Author + information
- Received January 25, 2017
- Revision received April 8, 2017
- Accepted April 20, 2017
- Published online February 5, 2018.
- Brett W. Sperry, MDa,
- Balaji K. Tamarappoo, MD, PhDa,
- Jorge D. Oldan, MDb,
- Omair Javed, MDa,
- Daniel A. Culver, DOc,
- Richard Brunken, MDd,
- Manuel D. Cerqueira, MDd and
- Rory Hachamovitch, MD, MSca,∗ ()
- aDepartment of Cardiovascular Medicine, Cleveland Clinic Foundation, Cleveland, Ohio
- bDepartment of Radiology, University of North Carolina, Chapel Hill, North Carolina
- cDepartment of Pulmonary Medicine, Cleveland Clinic Foundation, Cleveland, Ohio
- dDepartment of Nuclear Medicine, Imaging Institute, Cleveland Clinic Foundation, Cleveland, Ohio
- ↵∗Address for correspondence:
Dr. Rory Hachamovitch, Cardiovascular Medicine J1-5, Cleveland Clinic, Euclid Avenue, Cleveland, Ohio 44195.
Objectives This study sought to evaluate the incremental value of quantifying the extent and severity of myocardial perfusion and 18F-labeled fluorodeoxyglucose (FDG) abnormalities in predicting adverse outcomes among patients with suspicion for cardiac sarcoidosis (CS).
Background Positron emission tomography (PET) with FDG is a key component of the noninvasive assessment of patients with suspected CS. However, the optimal method for image interpretation has not been defined.
Methods A retrospective analysis was performed of 203 patients who underwent perfusion and FDG-PET imaging to evaluate for CS. Imaging findings were scored by conventional 3-category methods (normal perfusion and metabolism, abnormal perfusion or metabolism, abnormal perfusion and metabolism) and by summed scores using the 17-segment model to represent extent and severity of disease. Heterogeneity of metabolism was quantified using the coefficient of variation (SD divided by the mean) of FDG uptake. Multivariable Cox models were developed to assess associations between imaging findings and adverse events (death, heart transplant, or ventricular arrhythmia requiring defibrillation).
Results The indication for FDG-PET was ventricular arrhythmia in 69 (34%), heart block in 16 (8%), cardiomyopathy in 54 (27%), and other indications in 64 (32%). There were 63 patients who developed adverse events over a mean follow-up of 1.8 years. After robust adjustment, only the summed score in segments with a perfusion–metabolism mismatch and the coefficient of variation were important prognostically (p = 0.029 and p = 0.041, respectively).
Conclusions Quantitative measures of extent and severity of perfusion–metabolism mismatch and coefficient of variation of FDG uptake provide an incremental prognostic advantage in patients undergoing FDG-PET for CS. These results support the use of a more detailed analysis of imaging findings, as is conventional in coronary artery disease.
Sarcoidosis is a heterogeneous, granulomatous disorder of unknown etiology that may affect any organ, but has a predilection for lymph nodes and lung tissue. Cardiac involvement most commonly manifests as ventricular arrhythmias, heart block, and cardiomyopathy, and the identification of cardiac involvement has important prognostic and treatment implications (1).
Cardiac positron emission tomography (PET) with rubidium-82 (Rb-82) and 18F-labeled fluorodeoxyglucose (FDG) has been the mainstay of noninvasive radionuclide imaging for the detection of cardiac sarcoidosis (CS) and assessment of ongoing inflammation. Previous studies examining CS using FDG-PET have focused on categorical descriptions of test results. Specifically, abnormal perfusion accompanied by abnormal FDG (in any distribution) has been associated with adverse events when compared with normal perfusion and metabolism (2). However, whether the extent and severity of perfusion and FDG metabolism abnormalities have prognostic implications is as yet undefined. Additionally, heterogeneity of FDG uptake, as defined by the coefficient of variation (CoV) (SD divided by the mean) of FDG global segmental uptake values (SUV), has been described in CS (3), but its prognostic implications are untested.
