OPENACCESS Cardiac perfusion PET is increasingly used to assess ischemia and cardiovascular risk and Citation: Guerraty MA, Rao HS, Anjan VY, Szapary can also provide quantitative myocardial blood flow (MBF) and flow reserve (MBFR) values. H, Mankoff DA, Pryma DA, et al. (2020) The role of These have been shown to be prognostic biomarkers of adverse outcomes, yet MBF and resting myocardial blood flow and myocardial blood flow reserve as a predictor of major adverse MBFR quantification remains underutilized in clinical settings. We compare MBFR to tradi- cardiovascular outcomes. PLoS ONE 15(2): tional cardiovascular risk factors in a large and diverse clinical population (60% African- e0228931. https://doi.org/10.1371/journal. American, 35.3% Caucasian) to rank its relative contribution to cardiovascular outcomes. pone.0228931 Major adverse cardiovascular events (MACE), including unstable angina, non-ST and ST- Editor: Dalin Tang, Southeast University, CHINA elevation myocardial infarction, stroke, and death, were assessed for consecutive patients Received: August 13, 2019 who underwent rest-dipyridamole stress 82Rb PET cardiac imaging from 2012–2015 at the Accepted: January 26, 2020 Hospital of the University of Pennsylvania (n = 1283, mean follow-up 2.3 years). Resting MBF (1.1± 0.4 ml/min/g) was associated with adverse cardiovascular outcomes. MBFR Published: February 13, 2020 (2.1± 0.8) was independently and inversely associated with MACE. Furthermore, MBFR Copyright:© 2020 Guerraty et al. This is an open was more strongly associated with MACE than both traditional cardiovascular risk factors access article distributed under the terms of the Creative Commons Attribution License, which and the presence of perfusion defects in regression analysis. Decision tree analysis identi- permits unrestricted use, distribution, and fied MBFR as superior to established cardiovascular risk factors in predicting outcomes. reproduction in any medium, provided the original Incorporating resting MBF and MBFR in CAD assessment may improve clinical decision author and source are credited. making. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files. Funding: MG was supported NIH T32HL007843 and 1K08HL136890-01 from the National Institutes of Health (www.nih.gov). The funders had no role Introduction in study design, data collection and analysis, decision to publish, or preparation of the There is growing interest in coronary microvascular disease (CMVD) which has been associ- manuscript. ated with worsened cardiovascular outcomes  and may be involved in the pathogenesis of heart failure [2, 3]. PET (positron emission tomography) has been used extensively for myocar- Competing interests: The authors have declared that no competing interests exist. dial perfusion imaging (MPI) given superior imaging characteristics compared to SPECT PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 1 / 13 Myocardial Blood Flow Reserve and Cardiovascular Outcomes Abbreviations: CAD, Coronary Artery Disease; (single photon emission computed tomography) . In addition to enhanced MPI, cardiac CHF, Congestive Heart Failure; CKD, Chronic PET can provide quantitative myocardial blood flow (MBF) and flow reserve (MBFR) values, Kidney Disease; CMVD, Coronary Microvascular defined as the ratio of hyperemic MBF to resting MBF . In the absence of epicardial coronary Disease; MBF, Myocardial Blood Flow; MBFR, artery disease, MBFR is a measure of coronary microvascular function. Low MBFR has been Myocardial Blood Flow Reserve; MPI, Myocardial shown to correlate with worsened outcomes in various populations, including patients with Perfusion Imaging; PAD, Peripheral Artery Disease; PET, Positron Emission Tomography. diabetes, renal disease, and obesity and patients without coronary artery disease (CAD) [6–10]. With the availability of several commercial software packages to quantify MBF and MBFR, PET has reproducibly shown value as a prognostic biomarker of adverse outcomes [11–15]. Yet MBFR quantification remains clinically underutilized. Likely reasons for this include limited availability of imaging modalities that allow the quantification of MBFR; limited expe- rience in understanding the pathophysiology and interpretation of decreased MBFR; nascent understanding of the risk factors that contribute to MBFR; and unclear understanding of the role of MBFR as an independent cardiovascular risk factor. In a general population, the inter- pretation of MBFR is often confounded by epicardial coronary artery disease, which can affect regional radioactive tracer uptake, and therefore, regional blood flow calculations . To provide clinical context and understand the relationship between established cardiovas- cular risk factors and MBFR in a large and diverse clinical population, we examined consecu- tive patients referred for cardiac Rubidium-82 (82Rb) PET at the Hospital of the University of Pennsylvania, a tertiary referral center, over a 2.3 year period for risk factors and cardiovascu- lar outcomes. We hypothesized that MBFR has superior predictive value for cardiovascular events beyond