Comparison of two ELISA methods for measuring levels of the phosphorylated neurofilament heavy chain.

Recent studies suggest that the quantification of neurofilament subunits in cerebrospinal fluid (CSF), blood and amniotic fluid may reflect neuroaxonal damage and be of high clinical value. The present study aims to cross-validate two different independently developed ELISA techniques for the quantification of the phosphorylated axonal forms of the neurofilament heavy chain (pNfH). The London ELISA method is based on barbitone buffer and the commercially available SMI35 capture antibody. The Gainesville method uses Tris-buffered saline (TBS) and an affinity purified chicken polyclonal capture antibody (C-pNfH). Coded CSF from 50 patients with neurological diseases were analyzed in duplicate by both laboratories, each using both ELISA methods, but with each lab using their own detection antibody, tertiary antibody and chromogen. Methods were compared using Bland-Altman plots. Correlation and regression analyses were used to allow for transformation of values between both methods. The Bland-Altman plots demonstrated that 96% of all samples fell into the narrow 95% limits of agreement (0.04 units of OD). There was a high correlation (Spearman R=0.92, p<0.0001 and Pearson R=0.98, p<0.0001) between the Gainesville (Y) and the London (X) method with Y=0.132+1.104(*)X. The previously determined upper reference limit of 0.73 mg/l (London method) corresponds to 0.94 mg/l for the Gainesville method. CSF pNfH levels above the reference limit were observed in patients with encephalitis, encephalomyelitis, hydrocephalus, subarachnoid haemorrhage, spino-muscular atrophy, stroke and cancer with both methods agreeing in all cases. The two assays are in excellent agreement, suggesting that pNfH, which has a number of unusual protein chemical features, may be the biomarker of choice for the routine and robust detection of axonal injury and degeneration in both research and clinical contexts.