Improved Model Discrimination Research Articles

Adjusting for Congenital Heart Surgery Risk Using Administrative Data.

Congenital heart surgery (CHS) encompasses a heterogeneous population of patients and surgeries. Risk standardization models that adjust for patient and procedural characteristics can allow for collective study of these disparate patients and procedures. We sought to develop a risk-adjustment model for CHS using the newly developed Risk Stratification for Congenital Heart Surgery for ICD-10 Administrative Data (RACHS-2) methodology. Within the Kids' Inpatient Database 2019, we identified all CHSs that could be assigned a RACHS-2 score. Hierarchical logistic regression (clustered on hospital) was used to identify patient and procedural characteristics associated with in-hospital mortality. Model validation was performed using data from 24 State Inpatient Databases during2017. Of 5,902,538 total weighted hospital discharges in the Kids' Inpatient Database 2019, 22,310 pediatric cardiac surgeries were identified and assigned a RACHS-2 score. In-hospital mortality occurred in 543 (2.4%) of cases. Using only RACHS-2, the mortality mode had a C-statistic of 0.81 that improved to 0.83 with the addition of age. A final multivariable model inclusive of RACHS-2, age, payer, and presence of a complex chronic condition outside of congenital heart disease further improved model discrimination to 0.87 (P< 0.001). Discrimination in the validation cohort was also very good with a C-statistic of 0.83. We created and validated a risk-adjustment model for CHS that accounts for patient and procedural characteristics associated with in-hospital mortality available in administrative data, including the newly developed RACHS-2. Our risk model will be critical for use in health services research and quality improvement initiatives.

Safety and incremental prognostic value of stress CMR in patients with known chronic kidney disease

Abstract Background Cardiovascular disease (CVD) is the main cause of mortality in patients with chronic kidney disease (CKD). Although several studies have demonstrated the consistently high prognostic value of stress cardiovascular magnetic resonance (CMR), its prognostic value in patients with CKD is not well established. We aimed to assess the safety and the incremental prognostic value of vasodilator stress perfusion CMR in consecutive symptomatic patients with known CKD. Methods Between 2008 and 2021, we conducted a bi-center longitudinal study with all consecutive symptomatic patients with known CKD, defined by estimated glomerular filtration rate (eGFR) between 30 and 60 ml/min/1.73 m2, referred for vasodilator stress CMR. All patients with eGFR &amp;lt;30 ml/min/1.73 m² (n=62) were excluded due the risk of nephrogenic systemic fibrosis. All patients were followed for the occurrence of major adverse cardiovascular events (MACE) defined as cardiac death or recurrent nonfatal myocardial infarction (MI). Cox regressions were used to determine the prognostic value of stress CMR parameters. Results Of 825 patients with known CKD (71.4±8.8 years, 70% men), 769 (93%) completed the CMR protocol, and among those 702 (91%) completed the follow-up (median follow-up 6.4 (4.0–8.2) years). Stress CMR was well tolerated without occurrence of death or severe adverse event related to the injection of gadolinium or nephrogenic systemic fibrosis case. The presence of inducible ischemia was associated with the occurrence of MACE (hazard ratio [HR] 12.50; 95% confidence interval [CI]: 7.50–20.8; p&amp;lt;0.001). In multivariable analysis, ischemia and late gadolinium enhancement were independent predictors of MACE (HR 15.5; 95% CI: 7.72 to 30.9; and HR:4.67 [95%CI: 2.83-7.68]; respectively, both p&amp;lt;0.001). After adjustment, stress CMR findings showed the best improvement in model discrimination and reclassification above traditional risk factors (C-statistic improvement: 0.13; NRI=0.477; IDI=0.049). Conclusions In patients with known CKD, stress CMR is safe and its findings have an incremental prognostic value to predict MACE over traditional risk factors.Cumulative rates of MACECompeting risk analysis

Left atrioventricular coupling index assessed using cardiac CT as a prognostic marker of cardiovascular death

Abstract Background Although the left atrioventricular coupling index (LACI) measured by cardiovascular magnetic resonance (CMR) is a strong predictor of cardiovascular events, the availability to CMR remains limited in clinical routine. Therefore, it would be useful to validate LACI assessment using other imaging methods such as computed tomography (CT). However, the feasibility of LACI measurement using cardiac CT and its prognostic value are not established. Purpose We aimed to investigate the prognostic value of the LACI assessed by cardiac CT, to predict the occurrence of cardiovascular death in consecutive patients without known CVD referred for coronary CT angiography (CCTA). Methods Between 2010 and 2020, we conducted a single-center study with all consecutive patients without known CVD referred for CCTA. LACI was defined as the ratio of LA to LV end-diastolic volumes. The primary outcome was cardiovascular death. Cox regressions were used to evaluate the association of LACI with the primary outcome after adjustment for traditional risk factors and CCTA findings. Results In 1,444 patients (70±12 years, 43% men), 92 (6.4%) patients experienced all-cause death, including 67 (4.3%) patients with cardiovascular death after a median (IQR) follow-up of 6.8 (5.9-9.1) years. After adjustment for risk factors and CCTA findings, LACI was positively associated with the occurrence of cardiovascular death (adjusted hazard ratio [HR], 1.07 [95% CI, 1.05-1.09] per 1% increment, p&amp;lt;0.001), and all-cause death (adjusted HR, 1.05 [95% CI, 1.03-1.07] per 1% increment, p&amp;lt;0.001). After adjustment, a LACI ≥25% showed the best improvement in model discrimination and reclassification above traditional risk factors and CCTA findings (C-statistic improvement: 0.27; NRI=0.826; IDI=0.209, all p&amp;lt;0.001; LR-test p&amp;lt;0.001). Conclusions LACI measured by CCTA is independently associated with cardiovascular death and all-cause death in patients without known CVD referred for CCTA, with an incremental prognostic value over traditional risk factors and CCTA findings.LACI measurement using CCTASurvival curves stratified by LACI value

