Assessing Commercial Tissue-Based Assays for Autoimmune Neurologic Disorders (I): Antibodies to Intracellular Antigens

Abstract

Background and objectives: Current strategies to detect autoantibodies against intracellular neural antigens (IC-Abs) include tissue-based assays (TBAs) alongside line blots or cell-based assays (CBAs). Many clinical laboratories use commercially available TBAs as a screening test, but their diagnostic yield has not been assessed. We determined the performance of 2 commercial TBAs in detecting IC-Abs.

Methods: We analyzed samples from 100 patients with autoimmune or paraneoplastic neurologic syndromes harboring IC-Abs (confirmed by in-house TBAs and line blots or CBAs) and 50 negative controls. IC-Abs samples included serum (10 each for Hu, Yo, Ri, SOX1, CV2, Ma2, Tr, amphiphysin, and GAD65 antibodies) or CSF (10 with GFAP antibodies) samples. Two commercial indirect immunofluorescence (IIF) TBAs (INOVA and EUROIMMUN) were assessed by 2 experienced investigators and 3 less experienced raters, all blinded to antibody status. Discordant results were re-evaluated through interrater discussion and assessed using Cohen's kappa.

Results: The 2 experienced raters showed substantial agreement (85% for INOVA, 83% for EUROIMMUN) on negative/positive results, which increased to >95% after interrater discussion (Cohen's kappa 0.95 and 0.93, respectively). With IIF-INOVA, they correctly identified 118 of 150 samples (79%) and misclassified 28 of 150 (19%, 2 false positives and 26 false negatives) while results remained discordant in the remaining 4 of 150 samples (2%). With IIF-EUROIMMUN, they correctly identified 105 of 150 samples (70%) and misclassified 40 of 150 (27%, 6 false positives, 34 false negatives), with discordance in 5 of 150 samples (3%). Overall, the sensitivity was 73% for IIF-INOVA and 66% for IIF-EUROIMMUN. The specificity was 96% for IIF-INOVA and 88% for IIF-EUROIMMUN. Both TBAs showed low sensitivity in detecting CV2, SOX1, and amphiphysin antibodies while Ma2 antibodies were missed mainly by IIF-EUROIMMUN and Hu/Ri antibodies by IIF-INOVA. Antibody-specific immunostaining patterns were correctly identified in 62 of 100 positive samples with IIF-INOVA and 55 of 100 with IIF-EUROIMMUN (p = 0.34). Less experienced raters showed higher rates of false-positive results (up to 22% with IIF-EUROIMMUN).

Discussion: The performance of commercial IIF-TBAs for IC-Abs detection is suboptimal, exhibiting high false-negative rates of 25%-35%. Therefore, commercial TBAs should not be used alone for IC-Abs screening, but alongside more sensitive techniques, such as line blots. Discordant results between 2 techniques should prompt retesting in reference centers with in-house TBAs, particularly when the suspicion of an autoimmune or paraneoplastic syndrome is high.

Figure 1. Algorithm of the Study

Abs = antibodies; CBA = cell-based assay; IC-Abs = neural antibodies targeting intracellular targets; IHC-TBA = immunohistochemistry tissue-based assays; IIF-TBAs = indirect immunofluorescence tissue-based assays. *Confirmed by in-house IHC-TBA and a confirmatory test (line blot and/or CBA).

Figure 2. Proportion of True and False Positives/Negatives and of Discordant Results With INOVA (A) and EUROIMMUN (B) IIF-TBAs

IC-Abs = neuronal antibodies against intracellular targets; NC = negative controls.

Figure 3. Heatmap Illustrating the Identified Antibody-Specific Immunostaining Patterns for Each Sample Tested With the 2 Commercial IIF-TBAs

INOVA (A) and EUROIMMUN (B) positive samples are divided into different columns according to the specific antibody. For each antibody column, each of the 10 samples is color-coded according to the results after interrater agreement between the 2 expert raters. In case of samples considered positive, the identified antibody-specific pattern is specified (green: correctly identified as positive for the true antibody; light green: correctly identified as positive but for the wrong antibody; dark green: correctly identified as positive but without an identifiable pattern; wine color: wrongly identified as negative [ = false negative]; gray: discordant). (C) Negative controls tested with INOVA IIF-TBA (top row) and EUROIMMUN IIF-TBA (lower row) and the obtained results (green: correctly identified as negative; wine color: wrongly identified as positive; gray: discordant) and identified pattern of reactivity when the sample was considered positive.

Figure 4. Expected Cerebellar and Additional Tissue Staining in Commercial IIF-TBAs and Comparison Between INOVA and EUROIMMUN for Hu/Ri, Yo, GAD65, SOX1, and Ma2 Antibodies

GL = granular layer; ML = molecular layer; PC = Purkinje cell. Scale bar = 100 μm for the images in the first 4 rows; scale bar = 20 μm for the images in the last row.

Figure 5. Expected Cerebellar and Additional Tissue Staining in Commercial IIF-TBAs and Comparison Between INOVA and EUROIMMUN for Tr, Amphiphysin, CV2, and GFAP Antibodies

GL = granular layer; ML = molecular layer; PC = Purkinje cell. Scale bar = 100 μm.

Figure 6. Main Advantages of Additional Tissues in Both INOVA and EUROIMMUN IIF-TBAs

The presence of cerebral cortex in INOVA IIF-TBA allowed the recognition of Ma2-positive (A) and SOX1-positive (B) samples when the expected cerebellar staining was not clearly visible. The presence of intestinal tissue in EUROIMMUN IIF-TBA (C), similar to the presence of stomach tissue in INOVA IIF-TBA (not shown), allowed demonstration of reactivity with the myenteric plexus in Hu-positive samples. The presence of pancreatic tissue in EUROIMMUN IIF-TBA (D) allowed identification of GAD65 antibodies. Scale bar = 20 μm for (A) and (B); scale bar = 100 μm for (C) and (D).

Figure 7. Main Nonspecific Immunostainings Identified in Control (Negative) Samples That May Resemble the Patterns From Positive Samples

Neurofilament-like pattern (A) with diffuse staining of axons in the granular layer and in the white matter of the cerebellum; blood vessels staining (B) with diffuse staining of blood vessels' wall (white arrows); SOX1-like pattern (C) resembling the staining of nuclei of Bergmann glia cells in the Purkinje cell layer of the cerebellum (white arrows); and amphiphysin-like staining (D) with diffuse neuropil staining of the molecular layer. GL = granular layer; ML = molecular layer; PC = Purkinje cell; WM = white matter. Scale bar = 100 μm.