A total of 1,569 donors were managed by CTDN between January 1, 2002 and December 31, 2007, and 1,085 had stored ECGs available for analysis. Fourteen donors were excluded as their ECGs were of poor quality or had missing leads (e.g. 2-3 lead rhythm strips only). After excluding 68 donors<14 years and 21 donors>65 years of age, 980 ECGs were included in the final study cohort. There were 391 donors in the overall CTDN cohort who were 14-65 years of age and did not have stored ECGs available for interpretation. These donors were older, and had a higher incidence of hypertension, diabetes, and coronary artery disease compared to donors with ECGs (Supplementary Table
). They were less likely to receive hormonal therapy during the donor management period, and their hearts were less likely to be utilized for transplantation.
The characteristics of the donor cohort are summarized in . Mean donor age was 38 ± 14 years, and 63% were male. The most common causes of death were cerebrovascular (including subarachnoid hemorrhage and ischemic stroke, 47%), followed by head trauma (43%), and anoxia (9%). Twenty-six percent of donors had a history of hypertension, and 28% had a history of cocaine or methamphetamine use. One-third of donors had an elevated serum troponin level, defined in this study as a peak level ≥ 1.0 mcg/L, given the variety of assays (sandwich and immunoenzymatic) from multiple manufacturers used at different donor hospitals.
Ninety-three percent of donors in this cohort had at least one echocardiogram, and 16.5% had one or more additional echocardiograms, based on the discretion of the treating clinician. Associations between the ECG and first donor echocardiogram were studied. The median time elapsed between the ECG and echocardiogram was 98 minutes (IQR 29, 498). T he mean left ventricular ejection fraction (LVEF) was 62% ± 12%. Slightly more than half of donors had left ventricular hypertrophy (defined as LV septal or posterior wall thickness > 1.1 cm) and 20% had LV regional wall motion abnormalities.
When comparing clinical characteristics of donors whose hearts were or were not accepted for transplantation, we found that donors who died of cerebrovascular causes, who had a history of hypertension, diabetes, or coronary artery disease, or who had an elevated serum troponin level were less likely to be cardiac organ donors.
Characteristics of donor electrocardiograms
Electrocardiographic findings after brain death are summarized in . Mean heart rate was 102 ± 20 bpm, and 97% were in sinus rhythm. One or more ECG abnormalities were present in 51% of the ECGs studied. Atrial and ventricular ectopy were rare, as were atrioventricular block, conduction delays (including right and left bundle branch block), and fascicular block.
Characteristics of Donor Electrocardiograms
A notable finding was prolongation of the corrected QT interval: the mean QTc was 449 ± 48 msec, while 21% of donors had QTc>480 msec and 15% had QTc>500 msec. QT prolongation was significantly associated with cause of death: 28% of donors who died of cerebrovascular causes had a QTc>480 msec, compared to 23% of donors who died from anoxia, 20% of those who died from CNS tumors, and 14% of those who died from head trauma (p<0.001). This finding was more common in female donors (OR 2.7, 95% CI 2.0-3.7, p<0.001) compared to males. Among donors dying of cerebrovascular causes, 38% of females had a QTc>480 msec, and 28% had QTc>500 (versus 19% and 13% for males, respectively, p<0.001). Prolongation of the QTc interval was associated with lower serum potassium levels. The mean serum potassium level was 4.0 ± 0.6 mmol/L in donors with QTc<480 msec and 3.7 ± 0.5 mmol/L in donors with QTc≥480 msec (p<0.0001).
Also of note was the high prevalence of voltage criteria for LVH, present in 8% of potential organ donors. This finding was significantly more common in donors who died of cerebrovascular causes (14.7%), compared to those who died of head trauma (2.6%) or anoxia (1.1%), p<0.001, even after adjusting for donor history of hypertension (OR 10.9, 95% CI 1.5-80.5, p=0.02).
