Relative error and variability in blood flow measurements with radiolabeled microspheres.

The validity of blood flow measurements made using radiolabeled microspheres (reference sample technique) has been well established in a variety of experimental models. Previous studies of the variability of the method emphasize tissue sphere number as the principle source of random variation. However, blood flow measurements depend on sphere distribution to both tissue and reference samples as well as on quantitation of different isotopes in a single tissue sample. In this study we examined the major determinants of relative error and variability in flow rate measurements inherent in the reference sample technique. Error magnitude was predicted from a statistical model, simulated with numerical analogs, and measured in anesthetized dogs with 15-micrometers microspheres. Theoretically, with 2,000 spheres in the reference sample and greater than or equal to 475 spheres in a tissue sample, blood flow can be measured with 10% accuracy at the 95% confidence level and with duplicate variability of 14%. Random errors in isotope quantitation were influenced by the specific activity of the isotope and the fractional distribution of a given isotope over the various energy windows. Suspension of tissue and reference sample spheres at different heights (0.2 cm) resulted in systematic flow errors of up to 13%. When errors due to separation of isotopes and differences in sample height were minimized, the major determinant of variability in flow measurements during simultaneous injection of differently labeled microspheres was the number of spheres in both tissue and reference samples.