease, where at the beginning; their antibody test result is greater than the upper limits of the test. For example, of greater than 100. We have no idea how much greater than 100 is. As Dr. Snyder pointed out, it’s not linear at that level. And that does lead to some limitations in how we can use changes in the serologic test as a monitoring response. And I look at the degree of change. So I expect I usually repeat the testing in patients six months after going on a gluten-free diet and I expect to see a substantial drop in their serologic test results at that point. I usually pick just the one test so if they’re very positive of the tTG IgA, I just use that test to follow them. I don’t repeat all of their tests if I’m following up. I expect by a year most patients on a strict gluten-free diet, they will have become negative. So it’s really the degree of change that counts. One of the problems of course is that when change is analyzed, or change is the test method, or the test laboratory, then you may end up not being able to compare the pretreatment result with the posttreatment result and you’re really just left with going from positive to negative as a categorical value.
Do patients who have tested negative need to be retested again?
(Dr. Murray) I’ll take that question and certainly not everybody with celiac disease is sero-positive. There are some sero-negative individuals. Now some of that could be due to IgA deficiency and that’s been some of the impetus behind our introduction of these cascades is to address the issue of IgA deficiency. Three percent of patients with celiac disease have selective IGA deficiency. That means they either have no IgA at all or very low IgA so you cannot rely on a transglutaminase IgA test or endomysial antibody IgA test for detection. I’ve seen their IgA level to make sure that they would have sufficient IgA for a valid test will get rid of some of those false negatives, and even with that, there are some individuals who have a negative serologic test despite having celiac disease. The most common reason that I see is that the patient has already altered their diet. So for some individuals, reducing gluten or removing gluten from the diet can dramatically drop their serologic tests. Even in a matter of days to weeks.
So it is very important for the clinicians to be aware when they are undertaking testing for celiac disease, they have to ask the patient, “have you been on a gluten-restricted diet,” because if they have, it will reduce the sensitivity. That aside, there probably still are some individuals who are truly sero-negative. When that happens, so if I have somebody who is sero-negative but I strongly suspect celiac disease or malabsorption, for example based on their symptoms. I’m going to proceed and recommend referral to a gastroenterologist and biopsy anyway because of course there are other disorders of malabsorption that mimic celiac disease; tropical sprue, Whipple’s disease, and a number of other conditions that can damage the small intestine or can result in similar symptoms that need to be identified so that’s another circumstance.
But even with that, we’re still left with some that are truly sero-negative and when would you retest somebody. We really don’t know. There is very little data on what somebody’s future likelihood on getting celiac disease is. If the person starts out as a family member, of somebody with celiac disease, those people are more likely to have celiac disease and there is a number, probably a very small percentage. Once you’ve tested them once as an adult, for example, they’re rarely likely to get it again. If you test a young child however, 3 or 4 years of age, for example, then there is a much higher likelihood of that child in future develop celiac disease so one shouldn’t completely close that door of suspicion. Another circumstance are type I diabetics. Individuals with juvenile onset diabetes are particular high risk of celiac disease, maybe 5% to 6% of whom develop celiac and some advocate repeat testing over time. Another circumstance are children with Downs syndrome whose lifetime risks of celiac disease may be as high as 10%. So sometimes repetitive testing is necessary in those circumstances especially with the young people.
Dr. Snyder, did you want to comment further on that?
(Dr. Snyder) No, I think Dr. Murray covered everything pretty well. I guess the only other thing to mention is the HLA typing and in the context of a negative HLA result, it probably is not necessary to do serologic testing. And the other thing to keep in mind is that once an individual has an HLA type, there really is no need to do any sort of retyping or test them again for their HLA.