Assessing the extent and severity of perfusion and FDG defects is the standard of care in the radionuclide assessment of coronary artery disease (4–6). We hypothesized that a similar semiquantitative approach to defining the metrics of mismatch along with quantitative analysis of the CoV of FDG uptake would provide an incremental prognostic benefit over traditional image analysis in patients with suspicion for CS referred for cardiac FDG-PET.
A retrospective analysis was performed of consecutive patients seen at our institution from January 2005 to December 2013 who underwent perfusion and FDG-PET imaging as part of an evaluation of CS. All patients underwent a comprehensive clinical, laboratory, electrocardiographic, and echocardiographic evaluation. Patients were excluded in cases of an incomplete clinical assessment, obstructive coronary artery disease (defined as any documented lesion ≥70% or history of coronary revascularization), prior myocardial infarction, or uninterpretable FDG-PET imaging. Data were adjudicated based on the date of the index FDG-PET.
Electronic medical records were reviewed retrospectively for data within 3 months of the index FDG-PET. Moderate to severe dyspnea was defined as New York Heart Association functional class III or IV or Medical Research Council breathlessness scale 3 to 5. Indication for FDG-PET was divided into categories: ventricular arrhythmias, cardiomyopathy, heart block, or other signs and symptoms. Other signs and symptoms included atrial arrhythmias, syncope, and abnormal electrocardiogram with a history of pulmonary sarcoidosis. Patients were considered to have an abnormal resting electrocardiogram if there was any degree of heart block, bundle branch block, abnormal Q waves, or significant ST/T-wave abnormalities in ≥3 leads.
Transthoracic echocardiography was performed using commercially available Vivid 7 or Vivid 9 ultrasound systems (GE Healthcare, Milwaukee, Wisconsin) or EPIQ (Philips Medical, Amsterdam, the Netherlands). Echocardiographic parameters were measured prospectively in standard fashion as described by the American Society of Echocardiography guidelines (7) and reviewed for left ventricular dysfunction, regional wall motion abnormalities (involving ≥2 segments), and basal septal thinning (defined as a thickness of ≤5 mm).
Extracardiac involvement was defined by biopsy histology or hypermetabolic FDG uptake consistent with sarcoidosis noted on whole body FDG-PET. Cardiac biopsies were evaluated by 2 specialized cardiac pathologists experienced in the diagnosis of sarcoidosis. Patients were categorized as having definite or probable CS by the Japanese Ministry of Health and Welfare (JMHW) (8), Heart Rhythm Society (9), and World Association of Sarcoidosis and Other Granulomatous Diseases (9) criteria.
Whole body PET-CT was acquired with resting myocardial perfusion and metabolic imaging using Rb-82 and FDG, respectively. The protocol involved a prolonged fast of ≥12 h before the procedure to minimize myocardial FDG uptake. Beginning the morning before imaging, patients were instructed to have a “low-carbohydrate diet” consisting of foods that were generally high in fat and protein. Rb-82 was injected intravenously (30 to 50 mCi depending on body mass index) and resting perfusion images were acquired 2 min later. After Rb-82 imaging, 7 to 15 mCi of FDG was injected and images were acquired after a prolonged uptake period (>90 min later). Patients were imaged on Siemens Biography mCT 128 PET/CT scanners (Siemens Healthcare, Erlangen, Germany) and interpreted after attenuation correction using 4DM-SPECT software (4DM, INVIA Medical Imaging Solutions, Ann Arbor, Michigan). Radiation exposure for the Rb-82 and whole body FDG-PET was estimated at 12 mSv (10).
