Specificity and sensitivity of noninvasive measurement of pulmonary vascular protein leak

Noninvasive techniques employing external counting of radiolabeled protein have the potential for measuring pulmonary vascular protein permeability, but their specificity and sensitivity remain unclear. The authors tested the specificity and sensitivity of a double-radioisotope method by injecting radiolabeled albumin (131I) and erythrocytes (/sup 99m/Tc) into anesthetized dogs and measuring the counts of each isotope for 150 min after injection with an external gamma probe fixed over the lung. They calculated the rate of increase of albumin counts measured by the probe (which reflects the rate at which protein leaks into the extravascular space). To assess permeability the authors normalized the rate of increase in albumin counts for changes in labeled erythrocyte signal to minimize influence of changes in vascular surface area and thus derived an albumin leak index. They measured the albumin leak index and gravimetric lung water during hydrostatic edema (acutely elevating left atrial pressure by left atrial balloon inflation: mean pulmonary arterial wedge pressure = 22.6 Torr) and in lung injury edema induced by high- (1.0 g/kg) and low-dose (0.25 g/kg) intravenous thiourea. To test specificity hydrostatic and high-dose thiourea edema were compared. The albumin leak index increased nearly fourfold from control after thiourea injury (27.2 +/- 2.3 x 10-4 vs. 7.6 +/- 0.9 x 10-4 min-1) but did not change from control levels after elevating left atrial pressure (8.9 +/- 1.2 x 10-4 min-1) despite comparable increases in gravimetric lung water. To test sensitivity the authors compared low-dose thiourea with controls. Following low-dose thiourea, the albumin leak index nearly doubled despite the absence of a measurable increase in lung water