What is the role for testing of family members of patients diagnosed with celiac disease?(Dr. Murray) Several studies, including our own have shown that there is a substantial reservoir of hidden celiac disease among family members of people with celiac disease. So when I diagnose celiac disease in a patient, I will as part of my first evaluation, recommend brothers, sisters, children and parents, if living, should be tested for celiac disease. It doesn’t matter if they have symptoms or not, they should all be tested because of the very high likelihood of celiac disease in a family member. The risk for a sibling, a brother or a sister, or somebody with celiac disease may be as much as 15% or even 20%. Children, it’s a little less and parents maybe a little less more in the 5% to 10% rate but I would say about 50% of patients that you see as a first family member, the first member of the family with celiac disease, you will find another family member, and it’s also important if they live in the same household. Test them sooner rather than later because the diet in the household will change especially if there’s an affected child because there may be less and less gluten and that may make it harder to make the diagnosis in family members.
What is the most appropriate test for screening for celiac disease in adults? What about in children?
(Dr. Snyder) As far as screening tests go, I think that a lot of studies have certainly looked at this and probably most of them are quite consistent showing that the tTG IgA antibody is probably the most appropriate screening test for celiac disease, at least in adults. Again, as Dr. Murray mentioned, the IgA deficiency can complicate that, so it certainly is important because most people that you screen using the tTG IgA antibody, are going to screen negative, so you want to make sure you are not missing an individual because there is an IgA deficiency. So that is why the recommendation from Mayo would be to use IgA testing first and then base your serology testing from that IgA result.
Dr. Murray, what do you think about testing for children?
(Dr. Murray) I think that in children, in the past, we thought that in very young children under the age of 3, that tTG testing was less sensitive. In general, I think the more recent studies suggest that it probably preforms reasonably well. Now of course in these very young children, there are lots of other conditions that can mimic celiac disease. Other food protein intolerances can cause damage in the intestines. Lots of other symptoms that occur in that age group, so it is a little bit more challenging age group to make a diagnosis. However, tTG IgA antibodies possibly combined with some other approach including, as Dr. Snyder already mentioned, measuring total IgA is important. But one could make a case for adding in for example, deamidated gliadin peptide antibody tests to try and maximize sensitivity in that setting. But there really haven’t been a lot of studies in multiple tests analyzing in that particular very young age group. Children over the age of 3 or 4 behave very much like adults in terms of their serologic results so I would use the tissue transglutaminase IgA as the primary screening test in anyone over the age of 3 for example. You gain relatively little by adding in all the other tests, as putting them all together as a panel. You add very little in terms of sensitivity. Unless the patient is IgA deficient, in which case those other test take on much greater importance.
In what clinical settings is HLA typing for celiac disease most appropriate?
(Dr. Murray) HLA testing for celiac disease is really most powerful as a rule out test meaning that the individual who does not carry the HLA genes and they are actually gene pairs that encode for an HLA type, if they don’t have those, it makes celiac disease very, very unlikely. I didn’t say impossible because there are very rare exceptions. But in your clinic, if you are seeing a patient for whom celiac disease is a consideration and you do the HLA type and they don’t carry these HLA genes that encode for celiac susceptibility types, then you can say it is very, very, very unlikely this patient has celiac disease. As I said, there are rare exceptions. On the other hand, there are particular groups in which it is very helpful. So if I have a patient with Down syndrome for example, you’ve got a young individual with Down syndrome, they have the same prevalence of the HLA risk factors as the general population. You do the test on them, they don’t have the HLA type for celiac disease, you can virtually stop doing any testing for celiac disease on that group. You never have to retest them.
For family members it is a little different because if you are the family member of somebody with celiac disease your more likely to carry the HLA type and it may be as many as two-thirds of siblings of somebody with celiac disease have an HLA type putting them at risk of celiac disease. But only at most, 20 percent of them will get celiac disease. So that’s a circumstance where you end up with more false positives. You will have about a third of those siblings where you can say that celiac disease is never possible, so they are basically exonerated from any likely hood of celiac disease in that family context.