that provided by traditional cardiovascular risk factors (such as hypertension and diabetes) or perfusion defects. We performed regression analysis and unbiased decision- tree and binary-discretization analyses to understand the relationships between cardiovascular risk factors, MBFR, and cardiovascular outcomes and rank the relative contributions of MBFR and cardiovascular risk factors to outcomes. Materials and methods All patients undergoing clinical cardiac PET 82Rb myocardial perfusion imaging at the Hospi- tal of the University of Pennsylvania, an urban tertiary care center, between 2/2012 and 4/2015 were retrospectively examined in the study. Patients were included in the regression and out- comes analyses if they had both resting and stress MBF measurements. Heart transplant patients represent a unique population and were excluded from adjusted and survival analyses. The patient records were followed through 1/2016 with mean follow-up of 2.3 years. Cardio- vascular outcomes were assessed blinded to MBFR and by manual chart review of the Univer- sity of Pennsylvania Health Systems Electronic Health Record. They include unstable angina, non-ST elevation myocardial infarction, ST-elevation myocardial infarction, stroke, or death, and were identified in one of three ways: described in a physician visit note or discharge sum- mary, identified by brain and coronary imaging for acute stroke and MI, respectively; or assessed by ICD-9 code following the time of PET scan. The study was approved by the Uni- versity of Pennsylvania Institutional Review Board, and no informed consent was required for this retrospective study using data from the Electronic Health Record. Patients underwent a rest-dipyridamole stress 82Rb cardiac PET using Siemens Biograph mCT PET/CT scanner. Briefly, low dose CT images were acquired for photon attenuation correction. Rest images were obtained with a 6-minute list-mode dynamic PET acquisition imaging while 30mCi of 82Rb were injected intravenously as a fast bolus. Dipyridamole (0.56 mg/kg) was then administered, and three minutes after completion of dipyridamole infu- sion, dynamic PET imaging was repeated with an additional 30 mCi of 82Rb. Iterative recon- struction was performed with 2 iterations and matrix size 128 x 128. PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 2 / 13 Myocardial Blood Flow Reserve and Cardiovascular Outcomes Global and regional MBF and MBFR were calculated using syngo MBF software (Siemens Healthcare, Germany). The software uses the data from list-mode acquisition to determine time-activity curves for blood pool and myocardium. The data are fit into a one-compartment model of 82Rb kinetics with non-linear extraction curve to calculate global and regional MBF . This methodology including variable extraction fraction has been validated and adjusted using microsphere studies including direct comparison with 82Rb in a porcine model [17–19]. The syngo platform calculates the blood input function by identifying maximum activity points in the late summed image and performing subsequent LV motion correction. Assuming an intra-ventricular cylindrical-spherical shape model, tissue uptake time activity curves are generated from the maximum activity points obtained. There is minimal inter-and intra- observer variability in using the software . Absolute numbers and percentages are used to describe the patient population. Continuous variables are expressed as mean +/- standard deviation and were assessed for normality using Kolmogorov-Smirnov test [20, 21]. Variables that were not normally distributed were log- transformed. Student’s T-test was used to compare groups, and P-values<0.05 were consid- ered significant. Adjusted and unadjusted regression modeling and Kaplan Meier survival analysis were performed using R version 3.2.3 [22, 23]. Adjusting MBF by heart rate-blood pressure (HRBP) product did not alter the results, and therefore the unadjusted MBF values are presented. Discretization analysis was performed using entropy-based approach . Deci- sion tree analysis was performed using weka-3-13. Risk factors examined included age, gender, race, BMI, global rest MBF, global MBFR, the presence of perfusion defects, diabetes, hyper- cholesterolemia, family history of cardiac disease, obstructive sleep apnea, presence of hyper- tension, CKD, renal transplant, and tobacco use . Results Patient demographics are summarized in Table 1 and illustrate the age (58 ± 12.1), gender (54.9% female), and racial diversity (60% African-American) of the population. The popula- tion reflects institutional referral practices where PET myocardial imaging perfusion is pre- ferred to SPECT imaging for high BMI and high-risk patients in our center. This referral bias leads to a population with high BMI (36.4 ± 10.0), a high incidence of cardiovascular risk fac- tors, and many patients who had undergone renal or heart transplant. Additionally, a signifi- cant proportion had a history of CAD (41.1%) by ICD-9 diagnosis code. Given the higher rate of medical comorbidities, the rate of catheterization was 25% and the rate of