Time and event-specific deep learning for personalized risk assessment after cardiac perfusion imaging.

Standard clinical interpretation of myocardial perfusion imaging (MPI) has proven prognostic value for predicting major adverse cardiovascular events (MACE). However, personalizing predictions to a specific event type and time interval is more challenging. We demonstrate an explainable deep learning model that predicts the time-specific risk separately for all-cause death, acute coronary syndrome (ACS), and revascularization directly from MPI and 15 clinical features. We train and test the model internally using 10-fold hold-out cross-validation (n = 20,418) and externally validate it in three separate sites (n = 13,988) with MACE follow-ups for a median of 3.1 years (interquartile range [IQR]: 1.6, 3.6). We evaluate the model using the cumulative dynamic area under receiver operating curve (cAUC). The best model performance in the external cohort is observed for short-term prediction - in the first six months after the scan, mean cAUC for ACS and all-cause death reaches 0.76 (95% confidence interval [CI]: 0.75, 0.77) and 0.78 (95% CI: 0.78, 0.79), respectively. The model outperforms conventional perfusion abnormality measures at all time points for the prediction of death in both internal and external validations, with improvement increasing gradually over time. Individualized patient explanations are visualized using waterfall plots, which highlight the contribution degree and direction for each feature. This approach allows the derivation of individual event probability as a function of time as well as patient- and event-specific risk explanations that may help draw attention to modifiable risk factors. Such a method could help present post-scan risk assessments to the patient and foster shared decision-making.

Direct Risk Assessment From Myocardial Perfusion Imaging Using Explainable Deep Learning.

BACKGROUND: Myocardial perfusion imaging (MPI) is frequently used to provide risk stratification, but methods to improve the accuracy of these predictions are needed. OBJECTIVES: The authors developed an explainable deep learning (DL) model (HARD MACE [major adverse cardiac events]-DL) for the prediction of death or nonfatal myocardial infarction (MI) and validated its performance in large internal and external testing groups. METHODS: Patients undergoing single-photon emission computed tomography MPI were included, with 20,401 patients in the training and internal testing group (5 sites) and 9,019 in the external testing group (2 different sites). HARD MACE-DL uses myocardial perfusion, motion, thickening, and phase polar maps combined with age, sex, and cardiac volumes. The primary outcome was all-cause mortality or nonfatal MI. Prognostic accuracy was evaluated using area under the receiver-operating characteristic curve (AUC). RESULTS: During internal testing, patients with normal perfusion and elevated HARD MACE-DL risk were at higher risk than patients with abnormal perfusion and low HARD MACE-DL risk (annualized event rate, 2.9% vs 1.2%; P < 0.001). Patients in the highest quartile of HARD MACE-DL score had an annual rate of death or MI (4.8%) 10-fold higher than patients in the lowest quartile (0.48% per year). In external testing, the AUC for HARD MACE-DL (0.73; 95% CI: 0.71-0.75) was higher than a logistic regression model (AUC: 0.70), stress total perfusion deficit (TPD) (AUC: 0.65), and ischemic TPD (AUC: 0.63; all P < 0.01). Calibration, a measure of how well predicted risk matches actual risk, was excellent in both groups (Brier score, 0.079 for internal and 0.070 for external). CONCLUSIONS: The DL model predicts death or MI directly from MPI, by estimating patient-level risk with good calibration and improved accuracy compared with traditional quantitative approaches. The model incorporates mechanisms to explain to the physician which image regions contribute to the adverse event prediction.

Radiolabeled Thioflavin-T Derivative PET Imaging for the Assessment of Cardiac Amyloidosis.

PURPOSE OF REVIEW: Cardiac amyloidosis (CA) is an often under-recognized cause of heart failure with preserved ejection fraction. The goal of the current paper was to review imaging modalities available for detecting cardiac amyloidosis. We wished to determine what modalities are available for the diagnosis of cardiac amyloidosis and what modalities could be utilized in the future. RECENT FINDINGS: Early and delayed planar imaging of the chest currently plays a central role in the workup and diagnosis of CA. However, novel positron emission tomography (PET) tracers could play a large role in CA imaging in the future. There is an increasing body of literature supporting the use of targeted amyloid-binding PET radiotracers such as 11C-Pittsburgh compound B (11C-PIB), 18F-florbetapir, -flutemetamol, and -florbetaben for the detection of cardiac amyloid. While planar imaging currently plays a large role in the workup of CA, PET imaging could play an increasing important role in the future. The quantitative abilities of novel PET tracers could theoretically allow for the serial monitoring of patients and detection of response to therapy, and the sensitive nature of the tracers could allow for even earlier disease detection. Further work with large randomized controlled trial data is needed in the development and validation of PET tracers for cardiac amyloid and represents an exciting development within the realm of nuclear cardiology.

