Abstract
Objectives
Antibodies against glutamic acid decarboxylase isoform 65 (GAD-Ab) have been found in different severe neurological conditions associated with altered synthesis of γ-aminobutyric acid (GABA). Serum GAD-Ab can be found in up to 90 % of patients with type 1 diabetes mellitus (T1DM), mostly at relatively low concentrations, while high concentrations of GAD-ab are thought to be more frequently associate to a neurological condition, with levels 100-folds higher than those found in T1DM. Although CSF testing is recommended when suspecting a GAD-associated neurological syndrome, no commercial immunoassay is validated for this use and no cut-off is internationally recognized to support the diagnosis.
Methods
In this study we validated CSF testing of GAD-Ab on an automated chemiluminescence (CLIA) immunoassay that had previously shown good agreement with ELISA on serum.
Results
We tested 43 CSF from patients with typical GAD-associated neurological disorders and patients with other neurological conditions, identifying a clinical cut-off of 18 kIU/L that discriminated GAD-disease with an area under the curve (AUC) of 0.921. CLIA showed good analytical performances on repeatability and recovery tests in CSF and confirmed an excellent agreement with ELISA.
Conclusions
GAD-Ab associated neurological disorders are rare but CSF testing for GAD-Ab is a common request for neurologists when suspecting an insidious autoimmune central nervous system disease. CLIA platforms are expected to be increasingly adopted in clinical laboratories due to their flexibility and reliability, therefore studies on decisional levels should be implemented for improving the interpretation and utilization of laboratory data.
Introduction
Antibodies against glutamic acid decarboxylase isoform 65 (GAD-Ab) have been found in different severe neurological conditions associated to altered synthesis of γ-aminobutyric acid (GABA) from glutamate, where GAD is the rate-limiting enzyme [1]. GAD is an intracellular antigen found in the presynaptic neurons of central nervous system (CNS), β-cells of pancreatic islets, in the oviduct and the testes, and GAD-Ab can be found in up to 90 % of patients with type 1 diabetes mellitus (T1DM), mostly at relatively low concentrations [1, 2]. Neuronal disorders, now encompassed in a “GAD antibody-spectrum disorders” (GAD-SD) phenotype, particularly comprehend Stiff Person Syndrome (SPS) and its variants, including progressive encephalomyelitis with rigidity and myoclonus (PERM), cerebellar ataxia (CA), epilepsy and limbic encephalitis (LE) [3]. Classical SPS, first described in 1956 [4], is a rare disorder with a prevalence of 1–2 cases per million, more frequent in women and mostly develops in adults of 30–50 years of age [1, 2]. It has an insidious onset of axial muscle stiffness progressively involving proximal and distal limbs, resulting in abnormal postures, and superimposed painful spasms accompanied by dysautonomic and psychiatric symptoms [2]. CA is the second most frequent GAD-associated neurological disorder [5], characterized by gait and limb ataxia, dysarthria and nystagmus and also affecting women in 80 % of cases, usually with a previously diagnosed organ-specific autoimmune disease [1]. GAD-associated epilepsy affects most often the temporal lobe and shows pharmaco-resistance, and still is more common in women than in men [1], as it is LE, though the latter in younger age; GAD-LE can be either solely autoimmune and rarely is found as a paraneoplastic disease [2].
GAD-ab levels 100-folds higher than those found in T1DM are reported up to 80 % of SPS patients [6] and intrathecal synthesis in cerebrospinal fluid (CSF) is frequently described [7, 8] and support the correlation of autoantibodies with the neurological signs, though a direct pathogenic role is yet to be confirmed [1, 2].
Although high concentrations of GAD-ab are thought to be more frequently associated with a neurological condition [3], an international agreement on decisional level to support the diagnosis is still lacking [1].