Ai-based fully automated left atrioventricular coupling index as a prognostic marker in patients undergoing stress-CMR

Abstract Background The left atrioventricular coupling index (LACI) is a strong and independent predictor of heart failure (HF) in individuals without clinical cardiovascular disease. Its prognostic value is not established in patients with cardiovascular disease. Purpose To determine in patients undergoing stress CMR whether fully automated artificial intelligence-based LACI can provide incremental prognostic value to predict HF. Methods Between 2016 and 2018, we conducted a longitudinal study including all consecutive patients with abnormal (inducible ischemia or late gadolinium enhancement (LGE)) vasodilator stress CMR. Control subjects with normal stress CMR were selected using propensity score-matching. LACI was defined as the ratio of LA to LV end-diastolic volumes. The primary outcome included hospitalization for acute HF or cardiovascular death. Cox regression was used to evaluate the association of LACI with the primary outcome after adjustment for traditional risk factors. Results In 2,134 patients [65±12 years, 77% men, 1:1 matched patients (1,067 with normal and 1,067 with abnormal CMR)], LACI was positively associated with the primary outcome (median follow-up 5.2 (4.8-5.5) years) before and after adjustment for risk factors in the overall propensity-matched population (adjusted hazard ratio [HR], 1.18 [95%CI, 1.13-1.24] per 0.1% increment), patients with abnormal (adjusted HR, 1.22 [95%CI, 1.14-1.30] per 0.1% increment), and normal CMR (adjusted HR, 1.12 [95%CI, 1.05-1.20] per 0.1% increment; all p&amp;lt;0.001). After adjustment, a higher LACI of ≥25% showed the greatest improvement in model discrimination and reclassification over and above traditional risk factors and stress CMR findings (C-statistic improvement: 0.16; NRI=0.388; IDI=0.153, all p&amp;lt;0.001; LR-test p&amp;lt;0.001). Conclusion LACI is independently associated with hospitalization for HF and cardiovascular death in patients undergoing stress CMR, with an incremental prognostic value over traditional risk factors including inducible ischemia and LGE.LACI measurement method using CMRKaplan-Meier curves by LACI value

Prognostic impact of the extent, location, and pattern of late gadolinium enhancement in dilated cardiomyopathy

Abstract Background Several studies have shown that the presence and extent of late gadolinium enhancement (LGE) defined by cardiac magnetic resonance imaging (CMR) are strong prognosticators of death in patients with dilated cardiomyopathy (DCM). The interest of LGE location and pattern to improve risk stratification is not well established. Purpose To assess the incremental prognostic value of LGE extent, location and pattern for predicting all-cause death over traditional risk factors in DCM patients. Methods Between 2008 and 2021, we conducted a bi-centre longitudinal study including all consecutive DCM patients referred for CMR. DCM was confirmed excluding patients with ischemic heart disease, history of acute myocarditis, hypertrophic cardiomyopathy, and infiltrative disease. The outcome was all-cause death using the electronic French National Registry of Death. The prognostic value of LGE parameters was assessed using Cox analyses after adjustment for traditional prognostic factors: LGE presence, extent (number of segments), pattern (midwall or subepicardial), location and multiple areas. The incremental prognostic value of these LGE parameters was assessed using C-statistic increment, the continuous net reclassification improvement (NRI), the integrative discrimination index (IDI) and the global Chi-2. Results In 2,752 DCM patients (52±8 years, 56% male), 408 (15%) died after a median (IQR) follow-up of 9.2 (6.8-11.5) years. A total of 737 (27%) patients had LGE, including 664 (90%) LGE mid-wall and 73 (10%) LGE sub-epicardial. In the whole cohort (n=2,752), the LGE presence and extent were strong predictors of death (HR: 2.42 [95% CI: 1.67-3.51] and HR: 1.78 [95% CI: 1.64-1.93]; both p&amp;lt;0.001). In the subgroup of patients with LGE (n=737), the LGE multiple area was an independent predictor of death (HR: 5.12 with 95% CI: 3.53-7.41; p&amp;lt;0.001). The LGE pattern wasn't associated with death (subepicardial: HR: 0.88; 95% CI: 0.56-1.39, p=0.58). The location in both septal and free-wall was strongly associated with death compared to septal LGE only (HR: 10.1; 95% CI: 7.53-13.6; p&amp;lt;0.001), itself more at risk of death than free-wall LGE only (HR: 3.12; 95% CI: 2.03-4.78; p&amp;lt;0.001). In the 537 patients with LGE single area, LGE in septal segments was a strong prognostic marker (HR: 17.0; 95% CI: 5.88-49.4; p&amp;lt;0.001) compared to other locations (inferior: HR: 10.3; 95% CI: 3.45-30.5; p&amp;lt;0.001; lateral: HR: 6.13; 95% CI: 1.95-19.2; p&amp;lt;0.01; reference=anterior). A model combining the LGE extent, location and multiple area showed the best improvement in model discrimination and reclassification above traditional risk factors (C-statistic improvement: 0.19; NRI=11.8%; IDI=2.2%, all p&amp;lt;0.001). Conclusion In a large cohort of DCM patients, the LGE extent, location, and multiple area were independent prognosticators of death. A model combining these LGE findings had an incremental prognostic value over traditional prognostic factors to predict all-cause death.Incremental prognostic value of LGEHazard ratios regarding LGE segments