Other circumstances, the patient where they didn’t have serology done at the time of diagnosis but they had a biopsy that suggested it but they didn’t get better. HLA typing in that circumstance may help identify those who really had celiac disease or those who haven’t. People who come to us who have been on a gluten-free diet for a year for example and they want to know if they have celiac disease but they are not very enthusiastic about going back on a gluten challenge. HLA typing in that group may be very helpful. If you say you don’t have the HLA type, then they can’t have celiac disease.
So those are some of the circumstances in which I find HLA typing is useful and like the family members where it is a little more limited. When you get into complex family structures with multiple affected individuals and sometimes very complicated family structure in terms of parenthood for example, one has to be very cautious about undertaking this type of genetic testing in families because it really is a type of genetic testing in the family context and in that circumstance professional genetic counseling may be helpful in directing that care. It’s very useful if an individual does not have the HLA type it’s very unlikely there’s any point in doing any other testing. As I said, the most use for that, the individual already on a gluten-free diet.
Dr. Snyder, do you have anything further to add to that?
(Dr. Snyder) No, I don’t think so. I think Dr. Murray covered it completely.
What is the sensitivity and specificity of IgG tTG testing in IgA-deficient celiacs, as well as the sensitivity and specificity in non-IgA-deficient celiacs?
(Dr. Snyder) I can start off by addressing some of this question. As far as IgG tTG testing goes, I think probably a lot of the studies have looked in IgA deficient individuals and generally speaking the IgG anti-tTG antibodies assays tend to be significantly less sensitive than the IgA antibody tests. That’s why you really need to supplement that IgG antibody testing, not just perform the tTG but also use the deamidated gliadin in a patient who truly is IgA deficient. I think a few studies have shown that the deamidated gliadin IgG has significantly improved sensitivity in comparison to the IgG tTG while still maintaining a fairly good specificity. What I am not as sure about, and maybe Dr. Murray has some insight into this, in the non-IgA deficient individual, how do you think the sensitively and specificity is for the IgG antibodies?
(Dr. Murray) In our studies, looking at the tTG IgG in IgA sufficient individuals, the sensitivity is much lower. It’s probably close to 30%. And that’s because, in a normal individual with normal IgA terminal differentiation of their plasma cell response means they are going to end up with just producing IgA antibodies and they often don’t produce any IgG antibodies. So if you do both tests together, it’s pretty common in a celiac patient that they are positive for tTG IgA but negative for their tTG IgG test, so that’s not rare at all, in fact I would call that almost the norm. But what about the other side of the question which is specificity of the tTG IgG? I will certainly see false-positive tTG IgG in patients with other autoimmune diseases like thyroid disease or even type 1 diabetes so the specificity of that test drops significantly. I think at Mayo we’ve gone through a series over the last 15 years of changes where we started off just using the tTG IgA then added in the tTG IgG and now have we’ve come back to the idea of using your tTG IgA level as the best way to discriminate which is the best appropriate serologic test to do next, and in most cases of course it’s the tTG IgA and the tTG IgG adds very little except in the context of the IgA deficient individuals. So be cautious about interpreting somebody whose tTG IgG is positive, their tTG IgA is negative, but they’ve got normal IgA levels. That person in my experience often does not have celiac disease.
Do you have to stop doing anti-endomysial testing?
(Dr. Snyder) Well, I’m assuming that the question means do you have to stop in terms of if you’re performing tTG testing is there a role for anti-endomysial (EMA) testing? At Mayo we do still perform the EMA testing for the IgA antibody, not for the IgG but only the IgA isotype, and I think it can be useful in circumstances where perhaps the tTG testing is either equivocal or perhaps it is negative in a patient who you know is HLA positive. So I think it still can have a role perhaps not as a screening test due to the reproducibility issues and just the manual nature of the testing that we discussed before, but I think it still can have a role perhaps as secondary to the tTG antibody testing.
