Improved Specificity of Glutamate Decarboxylase 65 Autoantibody Measurement Using Luciferase-Based Immunoprecipitation System Assays

Abstract

Autoantibodies to glutamate decarboxylase (GADA) are widely used in the prediction and classification of type 1 diabetes. GADA radiobinding assays (RBAs) using N-terminally truncated antigens offer improved specificity, but radioisotopes limit the high-throughput potential for population screening. Luciferase-based immunoprecipitation system (LIPS) assays are sensitive and specific alternatives to RBAs with the potential to improve risk stratification. The performance of assays using the Nanoluc luciferase (Nluc)-conjugated GAD65 constructs, Nluc-GAD65(96-585) and full length Nluc-GAD65(1-585), were evaluated in 434 well-characterized serum samples from patients with recent-onset type 1 diabetes and first-degree relatives. Nonradioactive, high-throughput LIPS assays are quicker and require less serum than RBAs. Of 171 relatives previously tested single autoantibody positive for autoantibodies to full-length GAD65 by RBA but had not progressed to diabetes, fewer retested positive by LIPS using either truncated (n = 72) or full-length (n = 111) antigen. The Nluc-GAD65(96-585) truncation demonstrated the highest specificity in LIPS assays overall, but in contrast to RBA, N-terminus truncations did not result in a significant increase in disease-specificity compared with the full-length antigen. This suggests that binding of nonspecific antibodies is affected by the conformational changes resulting from addition of the Nluc antigen. Nluc-GAD65(96-585) LIPS assays offer low-blood-volume, high-specificity GADA tests for screening and diagnostics.

Figure 1

A: Diagram of the Nluc GAD₆₅ constructs, which were assessed for the sensitivity and specificity of GADA measurement by the LIPS assay. PLP, pyridoxal 5′phosphate. B: A plot of 25 low-risk, single GADA(96–585)–positive relatives who had not developed diabetes during follow-up (grey circle) and 11 patients with recent-onset T1D (black triangle) who were measured for GADA using ³⁵S-GAD₆₅(96–585) in radioimmunoassay and Nluc-GAD₆₅(1–585) and Nluc-GAD₆₅(96–585) in LIPS assays. Filled triangles and circles indicate positives by that construct by individual assay thresholds. Dotted line indicates the positivity threshold for ³⁵S-GADA(96–585). In patients, the median antibody levels (DK units/mL) for ³⁵S-GADA(96–585), Nluc-GADA(1–585), and Nluc-GADA(96–585) were 341.29 (range, 157.2–1267.3), 494.4 (range, 60.4–1072.5), and 357.8 (range, 73.8–988.1), respectively. In relatives, the median antibody levels were 42.0 (range, 18.3–368.5), 5.16 (range, 0.39–25.6), and 9.5 (range, 3.2–63), respectively.

Figure 2

A: A plot of ³⁵S-GADA(1–585) (grey triangles), Nluc-GADA(1–585) (orange triangles), and Nluc-GADA(96–585) (teal triangles) levels against ³⁵S-GADA(96–585) levels in 156 patients with recent-onset T1D. Overall, correlation of ³⁵S-GADA(1–585), Nluc-GADA(1–585), and Nluc-GADA(96–585) with ³⁵S-GADA(96–585) was excellent (r = 0.99, 0.91, and 0.87, respectively; P < 0.0001 for all). B: Receiver operator characteristic curve for ³⁵S-GADA(1–585) (black line), ³⁵S-GADA(96–585) (grey line), Nluc-GADA(1–585) (orange line), and Nluc-GADA(96–585) (teal line) measured by radioimmunoassay or LIPS based on data from 156 patients with newly diagnosed T1D and 221 healthy schoolchildren. The area under the curve (AUC) was 0.94 for ³⁵S-GADA(1–585), 0.93 for ³⁵S-GADA(96–585), 0.93 for Nluc-GADA(1–585), and 0.93 for Nluc-GADA(96–585). The partial AUC (at specificities >90%; within the grey box) was 0.082 for ³⁵S-GADA(1–585), 0.081 for ³⁵S-GADA(96–585), 0.084 for Nluc-GADA(1–585), and 0.082 for Nluc-GADA(96–585).

Figure 3

Kaplan-Meier survival curve for FDRs positive for ³⁵S-GADA(1-585) according to positivity for ³⁵S-GADA(96-585) (black lines), Nluc-GADA(1-585) (orange lines), and Nluc-GADA(96-585) (teal lines). ³⁵S-GADA(96-585), Nluc-GADA(1-585), and Nluc-GADA(96-585) identified relatives at increased risk of diabetes progression. Individuals positive for ³⁵S-GADA(96-585) had a 32% risk of developing diabetes within 15 years, and individuals positive for Nluc-GADA(1-585) or Nluc-GADA(96-585) had a 30% risk.