Assessment of left ventricular ejection fraction using low radiation dose computed tomography.

BACKGROUND: Cardiac CT is a non-invasive modality with the ability to estimate LVEF. However, given its limited temporal resolution and radiation, there has been initial resistance to use CT to measure LVEF. Developing an accurate, fast, low radiation dose protocol is desirable. OBJECTIVE: The objective of this study is to demonstrate that a 'low radiation dose' 64 slice cardiac computed tomography (CT) protocol is feasible and can accurately measure left ventricular ejection fraction (LVEF) while delivering a radiation dose lower than radionuclide angiography (RNA). METHODS: Patients undergoing RNA were prospectively screened and enrolled to undergo a 'low-dose' 64 slice CT LVEF protocol. LVEF measures, duration of each study and radiation dose between CT and RNA were compared. RESULTS: A total of 77 patients (mean age = 61.8 ± 12.2 years and 58 men) were analyzed. The mean LVEF measured by CT and RNA were 41.9 ± 15.2% and 39.4 ± 13.9%, respectively, (P = 0.154) with a good correlation (r = 0.863). Bland-Altman plot revealed a good agreement between the CT and RNA LVEF (mean difference of -2.4). There was good agreement between CT LVEF and RNA for identifying patients with LVEF ≤30% (kappa = 0.693) and LVEF ≥50% (kappa = 0.749). The mean dose estimated effective dose for CT and RNA were 4.7 ± 1.6 and 9.5 ± 1.0 mSv, respectively. The mean CT LVEF imaging duration (4:32 ± 3:05 minutes) was significantly shorter than the RNA image acquisition time (9:05 ± 2:36 minutes; p < 0.001). CONCLUSION: The results of our study suggest that low-dose CT LVEF protocol is feasible, accurate, and fast while delivering a lower radiation dose than traditional RNA.

Impact of SPECT myocardial perfusion imaging on cardiac care.

Single photon emission computed tomography myocardial perfusion imaging is a powerful modality for the assessment of coronary artery disease. It is useful in the diagnosis of CAD, prognostication of CAD and the determination of viability. It acts as guide for therapy and has the ability to assess effectiveness of therapy. The use of SPECT myocardial perfusion imaging has also been shown to be cost-effective compared to other modalities in cardiology.

Clinical interpretation standards and quality assurance for the multicenter PET/CT trial rubidium-ARMI.

UNLABELLED: Rubidium-ARMI ((82)Rb as an Alternative Radiopharmaceutical for Myocardial Imaging) is a multicenter trial to evaluate the accuracy, outcomes, and cost-effectiveness of low-dose (82)Rb perfusion imaging using 3-dimensional (3D) PET/CT technology. Standardized imaging protocols are essential to ensure consistent interpretation. METHODS: Cardiac phantom qualifying scans were obtained at 7 recruiting centers. Low-dose (10 MBq/kg) rest and pharmacologic stress (82)Rb PET scans were obtained in 25 patients at each site. Summed stress scores, summed rest scores, and summed difference scores (SSS, SRS, and SDS [respectively] = SSS-SRS) were evaluated using 17-segment visual interpretation with a discretized color map. All scans were coread at the core lab (University of Ottawa Heart Institute) to assess agreement of scoring, clinical diagnosis, and image quality. Scoring differences greater than 3 underwent a third review to improve consensus. Scoring agreement was evaluated with intraclass correlation coefficient (ICC-r), concordance of clinical interpretation, and image quality using κ coefficient and percentage agreement. Patient (99m)Tc and (201)Tl SPECT scans (n = 25) from 2 centers were analyzed similarly for comparison to (82)Rb. RESULTS: Qualifying scores of SSS = 2, SDS = 2, were achieved uniformly at all imaging sites on 9 different 3D PET/CT scanners. Patient scores showed good agreement between core and recruiting sites: ICC-r = 0.92, 0.77 for SSS, SDS. Eighty-five and eighty-seven percent of SSS and SDS scores, respectively, had site-core differences of 3 or less. After consensus review, scoring agreement improved to ICC-r = 0.97, 0.96 for SSS, SDS (P < 0.05). The agreement of normal versus abnormal (SSS ≥ 4) and nonischemic versus ischemic (SDS ≥ 2) studies was excellent: ICC-r = 0.90 and 0.88. Overall interpretation showed excellent agreement, with a κ = 0.94. Image quality was perceived differently by the site versus core reviewers (90% vs. 76% good or better; P < 0.05). By comparison, scoring agreement of the SPECT scans was ICC-r = 0.82, 0.72 for SSS, SDS. Seventy-six and eighty-eight percent of SSS and SDS scores, respectively, had site-core differences of 3 or less. Consensus review again improved scoring agreement to ICC-r = 0.97, 0.90 for SSS, SDS (P < 0.05). CONCLUSION: (82)Rb myocardial perfusion imaging protocols were implemented with highly repeatable interpretation in centers using 3D PET/CT technology, through an effective standardization and quality assurance program. Site scoring of (82)Rb PET myocardial perfusion imaging scans was found to be in good agreement with core lab standards, suggesting that the data from these centers may be combined for analysis of the rubidium-ARMI endpoints.

