Food elimination in IBS: the case for IgG testing remains doubtful

I read with interest the study of a diet for irritable bowel syndrome (IBS) based on serum IgG levels to foods (Gut 2004;53:1459–64).

In rigorous elimination diet studies, about one third of IBS patients are found not to have food intolerance.^1–3 Yet it appears that everyone tested for food specific IgG in this study had some positive reactions and was therefore subjected to dietary recommendations. This does not in itself suggest that serum IgG is a particularly useful test.

One notable finding of this study appears to be that 87% of patients gave a high level of IgG to yeast. In two large scale studies of IBS using diagnostic elimination diets, the percentages who had a symptomatic reaction to yeast when challenged were 5.5% (out of 73 unselected IBS patients)¹ and 12% (out of 122 unselected IBS patients).² It seems unlikely that yeast causes IBS symptoms in 87% of patients in Manchester but in only 5–12% of patients in Oxfordshire and Cambridgeshire. A logical implication is that high levels of IgG against yeast do not, in themselves, reveal anything significant in relation to IBS symptoms.

The same, in my view, would follow for several other foods. The numbers of patients with positive responses to eggs, cow’s milk, and cashew nuts, as judged by IgG levels, are much higher than one would expect from empirical dietary studies,^1,2 while the numbers testing positive to chocolate and oranges appear far too low. Again, it seems doubtful that IgG can reveal sensitivities accurately in IBS.

The percentage of patients showing substantial benefit from this diet is disappointing. In studies using a well conducted and rigorous elimination diet, the “number needed to treat” is between 1.5 and 2.2.^1–4 The “number needed to treat” in this study was 9. (The value of 2.5, calculated on the basis of those who fully complied with the diet, abrogates the intention to treat principle.)

This seemingly poor response to an IgG based diet confirms the widely held view to date that IgG testing for food intolerance is not of value.^5–7 These results suggest that if IgG testing identifies food intolerances at all, it does so fortuitously and with an apparent low degree of accuracy.

I conclude that the difference in outcome between the “true diet” and the “sham diet” groups can largely be explained, not by specific identification of food reactions, but by the gross differences between the two diets. The “true diet” excluded milk products for 84% of patients and wheat for 49% (both foods are known to be common offenders in IBS) while the total number of foods avoided by the group was 498 (value calculated from table 2). For the “sham diet” group, 1.3% avoided milk, 8% avoided wheat, and the total number of foods avoided was only 453. These overall differences between the diets could easily explain the modest difference in outcome between the two diet groups. The same diet sheets, distributed randomly to the patients in each group, regardless of IgG levels, would probably have produced the same overall result.

Similarly, I consider that the effectiveness of the blinding in this trial is questionable. The “nutritional advisor” giving support by telephone may have become aware of which patients were receiving the “sham diet” as this regularly excluded potatoes and rice, while the “true diet” rarely did so—the reverse being true for wheat, milk, and yeast. The views of the nutritional advisor on the likely effectiveness of the diets could inadvertently have been communicated to the patients, and unintentionally influenced their assessment of the outcome.

Before this trial was begun, in my view it would have made sense to try to answer the more basic research question: do high levels of IgG against a food predict an adverse reaction to that food? Only one very small trial has so far done this.⁸ It measured food specific serum IgG in individual IBS patients and compared the results with those from food challenges (following a period of avoidance): there was no correspondence between the foods identified. Such work needs to be repeated with larger sample sizes.

Despite the inconclusive results of this study, it has regrettably already been the subject of a press release and other publicity by the company that provided the IgG testing for this study, in order to promote IgG tests to the general public. On the company’s website, IgG testing is now described as “clinically proven” by the British Allergy Foundation on the basis of this study (The UK YorkTest website: www.yorktest.com). This blurring of the boundaries between what should be a disinterested scientific enquiry and the promotion of a commercial venture is regrettable.