Citations

Abstract Aims Spectrum bias can arise when a diagnostic test is derived from study populations with different disease spectra than the target population, resulting in poor generalizability. We used a real-world artificial intelligence (AI)-derived algorithm to detect severe aortic stenosis (AS) to experimentally assess the effect of spectrum bias on test performance. Methods and results All adult patients at the Mayo Clinic between 1 January 1989 and 30 September 2019 with transthoracic echocardiograms within 180 days after electrocardiogram (ECG) were identified. Two models were developed from two distinct patient cohorts: a whole-spectrum cohort comparing severe AS to any non-severe AS and an extreme-spectrum cohort comparing severe AS to no AS at all. Model performance was assessed. Overall, 258 607 patients had valid ECG and echocardiograms pairs. The area under the receiver operator curve was 0.87 and 0.91 for the whole-spectrum and extreme-spectrum models, respectively. Sensitivity and specificity for the whole-spectrum model was 80% and 81%, respectively, while for the extreme-spectrum model it was 84% and 84%, respectively. When applying the AI-ECG derived from the extreme-spectrum cohort to patients in the whole-spectrum cohort, the sensitivity, specificity, and area under the curve dropped to 83%, 73%, and 0.86, respectively. Conclusion While the algorithm performed robustly in identifying severe AS, this study shows that limiting datasets to clearly positive or negative labels leads to overestimation of test performance when testing an AI algorithm in the setting of classifying severe AS using ECG data. While the effect of the bias may be modest in this example, clinicians should be aware of the existence of such a bias in AI-derived algorithms.

AbstractIntroductionThe efficacy of cardiac resynchronization therapy (CRT) has been widely studied in the medical literature; however, about 30% of candidates fail to respond to this treatment strategy. Smart computational approaches based on clinical data can help expose hidden patterns useful for identifying CRT responders.MethodsWe retrospectively analyzed the electronic health records of 1664 patients who underwent CRT procedures from January 1, 2002 to December 31, 2017. An ensemble of ensemble (EoE) machine learning (ML) system composed of a supervised and an unsupervised ML layers was developed to generate a prediction model for CRT response.ResultsWe compared the performance of EoE against traditional ML methods and the state‐of‐the‐art convolutional neural network (CNN) model trained on raw electrocardiographic (ECG) waveforms. We observed that the models exhibited improvement in performance as more features were incrementally used for training. Using the most comprehensive set of predictors, the performance of the EoE model in terms of the area under the receiver operating characteristic curve and F1‐score were 0.76 and 0.73, respectively. Direct application of the CNN model on the raw ECG waveforms did not generate promising results.ConclusionThe proposed CRT risk calculator effectively discriminates which heart failure (HF) patient is likely to respond to CRT significantly better than using clinical guidelines and traditional ML methods, thus suggesting that the tool can enhanced care management of HF patients by helping to identify high‐risk patients.

Objectives: Clinicians need to identify which patients presenting with S. aureus bacteremia (SAB) have underlying cardiovascular implantable electronic device (CIED) infection and therefore need device extraction. The goal of this study was to identify clinical predictors of CIED infection in patients presenting with SAB. Methods: We retrospectively reviewed all cases of SAB in CIED recipients at Mayo Clinic from Jan 2001 to Dec 2011. CIED infection was defined using clinical, microbiologic and echocardiographic criteria. Patients who presented with clinical signs of CIED pocket infection were excluded. A multivariable model was developed to identify independent predictors of CIED infection. Model discrimination was summarized using the area under a receiver operating characteristics curve (AUC). Results: Among the 131 patients with SAB and CIED, 45 (34%) had underlying CIED infection. Permanent pacemaker (PPM) devices (OR 3.90, 95% CI 1.65-9.23, p=0.002), >1 device-related procedure (OR 3.30, 95% CI 1.23-8.86, p=0.018), and duration of SAB ≥4 days (OR 5.54, 95% CI 3.32-13.23, p <0.001) were independently associated with an increased risk of CIED infection. The AUC for this multivariable model was 0.79 indicating good discriminatory capacity to separate SAB patients with and without CIED infection. Various combinations of these three features and the corresponding predicted probabilities of CIED infection are summarized in Table. Conclusions: An individual patient’s risk of underlying CIED infection can be calculated based on the type of device, number of device-related procedures, and duration of SAB. Patients without any of these high-risk features have very low risk of underlying CIED infection and may be monitored closely without immediate device extraction.

