LAUSR

Neuron specific enolase (NSE) is a serum soluble tumor marker for certain neuroendocrine tumors and a marker of some types of brain trauma. Red blood cells (RBC) also contain substantial NSE concentrations, making even slight hemolysis a source of potential false positives and reason for sample rejection. Unfortunately, hemolysis can be common in specimens from patients with in vivo hemolysis due to trauma or medical intervention, or in vitro hemolysis due to difficult draws (e.g., pediatrics). While laboratories are appropriately reluctant to report false elevations of NSE, specimen rejection due to any evidence of hemolysis underserves these patients. Correction of NSE results in hemolyzed serum was previously reported as a possible solution; however, broad inter-individual differences in RBC NSE concentrations require measuring NSE in RBCs from a whole blood specimen submitted concurrently with a hemolyzed serum sample. Unfortunately, logistical challenges impede the effective implementation of this strategy. Our objective was to establish a reference distribution of NSE concentrations in RBCs (as a ratio of NSE/Hgb (ng/mg)) that would facilitate the derivation of a cutoff to distinguish true NSE elevations from those concentrations arising from in vivo or in vitro hemolysis. Serum NSE was measured by the Fujirebio NSE enzyme immunoassay kit, and H-index was measured using the Roche cobas c501 (previously shown to demonstrate good agreement with spectrophotometrically measured hemoglobin, Hgb). Purposely hemolyzed samples (n=80) were diluted to an H-index of ~100, and NSE was measured to establish the distribution of NSE/Hgb ratios. De-identified NSE results (n=5,441) previously analyzed in our laboratory were extracted from our data warehouse to determine the distribution of NSE values in non-hemolyzed clinical samples. R statistical software was used for linear regressions, power transform analysis, and subsequent inverse square root transformation to enable appropriate normalization of each distribution for curve fitting purposes. Median NSE/Hgb ratios in the hemolyzed and non-hemolyzed distributions were 21 and 125, respectively. Curve fitting of the transformed hemolyzed distribution provides an upper 99% value of 46 NSE/Hgb. To determine whether that cut-off is appropriate at different levels of hemolysis, NSE was measured in samples (n=60) from subjects (n=15) prepared at four levels of hemolysis (H-index ~25 to 400). The mean NSE/Hgb ratio did not change significantly across the range of H-index values: regression analysis documented a mean of 23 NSE/Hgb and an upper 99% prediction value of 41 NSE/Hgb. Additionally, the increase in NSE was related to the H-index as follows, NSE=-715.3 + 21.42*H-index. The slope (representing NSE/Hgb) was again in agreement with the mean ratio of 21 NSE/Hgb observed in the initial 80 hemolyzed samples. Overall, the derived reference distribution of NSE/Hgb facilitates the interpretation of clinically relevant NSE increases from elevated NSE concentrations due to in vivo or in vitro hemolysis.