DIAGNOSTIC METHODS NONINVASIVE GRAPHIC METHODS Detection of aortic porcine valve dysfunction by maximum entropy spectral analysis

A high-resolution method of spectral analysis, of the class generally called "maximum entropy method," was used in a study of aortic porcine valve closing sounds in 37 patients (ages 19 to 76). Spectra from 27 normal xenografts, implanted from 2 weeks to 61 months previously, were characterized by a dominant frequency peak, Fl, at 89 + 15 Hz (mean + SD), with a lower amplitude peak, F2, at 154 + 25 Hz. Eight of nine patients with aortic porcine valve dysfunction were proved surgically to have leaflet degeneration or infection and had either F1 (139 ± 54 Hz) and/or F2 (195 + 74 Hz) significantly higher than normal (p < .001). In two patients with paravalvar leak but no leaflet abnormality, F, and F2 were in the normal range. Estimation of F1 and F2 was highly reproducible and was unaffected by duration of implant up to 5 years. Spectral analysis of aortic porcine valve closing sounds by the maximum entropy method may be useful for detection of intrinsic xenograft dysfunction. Circulation 68, No. 1, 42-49, 1983. THE USE OF glutaraldehyde-preserved porcine xenografts for heart valve replacement has become widely accepted over the past decade. ' However, in common with all prosthetic valve types, they remain subject to infection and thrombosis.2-5 In addition, studies of gross and histologic anatomy have shown that porcine valve leaflets often undergo progressive ultrastructural changes that result in increasing stiffening with time.6' 7 Further degeneration may be accompanied by calcification and valve stenosis or valvar regurgitation from leaflet retraction or fracture.8-10 In patients who deteriorate clinically after heart valve replacement, it may be difficult to differentiate intrinsic valve dysfunction from left ventricular failure caused by myocardial disease and to identify valve infection when present. Although clinical evaluation will often establish the correct diagnosis, cardiac catheterization and angiography may be necessary in patients with suspected abnormal prosthetic valve function. These procedures are not without risk. Phonocardiographic, radiographic, radionuclide, From the Department of Medicine, Cardiac Unit, Massachusetts General Hospital, Boston, and the Massachusetts Institute of Technology, Cambridge. Supported by grants from the Ambrose Monell Foundation and American Edwards Laboratories. Address for correspondence: Robert S. Lees, M.D., New England Deaconess Hospital, 185 Pilgrim Rd., Boston, MA 02215. Received July 20, 1982; revision accepted March 24, 1983. 42 and ultrasound techniques have been used to diagnose mechanical and bioprosthetic valve dysfunction."-20 Recently, spectral analysis has been applied to valve sounds,2'-25 including the closing sounds of aortic porcine xenografts.24 In one of these studies, relatively high frequency content was found in the frequency spectra derived from two aortic porcine valves with abnormal function, which had been in place for 5 and 7 years; these spectral alterations were the same as those obtained from normally functioning aortic porcine valves implanted for a similar period.24 However, this study was based on the fast Fourier transform (FFT) method of spectral analysis. FFT is inherently limited by the uncertainty principle, which states in this case that frequency resolution is directly proportional to the duration of the signal. With most phonocardiographic techniques, aortic porcine valve closing sounds are estimated to be 15 to 25 msec in duration, which thereby imposes a 40 to 70 Hz resolution limit on FFTbased spectral analysis.26 A radically different method of spectral analysis is the autoregressive method, one of several closely related methods that have been grouped under the general term "maximum entropy method" (MEM).2$28 The specific technique used in this study was the "covariance method. 29 We will refer to data obtained with this method as the MEM estimate. This method is not limited in resolution by signal length and therefore may be particularly approCIRCULATION by gest on July 5, 2017 http://ciajournals.org/ D ow nladed from DIAGNOSTIC METHODS-NONINVASIVE GRAPHIC METHODS priate for analysis of signals of short duration, such as aortic porcine valve closing sounds. We report here the results of a study of MEM spectral analysis of aortic porcine valve closing sounds to discriminate between normal and abnormal valve function. Methods Patients. Phonocardiograms were performed on 37 patients, ages 19 to 76 years, who had undergone aortic valve replacement with a porcine xenograft from to 67 months previously (table 1). One patient (No. 27) had two separate aortic valve replacements and recordings from each valve were used. Of the 38 phonocardiograms available, 26 were from Hancock xenograft valves (sizes 19 to 29); 10 of these were modified by excision of the supportive muscle band.30 Twelve phonocardiograms were from Carpentier-Edwards xenografts (sizes 23 to 25). This study population was selected over a 12 month period from routine postoperative follow-up visits, from patients admitted to the hospital with suspected abnormalities of valve function, and from patients in the early recovery period after aortic porcine valve replacement. All patients had diastolic blood pressures within the normal range. This population was divided into three groups. Group 1 consisted of 27 patients (patients 1 to 26 and patient 27, study 2) with normal valve function as assessed by the clinical history, physical examination, and laboratory investigations, including serial electrocardiograms (ECGs) and chest x-rays. In two of these subjects, left heart catheterization and angiography indicated normal aortic porcine valve function. In 10 of these 27 patients examined by M mode and two-dimensional echocardiography, no abnormalities of the valve prostheses were shown. To obtain recordings from valves that had undergone only minimal ultrastructural changes of normal aging, 22 patients in this group were studied within the first 6 months after their aortic valve replacement. Four patients had undergone aortic valve replacement from 6 months to 5 years previously, and one patient had an aortic valve replacement more than 5 years before the study. Group 2 consisted of two patients (patient 27, study 1, and patient 28) who had aortic porcine paravalvar leak with normal leaflet anatomy. In patient 27 the presence of paravalvar leak was confirmed at valve surgery. In patient 28, a supravalvar aortogram demonstrated a paravalvar regurgitant jet of moderate severity, with normal aortic prosthetic valve leaflet motion. The finding of anatomically normal valve leaflet was supported in each patient by M mode and two-dimensional echocardiography. The two patients in this group had undergone valve replacement 23 and 20 months before the study. Group 3 consisted of nine patients with abnormal prosthetic valve function. Clinical findings and laboratory investigations, including serial ECGs and chest x-rays, were consistent with prosthetic valve dysfunction. Seven of the nine patients had M mode and two-dimensional echocardiography, and all nine patients underwent left heart catheterization and aortography. The valve leaflet abnormality was ultimately confirmed by surgery in eight patients. Four patients had intrinsic valve dysfunction with leaflet thickening and various degrees of calcification. All four had fracture of one cusp, with significant valvar regurgitation. Three patients had active bacterial endocarditis with vegetations, two having in addition significant valve regurgitation. Two patients had healed endocarditis with valve scarring and lesser degrees of regurgitation. Three patients in group 3 were studied within 6 months after valve replacement. Four patients had undergone aortic valve replacements from 6 months to 5 Vol. 68, No. 1, July 1983 years previously, and two patients had aortic valve replacement 5 years or longer before the study. Phonocardiograms. Phonocardiograms were taken with the patient lying supine at 45 degrees. Recordings were made from the left sternal border. To identify each component of the second heart sound and its variation with respiration, a standard