Evidence for the presence of antibodies against GAD in both serum and CSF of an SPS patient was first described in 1988, thanks to a combined approach of light-microscopy immunocytochemistry and Western blot [9]. A radioimmunoassay was later validated reporting an SPS-specific cut-off for high level of GAD-abs in serum (>20 nmol/L) [10] and several enzyme-linked immunosorbent assays (ELISA) are now available [1]. Collaborative efforts for standardization and harmonization of nonradioactive tests have been recently depicted by The Islet Autoantibody Standardization Program (IASP) where ELISA assays showed the most homogeneous and accurate performances [11].
Commercial immunoassays commonly adopted in clinical laboratories were specifically developed for serum samples to support T1DM diagnosis [1], so that results above the upper limit of detection (e.g. 2,000 kIU/L for the most common ELISA kits) are usually not quantified.
Chemiluminescence immunoassays (CLIA) have been recently developed as immunoassays were the label is a luminescent molecule, having a wide dynamic range, high specificity, random access and being easily automated [12].
A CLIA assay for GAD-ab has recently proved good analytical performances compared to ELISA and has been validated for clinical practice in automated test systems [13], with linearity up to 280 kIU/L, for the intended endocrinological target.
Regardless of the peculiar technique used, the immunoassays have been mostly developed to quantify GAD-ab in serum, so that results in CSF so far have been interpreted with caution, although CSF testing is recommended when suspecting a neurological syndrome associated with either high serum level of GAD-ab [1] or low or seronegative results [3].
This study aims to validate CSF testing on a CLIA platform for neurological disorders and to determine a cut-off of GAD-ab in CSF for typical neurological GAD-related diseases.
Materials and methods
We collected 43 CSF samples from neurological patients previously classified at Neurology Unit O.S.A. of University-Hospital of Padova with GAD-related syndromes in preliminary work between 2009 and 2015 (Zoccarato M., unpublished results) and CSF leftovers samples from diagnostic routine workup between December 2019 and April 2022 at the Department of Laboratory Medicine from patients with other neurological conditions previously anonymized as control group; CSF was stored at −20 °C. Each patient had already undergone standard diagnostic procedures that also encompassed testing with commercial cell-based assay (CBA) and immunoblotting (Euroimmun, Germany) to exclude coexistence of other antibodies (LG1, CASPR2, GABABR, AMPAR, NMDAR and onconeural antibodies).
The occurrence of typical features of GAD-ab phenotypes [14], [15], [16] was reported. Patients were classified as “typical” when the clinical diagnosis was: (a) SPS (and variants) or (b) cerebellar ataxia or (c) encephalitis or (d) epilepsy, and if bearing clinical and paraclinical features recognized as common for immune-mediated diseases. This evaluation was blind to GAD-ab testing. Twelve patients had clinical features typically related to GAD-Ab (SPS, CA, temporal epilepsy, LE); 31 patients underwent lumbar puncture for the diagnostic workup of dementia, other phenotypes of epilepsy, infectious diseases, cerebrovascular disorders, headache, NMDAR encephalitis, oncologic diseases, and were classified as having “not-typical” GAD-Ab features (control group).
Samples were thawed at room temperature immediately before GAD-ab analysis. GAD-ab testing was performed on an automated CLIA system (MAGLUMI 2000 Plus by Snibe, China); this commercial sandwich immunoassay is validated for in vitro diagnostic use for serum only; manufacturer declares a measuring range of 1.0–280.0 kIU/L with a limit of blank of 1 kIU/L, it has a traceable calibrator for GAD65 (WHO 1st Reference Reagent 97/550).
CSF samples were also tested on an ELISA commercial kit (RSR Limited, UK or Euroimmun, Germany) with an automated system (DSX by Technogenetics, Italy or Triturus by Grifols, Spain) and results were compared to CLIA assay.
Precision evaluation
CSF samples with low GAD-ab concentration were pooled and divided into three aliquots, calibrator 1 (13.7 kIU/L) and calibrator 2 (169.3 kIU/L) were added to an aliquot each; precision was estimated with triple measurements of each aliquot for three consecutive days, following a modified Clinical and Laboratory Standards Institute (CLSI) EP15-A3 protocol [17], and compared to the manufacturer declared results appropriately interpolated using a linear function. Recovery of spiked calibrators was calculated as %Rec = Measured concentration/Expected value * 100.
