Methods

Ninety-nine patients were included in this prospective study. All had underlying congenital heart disease with hemodynamically significant pulmonary valve lesions and underwent PPVI at Great Ormond Street Hospital, The Heart Hospital, or Harley Street Clinic (London, United Kingdom).

PPVI occurred from May 2001 to July 2007, enabling ≥1-year follow-up in all subjects. Clinical and morphologic inclusion criteria for PPVI have been described previously, but briefly, include RV systolic pressure >2/3 systemic with symptoms, RV systolic pressure >3/4 without symptoms, moderate to severe PR with either symptom, severe RV dysfunction or dilatation, or impaired exercise capacity. RV outflow tract diameter measurements had to be <22 × 22 mm and >14 × 14 mm.

Electrocardiograms recorded 24 hours before, 24 hours after, and 1 year after PPVI were analyzed. Standard original hardcopies were scanned for online analysis using CardioCalipers 3.3 for Windows (Iconico, New York, New York; available at: http://www.iconico.com ), which enabled magnification for greater measurement accuracy. Electrocardiograms were analyzed in random order by 1 observer (CP) who was blinded to date, diagnosis, lesion type, and outcome. QRS duration was averaged for each electrocardiogram after analysis of all 12 leads and defined as the first positive/negative deflection to the last sharp positive/negative deflection across the isoelectric line. QRS, QT, and JT dispersions were calculated by subtracting the narrowest interval from the widest across any 12 leads. End of the T wave was defined as the point of return to the isoelectric line. QT intervals were averaged across the 12 leads for each electrocardiogram and corrected for heart rate to obtain the corrected QT interval. Interobserver differences were compared to an independent observer (AK) who was blinded in a similar manner to a series of 20 sample study electrocardiograms. Interobserver differences between measured electrical parameters was <2.5%.

Preoperative and 1-year postoperative echocardiographic reports were reviewed. Images were obtained using a Vivid 7 (GE Vingmed, Milwaukee, Wisconsin) by operators experienced in scanning patients with congenital heart disease. RV systolic pressure was calculated from continuous-wave Doppler analysis of the tricuspid regurgitant jet. Sixteen percent (n = 16) had no discernible tricuspid regurgitant jet and were not included in this analysis; 89% of patients underwent 1-year follow-up echocardiography.

Cardiac magnetic resonance imaging was performed using a 1.5-T scanner (Symphony or Avanto, Siemens Medical System, Erlangen, Germany). Patients underwent scanning before the procedure and at 1 year after the procedure. In 17 patients (17%) cardiac magnetic resonance data were not acquired due to pacemaker or automated implantable cardioverter-defibrillator insertion (12%) or scan intolerance (5%). Cardiac magnetic resonance assessment of ventricular volumes and function has been previously described.

For this study, PR was defined quantitatively by cardiac magnetic resonance as a regurgitant fraction >25%, pulmonary stenosis (PS) was defined as an RV outflow velocity >3.8 m/s with a regurgitant fraction <25%, and mixed lesions comprised the remainder. The 17 patients not able to undergo magnetic resonance imaging had valve lesion severity assessed qualitatively by echocardiography and/or angiography.

Data are presented as mean ± SD. Student’s paired or unpaired t test or repeated measures analysis of variance was used to analyze data where appropriate. Repeated measures analysis of variance was followed by Fisher’s protected least significant difference post hoc analysis, with all t tests with Bonferroni correction as necessary. Linear regression analysis was used to ascertain any relation between electrocardiographic parameters and RV volume. A p value <0.05 was considered statistically significant. Statistical analysis was performed using StatView (SAS Institute, Cary, North Carolina) and all graphs were produced using Prism 5.00 for Windows (GraphPad Software, San Diego, California).

Statistical analysis consisted of examination of the group as a whole, which was then subdivided according to pulmonary valve lesion (PR, PS, mixed) and underlying structural pathology.

Results

Ninety-nine patients were identified for inclusion to this study. Mean age at time of procedure was 23.1 ± 10 years; 60% were men. The commonest pulmonary valvular lesion was predominantly stenotic (43%), followed by mixed (29%) and then predominantly regurgitant (27%). The most common congenital diagnosis within the group was pulmonary atresia with ventriculoseptal defect (32%) followed by repaired tetralogy of Fallot (28%). Others included transposition of the great arteries (11%), common arterial trunk (12%), congenital aortic or pulmonary valve disease (13%), and double-outlet right ventricle (3%). No patients died during follow-up.