cardiovascular events over the median follow-up of 27.6 months was 5%. Since CMVD often coexists with CAD, we included all patients with adequate MBFR mea- surements, regardless of their CAD status or the presence of perfusion defects, in examining the risk factors and disease processes that are associated with decreased MBFR (S1 Table). The unadjusted regression model reveals that traditional cardiovascular risk factors such as age, hypertension, diabetes, and hypercholesterolemia, as well as corresponding lab values, are associated with decreased MBFR. Gender was not associated with MBFR, though women had higher resting MBF than men (S1 Fig). Reduced ejection fraction and increased left-ventricular end-diastolic volume by gated acquisition were associated with decreased MBFR. Cardiovascu- lar diseases such as CAD, CHF, stroke, and PAD were also associated with decreased MBFR. Though the diagnosis of hypercholesterolemia was associated with decreased MBFR, higher total cholesterol and LDL levels were associated with higher MBFR. Statin usage may contrib- ute to this discrepancy. Of note, the presence of heart transplant was not associated with decreased MBFR. The risk factors associated with MBFR were used in an adjusted regression model (S2 Table). Traditional cardiovascular risk factors are associated with lower MBFR, as PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 3 / 13 Myocardial Blood Flow Reserve and Cardiovascular Outcomes Table 1. Patient characteristics. n (%) or mean +/- SD Age, y 58 ± 12.1 Gender Male 579 (45.1) Female 704 (54.9) Race African-American 768 (60) Caucasian 453 (35.3) Asian 23 (1.8) Hispanic 19 (1.5) Other/unknown 20 (1.6) Body Mass Index 36.4 ± 10.0 Hypertension 1065 (83) Diabetes Mellitus 581 (45.3) Hypercholesterolemia 820 (64.0) Coronary Artery Disease 528 (41.1) Congestive Heart Failure 362 (28.2) Stroke 115 (9.0) Peripheral Artery Disease 82 (6.4) Chronic Kidney Disease 445 (34.7) Family History of Heart Disease 101 (7.9) Renal Transplant 178 (13.9) Heart Transplant 202 (15.7) Tobacco Use Never 478 (37.2) Former 496 (38.7) Current 149 (11.6) Other/Unknown 160 (12.5) Indications Chest pain 684 (53.3) Pre-kidney transplant 112 (8.7) Pre-operative assessment 121 (9.4) Post-heart transplant 69 (5.4) Cardiomyopathy or CHF 33 (2.6) To evaluate known CAD 35 (2.7) Arrythmia 23 (1.8) Syncope/Dizziness 21 (1.6) Abnormal EKG 12 (1.0) Other/Unknown 169 (13.2) Laboratory Values Hemoglobin (g/dL) 12.7 ± 2 Hemoglobin A1c (%) 6.8 ± 1.8 Glucose (mg/dL) 125.5 ± 40.3 Creatinine (mg/dL) 1.6 ±1.8 Estimated GFR (ml/min/1.73m2) 33.9 ± 18.5 Pro-B Natriuretic Peptide (ng/L) 1392.1 ± 3095.9 Total Cholesterol (mg/dL) 168.8 ± 43.7 LDL Cholesterol (mg/dL) 95.7 ± 45.3 HDL Cholesterol (mg/dL) 46.3 ± 15.3 (Continued ) PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 4 / 13 Myocardial Blood Flow Reserve and Cardiovascular Outcomes Table 1. (Continued ) n (%) or mean +/- SD Non-HDL Cholesterol (mg/dL) 120.9 ±40.7 Triglycerides (mg/dL) 139.5 ± 90.6 Radioisotope Dose Rest Dose (mCi) 27.8 ± 1.6 Stress Dose (mCi) 27.7 ± 2.1 Perfusion No defect 996 (78.0) Fixed or reversible defect 281 (22.0) Global Myocardial Blood Flow Rest (ml/min/g) 1.1 ± 0.4 Stress (ml/min/g) 2.2 ± 0.8 Reserve 2.1 ± 0.8 https://doi.org/10.1371/journal.pone.0228931.t001 were CHF and PAD. Interestingly, non-Caucasian race was independently associated with higher MBFR in both the adjusted and unadjusted models (S1 and S2 Tables, S1 Fig). We then examined the relationship between MBFR, cardiovascular risk factors and disease, and outcomes. Decreased MBFR was associated with increased cardiovascular outcomes in an unadjusted analysis (Table 2, Fig 1A, and S3 Fig). Interestingly, increased resting MBF was Table 2. Unadjusted regression analysis of cardiovascular outcomes and risk factors. Risk Factor Odds Ratio P-value PET parameters MBFR 0.56±0.08 7.95e-5 Rest MBF 1.98±0.42 1.46e-3 Stress MBF 0.86±0.11 0.21 Perfusion 0.48±0.10 3.54e-4 Demographics Age 1.03±0.01 8.80e-5 Gender 1.52±0.29 0.03 Race 0.71±0.11 0.03 Body Mass Index 0.95±0.01 1.06e-6 Smoking Status 1.03±0.10 0.76 Cardiovascular Risk Factors Diabetes 1.62±0.31 0.01 Hypercholesterolemia 1.83±0.41 6.38e-3 Obstructive Sleep Apnea 1.09±0.22 0.67 Hypertension 3.49±1.39 1.64e-3 Family History of Cardiac Disease 1.21±0.41 0.57 Chronic Kidney Disease 2.21±0.43 4.31e-5 Renal Transplant 2.31±0.53 2.57e-4 Cardiovascular Diseases History of CAD 3.71±0.77 3.22e-10 Congestive Heart Failure 3.12±0.61 5.64e-9 History of Stroke 4.03±0.97 7.58e-9 Peripheral Artery Disease 2.58±0.77 1.44e-3 Heart Transplant 1.24±0.31 0.40 https://doi.org/10.1371/journal.pone.0228931.t002 PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 5 / 13 Myocardial Blood Flow Reserve and Cardiovascular Outcomes Fig 1. Relationship between MBFR and outcomes. Lower MBFR was associated with increased major adverse cardiovascular outcomes (MACE) (A). This relationship was preserved in patients with no history of CAD by ICD-9 code (B), and there was a trend towards increased MACE in patients with a history of CAD by ICD-9 code (C). https://doi.org/10.1371/journal.pone.0228931.g001 