(Dr. Murray) Circumstances in which I’ve used the endomysial antibody are patients whose tTG is positive but they’re not willing to undergo a biopsy. We know that in circumstances, say for example, where your pretest prevalence of disease is only 10% or even 5%, one in 20 patients that you are going to see with, say Type 1 Diabetes, are going to have celiac disease. Your tissue transglutaminate antibodies, maybe not strongly positive but kind of weakly positive, if your EMA is positive, then that patient almost certainly has celiac disease. Whereas, if their EMA is negative, it makes celiac disease much less likely in that circumstance. Sometimes that can replace, if somebody is not willing to undergo a small bowel biopsy, or there’s some other compelling reason, for example they are not medically fit for it, or it’s not available in a timely fashion, the endomysial antibody can be helpful in that circumstance.
The other application that I’ve used this for is when I’m testing a general population sample. So I have somebody who is tTG positive in the general population sample, I know the positive predictive value in that circumstance may be about 50%, not 100%. In that case I will do the endomysial antibody and if it’s positive it takes the positive predictive value up to maybe close to 100%. So it can be very useful when you don’t have ready access to small intestinal biopsy to confirm the diagnosis. It is subject to those issues, the technical difficulty. Our lab here does perform probably in the 10’s of 1000’s of samples a year so we are a high volume lab for that test. It does require a lot of quality control to watch out for things like observer variability in terms of interpretation of the test, and it is resource intensive, and it does mean a slower turnaround time because it takes a little longer to do that test than our standard ELISA tests.
Is a biopsy necessary even if the tTG is very strongly positive, say 5 times the upper limit of normal or above?
(Dr. Murray) That’s an excellent question. I think I’ll take that one. I should disclose my conflict of interest; I’m a gastroenterologist so I make my living by doing endoscopy, or at least my profession does. I would make the case that it depends on the certainty of the test, and that is somewhat dependent on what the pretest likelihood of the diagnosis is. So if the patient has a very low pretest likelihood, 1% (1 in 100 for example) and you have a test that is 98% specific, and you’re doing that in somebody with a very low likelihood of disease, then your post-test likelihood is only about 33%.
You’re going to have 2 false positives for every 1 true positive; so a one-in-three chance that you’re right. In that circumstance I think definitely one has to do a confirmatory test. As I mentioned, the endomysial antibody might be a way to get there, and still it’s a biopsy. For adults there is another importance for a biopsy, and that is that the biopsy is a baseline for which we follow up. We know about 50% of people who are diagnosed with celiac disease as an adult will not heal their intestine and failure to heal the intestine may be associated with a negative outcome. So in my practice, we not only take a biopsy to confirm the diagnosis at the beginning, but we take a repeat biopsy after a year or 2 of a gluten-free diet to insure the patient has healed. So that’s a little bit further in adults. In children they still require a biopsy; the professional guidelines require a biopsy to confirm what is essentially a lifelong diagnosis. There are occasional false positives in family members who haven’t yet developed the disease that have what is called latent or potential disease. These are people without symptoms where we find a positive tTG antibody, even a positive endomysial antibody, but their biopsy is negative. Those kids, usually children, need to be followed as they may become positive in the future, but they may not actually have a Celiac Disease at the time you initially study them.
(Dr. Snyder) I think this was a relevant question. I was at a Celiac Disease workshop at the last AACC meeting and this very question came up. Not only if the tTG is positive or strongly positive, but what if you have a tTG that is essentially a greater than value, you have a deaminated gliadin that’s a “greater than” value, perhaps even with an EMA result that’s positive, and the question again was, is a biopsy necessary? And I’m not a gastroenterologist so I wouldn’t be able to say one way or another but I think that’s certainly what a lot of people are thinking out there, can you, if you have enough positive serology, can you avoid it. But that doesn’t get past the issue of repeat testing….
(Dr. Murray) I think in that circumstance where you have all those positives, the certainty of Celiac Disease approaches 100%, so it’s not an issue of certainty of diagnosis, it’s more an issue of the management, subsequent management of the patients.