Anatomic versus physiologic assessment of coronary artery disease. Role of coronary flow reserve, fractional flow reserve, and positron emission tomography imaging in revascularization decision-making.

Angiographic severity of coronary artery stenosis has historically been the primary guide to revascularization or medical management of coronary artery disease. However, physiologic severity defined by coronary pressure and/or flow has resurged into clinical prominence as a potential, fundamental change from anatomically to physiologically guided management. This review addresses clinical coronary physiology-pressure and flow-as clinical tools for treating patients. We clarify the basic concepts that hold true for whatever technology measures coronary physiology directly and reliably, here focusing on positron emission tomography and its interplay with intracoronary measurements.

Lessons from the Tc-99m shortage: implications of substituting Tl-201 for Tc-99m single-photon emission computed tomography.

BACKGROUND: In 2009, the Chalk River nuclear reactor closed for repairs that led to a critical shortage of technetium-99m (Tc-99m). Several centers used thallium-201 (Tl-201) as an alternative radiotracer for myocardial perfusion imaging. Because Tl-201 is considered by many as a suboptimal radiotracer, we sought to understand the impact of using Tl-201 (during the Tc-99m shortage) on downstream resource utilization. METHODS AND RESULTS: We performed a retrospective study at the Ottawa Heart Institute of 7402 patients (60% men; mean age, 62.6 ± 11.8 years), patients were referred for myocardial perfusion imaging between May 2008 and January 2011 (PRE_Tc-99m [2938 patients]), during (DURING_Tl-201 [2959 patients]), and after (POST_Tc-99m [1505 patients]) the Tc-99m shortage. Patients were followed for 6 months after their index myocardial perfusion imaging to determine subsequent rates of cardiac catheterization or noninvasive imaging. More downstream testing was seen in the Tl-201 cohort (639 [21.4%] patients) than the Tc-99m cohort (537 [12.1%] patients; P<0.001). After adjustment using propensity scores, differences in downstream referral rates were maintained. The downstream investigations resulted in an estimated increase in per-patient costs ($165.22; 95% confidence interval, 153.00-177.42) in the DURING_Tl-201 cohort compared with the Tc-99m cohort ($90.97; 95% confidence interval, 83.42-98.90; P<0.001). As well, the mean effective radiation dose per-patient was higher in DURING_Tl-201 (23.57 mSv; 95% confidence interval, 23.16-23.96) than in Tc-99m (12.92 mSv; 95% confidence interval, 12.55-13.40; P<0.001). CONCLUSIONS: In this single-center study, the use of Tl-201 during the Tc-99m shortage was associated with an increase in downstream testing, cost, and patient radiation exposure, but these findings may not be generalizable to other centers. Although Tl-201 provided a short-term solution to the unexpected Tc-99m shortage, long-term cost-effective solutions should be areas of future study.

Quantitative PET/CT measures of myocardial flow reserve and atherosclerosis for cardiac risk assessment and predicting adverse patient outcomes.