Incremental prognostic value of coronary plaque features on CT angiography in symptomatic patients with obstructive CAD

Abstract Background Several coronary computed tomography angiography (CCTA) studies have shown the strong prognostic value of the coronary artery disease (CAD) burden and the potential interest of the plaque composition analysis. However, the incremental prognostic value of this plaque analysis above the CAD burden in patients with obstructive CAD is not well established. Purpose To assess the incremental prognostic value of the plaque composition analysis using CCTA to predict major adverse cardiovascular events (MACEs) above traditional risk factors and CAD burden in consecutive symptomatic patients with obstructive CAD. Methods In a large multicentric registry, consecutive symptomatic patients without known CAD referred for CCTA were screened between 2008-2020. All patients with obstructive CAD (at least 1 ≥50% stenosis on CCTA) were included. The primary composite outcome was MACE defined by cardiovascular death or nonfatal myocardial infarction. CCTA data were interpreted using multiplanar reconstruction according to the 16-segment coronary artery model. Coronary segments were scored visually for both degree of luminal stenosis and composition of coronary plaque which were classified as noncalcified, mixed, or calcified. To assess the prognostic value of each CCTA findings, different multivariable Cox regression models were used (Model 1: clinical [traditional risk factors]; Model 2: model 1 + CAD burden [number of proximal segments with stenosis ≥50% + number of vessels with obstructive CAD]; Model 3: model 2 + plaques feature [number of segments with noncalcified plaques]). The additional predictive value of each model for predicting MACEs was calculated using C-statistic increment, continuous net reclassification improvement (NRI), and the integrative discrimination index (IDI). Results Of 2,312 patients referred for CCTA, 2,038 (mean age 70±12 years, 46% men) completed follow-up (median 6.7 years; interquartile range: 5.9-9.1 years), and 319 experienced a MACE (14%). After adjustment for traditional risk factors, number of proximal segments with &amp;gt;50% stenosis, number of vessels with obstructive CAD, number of segments with noncalcified were all strongly associated with MACEs (all p&amp;lt;0.001). After adjustment, the addition of the analysis of plaque features (model 3) showed the best improvement in model discrimination and reclassification above a model including traditional risk factors and CAD burden (model 2) (C-statistic improvement=0.03; NRI=45.0%; IDI=9.8%, Chi-2 global=412, all p&amp;lt;0.001; LR-test p&amp;lt;0.001). Conclusions In a large cohort of symptomatic patients with obstructive CAD on CCTA, the analysis of plaque features had an incremental prognostic value to predict MACEs over and above a model combining traditional risk factors and CAD burden.Survival curves for MACEIncremental prognostic value of CCTA

Incremental prognostic value of late gadolinium enhancement extent, transmurality, and pattern in patients with ischaemic cardiomyopathy

