Interference and correction of slight hemolysis in the Roche electrochemiluminescence neuron-specific enolase...

Objective: neuron-specific enolase (NSE) is a key biomarker for neuroendocrine tumors and neurological injuries. However, hemolysis, even at slight levels, compromises the accuracy of NSE measurement. This study aimed to quantitatively evaluate the dose-dependent interference of slight hemolysis on serum NSE measurement and establish a mathematical correction model. Methods: thirty-four EDTA blood samples and their paired serum samples were analyzed in this study. First, hematocrit and hemoglobin levels were measured in the whole blood samples. Washed red blood cells (RBCs) were then completely lysed via repeated freeze-thaw cycles, and the resulting lysate was added to the matched serum at volume ratios ranging from 0.01% to 0.11%. NSE and hemolytic index (HI) were measured using a Roche cobas e801 and a cobas e702 analyzer. Next, we analyzed the correlation between NSE deviation and HI. Finally, a linear regression-based correction model was developed, and the model was evaluated via bland-Altman analysis and recovery rate experiments. Results: NSE interference increased in a significant, dose-dependent manner with hemolysis level (p < 0.05). Strong linear correlations were observed between NSE interference and hemolysis level (Y = 106.1X + 0.0049, R² = 0.9952), and between HI and hemolysis level (Y = 343.1X + 3.022, R² = 0.9960). This interference severely degraded clinical agreement, reducing Cohen's kappa from 1.00 (baseline) to 0.15 at 0.11% hemolysis. A correction model was established: corrected NSE = measured NSE - (0.3090 × HI - 0.9262). We evaluated the performance of this model using bland-Altman analysis, which yielded a recovery rate ranging from 85.61% to 111.75%, and the corrected NSE values remained stable across all tested concentrations. Seven-point within-specimen analysis showed inter-individual dispersion of ΔNSE/ΔHI slopes (CV 28.2%), yet population correction retained acceptable accuracy (1.75 µg L^-1 residual error) under the slight hemolysis conditions (HI ≤ 41). Conclusion: slight hemolysis causes significant, quantifiable interference in NSE immunoassays. The proposed mathematical correction model substantially reduces bias in the derivation cohort, and may offer a simple and practical solution for accurate NSE measurement in slightly hemolyzed samples, though it still awaits prospective clinical validation.