Background: Gated single-photon emission computed tomography (SPECT) is widely used for myocardial perfusion imaging and provides an automated assessment of left ventricular ejection fraction (LVEF). We prospectively tested the repeatability of serial SPECT-derived LVEF. This information is essential in order to inform the interpretation of a change in LV function on serial testing.
Methods: Consenting patients (n = 50) from among those referred for clinically indicated gated myocardial perfusion SPECT (MPs) were recruited. Following the clinical rest-stress study, patients were repositioned on the camera table for a second acquisition using identical parameters. Patient positioning, image acquisition and processing for the second scan were independently performed by a technologist blinded to the clinical scan. Quantitative LVEF was generated by Quantitative Gated SPECT and recorded as EF1 and EF2, respectively. Repeatability of serial results was assessed using the Bland-Altman method. The limits of repeatability and repeatability coefficients were generated to determine the maximum variation in LVEF that can be expected to result from test variability. Repeatability was tested across a broad range of LV systolic function and myocardial perfusion.
Results: The mean difference between EF1 and EF2 was 1.6% (EF units), with 95% limits of repeatability of +9.1% to -6.0% (repeatability coefficient 7.5%). Correlation between serial EF measurements was excellent (r = 0.9809). Similar results were obtained in subgroups based on normal or abnormal EF and myocardial perfusion. The largest repeatability coefficient of 8.1% was seen in patients with abnormal LV systolic function.
Conclusion: When test protocol and acquisition parameters are kept constant, a difference of >8% EF units on serial MPs is indicative of a true change 95% of the time.
Keywords: Gated-SPECT; LV function; ejection fraction; repeatability.