Linearity assessment
Three samples (patient 10 measured GAD-ab level of 212.5 kIU/L, patient 8,88.8 kIU/L, patient 11,74.9 kIU/L) were diluted with kit buffer (phosphate buffered saline containing bovine serum and NaN3 <0.1 %, previously tested for GAD-ab); different linear dilutions were prepared with a fixed final volume and tested in duplicate, average values were compared to expected GAD-ab concentrations as explained in the CLSI EP06 A: 2003 guideline [18].
Statistical analyses
Median and interquartile ranges (25th–75th percentiles, IQR) were used as descriptive statistics, while the non-parametric Kruskall–Wallis test was used for comparing GAD-ab level between the two studied groups. Spearman rank correlation was used to test linear association between variables. Clinical threshold was calculated with receiver-operator characteristic (ROC) analysis and Youden index cutoff with confidence (CI) intervals 95 %. Proportional and constant bias between CLIA and ELISA were already previously estimated [13], therefore we evaluated agreement at manufacturers’ declared cut-offs for serum (17 kIU/L for MAGLUMI, 10 kIU/L for RSR/Euroimmun) with Cohen’s kappa. Analysis were performed using Stata v 16.1 (StataCorp, Lakeway Drive, TX, USA) and GraphPad Prism v 9.1.
Results
Comprehensive results of CSF CLIA and ELISA testing, demographic data and clinical diagnosis of the 43 patients are reported in Table 1. None of the patients with typical GAD-ab features had coexistent other neural antibody positivity; one of the patients in the control group had NMDAR-positive encephalitis.
Patient | Age, years | Sex | GAD CLIA, kIU/L | GAD ELISA, kIU/L | Clinical diagnosis | Group |
---|---|---|---|---|---|---|
1 | 51 | F | 26.9 | 2,708 | Encephalitis | GAD-Ab typical features |
2 | 32 | F | >280 | 55,698 | Encephalitis | |
3 | 65 | F | >280 | 16,432 | SPS | |
4 | 27 | F | 38.0 | 2,963 | Encephalitis | |
5 | 32 | F | >280 | 11,060 | Encephalitis | |
6 | 70 | M | 35.6 | 187 | PERM | |
7 | 44 | M | 5.7 | 0 | PERM | |
8 | 25 | F | 88.8 | 985 | Epilepsy | |
9 | 74 | F | >280 | 3,710 | Ataxia | |
10 | 57 | M | 212.5 | 1,293 | Epilepsy | |
11 | 22 | F | 74.9 | 190 | Ataxia | |
12 | 52 | F | 259.7 | n.a. | SPS | |
13 | 51 | F | > 280 | 10,933 | Dysarthria | NOT GAD-Ab typical features |
14 | 56 | M | 8.8 | 0 | Corticobasal degeneration | |
15 | 60 | M | 109 | 9,059 | Epilepsy | |
16 | 35 | M | 5.5 | 0 | Encephalitis | |
17 | 64 | M | 5.8 | 0 | Encephalitis | |
18 | 67 | F | 5.6 | 0 | Encephalitis | |
19 | 56 | M | 6.2 | 0 | Epilepsy | |
20 | 47 | M | 5.7 | 0 | Epilepsy in obstructive sleep Apnea syndrome | |
21 | 68 | F | 5.7 | 0 | Status epilepticus | |
22 | 36 | M | 6.1 | 0 | NMDAR encephalitis | |
23 | 31 | M | 5.8 | 0 | Suspected vasculitis/Paraneoplastic encephalitis | |
24 | 18 | F | 5.2 | 0 | Headache | |
25 | 70 | M | 3.5 | n.a. | Behavioural changes | |
26 | 74 | M | 5.3 | 0 | Confusional, fever | |
27 | 48 | M | 6.1 | 0 | Cerebellar syndrome w colon cancer | |
28 | 68 | M | 5.3 | 0 | Alzheimer’s disease | |
29 | 71 | F | 5.5 | 0 | Alzheimer’s disease | |