Introduction: Prior cardiac surgery, especially the presence of coronary bypass grafts, is thought to preclude percutaneous epicardial access (EpiAcc) and, due to limited catheter maneuverability, mapping and ablation. The objective of our study was to evaluate the feasibility and safety of EpiAcc in patients with prior cardiac surgery. Methods: We retrospectively analyzed all patients who underwent EpiAcc as part of an ablation procedure between 01/01/2004-06/30/2013 at Mayo Clinic, Rochester, MN. All had documented VT or recurring, symptomatic PVCs. Baseline clinical characteristics and procedural characteristics were collected. Results: Of 162 patients that underwent EpiAcc, 18 had prior cardiac surgery (mean age 61 years, all male). This included 10 CABG, 2 epicardial ICD placement, 5 valve surgery (3 aortic valve, 2 mitral valve), 2 septal myectomy, 1 aortic arch replacement, 1 myocardial bridge unroofing and 1 myocardial perforation repair (3 patients had multiple procedures). Access was successful in 12 patients (66.7%); the inferior approach was utilized in 78%. Of patients with prior CABG, successful access was attained in 6 (60%). The average procedural time was 441.6 ± 141.3 minutes. Adhesiolysis was required in 10 patients with the sheath, access wire, pigtail catheter or ablation catheter. Intra-procedural coronary angiography was performed in 8 patients. There were a total of 45 VTs/PVCs ablated in 18 patients (average 2.5 ± 1.7), mean cycle length 463.6 ± 95.9ms. Ultimately, 12 patients (66.7%) underwent endocardial only ablation, 1 (5.5%) had epicardial only ablation while 5 (27.8%) had endocardial/epicardial ablation. The procedure was deemed successful in 13 patients (72.2%). Four patients had bleeding complications (pericardial effusion, pericardial hematoma, hemoperitoneum and pericardial tamponade [the latter required repair of the RV wall]). In patients with coronary grafts, there was no evidence of acute graft disruption. Conclusions: Percutaneous epicardial access is feasible in patients with previous cardiac surgery, including CABG. However, adhesiolysis is frequently required. Although the risk of coronary graft injury is low, life-threatening complications may occur.

Introduction: Long QT syndrome (LQTS) is a genetic disorder associated with abnormal ventricular repolarization and a predisposition for torsade de pointes and sudden cardiac death. Identification of markers to elucidate high risk patients is of major practical importance. Our aim was to use a novel, automated quantitative T wave analysis program to distinguish symptomatic from asymptomatic LQTS patients. Methods: We analyzed a genotyped cohort of 420 patients (22 ± 15.7years, 43% male) with either LQT1 (61%) or LQT2 (39%). ECG analysis was conducted using novel software that we created to automatically detect subtle changes in T wave morphology. Symptomatic patients were defined as those with a history of syncope or resuscitated cardiac arrest that was suspected of being LQTS triggered. Classification was performed using the top three T wave features in a linear discriminant classifier by 10x10 cross-validation. Results: Symptomatic LQT1 patients have a greater QTc duration in lead aVF (437 msec vs 416 msec, p <0.0001), a greater T wave center of gravity on the x axis (COGx) in lead V2 (0.304 vs 0.280, p <0.0001) and longer Tpeak-Tend/QT in lead II (0.208 vs 0.188, p < 0.0001) than asymptomatic patients with LQT1. Use of these 3 features enabled classification of symptomatic status in 72% of LQT1 cases, compared to 68% when QTc was used to classify alone. Symptomatic LQT2 patients had a longer QTc in Lead 1 (460 vs 418 msec, p < 0.0001), COGx in 4th segment of T wave was greater in lead V5 (0.399 vs 0.350, p<0.0001), and the Tpeak-Tend interval in V2 was longer (126 vs 99 msec, P =0.0004). Use of these 3 features enabled correct classification in 73% of LQT2 cases, compared to 67% when QTc was used alone. Conclusions: Application of an automated T wave analysis program was able to identify T wave features that distinguished the patients who presented as symptomatic LQTS, associated with syncope or resuscitated cardiac arrest. Use of these features could help assess arrhythmic risk and guide management strategies for patients with LQTS.

