GAD Antibodies in NIDDM: Ten-year follow-up from the diagnosis

Abstract

OBJECTIVE To study the frequency of antibodies to glutamic acid decarboxylase (GAD) and islet cell antibodies (ICAs) and their predictive value with respect to the development of insulin deficiency in 133 newly diagnosed middle-aged patients with non-insulin-dependent diabetes mellitus (NIDDM) and in 126 control subjects and to study the persistence of GAD antibodies in diabetic patients during the follow-up. RESEARCH DESIGN AND METHODS The study participants consisted of a well-characterized group of 133 middle-aged newly diagnosed patients with NIDDM and 126 control subjects. The follow-up examinations were performed 5 and 10 years after the baseline. The development of absolute and relative insulin deficiency was based on a stimulated C-peptide level that was undetectable or < 0.70 nmol/l, respectively. GAD antibodies were measured retrospectively from stored samples. RESULTS The overall prevalence of GAD antibody and ICA positivity at the time of diagnosis was 9.0 and 3.8% in diabetic patients and 1.6 and 0% in the control population, respectively. During the 10-year follow-up, 3 (2.3%) and 10 (7.5%) of the diabetic patients developed absolute and relative insulin deficiency, respectively. Of these, two (67%) and six (60%) had been GAD antibody-positive at the time of diagnosis. The sensitivity and specificity of the GAD antibody to predict absolute or relative insulin deficiency were 67 vs. 94% and 60 vs. 95%, while corresponding figures for ICA were 33 vs. 97% and 20 vs. 98%, respectively. The negative predictive value of GAD antibody testing was higher than positive predictive value (97 vs. 50%). During the follow-up, low-grade GAD antibody positivity showed an evanescent nature, whereas the high levels were quite persistent. CONCLUSIONS In an unselected population of newly diagnosed NIDDM patients, the prevalence of latent autoimmune diabetes in adults was <10%. While GAD antibody and ICA measured at the time of diagnosis of NIDDM are equally specific predictors of subsequent insulin dependency, the GAD antibody may have a higher sensitivity. Therefore, measurements of GAD antibody may aid the clinician in the choice of treatment of these patients.