The measurement of autoantibodies to insulin informs diagnosis of diabetes in a childhood population negative for other autoantibodies

Abstract

Aims: Some childhood type 1 diabetes cases are islet autoantibody negative at diagnosis. Potential explanations include misdiagnosis of genetic forms of diabetes or insufficient islet autoantibody testing. Many NHS laboratories offer combinations of three autoantibody markers. We sought to determine the benefit of testing for additional islet autoantibodies, including insulin (IAA) and tetraspanin 7 (TSPAN7A).

Methods: Radiobinding assays (RBAs) were used to test for four islet autoantibodies in children with newly diagnosed type 1 diabetes (n = 486; 54.1% male; median age 10.4 years [range 0.7-18.0]; median duration 1 day [range -183 to 14]). Islet autoantibody negative children were tested for TSPAN7A using a luminescence-based test. Where available, islet cell antibody (ICA) and human leucocyte antigen (HLA) data were considered.

Results: Using three autoantibody markers, 21/486 (4.3%) children were autoantibody negative. Testing for IAA classified a further 9/21 (42.9%) children as autoantibody positive. Of the remaining 12 (2.5%) autoantibody negative children, all were TPAN7A negative, seven were ICA negative and one was positive for the protective variant DQB1*0602. One was subsequently diagnosed with Maturity Onset of Diabetes in the Young, but follow-up was not available in all cases.

Conclusions: Using highly sensitive assays, testing for three autoantibodies fails to detect islet autoimmunity in approximately 1/20 children diagnosed with type 1 diabetes. Testing for IAA in children <5 years and GADA in those >10 years was the most effective strategy for detecting islet autoimmunity. The ability to test for all islet autoantibodies should inform clinical decisions and make screening for monogenic diabetes more cost-effective.

Keywords: autoimmunity; islet autoantibodies; type 1 diabetes.

FIGURE 1

Study flow diagram. *In eight children, due to low sample volume, IAA was measured in samples obtained within 2 weeks of diagnosis and GADA/IA‐2A/ZnT8A was measured by repeat sampling within 3 months of diagnosis (median duration 67.5 days [range 15–91]). ^§Available ICA data from historically conducted assays were considered in select analysis.

FIGURE 2

Comparison of GADA/IA‐2A/ZnT8A (a) and GADA/IA‐2A/ZnT8A/IAA (b) testing strategies. Testing for IAA provides an autoimmune classification of diabetes in 42.9% of children found autoantibody negative by GADA/IA‐2A/ZnT8A testing and identifies an additional 8.0% children positive for multiple (≥2) autoantibodies. The inclusion of IAA decreased the proportion of children found biochemical autoantibody from 4.3% to 2.5%.

FIGURE 3

The prevalence and sensitivity of autoantibody markers by age at diagnosis and the number/combination of tests performed. (a) The prevalence of most islet autoantibodies are associated with age at diagnosis; GADA increased with age > 10 years (p = 0.00016), IAA decreased with age > 10 years (p = 6.59 × 10⁻⁷) and is highest in children aged <5 years, and ZnT8A was higher in children aged >5 years (p = 0.011). There was no relationship between age at diagnosis and prevalence of IA‐2A (p = 0.486) or absence of islet autoantibodies under or over 10 years (p = 0.99). (b) Overall, the sensitivity increases and is generally highest when all four islet autoantibodies are tested independent of age. However, as a single test, IAA identifies the highest percentage of children <5 years. While GADA as a single test, identifies the highest percentage of children diagnosed >10 years.