itself associated with adverse outcomes, though stress MBF was not (Table 2, S2 Fig). Though this result was irrespective of HRBP product, there was an association between resting MBF and hemoglobin (S2 Fig). Since the presence of CAD can confound MBFR measurements, we examined the association between MBFR and outcomes in patients with and without CAD by PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 6 / 13 Myocardial Blood Flow Reserve and Cardiovascular Outcomes Table 3. Adjusted regression analysis for outcomes by strength of significant association. Odds Ratio P-value History of Stroke 2.58±0.68 3.24e-4 History of CAD 2.41±0.57 1.93e-4 Congestive Heart Failure 1.90±0.42 3.95e-3 Renal Transplant 1.83±0.46 0.02 MBFR 0.72±0.11 0.03 Body Mass Index 0.95±0.11 3.48e-5 Hypertension 1.99±0.90 0.13 https://doi.org/10.1371/journal.pone.0228931.t003 ICD-9 code. MBFR was associated with outcomes in patients without a history CAD, and we found a similar trend in patients with a history of CAD which was not statistically significant (Fig 1B and 1C). In the adjusted regression model, MBFR was independently and inversely associated with increased cardiovascular outcomes (Table 3, odds ratio 0.72 ± 0.11, p = 0.03). This effect was stronger than that seen with traditional cardiovascular risk factors, such as hypertension and high BMI, but weaker than that of known cardiovascular disease such as CAD, PAD, or stroke. Fur- thermore, the presence of perfusion defects, which was associated with cardiovascular outcomes in the unadjusted model, was not statistically significant when adjusted for a history of CAD. Discretization analysis identified an optimal MBFR cutoff of 1.35 for our cohort of patients which was associated with increased cardiovascular outcomes. Kaplan-Meier curves for MBFR <1.35 and > 1.35 highlight the difference in event-free survival for our cohort (Fig 2A). Cost analysis further illustrates how using discretization analysis identifies an MBFR cut-point that maximizes the number of correct predictions (Fig 2B). MBFR of 2 has been used as a cutoff between normal and abnormal MBFR, and there was a significant difference in outcomes with MBFR of 2 (S3 Fig) . Fig 2. Kaplan-Meier and cost analysis for MBFR thresholds. Kaplan-Meier curve showing decreased freedom from events in patients with MBFR < 1.35 relative to MBFR > 1.35 based on discretization analysis (A). Cost analysis illustrates how the discretization analysis cutpoint maximizes the number of true positives and true negatives (B). https://doi.org/10.1371/journal.pone.0228931.g002 PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 7 / 13 Myocardial Blood Flow Reserve and Cardiovascular Outcomes Fig 3. Decision tree analysis. Several risk factors and imaging characteristics were used for the analysis (A). Global reserve was identified in an unbiased way as the first branch off point between outcomes and no outcomes (B). The second most significant factor in outcomes was a history of renal transplant. Rectangles reflect the number of cases that are separated with each branch point node and the number incorrectly classified, if any. Renal transplant = 1 history of renal transplant. Gender = 1 represents male and Gender = 0 represents female. Smoking status is as follows: non- smoker = 0, former smoker = 1, current smoker = 2, unknown smoking status = 3. https://doi.org/10.1371/journal.pone.0228931.g003 Decision tree analysis is a machine learning method which ranks the contribution of risk factors to an outcome. PET imaging parameters, demographics, and risk factors were used for the analysis (Fig 3A). The presence of overt cardiovascular disease such as CAD, CHF, PAD, and stroke were excluded for the analysis since regression analysis showed that they are more strongly associated with outcomes than risk factors. The analysis identified that an MBFR cutoff of 1.35 was the most divisive risk factor between participants that did have an adverse outcome and those that did not in our cohort (Fig 3B). That is, MBFR more accurately differ- entiated between patients that had cardiovascular outcomes and those that did not compared to all other risk factor. The next branch point was whether a patient had a history of renal transplant. For patients that had low MBFR and had a history of renal transplant, factors such as increased age, male gender, and smoking status were associated with outcomes. Further- more, other risk factors known to be associated with cardiovascular outcomes, such as hyper- tension, diabetes, and perfusion defects on myocardial perfusion imaging, did not appear in the decision tree. Discussion We developed a large database of patients (n = 1283) who underwent clinically-indicated car- diac 82Rb PET to examine the association between MBFR, cardiovascular risk factors, and car- diovascular outcomes. MBFR has been difficult to understand and contextualize because of its dichotomous nature; reduced MBFR can represent both a cardiovascular disease, as in ische- mia with non-obstructive coronary arteries (INOCA), and a cardiovascular risk factor that