Mean preprocedure QRS duration was prolonged at 137 ± 29 ms with no significant change at any time point in the total group (24 hours, 133 ± 28 ms; 1 year, 134 ± 29 ms, p = 0.7; Figure 1 ). The subgroup of patients with a preprocedure QRS duration ≥150 ms also showed no significant change in global or precordial lead QRS duration immediately or at 24 hours or 1 year after PPVI (162 ± 9, 159 ± 12, 160 ± 12 ms, p = 0.28; Figure 1 ).

No significant change in total group QRS duration (A) or in those with preprocedure QRS >150 ms (B) at any time point during follow-up.

When QRS duration was assessed by underlying pulmonary valvular lesion, there was a significant decrease in the regurgitant group at 1 year (135 ± 27 to 128 ± 29 ms, p = 0.007) ( Figure 2 ). Patients with underlying PS or mixed valvular lesions showed no significant decrease in QRS duration (p = 0.98, p = 0.16).

Changes in QRS duration with follow-up in patients with pulmonary regurgitation (p = 0.007) (circles) , pulmonary stenosis (p = 0.98) (squares) , or mixed (p = 0.16) (triangles) at 1 year. Patients with pulmonary regurgitation showed a significant difference at 1 year.

In the entire group, corrected QT interval and QT, QRS, and JT dispersions shortened significantly at 1 year (p ≤0.001; Figure 3 ). These 4 parameters were prolonged immediately after the procedure (JT and QT dispersions achieving statistical significance, p = 0.04, p = 0.002) before shortening.

Total means ± SEs (bars) for QRS dispersion (QRSd), JT dispersion (JTd), and QT dispersion (QTd) (left axis) and corrected QT interval (QTc) (right axis) . JT and QT dispersions showed significant increases immediately after percutaneous pulmonary valve implantation.

When comparing patients according to valve lesion, those with obstructive lesions had a significant improvement in all dispersion values at 1 year compared to those with regurgitant or mixed lesions ( Figure 4 ). Corrected QT interval was significantly shorter at 1 year in the PR (450 ± 30 to 435 ± 36 ms) and PS (448 ± 31 to 433 ± 26 ms) groups (p = 0.02 for the 2 comparisons).

Dispersion values in the obstructive (QRS dispersion, p = 0.03; JT dispersion, p = 0.50; QT dispersion, p = 0.10) (squares) and regurgitant (QRS dispersion, p = 0.02; JT dispersion, p = 0.03; QT dispersion, p = 0.05) (circles) groups. Obstructive lesions showed significant changes at 1 year. Abbreviations as in Figure 3 .

There was no significant difference in any of the electrocardiographic parameters within the different underlying congenital diagnoses.

Echocardiographic determination of RV systolic pressure revealed significant decreases 1 year after PPVI in all patients (66.3 ± 22.7 to 51.7 ± 18.6 mm Hg, p <0.0001; Table 1 ). When this is broken down to underlying lesion type, those with predominant stenosis had the greatest decrease in pressure acutely (76.2 to 54.4 mm Hg, p <0.0001) and at 1 year (76.2 to 57.8 mm Hg, p = 0.0004).

Indexed RVEDV and RV end-systolic volume decreased significantly across the entire group at 1 year (−17.7% change, p <0.0001; −12.8%, p = 0.0005, respectively; Table 1 ). There was no significant change in RV ejection fraction (p = 0.29). Patients with PR had a significantly larger starting RVEDV than those with PS (111.8 ± 92.1 and 91 ± 26 ml/m ² , p = 0.017); however, their 1-year values were not significantly different (92.1 ± 31 and 82 ± 27 ml/m ² ).

Preprocedure QRS duration correlated with preprocedure RVEDV (r = 0.34, p = 0.002). There was no such correlation at 1 year; however, RVEDV at 1 year did correlate with corrected QT interval at 1 year (r = 0.4, p = 0.007). When analyzed by valvular lesion there was a significant correlation at 1 year with changes in RVEDV and QT dispersion in the PS group (r = 0.4, p = 0.03). Similarly, change in RVEDV correlated with change in corrected QT interval (r = 0.6, p = 0.04) and change in JT dispersion (r = 0.6, p = 0.03), respectively, in the PR group.