Introduction: As the US population ages and indications for cardiac device implantation expand, there is a concomitant increase in patients undergoing lead removal/extraction (LR/E).. We sought to examine trends in outcomes and complications associated with LR/E. Methods: We conducted a retrospective analysis using data from a large U.S. commercial insurance database (privately insured and Medicare Advantage patients) identifying all patients undergoing LR/E between 2004 and 2013 with at least one year continuous medical plan enrollment prior to LR/E (to identify baseline comorbidities). Using ICD9 codes and CPT codes, major complications (hemorrhage requiring transfusion, vascular injury, pericardial complications, respiratory complications, cerebrovascular complications, and requirement for cardiac surgery) associated with LR/E and 30-day readmission rates were evaluated. Results: Among 42,878 patients who underwent LR/E (median age of 62 ( 54, 72) years, 70% male); approximately 1 in 40 experienced a major adverse event (2.4%) during the study period. Respiratory complications were the most common complication and 311 patients required cardiac surgery. 30-day readmission occurred in 11.1% of the total population. The most common reasons for readmission were heart failure exacerbation (18.7%) and device related complications (13.2%). Conclusion: The observed complication rates after LE/R in clinical practice are similar to those seen in recent clinical trials and have been relatively stable over the past 10 years.

Introduction: Percutaneous epicardial access (EAc) is increasingly performed as part of VT/PVC ablation, especially in those that have failed prior ablation. We aimed to delineate the outcomes of EAc. Methods: We retrospectively analyzed individuals undergoing EAc for arrhythmia ablation between June 2004 and July 2014. Acute procedural success was defined as elimination/non-inducibility of the clinical arrhythmia after ablation; clinical success was defined as absence of symptoms/arrhythmia. Results: Of 175 patients, 126 (72%) were male with a mean age of 52.8±15.7 years. Eighty-three (47.4%) patients had NICM, 25 (14.3%) had ICM and 34 (19.4%) had normal hearts. Baseline EF was 44±16%. Patients had failed an average of 1.7±1.2 antiarrhythmic drugs; 139 patients (79%) had failed prior endocardial (endo) ablation (average 1.3±1.1). Six (3.4%) had prior EAc and 18 (10.3%) prior cardiac surgery. The indication for ablation was VT or PVC in 113 (64.6%) and 60 (34.3%) patients, respectively. Successful access was obtained in 168 (98%) patients. Anterior access, inferior access, or both, was utilized in 89 (52.7%), 77 (45.6%) and 2 (1.2%) patients, respectively. There were an average of 2.7±2.2 VT/PVCs identified, and 1.6±1.1 ablated. Endo, epicardial (epi) or combined ablation was performed in 58 (33.1%), 14 (8%), and 100 (57.1%) patients, respectively; 3 did not undergo ablation. Procedural success was obtained in 158 of 173 (91.3%) patients for whom data was available. At 12 months, clinical success was maintained in 62% with no difference between endo, epi or combined ablation (figure). Cardiac tamponade requiring repeat EAc occurred in 12 (6.9%) patients. There were no procedural related deaths. Conclusion: Epicardial access is feasible in almost all patients with no prior cardiac surgery, and permits acute procedural success in 91.3% of patients, most of whom had failed previous ablation. Furthermore, complications are relatively infrequent in this complex cohort.

Introduction: Patients with congenital heart disease (CHD) frequently require implantable cardiac devices. Device infection or malfunction may necessitate lead extraction. Extraction may be challenging due to long lead dwell times, anatomic abnormalities, and prior cardiac surgery. Little is known about extraction outcomes in the CHD population. Methods and Results: This retrospective study included 41 CHD patients and 82 age and gender matched controls that underwent lead extractions at two centers between 2001-2014. Only patients with leads older than 12 months were included. There were 79 leads in CHD patients and 150 in controls. Patients with CHD had a mean age of 39±17 years at extraction and on average their leads had been implanted for 81±84 months, this was not significantly different from the control group. Eighty eight percent of CHD patients had 1 or more cardiac surgeries as compared to only 22% of controls (p<0.001). The number of abandoned leads was also significantly different with 16 abandoned leads in the CHD group and 3 in controls (p<0.001). There was no statistically significant difference in extraction techniques between the groups. Complete extraction was achieved in 94% of patients in both groups. There were no CHD group complications. Control group complications included 3 SVC lacerations requiring sternotomy (one was fatal), and 1 ventricular perforations with tamponade requiring sternotomy. None of these patients had a history of cardiac surgery. The average age of leads in patients with a complication was 15.3 years which was older than the mean lead age seen in the CHD and control populations which was 6.75 and 4.6 years respectively. Conclusions: Lead extraction can be safely performed in patients with CHD. Despite anatomic abnormalities and previous cardiac surgery, the outcome of lead extraction in patients with CHD is comparable to controls. Prior cardiac surgery may lower the risk for extraction complications.