modifies prognosis of cardiovascular disease . In the absence of epicardial coronary artery disease, MBFR represents microvascular function and low MBFR represents CMVD. We examined risk factors that affect MBFR and PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 8 / 13 Myocardial Blood Flow Reserve and Cardiovascular Outcomes identified hypertension, diabetes, and age as significant risk factors that affect MBFR in an adjusted model. These are known risk factors for CMVD . Additionally, our data showed that African-Americans had increased MBFR relative to Caucasians. There have been similar signals in other studies. For example, African Americans were found to have invasive coronary flow reserve measurements of 4.4 ± 2.3 whereas Caucasian had coronary flow reserve measure- ments of 4.1 ± 2 . However, this is the first study to robustly show this difference with large number of African American patients using cardiac perfusion PET measurements. The signifi- cance of this finding on long-term outcomes in African-Americans will require additional studies. We also found that women had increased resting MBF relative to men, which is con- sistent with prior reports . Given the association between resting MBF and outcomes, this result warrants future work to understand the pathophysiology of this finding and relevance to long-term outcomes. In unadjusted analyses for MBFR, CHF was strongly associated with MBFR. Though our study did not differentiate between heart failure with preserved ejection fraction (HFPEF) and heart failure with reduced ejection fraction (HFrEF), there is evidence that microvascu- lar health contributes to the pathogenesis of heart failure [2, 30]. There is likely an interplay between CMVD contributing to heart failure and the disease processes that lead to heart fail- ure causing CMVD. MBFR has also been shown to be correlated to outcomes in patients with heart failure . Further quantification and understanding of microvascular disease in these specific populations will provide additional insight into disease pathogenesis and prognosis. Depressed MBFR was also found to be independently and strongly associated with adverse cardiovascular outcomes, in agreement with multiple prior studies that have established the association between MBFR and cardiovascular outcomes in various populations [6–10]. In our population, this result was driven predominantly by increased all-cause mortality, which may be a reflection of the high burden of medical comorbidities in this population. The etiol- ogy of the link between decreased MBFR and outcomes is still unclear. Though it is possible that that low MBFR increased the risk of death via a direct cardiovascular process, such as myocardial infarction, it is also possible that decreased MBFR serves as a systemic marker of disease, capturing lifetime exposure to systemic risk factors and the association is therefore indirect. This dichotomy is reflected in the strength of association with outcomes, where MBFR sits between known cardiovascular disease and traditional risk factors. More specifically, in adjusted regression analysis, MBFR was more strongly associated with outcomes than tradi- tional cardiovascular risk factors and less so than having overt atherosclerotic disease. A poten- tial hypothesis for this finding is that MBFR may reflect the cumulative effect of exposure to certain risk factors such as age, diabetes, hypertension, and hypercholesterolemia. MBFR remained independently associated with outcomes after adjusting for history of CAD. Though perfusion defects were associated with outcomes in the unadjusted model, the association dis- appeared when the analysis adjusted for history of CAD. This suggests that MBFR gives addi- tional information about the risk of outcomes and complements a diagnosis of CAD or the presence of perfusion defects. Additionally, we found that increased rest MBF was also associated with adverse outcomes. There is evidence that the decreased MBFR often seen in CMVD may be due to increased resting MBF rather than decreased stress MBF . We found this to be true in our data as well, even after correcting for HRBP product. One potential explanation is that patients with increased resting MBF had lower hemoglobin levels, which is itself associated with poor out- comes. A second potential hypothesis is that microvascular disease, which includes luminal obstruction and basement membrane thickening , impairs adequate oxygen transport PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 9 / 13 Myocardial Blood Flow Reserve and Cardiovascular Outcomes from the vessel lumen to the cardiomyocytes even under basal conditions. The cardiomyocytes may compensate by increasing MBF in hopes of improving oxygen transport, and elevated resting MBF may be evidence of these structural changes. We next used two unbiased approaches to examine the relationship between MBFR and risk factors. Decision tree analysis determined that MBFR is the most divisive risk factor rela- tive to other cardiovascular risk factors in separating patients who will and will not have