30 | 82 | M | 5.2 | 0 | Cerebrovascular syndrome | |
31 | 50 | M | 5.4 | 0 | Epilepsy | |
32 | 58 | M | 5.1 | 0 | Epilepsy in glioma | |
33 | 79 | M | 5.0 | 0 | Epilepsy in cerebrovascular syndrome | |
34 | 60 | F | 4.6 | 0 | Dementia | |
35 | 69 | M | 5.4 | 0 | Epilepsy in cerebrovascular syndrome | |
36 | 47 | F | 6.3 | 0 | Viral cerebellitis | |
37 | 50 | M | 5.8 | 0 | Talamic stroke | |
38 | 72 | M | 6.1 | 0 | Viral encephalitis | |
39 | 76 | F | 6.4 | 0 | Infectious radiculopathy | |
40 | 45 | M | 5.3 | 0 | Cerebral venous thrombosis | |
41 | 82 | F | 5.6 | 0 | Alzheimer’s disease | |
42 | 67 | M | 5.4 | 0 | Suspected neurodegenerative | |
43 | 49 | M | 5.1 | 0 | Epilepsy in glioma |
-
Positive GAD-Ab typical features, studied group; negative GAD-Ab typical features, control group. kIU, kilo International Units; SPS, Stiff Person Syndrome; PERM, progressive encephalomyelitis with rigidity and myoclonus; NMDAR, N-methyl-D-aspartate receptor.
GAD-ab concentrations were different between the group of patients with typical features and the control group with both assays. For CLIA median concentration was 150.7 kIU/L (36.8–280 IQR) vs. 5.6 kIU/L (5.3–6.1, χ2=18.651, p=0.0001); for ELISA the median was 2,708 kIU/L (190–11,060) vs. 0 (χ2=15.779, p=0.0001). Also, women had higher median concentrations both with CLIA (32.4 kIU/L vs. 5.7) and ELISA 190 kIU/L vs. 0. GAD-ab concentrations were not correlated to age (p= n.s. for both CLIA and ELISA).
Precision
Repeatability and intermediate precision of CLIA assay in a 3-days analysis are reported in Table 2, intra-assay and total CV% were both ≤11 %. Recovery of calibrator 1 ranged from 98 to 106 %, calibrator 2,124–128 %.
Design | Level, kIU/L | Measured repeatability CV% | Measured intermediate precision CV% | Manufacturer obtained repeatabilitya CV% | Manufacturer obtained intermediatea precision CV% |
---|---|---|---|---|---|
3 × 3 CLSI EP15-A3 | 5 | 7.41 | 10.95b | 5.33 | 8.26 |
9 | 3.38 | 3.38 | 5.18 | 7.86 | |
102 | 1.16 | 1.16 | 2.17 | 2.5 |
-
aObtained from the MAGLUMI™ GAD65 (CLIA) insert, N 069 GAD65-en-EU, V12.0, 2020-02; precision results were interpolated by using a linear function. CLSI, Clinical and Laboratory Standards Institute. bUpper verification limit 8.57. kIU, kilo International Units; CV, coefficient of variation.
Linearity
Dilution linearity for three samples is shown in Figure 1.
Kit buffer, phosphate-buffered saline (PBS) and a pool of GAD-Ab negative CSF samples were previously tested as dilution means and Passing–Bablok regression and Bland–Altman analysis showed no constant bias between the three specimens (data not shown).
Clinical cut-off
CLIA ROC curve for outcome (GAD-typical features: SPS, CA, Epilepsy and LE, PNS vs. non-GAD pathology) had AUC=0.921 (95 % CI: 0.832–1.000) (Figure 2). Youden’s cut-off identified was 18 kIU/L, sensitivity was 90.9 % (95 % CI: 58.7–99.8 %), specificity 93.5 % (95 % CI: 78.6–99.2 %); positive likelihood ratio 14.09 (95 % CI: 3.64–54.54), negative likelihood ratio 0.10 (95 % CI: 0.01–0.63).