Finally, repolarization changes were present in 22% of the donor ECGs examined. Two percent of donor ECGs met criteria for pathologic ST elevations, 10% of donor ECGs demonstrated significant ST depressions, and 12% had significant T wave inversions, while 18% had non-specific ST-T wave abnormalities. Q waves suggestive of prior myocardial infarction were found in 7% of donor ECGs. Overall, 51% of donor ECGs were classified as “abnormal” due to one or more of the above findings.
Correlations between donor ECG and echocardiographic findings
Left ventricular hypertrophy was present on 8% of donor ECGs and 54% of echocardiograms. Given this disparity, LVH on ECG was found to have high specificity (97%) for increased LV wall thickness, but low sensitivity (11%). The presence of LVH on ECG increased the odds of increased LV wall thickness by 3.5-fold (95% CI 1.9-6.6, p<0.001).
The finding of an elevated serum troponin level ≥1.0 mcg/L was not associated with repolarization abnormalities on ECG, as defined by the presence of significant ST elevations, ST depressions, or T wave inversions. Specifically, having an elevated troponin level increased the odds of repolarization abnormalities by only 1.3-fold (95% CI 0.9-1.7, p=0.2) and had a sensitivity of 38% and a specificity of 67% for the presence of significant ST-T wave abnormalities.
Finally, the presence of pathologic Q waves on ECG had a high specificity for reduced LV ejection fraction (defined as LVEF<50%, specificity=97%) and LV regional wall motion abnormalities (RWMA, specificity=96%), albeit sensitivity was low (12% for reduced LVEF, 15% for RWMA).
Donor ECG findings and cardiac allograft utilization for heart transplantation
Fifty seven percent (N=560) of donor allografts in this cohort were accepted for heart transplantation. The results of multivariable analyses examining associations between donor ECG predictors and cardiac allograft utilization are presented in . These models were adjusted for donor age, sex, cause of death, blood type, race, height, and diagnosis of hypertension, diabetes, and coronary artery disease. Our analyses demonstrate that prolongation of the PR and QRS intervals are associated with decreased allograft utilization. Specifically, for every 10 msec increase in the PR interval, the odds of allograft utilization decreases by 10% (OR 0.9, 95% CI 0.84-0.98, p=0.01). Similarly, for every 10 msec increase in the QRS interval, the odds of allograft utilization decreases by 18% (OR 0.82, 95% CI 0.72-0.93, p=0.002). Notably, prolongation of the QT interval was not associated with reduced allograft utilization, after adjusting for relevant covariates.
Electrocardiographic Predictors of Cardiac Allograft Utilization
As a general category, repolarization abnormalities on ECG (defined as the presence of pathologic ST elevations, ST depressions, and/or T wave inversions) were not associated with reduced allograft utilization; however, changes in individual leads did reveal significant associations. Specifically, ST segment changes in leads I, V1, V2, and V3 and T wave inversions in leads I, II, and aVR were associated with reduced allograft utilization. Finally, the presence of pathologic Q waves in leads V1 and V2, suggestive of prior anteroseptal myocardial infarction, were also associated with reduced allograft utilization. When grouped by the presence of any ST segment, T wave, or Q wave abnormality on the 12-lead ECG, pathologic Q waves and T wave inversions were associated with allograft non-utilization in unadjusted models. However, after adjusting for donor demographic variables these results were attenuated towards the null, with only pathologic Q waves remaining associated with non-utilization. Finally, after adjusting for echocardiographic abnormalities (left ventricular ejection fraction<50%, regional wall motion abnormalities, and LVH), no significant associations between ECG abnormalities and allograft utilization remained ().
Abnormal ST segments, T waves, and Q waves as Predictors of Cardiac Allograft Utilization: Unadjusted and Adjusted Models
Finally, voltage criteria for LVH on the 12-lead ECG was a strong predictor of allograft non-utilization. Specifically, the presence of LVH reduced the odds of graft acceptance for transplantation by 77% (univariate OR 0.23, 95% CI 0.13-0.38, p<0.001). This association remained highly significant after adjusting for donor demographic variables and echocardiographic abnormalities (multivariate OR 0.36, 95% CI 0.18-0.72).