The Rb-82 and FDG imaging results were interpreted retrospectively jointly by the consensus of 2 experienced physicians trained in the interpretation of nuclear imaging while blinded to clinical information. FDG-PET tomograms were reconstructed and reoriented using an automatic algorithm (11) and manual reorientation when needed. Rb-82 and FDG data were reviewed side by side and initially scored using a 3-category scoring system (normal perfusion and metabolism, abnormal perfusion or metabolism, abnormal perfusion and metabolism) as described (2). Subsequently, results were scored using a 5-point semiquantitative system (12). Briefly, a 17-segment model of the left ventricular myocardium was used to score each segment from 0 to 4, with 0 representing normal perfusion and 4 an absence of perfusion. FDG images were analyzed using a similar 17-segment semiquantitative approach; a score of 4 represented dense FDG uptake and 0 represented absence of FDG uptake (compared with blood pool) (10). The pattern of FDG-PET findings in each segment was subcategorized as follows: normal perfusion and absence of FDG uptake, abnormal perfusion and absence of FDG uptake (matched pattern), abnormal perfusion and increased FDG uptake (mismatched pattern), and normal perfusion and increased FDG uptake. Patients with diffuse FDG uptake were scored as normal metabolism and due to incomplete myocardial glucose suppression. A summed rest score (SRS) was calculated by adding the perfusion scores in each of the 17 segments. Similarly, an SRS in matched and mismatched segments was calculated.
FDG uptake was quantified by measuring the SUV of each of the 17 segments. The maximum, minimum, mean, and SD of SUV were calculated. The CoV of the entire left ventricular myocardium was defined as the average of each segmental SD divided by the average of each segmental mean, as described (3). Ten patients representing an even distribution of CoV values were remeasured >1 month apart by 2 experienced physicians and interobserver and intraobserver variability was assessed using interclass correlation coefficients.
Mortality was determined from the electronic medical records and the national death index. Ventricular tachycardia requiring defibrillation was adjudicated based on clinical history or device interrogation. Occurrence of heart transplantation was assessed from the electronic medical records.
Categorical variables are presented as frequency and percent, and continuous variables as mean ± SD. Categorical variables were analyzed using the Fisher exact test, and continuous variables using a 2-tailed Student t test.
Survival analysis was performed as a time to first event analysis using a composite endpoint of all-cause mortality, heart transplantation, or ventricular arrhythmia requiring defibrillation. Patients were censored at 5 years or last follow-up.
The goal of our multivariable survival analysis was to determine the association between FDG-PET variables and adverse events after adjusting for potential confounders. To this end, our approach incorporated several steps. To avoid model overfitting, we initially performed a series of multivariable Cox proportional hazards analyses to construct composite variables to parsimoniously capture pre-imaging data. This process led to the development of a clinical risk score incorporating general clinical risk factors, and a medication score incorporating baseline medications. Variables preselected for the clinical risk score included age, sex, race, severe dyspnea, prior history of smoking, and diabetes. Baseline use of diuretics, beta-blocker, renin-angiotensin–aldosterone system blocker, antiarrhythmic, and immunosuppression (including steroids) were the components of the medication score. Our multivariable models included adjustments for these risk scores as well as left ventricular ejection fraction, history of ventricular dysrhythmia, and whether the patient met JMHW criteria for CS. Variables included in these adjustments were chosen based on previous reports and clinical judgment. The JMHW criteria were used as an adjustment factor; other criteria (Heart Rhythm Society and World Association of Sarcoidosis and Other Granulomatous Diseases) include abnormal FDG-PET as a variable.
Subsequently, we developed a series of multivariable Cox models to assess the association between FDG-PET imaging findings and the composite endpoint after adjusting for the above defined variables and stratifying by the presence of an implantable cardioverter-defibrillator (ICD). All multivariable models included a metric of perfusion and FDG measurement. Harrell’s C statistic was used to assess model discrimination. Figures were generated to demonstrate risk-adjusted outcomes based on FDG-PET findings.
All statistical tests were 2-sided and p < 0.05 was considered significant. All model assumptions were examined including linearity, collinearity, additivity, and proportional hazards. Statistical analysis was performed using Stata version 13 (StataCorp LP, College Station, Texas) and S-PLUS (Insightful Corp., Seattle, Washington). The study was approved by the Cleveland Clinic Institutional Review Board.