Abstract Background Presence and extent of late gadolinium enhancement (LGE) using cardiac magnetic resonance (CMR) are strong prognosticators of outcomes in ischaemic cardiomyopathy (ICM), but the prognostic value of multiple location of LGE, LGE transmurality, and midwall LGE area in addition to ischaemic LGE is not well established. Purpose To assess the incremental prognostic value of presence, extent, transmurality, and pattern of LGE in patients with ICM and reduced left ventricular ejection fraction (LVEF)­­. Methods Between 2008 and 2022, all consecutive patients referred for myocardial viability assessment using CMR and a history of ICM (≥70% stenosis in ≥1 epicardial coronary vessel on angiography and/or history of myocardial infarction and/or coronary revascularization) and LVEF value &amp;lt;50% were prospectively recruited. The outcome was all-cause death using the electronic French National Registry of Death. Nested Cox proportional hazard models were used to assess the prognostic value of the following LGE features: the extent, the transmurality of subendocardial ischaemic LGE (&amp;lt;50%, ≥50%, or transmural), the presence of LGE in multiple areas, and the presence of midwall LGE. The additional predictive value of the LGE features was assessed by the C-statistic increment, the continuous net reclassification improvement (NRI), the integrative discrimination index (IDI) and the global Chi-2. Results Among the 6,082 included patients (mean age 65±12 years, 73% male, mean LVEF 43.8±5.7%), 3,591 (59%) had ischaemic LGE; 652 (10.7%) patients died during a median follow up of 9 [6.6-11.5] years. After adjustment for risk factors in the overall population, the presence of ischaemic LGE was associated with a higher risk of death (adjusted HR=3.57; 95%CI: 2.89-4.40, p&amp;lt;0.001). Figure 1 shows survival according to the different LGE features. Patients with both multiple localisations of ischaemic LGE area and additional LGE midwall exhibited the highest risk of death (p&amp;lt;0.001). After adjustment in the subgroup of patients with ischaemic LGE (n=3,591), LGE extent (HR=1.55; 95%CI: 1.47-1.63), multiple localisations of ischaemic LGE (HR 1.61 95%CI: 1.32-1.97), LGE transmurality (HR=4.48; 95%CI: 3.39-5.93), and presence of additional midwall LGE (HR=1.77; 95%CI: 1.44-2.17, all p&amp;lt;0.001) were independently associated with a higher risk of death. A model combining all LGE features showed the best improvement in model discrimination and reclassification above traditional factors (C-statistic improvement: 0.17; NRI=54.1%; IDI=12.6%, Chi-2 global=1133, all p&amp;lt;0.001; LR-test p&amp;lt;0.001, Figure 2). Conclusion In a large cohort of ICM patients with reduced LVEF, ischaemic LGE extent, transmurality, multiple ischaemic LGE localisations, and additional presence of midwall LGE were independent prognosticators of death. A model combining all these LGE features had an incremental prognostic value over traditional prognostic factors to predict all-cause death.Kaplan-Meier survival curvesIncremental prognostic value

Abstract 12716: Left Ventricular Hypertrophy Phenotype to Predict Incident Atrial Fibrillation: The Multi-Ethnic Study of Atherosclerosis

Introduction: Left ventricular hypertrophy (LVH) has been shown to be associated with the occurrence of atrial fibrillation (AF). However, due to the heterogeneous nature of LVH, the predictive value of the LVH phenotype for incident AF remains uncertain. Methods: We utilized data from the Multi-Ethnic Study of Atherosclerosis (MESA). In this study, participants who underwent cardiac magnetic resonance imaging examination were included for analysis. LVH was defined as LV mass index ≥ 85.3 g/m 2 for females and ≥ 107.8 g/m 2 for males. Isolated LVH was determined as LVH without elevated cardiac biomarkers, and malignant LVH was determined as LVH with elevated hs-TnT level of ≥14 ng/L, or/and NT Pro-BNP level of ≥125 pg/mL. Results: A total of 4983 participants were included in the final analysis, with a mean age of 61.5 years. 272 (5.6%) had isolated LVH, and 222 (4.5%) had malignant LVH. During a median follow-up of 8.5 years, 272 cases of AF were observed. Compared to participants without LVH or elevated cardiac biomarkers, those with isolated LVH (HR, 1.82; 95% CI, 1.03-3.20) and malignant LVH (HR, 4.13; 95% CI, 2.77-6.16) had higher AF risks after adjusted for age, sex, race, body mass index, systolic blood pressure, antihypertensive medication use, smoking status, and diabetes. Malignant LVH carried a 1.5-fold increased risk (HR: 2.48, 95% CI: 1.30-4.73) of AF compared to isolated LVH (Panel A). Kaplan-Meier cumulative AF incidence curves showed that malignant LVH had a significantly higher incidence of AF than those in other phenotypes (Panel B). Including the LVH phenotype in the CHARGE-AF model improved model discrimination (areas under the curves[AUC] increase: 0.03, p&lt;0.001) (Panel C). Conclusions: The risks of AF incidence varied across LVH phenotypes, showing that malignant LVH carried a higher risk compared to other LVH phenotypes. LVH phenotype could provide additional predictive power for AF occurrence above the CHARGE-AF model.

Abstract 12346: Atherosclerotic Cardiovascular Disease Risk Associated With Social Determinants of Health at Individual and Area Levels

Introduction: Adverse social determinants of health (SDOH) are associated with an increased risk for atherosclerotic cardiovascular disease (ASCVD). However, it is unclear if the ASCVD risks associated with area-level SDOH are similar to those with individual-level measures. Further, SDOH are not included in the Pooled Cohort Equations (PCEs) when estimating 10-year ASCVD risk. Studies have shown that PCEs systematically underestimate risk for socially deprived individuals. Aims To examine the association of individual- and area-level SDOH including education, income, and employment with incident ASCVD. Also, we assessed if adding SDOH to the PCEs improves risk prediction accuracy. Methods: We pooled data from four cohorts: Jackson Heart Study, Multi-Ethnic Study of Atherosclerosis, Framingham Heart Study Offspring Cohort, and Reasons for Geographic and Racial Differences in Stroke Study. We excluded participants with a history of ASCVD. Individual-level SDOH were based on participants’ self-report and area-level SDOH were derived by linking participants’ addresses to US Census tract data. We used Cox proportional hazards models to assess the associations between SDOH and ASCVD. We also assessed the changes in discrimination (Harrell’s C-index) and calibration (Integrated Brier Score) after adding SDOH to risk prediction models. Results: We analyzed data from 27,603 participants (mean age 61.4 years, 59% women, and 49% White). During a median follow-up of 13 years, 2,839 incident ASCVD events occurred. Both individual- and area-level SDOH were associated with ASCVD ( Figure, Panel A ). Adding both individual- and area-level SDOH to risk prediction modestly improved model discrimination and calibration in Black men and women ( Figure, Panels B &amp; C ), and also improved calibration in non-Black women. Conclusions: The findings suggest that SDOH may be considered in the development of future ASCVD risk assessment tools, particularly among Black individuals.

