Many biochemical auto-analyzers have methods that measure the hemolysis index (HI) to quantitatively assess the degree of hemolysis. Past reports on HI are mostly in vitro studies. Therefore, we evaluated the optimal wavelength of HI measurement ex vivo using clinical samples. Four different wavelengths (410/451 nm: HI-1, 451/478 nm: HI-2, 545/596 nm: HI-3 and 571/596 nm: HI-4) were selected for HI measurement, and correlations were examined from the measurement results of 3890 clinical samples. Another set of 9446 clinical samples was used to examine the correlation of HI with lactate dehydrogenase (LDH), aspartate aminotransferase (AST) and potassium (K). Strong correlations were found between HI-4 and HI-1 and between HI-4 and HI-3. HI-1 and HI-2 cannot correctly assess hemolysis for high bilirubin samples, and HI-3 cannot correctly assess hemolysis for high triglyceride samples. LDH, AST and K correlated positively with HI-4 in clinical samples. For every 1-unit increase in HI-4, LDH increased by 19.51 U/L, AST by 1.03 U/L and K by 0.061 mmol/L, comparable to reports of other studies. In clinical samples, HI-4 was less susceptible to bilirubin and chyle and reflected well the changes in LDH, AST and K caused by hemolysis. This suggested that the optimal wavelength for HI measurement is 571 nm.

Aim: With the increasing incidence of nonviral hepatocellular carcinoma (HCC), current surveillance markers such as alpha‐fetoprotein (AFP) and des‐gamma carboxyprothrombin (DCP) show limitations. We investigated the percentage of glycosylated ferritin relative to total ferritin (%GF) as a novel diagnostic biomarker for HCC. Methods: Samples from 660 patients (chronic liver disease [CLD], n = 278 and HCC, n = 382) were divided into training and validation cohorts (7:3). Glycosylated ferritin was measured using concanavalin A, and %GF was calculated as 100 × (glycosylated ferritin/total ferritin). A multivariate logistic regression analysis was carried out in the training cohort and validated in the validation cohort to improve diagnostic accuracy. Results: In all HCC stages, %GF levels were significantly lower than CLD. The sensitivity and specificity of %GF were 73.7% and 70.8%, respectively, in the training cohort, and 74.1% and 69.8%, respectively, in the validation cohort. The area under the curve (AUC) values of 0.782 (training) and 0.800 (validation) were comparable to those of AFP and DCP. We developed the GFAD index, integrating age, sex, %GF, AFP, and DCP. The GFAD index showed superior diagnostic performance with AUCs of 0.939 (training) and 0.923 (validation). The sensitivity and specificity were 81.9% and 92.7%, respectively (training), and 78.6% and 87.2%, respectively (validation). For early stage HCC, the sensitivity was 67.0% (training) and 69.2% (validation), regardless of the etiology of HCC. Conclusions: The diagnostic performance of %GF for HCC was comparable to AFP and DCP. Moreover, the GFAD index showed excellent performance, even in early stage HCC, regardless of etiology.