out- comes. Furthermore, other risk factors known to be associated with outcomes did not appear in the decision tree. This suggests that traditional cardiovascular risk factors are not as strongly associated with outcomes as MBFR and their effect is captured within the association between MBFR and outcomes. We also used discretization analysis to identify an MBFR cut-point that best divides patients that did and did not have cardiovascular outcomes. This analysis is independent of the deci- sion tree algorithm and identified the same cutoff of 1.35. Our analysis, as well as others, show that lower MBFR is associated with worsened outcomes regardless of cut-point . Though the discretization analysis is specific to both our patient population and a mean follow-up of 2.3 years, it represents the application of unbiased analysis to partition a continuous variable into two groups. If prospectively validated, this tool could be used by individual institutions to determine appropriate cutoffs to report increased risk of cardiovascular outcomes. It can additionally be used to identify thresholds of MBFR for future studies that would select at risk populations for interventions or therapies. When depressed MBFR is viewed as a disease, determination of a threshold for intervention is important to inform the risk and benefits of directing patient care. MBF and MBFR were assessed using 82Rb, which is a convenient tracer for clinical use given short half-life and the ability to purchase generator which obviates need of nearby cyclo- tron. However, it has longer positron range and lower extraction fraction relative to oxygen-15 and N-13 ammonia . Additional limitations include single center and retrospective nature of the study, a referral bias that enriches the study population for cardiovascular disease and therefore cardiovascular outcomes. Future studies will examine the value of MBFR as a pro- spective predictor of outcomes and determine whether using an MBFR-guided strategy for treatment and cardiovascular risk reduction could affect outcomes. Studies of serial MBFR measurements would also contribute to a better understanding of the progression of CMVD and inform treatment efficacy. For example, statins have been shown to improve MBFR in a short term study , but the clinical implications of this improvement, and the benefit of lon- ger term pharmacologic treatment is still unknown. Though obtaining MBFR measurements clinically is currently limited to PET scanners, there is potential to use current SPECT technol- ogy to obtain analogous measurements which would further broaden the clinical utility of MBFR . In summary, we used a large and racially diverse clinical population that underwent 82Rb cardiac PET to examine risk factors for decreased MBFR and the association between MBFR and cardiovascular outcomes. We found that rest MBF and MBFR are associated with cardio- vascular outcomes. We further stratified MBFR as more strongly associated with outcomes than other established cardiovascular risk factors and less strongly associated with outcomes than established cardiovascular disease. These findings expand the existing literature on MBF and MBFR to a diverse and clinically sicker population and also show that decreased MBFR is more strongly associated with adverse cardiovascular outcomes than traditional risk factors like hypertension, diabetes, and obesity. This suggests that obtaining MBFR measurements may provide additional prognostic information beyond traditional cardiovascular risk factors. Future studies are needed to understand and parse out the contributions of CAD and CMVD to MBFR measurements. PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 10 / 13 Myocardial Blood Flow Reserve and Cardiovascular Outcomes Supporting information S1 Fig. Gender and racial differences in MBF and MBFR. Women had higher resting MBF than men (A). Caucasian participants had lower MBFR than black participants (B). (TIF) S2 Fig. The association between rest MBF, outcomes, and hemoglobin. Elevated Rest MBF was associated with increased cardiovascular outcomes (A) and lower levels of hemoglobin (B). (TIF) S3 Fig. MBFR Kaplan Meier curves. There is a stepwise change in survival between patients with lower MBFR of 1.6 (A) and 2.0 (B). MBFR cutoff of 2 is often used to differentiation between normal and abnormal, and 1.6 is often used to determine significant disease. (TIF) S1 Table. Unadjusted regression analysis of risk factors and laboratory values associated with MBFR. (PDF) S2 Table. Adjusted regression model of MBFR and risk factors shows the cardiovascular risk factors and cardiovascular diseases that are associated with decreased MBFR. (PDF) S3 Table. Overview of adverse outcomes and association with MBFR. (PDF) Author Contributions Conceptualization: Marie A. Guerraty, Venkatesh Y. Anjan, David A. Mankoff, Daniel A. Pryma, Daniel J. Rader, Jacob G. Dubroff. Data curation: Marie A. Guerraty, Hannah Szapary. Formal analysis: Marie A. Guerraty, H. Shanker Rao, Jacob G. Dubroff. Investigation: Marie A. Guerraty, Venkatesh Y. Anjan, Hannah Szapary. Methodology: H. Shanker Rao, Jacob G. Dubroff. Project administration: Marie A. Guerraty. Supervision: Jacob G. Dubroff. Validation: Marie A. Guerraty. Visualization: H. Shanker Rao. Writing – original draft: Marie A. Guerraty, Jacob G. Dubroff. Writing – review & editing: Marie A. Guerraty, H. Shanker Rao, Venkatesh Y. Anjan, Han- nah Szapary, David A. Mankoff, Daniel A. Pryma, Daniel J. Rader, Jacob G. Dubroff. References 1. Bairey Merz CN, Pepine CJ, Walsh MN, Fleg JL. Ischemia and No Obstructive Coronary Artery Disease (INOCA): Developing Evidence-Based Therapies and Research Agenda for the Next Decade. Circula- tion. 2017; 135(11):1075–92. Epub 2017/03/16. https://doi.org/10.1161/CIRCULATIONAHA.116. 