Decreased diastolic hydraulic forces incrementally associate with survival beyond conventional measures of diastolic dysfunction

Decreased hydraulic forces during diastole contribute to reduced left ventricular (LV) filling and heart failure with preserved ejection fraction. However, their association with diastolic function and patient outcomes are unknown. The aim of this retrospective, cross-sectional study was to determine the mechanistic association between diastolic hydraulic forces, estimated by echocardiography as the atrioventricular area difference (AVAD), and both diastolic function and survival. Patients (n = 5176, median [interquartile range] 5.5 [5.0–6.1] years follow-up, 1213 events) were selected from the National Echo Database Australia (NEDA) based on the presence of relevant transthoracic echocardiographic measures, LV ejection fraction (LVEF) ≥ 50%, heart rate 50–100 beats/minute, the absence of moderate or severe valvular disease, and no prior prosthetic valve surgery. NEDA contains echocardiographic and linked national death index mortality outcome data from 1985 to 2019. AVAD was calculated as the cross-sectional area difference between the LV and left atrium. LV diastolic dysfunction was graded according to 2016 guidelines. AVAD was weakly associated with E/e’, left atrial volume index, and LVEF (multivariable global R2 = 0.15, p < 0.001), and not associated with e’ and peak tricuspid regurgitation velocity. Decreased AVAD was independently associated with poorer survival, and demonstrated improved model discrimination after adjustment for diastolic function grading (C-statistic [95% confidence interval] 0.644 [0.629–0.660] vs 0.606 [0.592–0.621], p < 0.001) and E/e’ (0.649 [0.635–0.664] vs 0.634 [0.618–0.649], p < 0.001), respectively. Therefore, decreased hydraulic forces, estimated by AVAD, are weakly associated with diastolic dysfunction and demonstrate an incremental prognostic association with survival beyond conventional measures used to grade diastolic dysfunction.

A deep learning system for retinal vessel calibre improves cardiovascular risk prediction in Asians with chronic kidney disease.

Cardiovascular disease (CVD) and mortality is elevated in chronic kidney disease (CKD). Retinal vessel calibre in retinal photographs is associated with cardiovascular risk and automated measurements may aid CVD risk prediction. Retrospective cohort study of 860 Chinese, Malay and Indian participants aged 40-80 years with CKD [estimated glomerular filtration rate (eGFR) <60ml/min/1.73m2] who attended the baseline visit (2004-2011) of the Singapore Epidemiology of Eye Diseases Study. Retinal vessel calibre measurements were obtained by a deep learning system (DLS). Incident CVD [non-fatal acute myocardial infarction (MI) and stroke, and death due to MI, stroke and other CVD] in those who were free of CVD at baseline was ascertained until 31 December 2019. Risk factors (established, kidney, and retinal features) were examined using Cox proportional hazards regression models. Model performance was assessed for discrimination, fit, and net reclassification improvement (NRI). Incident CVD occurred in 289 (33.6%) over mean follow-up of 9.3 (4.3)years. After adjusting for established cardiovascular risk factors, eGFR [adjusted HR 0.98 (95% CI: 0.97-0.99)] and retinal arteriolar narrowing [adjusted HR 1.40 (95% CI: 1.17-1.68)], but not venular dilation, were independent predictors for CVD in CKD. The addition of eGFR and retinal features to established cardiovascular risk factors improved model discrimination with significantly better fit and better risk prediction according to the low (<15%), intermediate (15-29.9%), and high (30% or more) risk categories (NRI 5.8%), and with higher risk thresholds (NRI 12.7%). Retinal vessel calibre measurements by DLS were significantly associated with incident CVD independent of established CVD risk factors. Addition of kidney function and retinal vessel calibre parameters may improve CVD risk prediction among Asians with CKD.

Artificial intelligence evaluation of coronary computed tomography angiography for coronary stenosis classification and diagnosis.

Ruling out obstructive coronary artery disease (CAD) using coronary computed tomography angiography (CCTA) is time-consuming and challenging. This study developed a deep learning (DL) model to assist in detecting obstructive CAD on CCTA to streamline workflows. In total, 2929 DICOM files and 7945 labels were extracted from curved planar reformatted CCTA images. A modified Inception V3 model was adopted. To validate the artificial intelligence (AI) model, two cardiologists labelled and adjudicated the classification of coronary stenosis on CCTA. The model was trained to differentiate the coronary artery into binary stenosis classifications <50% and ≥50% stenosis. Using the quantitative coronary angiography (QCA) consensus results as a reference standard, the performance of the AI model and CCTA radiology readers was compared by calculating Cohen's kappa coefficients at patient and vessel levels. The net reclassification index was used to evaluate the net benefit of the DL model. The diagnostic accuracy of the AI model was 92.3% and 88.4% at the patient and vessel levels, respectively. Compared with CCTA radiology readers, the AI model had a better agreement for binary stenosis classification at both patient and vessel levels (Cohen kappa coefficient: .79 vs. .39 and .77 vs. .40, p < .0001). The AI model also exhibited significantly improved model discrimination and reclassification (Net reclassification index = .350; Z = 4.194; p < .001). The developed AI model identified obstructive CAD, and the model results correlated well with QCA results. Incorporating the model into the reporting system of CCTA may improve workflows.