A total of 269 patients were screened for inclusion. The final study cohort consisted of 203 patients after exclusion criteria were satisfied (Figure 1). Obstructive coronary artery disease was ruled out in all patients (81% with cardiac catheterization and 19% with rest–stress perfusion imaging). Indication for FDG-PET was ventricular arrhythmia in 69 (34%), heart block in 16 (8%), cardiomyopathy in 54 (27%), and other indications in 64 (32%). Baseline clinical characteristics are provided in Table 1. In the overall cohort, 72% of patients had evidence of extracardiac sarcoidosis. Definite or probable CS was characterized by societal criteria in 93 (46%) of patients as follows: 93 patients (46%) by World Association of Sarcoidosis and Other Granulomatous Diseases, 83 (41%) by Heart Rhythm Society, and 30 (15%) by JMHW criteria.
Table 2 details FDG-PET findings; in patients with normal studies, 29 (14% of total) had diffuse FDG uptake that was classified as normal due to incomplete myocardial glucose suppression. More than one-half of the total cohort had a perfusion or metabolism abnormality (n = 109), with 17 having only metabolism, 37 having only perfusion, and 55 having both defects. Fasting glucose level was 94.1 ± 19.4 with no difference in glucose values based upon presence of abnormal FDG uptake (n = 172; p = 0.922). The mean CoV was 0.102 ± 0.053. CoV measurements were highly reproducible; the interclass correlation coefficients for intraobserver and interobserver variability were 0.984 and 0.973, respectively (p < 0.001 for both).
In addition, more than one-half of patients had significant FDG uptake on whole body FDG-PET, suggestive of active extracardiac sarcoidosis. Patients who developed adverse events were more likely to have a higher SRS, summed FDG score, summed score in matched and mismatched segments, SD of global SUV, and CoV. There was no difference in the mean or maximum global SUV in patients with or without subsequent events.
All patients were followed for a mean of 1.8 years, and 63 patients (31%) met the primary endpoint. There were 24 deaths, 10 heart transplantations, and 37 patients with ventricular arrhythmias requiring defibrillation during the follow-up period. Patients with an abnormal FDG-PET were more likely to meet the composite endpoint than those with normal imaging findings (p = 0.005).
The clinical risk score, medication score, ejection fraction, prior history of ventricular arrhythmia, and JMHW criteria were used as adjustment factors for the final models containing FDG-PET imaging variables. The clinical risk score was significantly associated with the composite outcome (chi-square = 23.5; p < 0.001), and there was a trend toward an association of the medication score (chi-square = 10.7; p = 0.06) (Online Table 1).
Table 3 details the results of sequential models after adjusting for the above factors and stratifying by ICD. Conventional 3-category scoring of FDG-PET did not add predictive value after robust adjustment. Similarly, metrics related to the degree of FDG uptake were not incrementally prognostic after adjustment for the above factors and summed perfusion score (summed FDG score, p = 0.130; mean SUV, p = 0.433; maximum SUV, p = 0.270). Rather, the heterogeneity of FDG uptake (CoV) was associated with adverse events (p = 0.009). Additionally, when the SRS was divided into scores based on the presence of abnormal FDG uptake in the same segment, it was found that only the SRS in segments with abnormal FDG (mismatched segments) was associated significantly with adverse events.
The final model combining these measures demonstrated that both the SRS in mismatched segments and the CoV were incrementally prognostic, with Harrell’s C statistic of 0.70 for the final model (Table 3, Figure 2). C statistic of the model without FDG-PET variables was 0.66. Figure 3 demonstrates the relationship between CoV and SRS in mismatch segments with respect to risk-adjusted outcomes based on the results of multivariable survival modeling. An illustrative example of patients with mismatched segments and a high CoV is seen in Figure 4, along with the additive association of these variables with adverse events.