024534 PMID: 28289007 PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 11 / 13 Myocardial Blood Flow Reserve and Cardiovascular Outcomes 2. Mohammed SF, Hussain S, Mirzoyev SA, Edwards WD, Maleszewski JJ, Redfield MM. Coronary microvascular rarefaction and myocardial fibrosis in heart failure with preserved ejection fraction. Circu- lation. 2015; 131(6):550–9. https://doi.org/10.1161/CIRCULATIONAHA.114.009625 PMID: 25552356 3. Schiattarella GG, Altamirano F, Tong D, French KM, Villalobos E, Kim SY, et al. Nitrosative stress drives heart failure with preserved ejection fraction. Nature. 2019; 568(7752):351–6. Epub 2019/04/12. https://doi.org/10.1038/s41586-019-1100-z PMID: 30971818. 4. Mc Ardle BA, Dowsley TF, deKemp RA, Wells GA, Beanlands RS. Does rubidium-82 PET have supe- rior accuracy to SPECT perfusion imaging for the diagnosis of obstructive coronary disease?: A system- atic review and meta-analysis. J Am Coll Cardiol. 2012; 60(18):1828–37. https://doi.org/10.1016/j.jacc. 2012.07.038 PMID: 23040573. 5. Kajander SA, Joutsiniemi E, Saraste M, Pietila M, Ukkonen H, Saraste A, et al. Clinical value of absolute quantification of myocardial perfusion with (15)O-water in coronary artery disease. Circ Cardiovasc Imaging. 2011; 4(6):678–84. https://doi.org/10.1161/CIRCIMAGING.110.960732 PMID: 21926262. 6. Murthy VL, Naya M, Taqueti VR, Foster CR, Gaber M, Hainer J, et al. Effects of sex on coronary micro- vascular dysfunction and cardiac outcomes. Circulation. 2014; 129(24):2518–27. https://doi.org/10. 1161/CIRCULATIONAHA.113.008507 PMID: 24787469 7. Shah NR, Charytan DM, Murthy VL, Skali Lami H, Veeranna V, Cheezum MK, et al. Prognostic Value of Coronary Flow Reserve in Patients with Dialysis-Dependent ESRD. J Am Soc Nephrol. 2016; 27 (6):1823–9. Epub 2015/10/16. https://doi.org/10.1681/ASN.2015030301 PMID: 26459635 8. Chow BJ, Dorbala S, Di Carli MF, Merhige ME, Williams BA, Veledar E, et al. Prognostic value of PET myocardial perfusion imaging in obese patients. JACC Cardiovasc Imaging. 2014; 7(3):278–87. Epub 2014/02/25. https://doi.org/10.1016/j.jcmg.2013.12.008 PMID: 24560212. 9. Murthy VL, Naya M, Foster CR, Gaber M, Hainer J, Klein J, et al. Association between coronary vascu- lar dysfunction and cardiac mortality in patients with and without diabetes mellitus. Circulation. 2012; 126(15):1858–68. https://doi.org/10.1161/CIRCULATIONAHA.112.120402 PMID: 22919001 10. Murthy VL, Naya M, Foster CR, Hainer J, Gaber M, Di Carli G, et al. Improved cardiac risk assessment with noninvasive measures of coronary flow reserve. Circulation. 2011; 124(20):2215–24. https://doi. org/10.1161/CIRCULATIONAHA.111.050427 PMID: 22007073 11. Herzog BA, Husmann L, Valenta I, Gaemperli O, Siegrist PT, Tay FM, et al. Long-term prognostic value of 13N-ammonia myocardial perfusion positron emission tomography added value of coronary flow reserve. J Am Coll Cardiol. 2009; 54(2):150–6. Epub 2009/07/04. https://doi.org/10.1016/j.jacc.2009. 02.069 PMID: 19573732. 12. Fukushima K, Javadi MS, Higuchi T, Lautamaki R, Merrill J, Nekolla SG, et al. Prediction of short-term cardiovascular events using quantification of global myocardial flow reserve in patients referred for clini- cal 82Rb PET perfusion imaging. J Nucl Med. 2011; 52(5):726–32. Epub 2011/04/19. https://doi.org/10. 2967/jnumed.110.081828 PMID: 21498538. 13. Farhad H, Dunet V, Bachelard K, Allenbach G, Kaufmann PA, Prior JO. Added prognostic value of myo- cardial blood flow quantitation in rubidium-82 positron emission tomography imaging. Eur Heart J Cardi- ovasc Imaging. 2013; 14(12):1203–10. Epub 2013/05/11. https://doi.org/10.1093/ehjci/jet068 PMID: 14. Valenta I, Dilsizian V, Quercioli A, Ruddy TD, Schindler TH. Quantitative PET/CT measures of myocar- dial flow reserve and atherosclerosis for cardiac risk assessment and predicting adverse patient out- comes. Curr Cardiol Rep. 2013; 15(3):344. Epub 2013/02/12. https://doi.org/10.1007/s11886-012- 0344-0 PMID: 23397541. 15. Nesterov SV, Deshayes E, Sciagra R, Settimo L, Declerck JM, Pan XB, et al. Quantification of myocar- dial blood flow in absolute terms using (82)Rb PET imaging: the RUBY-10 Study. JACC Cardiovasc Imaging. 2014; 7(11):1119–27. Epub 2014/10/13. https://doi.org/10.1016/j.jcmg.2014.08.003 PMID: 16. Gerber BL, Melin JA, Bol A, Labar D, Cogneau M, Michel C, et al. Nitrogen-13-ammonia and oxygen- 15-water estimates of absolute myocardial perfusion in left ventricular ischemic dysfunction. J Nucl Med. 1998; 39(10):1655–62. Epub 1998/10/17. PMID: 9776263. 