Left atrioventricular coupling index assessed using cardiac CT as a prognostic marker of cardiovascular death

PurposeThe purpose of this study was to investigate the prognostic value of left atrioventricular coupling index (LACI) assessed by cardiac computed tomography (CT), to predict cardiovascular death in consecutive patients referred for cardiac CT with coronary analysis. Materials and methodsBetween 2010 and 2020, we conducted a single-centre study with all consecutive patients without known cardiovascular disease referred for cardiac CT. LACI was defined as the ratio of left atrial to left ventricle end-diastolic volumes. The primary outcome was cardiovascular death. Cox regressions were used to evaluate the association between LACI and primary outcome after adjustment for traditional risk factors and cardiac CT angiography findings. ResultsIn 1,444 patients (mean age, 70 ± 12 [standard deviation] years; 43% men), 67 (4.3%) patients experienced cardiovascular death after a median follow-up of 6.8 (Q1, Q3: 5.9, 9.1) years. After adjustment, LACI was positively associated with the occurrence of cardiovascular death (adjusted hazard ratio [HR], 1.07 [95% CI: 1.05–1.09] per 1% increment; P < 0.001), and all-cause death (adjusted HR, 1.05 [95% CI: 1.03–1.07] per 1% increment; P <0.001). After adjustment, a LACI ≥ 25% showed the best improvement in model discrimination and reclassification for predicting cardiovascular death above traditional risk factors and cardiac CT findings (C-statistic improvement: 0.27; Nnet reclassification improvement = 0.826; Integrative discrimination index =0.209, all P < 0.001; likelihood-ratio-test, P < 0.001). ConclusionLACI measured by cardiac CT is independently associated with cardiovascular death and all-cause death in patients without known cardiovascular disease referred for cardiac CT, with an incremental prognostic value over traditional risk factors and cardiac CT findings.

Incremental prognostic value of fully-automatic LVEF by stress CMR using machine learning

Abstract Funding Acknowledgements Type of funding sources: None. Background Although some recent reports showed that fully automated artificial intelligence (AI)-based left ventricular ejection fraction (LVEF) measured at stress has good performance, its prognostic value during a stress CMR exam to predict outcomes is not well established. Aim To determine in patients undergoing stress CMR whether fully automated AI-based LVEF (LVEF-AI) measured at stress can provide incremental prognostic value to predict death. Methods Between 2016 and 2018, we conducted a longitudinal study including all consecutive patients referred for vasodilator stress CMR. LVEF-AI was assessed using AI-algorithm combines multiple deep learning networks for LV segmentation. The primary outcome was all-cause death defined by the French National Registry of Death. Cox regression was used to evaluate the association of stress LVEF-AI with death after adjustment for traditional risk factors. Results In 9,712 patients (66±15 years, 67% men), there was an excellent correlation between stress LVEF-AI measurement and LVEF measured by expert (LVEF-expert) (r=0.94, p&amp;lt;0.001). Using Bland–Altman analysis, we found that the difference between the mean LVEF-expert and the LVEF-AI group was −0.1% (−0.066–0.067), that was not statistically significant (p = 0.46). Stress LVEF-AI was associated with death (median [IQR] follow-up 4.5 [3.7–5.2] years) before and after adjustment for risk factors (adjusted hazard ratio [HR], 0.84 [95% CI, 0.82–0.87] per 5% increment, p&amp;lt;0.001). Stress LVEF-AI had similar significant association with death occurrence compared with LVEF-expert. After adjustment, a lower stress LVEF-AI showed the greatest improvement in model discrimination and reclassification over and above traditional risk factors and stress CMR findings (C-statistic improvement: 0.11; NRI=0.250; IDI=0.049, all p&amp;lt;0.001; LR-test p&amp;lt;0.001), with a superior additional prognostic value than the LVEF-AI measured at rest. Conclusion AI-based fully automated LVEF measured at stress is independently associated with the occurrence of death in patients undergoing stress CMR, with an additional prognostic value above traditional risk factors, inducible ischemia and LGE.

