May 10, 2003
Diagnosis and Symptoms
Question from a physician in South Carolina, USA:
What is the latest update as to the sensitivity and specificity of pancreatic antibodies (ICA 512/IA , GAD 65, insulin autoantibodies or combinations thereof) to "best" establish a diagnosis of type 1A diabetes? What are the false positive and false negative rates?
I asked a colleague about this, since I had realised for some time that our antibody results here are reported out as ‘positive’ (99th percentile) or ‘negative’ (<99th percentile) and not in terms of a number. I was told that some time ago, because of wide variations in how antibody results were reported, the Immunology in Diabetes Mellitus Society held a workshop to set certain standards, recognising that there were always differences in technique between laboratories. In effect, this meant that each certified lab had to meet certain defined criteria and were responsible for setting their own 99th percentile levels for the normal population in their area. This arrangement seems to have worked well both for distinguishing type 1A (autoimmune) diabetes in people with diabetes and the risk of getting clinical diabetes in those who do not have it. This means also that individual labs have the responsibility of assessing their own sensitivity and specificity levels which probably depend mostly on who is doing the test. DOB