ELISA ROC curve had AUC=0.909 (95 % CI 0.901 to 1.000) (Figure 2); at best cut-off of 93.5 kIU/L sensitivity was 90.9 % (95 % CI: 58.7 and 99.8 %) and specificity 93.3 % (77.9–99.2 %).
CLIA vs. ELISA
Agreement on positive or negative samples was excellent (100 %), with a Cohen’s kappa of 1.00, SE=0.156, p<0.001.
Discussion
In this study we validated a commercial CLIA assay for GAD-ab testing in CSF as, at our knowledge, no commercial immunoassay has been previously validated for this use, being usually designed to detect low concentrations of GAD-ab in serum of T1DM patients [19]. CLIA showed good performances on CSF with intra-assay and total CV% comparable to what manufacturer declares for serum, with only a minor caveat on concentration level 5 of kIU/L, well below the reference upper level of 17 kIU/L, where imprecision results were 7.41 and 10.95 % compared to manufacturer’s 5.33 and 8.26 %, respectively.
Additionally, the analytical recovery using the traditional spike addition test was acceptable (98–106 % at 13.7 kIU/L and 124–128 % at 169.3 kIU/L); based on these results, CSF seems a suitable matrix for this CLIA automated assay and GAD-ab requests might be implemented in routine testing.
However, data from the serial dilution test did not provide a satisfactory linearity, except for sample patient 10; therefore, sample dilution in routine practice should be attempted cautiously.
A recent comprehensive review on GAD-ab neurological disorders enclose CSF GAD testing in the algorithm to determine the likelihood of an autoimmune cause when GAD-ab are found in serum, recommending that CSF should be tested when high levels are detected in serum, unless other clear alternative diagnosis can be defined. Additionally, for patients with classic SPS, the detection of GAD antibodies in the CSF should suffice to establish a definite autoimmune diagnosis, while for the other phenotypes the finding of GAD-ab in the CSF should be accompanied by a demonstration intrathecal synthesis to establish a pathogenic link that is crucial for therapeutic management [1].
Some authors, however, suggest that CSF testing might not be necessary when GAD-abs concentration in serum is >10,000 kIU/L, while it could be useful at lower and/or negative concentrations [19]. The latter circumstances might benefit from the demonstration of intrathecal GAD-abs synthesis through a modified Link’s formula that has already been proposed [7]. However, even if intrathecal antibody synthesis is not demonstrated, the mere presence of anti-GAD in CSF allows the definition of a probable autoimmune disease [1].
While previous studies excluded an association between serum GAD-abs concentration and disease severity or degree of clinical improvement after therapy [19, 20], Munoz-Lopetegi identified “high level antibody concentrations” with a cut-off of 100 kIU/L in CSF and 10,000 kIU/L in serum with an ELISA method and immunohistochemistry and cell-based assay as confirmatory qualitative methods for GAD-ab-associated neurologic syndromes that might favorably respond to immunotherapy [21]. We confirmed this report in our cohort as we identified a CSF best cut-off of 93.5 kIU/L for ELISA for typical GAD-related clinical features.
The corresponding CLIA best cut-off was 18 kIU/L, consistent with the established serum value of 17 kIU/L.
Finally, we demonstrated that CLIA had an excellent agreement with ELISA (Cohen’s kappa=1) and ROC curve had similar AUC (0.921 vs 0.909), and, with the defined cut-offs, similar sensitivity and specificity, being the first report of harmonization and result comparability between two methods on CSF samples.
Although statistical power was reached, the relatively low number of subjects included is a limitation of this study; this is a common issue when addressing rare disorders that might be overcome in further multicentric evaluations.
A further limitation is that due to low sample volume CLIA results >280 kIU/L could not be diluted to fall within the measurement range.