Safety and incremental prognostic value of stress CMR in patients with known chronic kidney disease

Abstract Funding Acknowledgements Type of funding sources: None. Background Cardiovascular disease (CVD) is the main cause of mortality in patients with chronic kidney disease (CKD). Although several studies have demonstrated the consistently high prognostic value of stress cardiovascular magnetic resonance (CMR), its prognostic value in patients with CKD is not well established. We aimed to assess the safety and the incremental prognostic value of vasodilator stress perfusion CMR in consecutive symptomatic patients with known CKD. Methods Between 2008 and 2021, we conducted a bi-center longitudinal study with all consecutive symptomatic patients with known CKD, defined by estimated glomerular filtration rate (eGFR) between 30 and 60 ml/min/1.73 m2, referred for vasodilator stress CMR. All patients with eGFR &amp;lt;30 ml/min/1.73 m² (n=62) were excluded due the risk of nephrogenic systemic fibrosis. All patients were followed for the occurrence of major adverse cardiovascular events (MACE) defined as cardiac death or recurrent nonfatal myocardial infarction (MI). Cox regressions were used to determine the prognostic value of stress CMR parameters. Results Of 825 patients with known CKD (71.4±8.8 years, 70% men), 769 (93%) completed the CMR protocol, and among those 702 (91%) completed the follow-up (median follow-up 6.4 (4.0–8.2) years). Stress CMR was well tolerated without occurrence of death or severe adverse event related to the injection of gadolinium or nephrogenic systemic fibrosis case. The presence of inducible ischemia was associated with the occurrence of MACE (hazard ratio [HR] 12.50; 95% confidence interval [CI]: 7.50–20.8; p&amp;lt;0.001). In multivariable analysis, ischemia and late gadolinium enhancement were independent predictors of MACE (HR 15.5; 95% CI: 7.72 to 30.9; and HR:4.67 [95%CI: 2.83–7.68]; respectively, both p&amp;lt;0.001). After adjustment, stress CMR findings showed the best improvement in model discrimination and reclassification above traditional risk factors (C-statistic improvement: 0.13; NRI=0.477; IDI=0.049). Conclusions In patients with known CKD, stress CMR is safe and its findings have an incremental prognostic value to predict MACE over traditional risk factors.

Dysphagia at 1 Year is Associated With Mean Dose to the Inferior Section of the Brain Stem

Dysphagia is a common toxicity after head and neck (HN) radiation therapy that negatively affects quality of life. We explored the relationship between radiation therapy dose to normal HN structures and dysphagia 1 year after treatment using image-based datamining (IBDM), a voxel-based analysis technique. We used data from 104 patients with oropharyngeal cancer treated with definitive (chemo)radiation therapy. Swallow function was assessed pretreatment and 1 year posttreatment using 3 validated measures: MD Anderson Dysphagia Inventory (MDADI), performance status scale for normalcy of diet (PSS-HN), and water swallowing test (WST). For IBDM, we spatially normalized all patients' planning dose matrices to 3 reference anatomies. Regions where the dose was associated with dysphagia measures at 1 year were found by performing voxel-wise statistics and permutation testing. Clinical factors, treatment variables, and pretreatment measures were used in multivariable analysis to predict each dysphagia measure at 1 year. Clinical baseline models were found using backward stepwise selection. Improvement in model discrimination after adding the mean dose to the identified region was quantified using the Akaike information criterion. We also compared the prediction performance of the identified region with a well-established association: mean doses to the pharyngeal constrictor muscles. IBDM revealed highly significant associations between dose to distinct regions and the 3 outcomes. These regions overlapped around the inferior section of the brain stem. All clinical models were significantly improved by including mean dose to the overlap region (P ≤ .006). Including pharyngeal dosimetry significantly improved WST (P=.04) but not PSS-HN or MDADI (P ≥ .06). In this hypothesis-generating study, we found that mean dose to the inferior section of the brain stem is strongly associated with dysphagia 1 year posttreatment. The identified region includes the swallowing centers in the medulla oblongata, providing a possible mechanistic explanation. Further work including validation in an independent cohort is required.

Vital signs-based deterioration prediction model assumptions can lead to losses in prediction performance

ObjectiveVital signs-based models are complicated by repeated measures per patient and frequently missing data. This paper investigated the impacts of common vital signs modeling assumptions during clinical deterioration prediction model development. Study Design and SettingElectronic medical record (EMR) data from five Australian hospitals (1 January 2019–31 December 2020) were used. Summary statistics for each observation's prior vital signs were created. Missing data patterns were investigated using boosted decision trees, then imputed with common methods. Two example models predicting in-hospital mortality were developed, as follows: logistic regression and eXtreme Gradient Boosting. Model discrimination and calibration were assessed using the C-statistic and nonparametric calibration plots. ResultsThe data contained 5,620,641 observations from 342,149 admissions. Missing vitals were associated with observation frequency, vital sign variability, and patient consciousness. Summary statistics improved discrimination slightly for logistic regression and markedly for eXtreme Gradient Boosting. Imputation method led to notable differences in model discrimination and calibration. Model calibration was generally poor. ConclusionSummary statistics and imputation methods can improve model discrimination and reduce bias during model development, but it is questionable whether these differences are clinically significant. Researchers should consider why data are missing during model development and how this may impact clinical utility.