Cardiac sarcoidosis, a disease process associated with poor clinical outcomes, is commonly investigated using cardiac imaging with FDG-PET to detect myocardial damage and ongoing inflammation. Prior studies have shown that abnormal perfusion and metabolism as detected by FDG-PET may be a prognostic marker of adverse cardiac events (2). However, these analyses have not examined how the results of the perfusion and FDG portions of the test should be coded to optimize risk assessment. In this study, after robust multivariable adjustment, we found that the SRS in segments with abnormal FDG (mismatched pattern) and the CoV were associated significantly with adverse events, and were superior predictors of outcomes compared with the other metrics examined. Therefore, both the extent and severity of mismatched defects and the heterogeneity of FDG uptake are prognostically important in CS. These imaging findings help to augment our knowledge of the disease process in this elusive condition.
CS is truly a disease that may present to and be managed by any subspecialty of cardiology: electrophysiology, heart failure, imaging, or the general cardiologist. The indications for cardiac evaluation have not been discussed in prior large studies; the most common indication in our cohort was ventricular arrhythmia (34%), although cardiomyopathy represented 27%. All patients had significant suspicion for the disease, with 72% having evidence of extracardiac sarcoidosis; yet, only 15% met JMHW criteria. The composite outcome of all-cause mortality, heart transplantation, or ventricular arrhythmia requiring defibrillation was met by 31% of patients over 1.8 years, which is similar to prior studies (2,13,14).
Several interesting messages can be gained from this analysis. First, the combination of perfusion and metabolism defects are important prognostically in the analysis of FDG-PET imaging in CS. The overall SRS, representing the extent and severity of perfusion defects, noted a trend toward significant association with adverse events in preliminary models. When this overall score was subdivided based on matched or mismatched scores, we found that the mismatched segments were driving the association with outcomes.
Second, both the extent and severity of perfusion defects are important prognostically. Perfusion defects became more prognostically predictive as they were analyzed in more granular detail, that is, from a binary assessment, to extent and severity (SRS), to extent and severity in segments with abnormal FDG uptake. In contrast, segmental analysis of FDG uptake was only useful to identify mismatched segments with an associated perfusion defect. Neither the grading of FDG uptake, nor the severity as assessed by mean or maximum SUV, were incrementally prognostic.
Third, the prognostic importance in quantifying myocardial FDG metabolism lies in the CoV. This measure of heterogeneity in myocardial glucose uptake can now be considered a unique metric that is both diagnostic (3) and prognostic in this disease process.
Using a semiquantitative SRS is a familiar approach for the nuclear community in the interpretation of both single photon emission computed tomography and PET imaging, because it has been used for coronary artery disease for 3 decades in the prediction of adverse outcomes (4,12,15). As in coronary artery disease, we found that using elements of this detailed semiquantitative scoring system provides incremental prognostic data in sarcoidosis. It is important to note that the exclusion of obstructive coronary artery disease in CS is essential, because perfusion and metabolism abnormalities are more classically seen in ischemic heart disease.
Early studies demonstrated fasting FDG-PET abnormalities in patients with CS (16) as well as a higher sensitivity than gallium, technetium, and cardiac magnetic resonance imaging (CMR) in detecting disease with JMHW criteria as the reference standard (17,18). Using the 17-segment model, our institution reported previously that patients with a mismatched defect in >6% of the myocardium had clinically active disease (19). However, only 1 study has described subsequent cardiovascular outcomes in patients having undergone FDG-PET imaging. Blankstein et al. (2) studied 118 consecutive patients who underwent FDG-PET for suspected CS and graded imaging findings in a 3-category system, as described. They found that FDG-PET demonstrating abnormal perfusion and metabolism in any distribution was significantly associated with outcomes as compared with normal FDG-PET. However, this study is limited due to the binary coding of imaging findings, sample size, and lack of robust multivariable adjustment. Particularly, we used stratified models by the presence of an ICD. ICD stratification is needed in this type of analysis, because the baseline hazard of developing an adverse event (one of which is aborted ventricular arrhythmia) is expected to be different among patients with and without an ICD.
