LAUSR

To the Editor: We have read the article written by Uphaus et al with great interest and appreciation. This study revealed sNFL (serum neuron–specific neurofilament light chain) is a good biomarker for functional independence 90 days after acute ischemic stroke (AIS) and predicts long-term cardiovascular outcome. However, results of the study should be interpreted with the following considerations. First, therapy was a critical factor for patients with AIS. This study did show the details of therapy of AIS in hospital and out of hospital, including the time from onset to admission, thrombolysis, thrombectomy, stent implantation, statins, antiplatelet, or anticoagulation, which were important for sNFL to predict shortterm functional outcome and long-term cardiovascular outcome. Second, in this study, sNfL was measured 24 hours and 12 months after AIS; the authors could not get the time dynamic curve of this biomarker after AIS, especially within 12 hours. Within 6 hours after AIS, there may be an effect on the expression of sNfL after receiving thrombolysis, thrombectomy, and stent implantation. Therefore, it was necessary to measure the levels of sNfL at different time points during early phase of AIS or conducted a subgroup analysis. Third, the result showed the power of sNfL identifying highrisk patient with long-term outcome was superior over biomarkers, such as NT pro-BNP (N-terminal pro-B-type natriuretic peptide), ANP (atrial natriuretic peptide), and FABP (fatty acidbinding protein), which were previously used for this purpose. However, some inflammatory biomarkers, such as neutrophilto-lymphocyte ratio, LP-PLA2 (lipoprotein-associated phospholipase A2), and CRP (C-reactive protein), also had good predictive value for outcomes in patients with AIS, thus these conventional biomarkers should be measured in this study, and conducted a comparison with sNfL. Fourth, the National Institutes Health Stroke Scale score was a conventional tool to stratify high-risk patients with AIS. In this study, the area under curve for sNfL predicting cardiovascular end point was 0.729, similar to that of National Institutes Health Stroke Scale. Considering this, why the authors do not just use the much cheaper conventional National Institutes Health Stroke Scale, which seems just as good? In addition, the additional use of sNfL resulted in an improvement of the area under curve for the combination of the clinical variables (diabetes mellitus, National Institutes Health Stroke Scale at admission, and age-related white matter changes rating) from 0.792 to 0.820 (P=0.052), no difference was found between these 2 area under curves. Therefore, these disadvantages of sNfL would restrict its clinical practice in predicting outcome in patients with AIS.