Introduction Multi-detector computed tomography (MDCT) is commonly used to identify pulmonary vein and left atrial anatomy prior to catheter ablation for atrial fibrillation (CAAF). Additionally, transesophageal echocardiography (TEE) is utilized to exclude left atrial/left atrial appendage thrombus (ICT). Whether MDCT alone can be used to exclude thrombus obviating the need for TEE is unknown. Methods All Pts referred for CAAF at Mayo Clinic between March 2004 and October 2006 were included. Clinical data (including CHADS score), 64 slice MDCT (non-ECG gated), TEE and intra-cardiac echocardiograph (ICE) were all performed and recorded. Image data was reviewed independently and blinded. The left atrial appendage (LAA) was defined as fully opacified or underfilled. An underfilled LAA was further interrogated using a region of interest over the filling defect and mean attenuation in Hounsfield Units (HU) were recorded. Results In total 400 Pts (mean age 56 ± 10 yrs; 76% male; EF 56 ± 10 % ) were included. 362 had no ICT by TEE nor filling defect by MDCT. 38 patients had an underfilled LAA in whom ICT was confirmed by TEE in 7. Sensitivity and specificity of MDCT for ICT was 100% and 92% respectively with a NPV of 100%. In 19 Pts with underfilled LAA’s, a mean LAA emptying velocity was found to be significantly reduced at 26 cm/s (6 – 61 cm/s). Those with a thrombus had a mean attenuation value below 70 HU. CHADS2 score was higher (2.0 versus 1.0) in Pts with LAA filling defect versus those without LAA filling defect. All Pts with confirmed thrombus were taking Warfarin. ICT was not seen in any Pts aged <52 yrs with CHADS score ≤1. Conclusions Sensitivity and specificity of MDCT for detection of LAA thrombus is high. LAA filling defects on MDCT correlate with low emptying velocities by TEE and low attenuation value in HU. These data would suggest that in the absence of MDCT filling defect, no further imaging is required. TEE should be utilized prior to CAAF to investigate MDCT filling defects. Multidetector Computed Tomography for the Detection of LAA Thrombus

Introduction Left ventricular non-compaction (LVNC) is a rare cardiomyopathy thought to result from intrauterine arrest of myocardial compaction which has been associated with and increased risk of sudden cardiac death (SCD). However, appropriate identification of at-risk individuals and risk-stratification is lacking, particularly in patients diagnosed during adulthood. We sought to identify high-risk patients wtih LVNC in order to guide therapy. Hypothesis That patients with decreased left ventricular ejection fraction (LVEF) would have worse outcomes than those with preserved left ventricular function. Methods We reviewed outcomes in all patients diagnosed with LVNC at Mayo Clinic between 2001 through 2006. Data were prospectively entered into a centralized database and retrospectively analyzed. Mortality was compared to age and sex matched population rates. Results A total of 80 patients were diagnosed with LVNC during this interval. After exclusion of pediatric patients (n = 21), 59 adult patients (62% male; age 46 ± 19 years, range 18–83) were analyzed. There were 4 deaths in this group (none due to stroke, mortality 18% at 2 years) during follow-up (mean 13.0 ± 13.5 months, range 0–52). Overall survival of patients with LVNC was worse than age and sex matched controls (Minnesota population p < 0.001). This was also noted in patients with LVEF > 35% (n = 35, p = 0.02). Overall LVEF was 41 ± 16% but was lower in patients who died compared to survivors (26 ± 10% vs. 42 ± 15%; Hazard ratio = 2.0 for a decrease in ejection fraction of 10%, p = 0.12). There were no deaths in patients with LVEF > 50% (n = 17). Conclusion Survival of adults with LVNC is worse than expected for age and sex. A decrease in ejection fraction is associated with increased risk of death in this population. These data suggest that patients diagnosed with LVNC and reduced LVEF may benefit from ICD implantation, whereas patients with a normal LVEF may be observed.