Diagnostic accuracy of indium-111 platelet scintigraphy in identifying left ventricular thrombi.

This study defines the optimum imaging time window after injection of labeled platelet suspension for detection on left ventricular (LV) thrombi, identifies the most useful imaging views, and determines the reproducibility of this technique. A total of 662 images obtained from 64 patients were analyzed retrospectively on 2 separate occasions by 3 observers blinded as to patient identity, view (right anterior oblique, anterior, left anterior oblique, and left lateral), and time after injection of the platelet suspension that the images were obtained (0 to 2, 3 to 4, and 5 to 6 days). Images were categorized as either positive or negative. In every case surgical or autopsy verification of the presence or absence of LV thrombus was possible. The best combination of sensitivity, specificity, and diagnostic accuracy was found in the 3- to 4-day period in the left anterior oblique view and was 54 +/- 5% (mean +/- standard deviation), 98 +/- 1%, and 85 +/- 2%, respectively. Sensitivity, specificity, and diagnostic accuracy were not enhanced by adding additional views (right anterior oblique, left lateral, and anterior) to the left anterior oblique view in the 3- to 4-day time period. However, using multiple views, localization of thrombi to the left ventricle was facilitated. In a second retrospective analysis, a comparison of day 0 with day 3 to 4 images enhanced sensitivity and accuracy to 65 (p less than 0.001) and 90% (not significant), respectively. Specificity was unchanged at 99%. Mean intra- and interobserver agreement was 91 and 88%, respectively. Thus, (1) indium-111 platelet scintigraphy is a reproducible and specific technique for identifying LV thrombus, and (2) we advise imaging on day 0 and again 3 to 4 days after injection of the platelet suspension in right anterior oblique, left anterior oblique, left lateral, and anterior views to maximize accuracy and to facilitate localization of LV thrombus.

Influence of attenuation on radionuclide stroke volume determinations.

Using a method for determination of absolute volumes, including correcting for attenuation, we have explored the ability of the method to determine stroke volume in humans by radionuclide techniques. Thermodilution cardiac output determinations and multigated equilibrium blood-pool scintigraphy in the LAO view were performed simultaneously in twenty patients in which no evidence of intracardiac shunts or valvular disease was present. The correlation was good between the attenuated radionuclide and thermodilution stroke volume (r = 0.80, s.e.e. of estimate = 12 ml; SVtd = 2.31 x SVr + 18 ml). When correction for attenuation was made, the correlation improved (r = 0.96, s.e.e. = 6 ml) and approached the line of identity (SVtd = 0.99 x SVr + 1.2 ml). The correlation was also good between radionuclide cardiac output, corrected for attenuation, and the thermodilution cardiac output (r = 0.89, s.e.e. = 0.36 l/min; COtd = 0.86 x COr + 0.67 l/min). Thus our method of correction for attenuation in the determination of absolute left-ventricular volumes has been shown to provide a reliable, noninvasive means of calculating stroke volume and cardiac output in humans, without the use of geometric assumptions or regression equations.