Conventional scintigraphic myocardial perfusion imaging with SPECT/CT or with PET/CT has evolved as an important clinical tool for the diagnostic assessment of flow-limiting epicardial lesions and risk stratification of patients with suspected CAD. By determining the relative distribution of radiotracer-uptake in the left-ventricular (LV) myocardium during stress, the presence of flow-limiting CAD lesions can be identified. While this approach successfully identifies epicardial coronary artery lesions, the presence of subclinical and non-obstructive CAD may go undetected. In this direction, the concurrent ability of PET/CT to assess absolute myocardial blood flow (MBF) in ml/g/min, rather that relative regional distribution of radiotracer-uptake, and myocardial flow reserve (MFR), expands the scope of conventional myocardial perfusion imaging from the identification of more advanced and flow-limiting epicardial lesions to (1) subclinical CAD, (2) an improved characterization of the extent and severity of CAD burden, and (3) the discovery of "balanced" reduction in myocardial blood flow as a consequence of 3 vessel CAD. Concurrent to the PET data, the CT component of the hybrid PET/CT allows the assessment of coronary artery calcification as an indirect surrogate for CAD burden, without contrast, or with contrast angiography to directly denote coronary stenosis and/or plaque morphology with CT. Hybrid PET/CT system, therefore, has the potential to not only identify and characterize flow-limiting epicardial lesions but also subclinical stages of functional and/or structural stages of CAD. Whether the application of PET/CT for an optimal assessment of coronary pathology, its downstream effects on myocardial perfusion, and coronary circulatory function will in effect lead to changes in clinical decision-making process, investiture in preventive health care, and improved long-term outcome, awaits scientific verification.

Potential clinical and economic consequences of noncardiac incidental findings on cardiac computed tomography.

OBJECTIVES: We sought to determine the incidence, clinical significance, and potential financial impact of noncardiac incidental findings (IF) identified with cardiac computed tomography (CT). BACKGROUND: Cardiac CT is gaining acceptance and may lead to the frequent discovery of extracardiac IF. METHODS: Consecutive patients undergoing cardiac CT had noncardiac structures evaluated after full field of view (32 to 50 cm) reconstruction. IF were categorized as clinically significant (CS), indeterminate, or clinically insignificant. Patient follow-up was performed by telephone, and verified with hospital records and/or communication with physicians. RESULTS: Of 966 patients (58 +/- 16 years of age, 55.4% men, >98% outpatients), 401 (41.5%) patients had noncardiac IF. A total of 12 (1.2%) patients had CS findings, and 68 (7.0%) patients had indeterminate findings. At follow-up (18.4 +/- 7.6 months), none of the indeterminate findings became CS. Although 3 patients with indeterminate findings were diagnosed with malignant lesions, they were unrelated to the IF. After adjusting for age, IF were not an independent predictor of noncardiac death. Noncardiac death and cancer death in patients with and without IF were not statistically different. One patient suffered a major complication related to the investigation of an IF. The total direct cost associated with investigating IF was Canadian $57,596 (U.S. $83,035). CONCLUSIONS: Although noncardiac IF are common, clinically significant or indeterminate IF are less prevalent. Rates of death were similar in patients with and without IF, and IF was not an independent predictor of noncardiac death. The investigation of IF is not without cost or risk. Larger studies are required to assess the potential mortality benefit of identifying IF.

Evaluation of outcome and cost-effectiveness using an FDG PET-guided approach to management of patients with coronary disease and severe left ventricular dysfunction (PARR-2): rationale, design, and methods.

Patients with severe ventricular dysfunction and coronary disease have high morbidity and mortality. They may benefit from revascularization but have significant perioperative morbidity and mortality. Positron emission tomography (PET) imaging with F-18-fluorodeoxyglucose (FDG) can detect viable myocardium that may recover from revascularization in such patients. It is unclear whether use of FDG PET in this population improves outcome or is cost-effective. The principal aim of this study is to determine whether FDG PET-guided therapy improves clinical outcome compared to standard care. Secondary objectives are to determine whether FDG PET-guided therapy improves left ventricular (LV) function, improves quality of life, and provides a cost benefit versus standard care. Included in this multicenter randomized controlled trial are patients with coronary artery disease and severe LV dysfunction who are referred for revascularization, heart failure, or cardiac transplantation or in whom FDG PET is potentially useful. Consenting subjects will be randomized to therapy directed by FDG PET or standard care. The primary outcome is the composite cardiovascular endpoint of cardiac death, myocardial infarction, transplantation, or rehospitalization for unstable angina or heart failure. Secondary outcomes include health-related quality of life, costs, mortality, cardiovascular events, and LV function. Assuming two-sided alpha=0.05, power=80%, a sample size of 206 patients per group is required to detect a 15% absolute difference in the primary outcome between PET-directed therapy compared to standard care. Analyses will be conducted on an intention-to-treat basis. To our knowledge, this is the first large trial to evaluate whether FDG PET-directed therapy is effective and provides a cost benefit in patients with severe LV dysfunction. If so, thousands of such patients can be risk-stratified to select who is likely to benefit from revascularization.