Signal-averaged electrocardiography and detection of heart transplant rejection: comparison of time and frequency domain analyses.

To evaluate the role of the signal-averaged electrocardiogram (ECG) in the detection of heart transplant rejection, findings on 277 ECGs were compared with those in 218 endomyocardial biopsy specimens in 25 patients followed up for a median duration of 5.2 months (range 7 days to 17.5 months). Signal-averaged ECGs obtained at intervals of 16.4 +/- 22.3 days were analyzed in the time domain before and after high pass filtering at 25 and 70 Hz. Frequency domain analysis was performed with use of a fast Fourier transform algorithm. Sixteen severe rejection episodes requiring treatment were observed. These episodes induced significant decreases in peak and root-mean-square voltages of both filtered and unfiltered QRS complexes, as well as in the total spectral area. Conversely, QRS duration and 50- to 250-Hz or 70- to 110-Hz spectral areas were not significantly altered. In 14 cases mild rejection episodes were observed that did not significantly alter any of the variables studied. The root-mean-square voltage of the 70-Hz high pass filtered QRS complex was found to be the most accurate variable in detecting rejection. Moreover, this variable was also the most reproducible in 10 healthy control subjects. The optimal rejection criterion was defined as an 11% decrease in voltage between two consecutive recordings. It provided 87.5% sensitivity with 78.4% specificity. In conclusion, the signal-averaged ECG is helpful in the management of heart transplant rejection. Frequency domain analysis of the QRS complex does not increase the accuracy of the technique compared with the time domain approach.