Quantitative gated SPECT-derived phase analysis on gated myocardial perfusion SPECT detects left ventricular dyssynchrony and predicts response to cardiac resynchronization therapy

Abstract

The significance of left ventricular (LV) dyssynchrony for the prediction of response to cardiac resynchronization therapy (CRT) has been demonstrated. Parameters reflecting LV dyssynchrony (phase SD, histogram bandwidth) can be derived from gated myocardial perfusion SPECT (GMPS) using phase analysis. The feasibility of LV dyssynchrony assessment with phase analysis on GMPS using Quantitative Gated SPECT (QGS) software has not been demonstrated in patients undergoing CRT. The aim of the present study was to validate the QGS algorithm for phase analysis on GMPS in a direct comparison with echocardiography using tissue Doppler imaging (TDI) for LV dyssynchrony assessment. Also, prediction of response to CRT using GMPS and phase analysis was evaluated.

Methods: Patients (n = 40) with severe heart failure (New York Heart Association class III-IV), an LV ejection fraction of no more than 35%, and a QRS complex greater than or equal to 120 ms were evaluated for LV dyssynchrony using GMPS and echocardiography with TDI. At baseline and after 6 mo of CRT, clinical status, LV volumes, and LV ejection fraction were evaluated. Patients with functional improvement were classified as CRT responders.

Results: Both histogram bandwidth (r = 0.69, r(2) = 0.48, SEE = 25.4, P < 0.01) and phase SD (r = 0.65, r(2) = 0.42, SEE = 26.8, P < 0.01) derived from GMPS correlated significantly with TDI for assessment of LV dyssynchrony. At baseline, CRT responders showed a significantly larger histogram bandwidth (94 degrees +/- 23 degrees vs. 68 degrees +/- 21 degrees , P < 0.01) and a larger phase SD (26 degrees +/- 6 degrees vs. 18 degrees +/- 5 degrees , P < 0.01) than did nonresponders. Receiver-operating-characteristic curve analysis identified an optimal cutoff value of 72.5 degrees for histogram bandwidth to predict CRT response, yielding a sensitivity of 83% and a specificity of 81%. For phase SD, sensitivity and specificity similar to those for histogram bandwidth were obtained at a cutoff value of 19.6 degrees .

Conclusion: QGS phase analysis on GMPS correlated significantly with TDI for the assessment of LV dyssynchrony. Moreover, a high accuracy for prediction of response to CRT was obtained using either histogram bandwidth or phase SD.