Measurement of methemoglobin in neonatal samples containing fetal hemoglobin

Because of the potential for methemoglobinemia during nitric oxide therapy in newborns, methods are needed to accurately quantify methemoglobin (MetHb) in the presence of the high concentrations of fetal hemoglobin (Hb F), bilirubin, and lipids seen in these patients. Spectral differences between fetal and adult Hbs invalidate assumptions of conventional multiwavelength Hb photometry, so we evaluated an "overdetermined" system (Ciba-Corning Model 270), in which absorbances at seven wavelengths are measured to quantify four Hb derivatives. Adult and umbilical cord blood (Hb F 96%) samples were prepared to contain known MetHb fractions. Measured MetHb was linear in cord blood to > or = 15% MetHb. Within-run precision (CV) was < 2.2% (n = 10) at each of seven MetHb fractions between 5% and 100%. Measured (y) and expected (x) MetHb fractions in cord blood were in good agreement (y = 1.0200x + 0.100, Sylx = 0). Added bilibrubin (200 mg/L serum) and lipid (30 g/L) did not interfere. No significant differences were seen for adult and cord blood samples with identical MetHb fractions (P = 0.72), whereas a significant difference was noted with an exactly determined system (P = 0.0033). At clinically relevant MetHb fractions (< 15%), a trend towards increased values in cord blood was noted with an exactly determined system (y = 1.0520x + 0.7600). We conclude that this overdetermined system measures MetHb accurately in samples from patients with large concentrations of Hb F.