Conclusions
Validation and standardization of GAD-ab testing for CSF are long-awaited goals for neurological clinical practice, due to the crucial impact of intrathecal synthesis on therapeutic management and long-term follow-up.
CLIA systems are expected to increase in clinical laboratories due to their performance advantages and their flexibility [22], agreement analysis and establishment of specific decisional levels should be pursued to facilitate clinical management particularly of complex patients, for whom a differential diagnosis could be challenging. The relative infrequency of these neurological syndromes might limit the statistical power of our results, so a more extensive application of a cut-off for CSF testing should prove its reliability.
-
Research funding: None declared.
-
Author contributions: All authors have accepted responsibility for the entire content of this manuscript and approved its submission.
-
Competing interests: Authors state no conflict of interest.
-
Informed consent: Not applicable.
-
Ethical approval: Not applicable.
References
1. Graus, F, Saiz, A, Dalmau, J. GAD antibodies in neurological disorders – insights and challenges. Nat Rev Neurol 2020;16:353–65. https://doi.org/10.1038/s41582-020-0359-x.Search in Google Scholar PubMed
2. Baizabal-Carvallo, JF. The neurological syndromes associated with glutamic acid decarboxylaseantibodies. J Autoimmun 2019;101:35–47. https://doi.org/10.1016/j.jaut.2019.04.007.Search in Google Scholar PubMed
3. Dalakas, M. Stiff-person syndrome and GAD antibody-spectrum disorders: GABAergic neuronal excitability, immunopathogenesis and update on antibody therapies. Neurotherapeutics 2022;19:832–47. https://doi.org/10.1007/s13311-022-01188-w.Search in Google Scholar PubMed PubMed Central
4. Moersch, FP, Woltman, H. Progressive fluctuating muscular rigidity and spasm (stiff-man syndrome): report of a case and some observations in 13 other cases. Mayo Clin Proc 1956;31:421–7.Search in Google Scholar
5. Budhram, A, Sechi, E, Flanagan, EP, Dubey, D, Zekeridou, A, Shah, SS, et al.. Clinical spectrum of high-titre GAD65 antibodies. J Neurol Neurosurg Psychiatry 2021;92:645–54. https://doi.org/10.1136/jnnp-2020-325275.Search in Google Scholar PubMed PubMed Central
6. Saiz, A, Blanco, Y, Sabater, L, Gonzalez, F, Bataller, L, Casamitjana, R, et al.. Spectrum of neurological syndromes associated with glutamic acid decarboxylase antibodies: diagnostic clues for this association. Brain 2008;131:2553–63. https://doi.org/10.1093/brain/awn183.Search in Google Scholar PubMed
7. Dalakas, M, Li, M, Fujii, M, Jacobowitz, D. Stiff person syndrome: quantification, specificity, and intrathecal synthesis of GAD65 antibodies. Neurology 2001;57:780–4. https://doi.org/10.1212/wnl.57.5.780.Search in Google Scholar PubMed
8. Gresa-Arribas, N, Ariño, H, Martínez-Hernández, E, Petit-Pedrol, M, Sabater, L, Saiz, A, et al.. Antibodies to inhibitory synaptic proteins in neurological syndromes associated with glutamic acid decarboxylase autoimmunity. PLoS One 2015;10:e0121364. https://doi.org/10.1371/journal.pone.0121364.Search in Google Scholar PubMed PubMed Central
9. Solimena, M, Folli, F, Denis-Donini, S, Comi, G, Pozza, G, De Camilli, P, et al.. Autoantibodies to glutamic acid decarboxylase in a patient with stiff-man syndrome, epilepsy, and type I diabetes mellitus. N Engl J Med 1988;318:1012–20. https://doi.org/10.1056/nejm198804213181602.Search in Google Scholar PubMed
