To lend some clarity, a group of researchers in Romania analyzed the significance of genetic tests. It has been estimated that 98% of people with celiac disease have the DQ2 and DQ8 HLA haplotypes. But because these alleles are found in as many as 40% of the general population, they are considered more a prerequisite for developing celiac than a true positive indicator. However, they have high negative predictive value: their absence has traditionally been used to rule out a diagnosis of celiac.
Samasca et al. looked at the HLA types of thirty-seven children with celiac disease confirmed by duodenal biopsy. Some also had confounding conditions like malabsorption syndromes, type I diabetes, and juvenile rheumatoid arthritis. Nine of the children were negative for the HLA DQ2 and DQ8 haplotypes. Of these nine, only two tested positive for the presence of anti tissue transglutaminase antibodies. One of them, a four year old boy, improved on a gluten free diet; the other, a 9 year old girl, did not comply with her gluten free diet, making it difficult to draw conclusions from her symptoms (notably an unsatisfying weight curve).
Based on these meager data, the authors conclude that the absence of DQ2 and DQ8 should not be used to exclude a diagnosis of celiac disease. Yet another, more plausible explanation might be that the few children in their study who lack these haplotypes as well as antibodies against tissue transglutaminase but still exhibit unspecified "changes on duodenal biopsy" might be suffering from gluten intolerance, which is mediated by an immune pathway less defined than the one leading to celiac disease. Still, their point that diagnosing celiac disease can be a complicated business is certainly a valid one.