Disease mechanisms and future directions
These results aid in our understanding of the mechanism of adverse events in CS. The lifecycle of a sarcoid granuloma as captured by FDG-PET cardiac imaging likely begins with early active inflammation, visualized as abnormal FDG uptake and normal perfusion. We see this anecdotally, because areas of mismatch tend to be surrounded by a rim of normally perfused tissue with abnormal FDG uptake. The second stage of progression is the mismatch pattern. The third stage is a burned-out granuloma or “scar” with decreased perfusion and lack of FDG uptake. Our finding of prognostic importance in patients with this second stage of perfusion–metabolism mismatch and a high CoV may be due to the arrhythmogenicity of actively inflamed myocardium (because most events were ventricular arrhythmia requiring defibrillation), although this study was not designed to test that hypothesis specifically.
Although segments of abnormal FDG uptake without corresponding perfusion defect have been postulated to represent early active inflammation (2), a summed score of these segments was not predictive of outcomes in our cohort. Neither was the mean or maximum global myocardial SUV. It is likely that the inability to suppress myocardial glucose metabolism sometimes accounts for “false-positive” FDG uptake, particularly when the SRS is low. Amendments to the standard viability protocol were undertaken to minimize intrinsic FDG uptake. We noted that often the lateral wall shows increased FDG signal in the setting of normal perfusion, which was also described in prior studies and is of unclear significance (2,20,21). This theory supports the greater prognostic value of the CoV metric compared with semiquantitative or quantitative assessments of global SUV. Patients with heterogeneous FDG uptake likely have truly pathologic myocardial glucose metabolism as opposed to false-positive uptake from inadequate glucose suppression.
These FDG-PET findings may identify high-risk patients who would benefit from an increase in medical therapy and potentially an ICD for primary prevention. However, a normal study without perfusion or metabolism abnormalities should not be used to rule out potential future events or the need for ICD. In our cohort, 20% of patients with normal FDG-PET findings went on to have adverse events, speaking to the inherent high-risk nature of the population undergoing testing. Clinical discretion should be used with the knowledge that these particular FDG-PET findings are associated with higher event rates.
Additional uses for FDG-PET imaging in sarcoidosis are evolving. The true sensitivity, specificity, and diagnostic accuracy of FDG-PET in CS remains unknown, because a reliable gold standard for diagnosis does not exist. In patients with biopsy-proven CS and heart failure, FDG-PET and CMR were all abnormal and consistent with the disease (22); however, these results should not be extrapolated to patients with more subtle forms of cardiac involvement. FDG-PET may also be useful in following disease patterns over time and in response to particular therapies. In addition, combined PET and CMR may be useful for scar quantification by late gadolinium enhancement while delivering a lower overall radiation burden for patients (17).
This study is a single-center cohort and generalizability is limited due to local experience, image interpretation, and imaging protocols. Although our cohort consisted of patients for whom there was a clinical concern for CS, not all patients seen in our center underwent FDG-PET imaging. Patients may have been managed medically without imaging or with CMR as the imaging approach. Practice patterns commonly referred patients to FDG-PET in cases of an unclear diagnosis, if initial management was unsuccessful, or to determine the success of medical management. FDG is a highly nonspecific marker for inflammation related to any etiology (i.e., myocarditis) and may be associated with death, heart transplantation, or ventricular arrhythmias in these patients as well. Although our study represents the largest to date, this cohort remains relatively small due to the uncommon nature of the disease and limits robust variable adjustment. Although several methods to assess FDG uptake were used (summed score, maximum SUVs, mean SUVs, CoV), other groups have tested left ventricular FDG volume, activity, and volume above certain SUV thresholds (23,24).
