Remark on utility and error rates of the allopurinol test in detecting mild ornithine transcarbamylase deficiency

Abstract

Carriers of X-linked ornithine transcarbamylase deficiency (OTCD) are themselves mildly affected. The allopurinol test is quite sensitive (92.7%) and very specific in detecting these carriers. Consequently, it has also been recommended for the diagnosis of mild OTCD in the general population. However, there is a controversy on its utility since OTCD could not be demonstrated in several patients with positive test results but negative family histories. We show that this controversy is due to an improper use of statistical concepts, i.e., to the postulate of a specificity of “100%,” and to the confusion of specificity with type I error rate. Spontaneous orotic aciduria implies a positive allopurinol test and limits the specificity of the test to a maximum of 99.7%. Therefore, according to Bayes' theorem, almost all positive test results in the general population must turn out to be type I errors, due to the minute prevalence (1/32,000) of mild OTCD (i.e., asymptomatic carriers and male patients with inapparent disease). Family history seems to be the only preselective parameter that can sufficiently raise the prevalence in the group to be tested. Bayesian analysis also yields the rate of type II errors (OTCD inspite of a negative test) which is high in closely related at-risk females (22.6% in mothers of male patients) but minimal in the general population. Conclusion. The allopurinol test is useful for the exclusion but not for the diagnosis of inapparent OTCD in sporadic individuals. Test results in possible carriers should be interpreted with caution.