Prognostic impact of artificial intelligence-based fully automated global circumferential strain in patients undergoing stress CMR

To determine whether fully automated artificial intelligence-based global circumferential strain (GCS) assessed during vasodilator stress cardiovascular magnetic resonance (CMR) can provide incremental prognostic value. Between 2016 and 2018, a longitudinal study included all consecutive patients with abnormal stress CMR defined by the presence of inducible ischaemia and/or late gadolinium enhancement (LGE). Control subjects with normal stress CMR were selected using a propensity score-matching. Stress-GCS was assessed using a fully automatic machine learning algorithm based on featured-tracking imaging from short-axis cine images. The primary outcome was the occurrence of major adverse clinical events (MACE) defined as cardiovascular mortality or nonfatal myocardial infarction. Cox regressions evaluated the association between stress-GCS and the primary outcome after adjustment for traditional prognosticators. In 2670 patients [65 ± 12 years, 68% men, 1:1 matched patients (1335 with normal and 1335 with abnormal CMR)], stress-GCS was associated with MACE [median follow-up 5.2 (4.8–5.5) years] after adjustment for risk factors in the propensity-matched population (adjusted hazard ratio [HR]: 1.12 [95% CI: 1.06–1.18]) and patients with normal CMR (HR: 1.43 [95% CI: 1.30–1.57], both P < 0.001), but not in patients with abnormal CMR (P = 0.33). In patients with normal CMR, an increased stress-GCS > −10% showed the best improvement in model discrimination and reclassification above traditional and stress CMR findings (C-statistic improvement: 0.27; NRI = 0.538; IDI = 0.108, all P < 0.001; LR-test P < 0.001). Stress-GCS is independently associated with MACE in patients undergoing stress CMR, with an incremental prognostic value over traditional risk factors and stress CMR findings in the group of patients with normal CMR. Prognostic value of AI-based Stress-GCS (Fig. 1).

AI-based fully automated left atrioventricular coupling index as a prognostic marker in patients undergoing stress-CMR

The left atrioventricular coupling index (LACI) is a strong and independent predictor of heart failure (HF) in individuals without clinical cardiovascular disease. To determine in patients undergoing stress CMR whether fully automated artificial intelligence-based LACI can provide incremental prognostic value to predict HF. Between 2016 and 2018, we conducted a study including all consecutive patients with abnormal vasodilator stress CMR [inducible ischemia or late gadolinium enhancement (LGE)]. Control subjects with normal CMR were selected using propensity score-matching. LACI was defined as the ratio of LA to LV end-diastolic volumes. The primary outcome included hospitalization for acute HF or cardiovascular death using Cox regression. In 2662 patients [65 ± 12 years, 68% men, 1:1 matched patients (1331 with normal and 1331 with abnormal CMR)], LACI was positively associated with the primary outcome [median follow-up 5.2 (4.8–5.5) years] before and after adjustment for risk factors in the overall propensity-matched population [adjusted hazard ratio (HR), 5.94 (95%CI, 3.74–9.45) per 0.1% increment], patients with abnormal [adjusted HR, 6.38 (95%CI, 3.77–10.8) per 0.1% increment], and normal CMR [adjusted HR, 6.15 (95%CI, 2.97–12.7) per 0.1% increment; all P < 0.001]. After adjustment, a higher LACI of ≥ 25% showed the greatest improvement in model discrimination and reclassification over and above traditional risk factors and stress CMR findings (C-statistic improvement: 0.15; NRI = 0.705; IDI = 0.398, all P < 0.001; LR-test P < 0.001). LACI is independently associated with hospitalization for HF and cardiovascular death in patients undergoing stress CMR, with an incremental prognostic value over traditional risk factors including inducible ischemia and LGE (Fig. 1).

Left atrioventricular coupling index assessed using cardiac CT as a prognostic marker of cardiovascular death

Although the left atrioventricular coupling index (LACI) measured by Cardiac MRI is a strong predictor of CV events, the availability to CMR remains limited in clinical routine. Therefore, it would be useful to validate LACI assessment using other imaging methods such as computed tomography (CT). To investigate the prognostic value of the LACI assessed by cardiac CT, to predict the occurrence of CV death in consecutive patients without known CVD referred for CCTA. Between 2010 and 2020, we conducted a single-center study with all consecutive patients without known CVD referred for CCTA. LACI was defined as the ratio of LA to LV end-diastolic volumes. The primary outcome was cardiovascular death. Cox regressions were used to evaluate the association of LACI with the primary outcome after adjustment for traditional risk factors and CCTA findings. In 1444 patients (70 ± 12 years, 43% men), 92 (6.4%) patients experienced all-cause death, including 67 (4.3%) patients with cardiovascular death after a median (IQR) follow-up of 6.8 (5.9–9.1) years. After adjustment for risk factors and CCTA findings, LACI was positively associated with the occurrence of cardiovascular death (adjusted hazard ratio [HR]: 1.07 [95% CI: 1.05–1.09] per 1% increment, P < 0.001), and all-cause death (adjusted HR, 1.05 [95% CI: 1.03–1.07] per 1% increment, P < 0.001). After adjustment, a LACI ≥ 25% showed the best improvement in model discrimination and reclassification above traditional risk factors and CCTA findings (C-statistic improvement: 0.27; NRI = 0.826; IDI = 0.209, all P < 0.001; LR-test P < 0.001). LACI measured by CCTA is independently associated with cardiovascular death and all-cause death in patients without known CVD referred for CCTA, with an incremental prognostic value over traditional risk factors and CCTA findings. Incremental prognostic value of LACI using CT (Fig. 1).