10. Walikonis, J, Lennon, V. Radioimmunoassay for glutamic acid decarboxylase (GAD65) autoantibodies as a diagnostic aid for stiff-man syndrome and a correlate of susceptibility to type 1 diabetes mellitus. Mayo Clin Proc 1998;73:1161–6. https://doi.org/10.4065/73.12.1161.Search in Google Scholar PubMed
11. Lampasona, V, Pittman, DL, Williams, AJ, Achenbach, P, Schlosser, M, Akolkar, B, et al.. Islet autoantibody standardization Program 2018 workshop: interlaboratory comparison of glutamic acid decarboxylase autoantibody assay performance. Clin Chem 2019;65:1141–52. https://doi.org/10.1373/clinchem.2019.304196.Search in Google Scholar PubMed PubMed Central
12. Cinquanta, L, Fontana, DE, Bizzaro, N. Chemiluminescent immunoassay technology: what does it change in autoantibody detection? Auto Immun Highlights 2017;8:9. https://doi.org/10.1007/s13317-017-0097-2.Search in Google Scholar PubMed PubMed Central
13. Cosma, C, Padoan, A, Plebani, M. Evaluation of precision, comparability and linearity of MAGLUMI TM 2000 Plus GAD65 antibody assay. J Lab Precis Med 2019;4:31. https://doi.org/10.21037/jlpm.2019.09.03.Search in Google Scholar
14. Dalakas, M, Fujii, M, Li, M, McElroy, B. The clinical spectrum of anti-GAD antibody-positive patients with stiff-person syndrome. Neurology 2000;55:1531–5. https://doi.org/10.1212/wnl.55.10.1531.Search in Google Scholar PubMed
15. Baizabal-Carvallo, JF, Alonso-Juarez, M. Cerebellar disease associated with anti-glutamic acid decarboxylase antibodies: review. J Neural Transm 2017;124:1171–82. https://doi.org/10.1007/s00702-017-1754-3.Search in Google Scholar PubMed
16. Graus, F, Titulaer, MJ, Balu, R, Benseler, S, Bien, CG, Cellucci, T, et al.. A clinical approach to diagnosis of autoimmune encephalitis. Lancet Neurol 2016;15:391–404. https://doi.org/10.1016/s1474-4422(15)00401-9.Search in Google Scholar
17. C.L.S.I. User verification of precision and estimation of bias; approved guideline—third edition. CLSI document EP15-A3. Wayne, PA, USA: Clinical and Laboratory Standards Institute; 2014.Search in Google Scholar
18. C.L.S.I. Evaluation of the linearity of quantitative measurement procedures: a statistical approach. CLSI EP06-A, 1st ed. Wayne, PA, USA: Clinical and Laboratory Standards Institute; 2003.Search in Google Scholar
19. Tsiortou, P, Alexopoulos, H, Dalakas, M. GAD antibody-spectrum disorders: progress in clinical phenotypes, immunopathogenesis and therapeutic interventions. Ther Adv Neurol Disord 2021;14:1–17. https://doi.org/10.1177/17562864211003486.Search in Google Scholar PubMed PubMed Central
20. Dalakas, M, Fujii, M, Li, M, Lutfi, B, Kyhos, J, McElroy, B. High-dose intravenous immune globulin for stiff-person syndrome. N Engl J Med 2001;345:1870–6. https://doi.org/10.1056/nejmoa01167.Search in Google Scholar PubMed
21. Muñoz-Lopetegi, A, de Bruijn, MA, Boukhrissi, S, Bastiaansen, AE, Nagtzaam, MM, Hulsenboom, ES, et al.. Neurologic syndromes related to anti-GAD65: clinical and serologic response to treatment. Neurol Neuroimmunol Neuroinflamm 2020;7:e696. https://doi.org/10.1212/nxi.0000000000000696.Search in Google Scholar
22. Bizzaro, N, Tozzoli, R, Villalta, D. Autoimmune diagnostics: the technology, the strategy and the clinical governance. Immunol Res 2015;61:126–34. https://doi.org/10.1007/s12026-014-8587-z.Search in Google Scholar PubMed
© 2023 the author(s), published by De Gruyter, Berlin/Boston
This work is licensed under the Creative Commons Attribution 4.0 International License.