As mentioned, some patients may not have adhered to dietary recommendations, causing increased myocardial glucose uptake and decreased image scoring accuracy. However, this limitation is in line with what occurs in standard clinical practice. Intravenous heparin may aid in converting the myocardium to fatty acid metabolism, thus decreasing spurious FDG uptake. However, this was not part of our protocol because heparin is not rigorously studied and data are mixed. Therefore, the particular CoV values described in this study may be influenced by different fasting and patient preparation protocols.
In the analysis of FDG-PET imaging in sarcoidosis, defining the extent and severity of disease is important prognostically and superior to standard 3-category scoring. Both the SRS in segments with abnormal FDG uptake (mismatched pattern) and the CoV of FDG uptake (representing heterogeneity of myocardial metabolism) are significantly associated with the composite outcome of all-cause mortality, heart transplantation, or ventricular arrhythmia requiring defibrillation. Our results support the use of a more detailed analysis of FDG-PET in the assessment of CS, similar to the conventional method used for interpretation in coronary artery disease.
COMPETENCY IN MEDICAL KNOWLEDGE: The extent and severity of perfusion and metabolism mismatch in patients undergoing FDG-PET for suspected CS is important prognostically after risk adjustment. In addition, the heterogeneity of FDG uptake, as measured by the CoV, is also associated with adverse events.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: These results support the use of a more detailed analysis of the extent and severity of FDG-PET defects in the assessment of CS, similar to the conventional method used for the interpretation of radionuclide imaging in coronary artery disease.
TRANSLATIONAL OUTLOOK: Because there are several potential therapeutic treatments for CS, future studies may attempt to assess the differential prognostic usefulness among regimens.
Dr. Tamarappoo is currently affiliated with the Department of Cardiovascular Medicine, Cedars-Sinai Heart Institute, Los Angeles, California.
Dr. Culver has received clinical trial support from Mallinkrodt; and has received consultant fees from Gilead. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Sperry, Tamarappoo, and Oldan contributed equally to this work.
- Abbreviations and Acronyms
- cardiac magnetic resonance
- coefficient of variation
- cardiac sarcoidosis
- 18F-labeled fluorodeoxyglucose
- implantable cardioverter-defibrillator
- Japanese Ministry of Health and Welfare
- positron emission tomography
- summed rest score
- segmental uptake values
- Received January 25, 2017.
- Revision received April 8, 2017.
- Accepted April 20, 2017.
- 2018 American College of Cardiology Foundation
- Birnie D.H.,
- Nery P.B.,
- Ha A.C.,
- Beanlands R.S.
- Tahara N.,
- Tahara A.,
- Nitta Y.,
- et al.
- Hachamovitch R.,
- Hayes S.W.,
- Friedman J.D.,
- Cohen I.,
- Berman D.S.
- Lang R.M.,
- Badano L.P.,
- Mor-Avi V.,
- et al.
- Soejima K.,
- Yada H.
- Dilsizian V.,
- Bacharach S.L.,
- Beanlands R.S.,
- et al.
- Ficaro E.P.,
- Lee B.C.,
- Kritzman J.N.,
- Corbett J.R.
- Cerqueira M.D.,
- Weissman N.J.,
- Dilsizian V.,
- et al.
- Schuller J.L.,
- Zipse M.,
- Crawford T.,
- et al.
- Hachamovitch R.,
- Berman D.S.,
- Kiat H.,
- et al.
- Yamagishi H.,
- Shirai N.,
- Takagi M.,
- et al.
- Maurer A.H.,
- Burshteyn M.,
- Adler L.P.,
- Gaughan J.P.,
- Steiner R.M.
- Gropler R.J.,
- Siegel B.A.,
- Lee K.J.,
- et al.
- Sperry B.W.,
- Oldan J.,
- Hachamovitch R.,
- Tamarappoo B.K.
- Ahmadian A.,
- Brogan A.,
- Berman J.,
- et al.
- Osborne M.T.,
- Hulten E.A.,
- Singh A.,
- et al.