17. Lortie M, Beanlands RS, Yoshinaga K, Klein R, Dasilva JN, DeKemp RA. Quantification of myocardial blood flow with 82Rb dynamic PET imaging. Eur J Nucl Med Mol Imaging. 2007; 34(11):1765–74. Epub 2007/07/10. https://doi.org/10.1007/s00259-007-0478-2 PMID: 17619189. 18. Lautamaki R, George RT, Kitagawa K, Higuchi T, Merrill J, Voicu C, et al. Rubidium-82 PET-CT for quantitative assessment of myocardial blood flow: validation in a canine model of coronary artery steno- sis. Eur J Nucl Med Mol Imaging. 2009; 36(4):576–86. https://doi.org/10.1007/s00259-008-0972-1 PMID: 18985343. 19. Pan XB, Declerk, J. Validation syngo. PET Myocardial Blood Flow. https://static.healthcare.siemens. com/siemens_hwem-hwem_ssxa_websites-context-root/wcm/idc/groups/public/@us/documents/ PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 12 / 13 Myocardial Blood Flow Reserve and Cardiovascular Outcomes download/mda1/mtax/~edisp/mi-validation_syngo_pet_myocardial_blood_flow_100453198_1- 01986972.pdf. 20. Komsta L, Movomestky, F. moments: Moments, cumulants, skewness, kurtosis, and related tests. R package Version 0.14. http://CRAN.R-project.org/package=moments2015. 21. Gross J, Ligges, U. nortest: Tests for Normality. R package version 1.0–4. http://CRAN.R-project.org/ package=nortest2015. 22. Team RC. R: A language and environment for statistical computing. R Foundation for Statistical Com- puting. Vienna, Austria 2014. 23. Therneau T. A Package for Survival Analysis in S. version 2.38. 2015. 24. Grzymala-Busse JW. Discretization Based on Entropy and Multiple Scanning. Entropy-Switz. 2013; 15 (5):1486–502. 25. Frank F, Hall M.A., Witten I.H. Data Mining: Practical Machine Learning Tools and Techniques. 4th ed ed. Kaufmann M, editor 2016. 26. Gibson CM, Cannon CP, Daley WL, Dodge JT Jr., Alexander B Jr., Marble SJ, et al. TIMI frame count: a quantitative method of assessing coronary artery flow. Circulation. 1996; 93(5):879–88. https://doi.org/ 10.1161/01.cir.93.5.879 PMID: 8598078. 27. Camici PG, Crea F. Coronary microvascular dysfunction. The New England journal of medicine. 2007; 356(8):830–40. https://doi.org/10.1056/NEJMra061889 PMID: 17314342. 28. Houghton JL, Prisant LM, Carr AA, Flowers NC, Frank MJ. Racial differences in myocardial ischemia and coronary flow reserve in hypertension. J Am Coll Cardiol. 1994; 23(5):1123–9. Epub 1994/04/01. https://doi.org/10.1016/0735-1097(94)90600-9 PMID: 8144778. 29. Kobayashi Y, Fearon WF, Honda Y, Tanaka S, Pargaonkar V, Fitzgerald PJ, et al. Effect of Sex Differ- ences on Invasive Measures of Coronary Microvascular Dysfunction in Patients With Angina in the Absence of Obstructive Coronary Artery Disease. JACC Cardiovasc Interv. 2015; 8(11):1433–41. Epub 2015/09/26. https://doi.org/10.1016/j.jcin.2015.03.045 PMID: 26404195 30. Owan TE, Hodge DO, Herges RM, Jacobsen SJ, Roger VL, Redfield MM. Trends in prevalence and outcome of heart failure with preserved ejection fraction. The New England journal of medicine. 2006; 355(3):251–9. https://doi.org/10.1056/NEJMoa052256 PMID: 16855265. 31. Majmudar MD, Murthy VL, Shah RV, Kolli S, Mousavi N, Foster CR, et al. Quantification of coronary flow reserve in patients with ischaemic and non-ischaemic cardiomyopathy and its association with clini- cal outcomes. Eur Heart J Cardiovasc Imaging. 2015; 16(8):900–9. Epub 2015/02/27. https://doi.org/ 10.1093/ehjci/jev012 PMID: 25719181 32. Kato S, Saito N, Nakachi T, Fukui K, Iwasawa T, Taguri M, et al. Stress Perfusion Coronary Flow Reserve Versus Cardiac Magnetic Resonance for Known or Suspected CAD. J Am Coll Cardiol. 2017; 70(7):869–79. Epub 2017/08/12. https://doi.org/10.1016/j.jacc.2017.06.028 PMID: 28797357. 33. Murthy VL, Lee BC, Sitek A, Naya M, Moody J, Polavarapu V, et al. Comparison and prognostic valida- tion of multiple methods of quantification of myocardial blood flow with 82Rb PET. J Nucl Med. 2014; 55 (12):1952–8. Epub 2014/11/28. https://doi.org/10.2967/jnumed.114.145342 PMID: 25429160. 34. Maddahi J, Packard RR. Cardiac PET perfusion tracers: current status and future directions. Semin Nucl Med. 2014; 44(5):333–43. Epub 2014/09/23. https://doi.org/10.1053/j.semnuclmed.2014.06.011 PMID: 25234078 35. Baller D, Notohamiprodjo G, Gleichmann U, Holzinger J, Weise R, Lehmann J. Improvement in coro- nary flow reserve determined by positron emission tomography after 6 months of cholesterol-lowering therapy in patients with early stages of coronary atherosclerosis. Circulation. 1999; 99(22):2871–5. Epub 1999/06/09. https://doi.org/10.1161/01.cir.99.22.2871 PMID: 10359730. 36. Petretta M, Soricelli A, Storto G, Cuocolo A. Assessment of coronary flow reserve using single photon emission computed tomography with technetium 99m-labeled tracers. J Nucl Cardiol. 2008; 15(3):456– 65. https://doi.org/10.1016/j.nuclcard.2008.03.008 PMID: 18513652. PLOS ONE | https://doi.org/10.1371/journal.pone.0228931 February 13, 2020 13 / 13
PLoS ONE – Public Library of Science (PLoS) Journal
Published: Feb 13, 2020
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera