Abstract
A characteristic feature of obstructive coronary artery disease (CAD) is impaired myocardial blood flow (MBF). Previous studies have shown that atrial fibrillation (AF) is also associated with reduced hyperemic MBF. However, it is unknown whether an independent association exists between AF and MBF in symptomatic patients with suspected obstructive CAD. We studied 874 patients at Kuopio University Hospital (Finland) who underwent coronary computed tomography angiography (CCTA) and [15O]H2O myocardial perfusion positron emission tomography due to suspected obstructive CAD in 2017–2023. The final cohort comprised 31 AF patients and 120 propensity score-matched sinus rhythm (SR) patients. The matching criteria were age, sex, body mass index, number of traditional risk factors for CAD (0–5), calcium score, and total CCTA status (i.e., the overall severity of CCTA findings). Global hyperemic MBF was lower in AF patients than in SR patients (median 1.9 mL/g/min [IQR 1.4–2.6 mL/g/min] vs. 3.0 mL/g/min [IQR 2.4–3.6 mL/g/min], p < 0.001). Impaired hyperemic MBF (< 2.3 mL/g/min in ≥ 1 segment) was more frequent in AF patients (n = 25/31, 81%) than in SR patients (n = 50/120, 42%, p < 0.001). Furthermore, the presence of AF was independently associated with reduced hyperemic MBF in three multivariate general linear models (p < 0.001 in each model). To summarize, AF was independently associated with impaired hyperemic MBF in patients with suspected obstructive CAD. This association should be considered in clinical practice when interpreting the results of myocardial perfusion imaging in patients with AF.
Introduction
Despite advances in coronary artery disease (CAD) management, it is still a considerable health burden globally [1]. In recent years, the primary diagnostic methods of CAD have changed. For the initial diagnosis of CAD, the latest American and European guidelines recommend noninvasive imaging in most scenarios [2,3]. Among these methods, coronary computed tomography angiography (CCTA) has emerged as a widely used and practical modality. CCTA enables direct visualization of epicardial coronary arteries and shows the distribution and severity of coronary atherosclerosis. Due to its high sensitivity and negative predictive value, CCTA excels in ruling out CAD even in patients with high pretest probability [4]. CCTA demonstrates lower specificity for confirming obstructive CAD, however [4]. To increase diagnostic accuracy, CCTA can be combined with myocardial perfusion imaging (MPI). Among the various MPI modalities, positron emission tomography (PET) with [15O]H2O is considered the reference standard for absolute quantification of myocardial blood flow (MBF) [5]. However, impaired MBF is not specific to obstructed epicardial vessels. Instead, [15O]H2O PET MPI also reflects the perfusion of the coronary microvasculature [6].
Besides CAD, atrial fibrillation (AF), the most common sustained arrhythmia, causes significant morbidity, hospitalization, and mortality [7]. CAD and AF share risk factors and, additionally, CAD is a risk factor for the development of AF [8,9]. Even patients without CAD may present with symptoms and clinical findings suggesting myocardial ischemia if they have AF [10,11]. This may relate to several mechanisms including coronary microvascular dysfunction [12]. Previous studies in various patient populations have reported that when compared to patients with sinus rhythm (SR), those with AF have lower myocardial perfusion [13–18]. Still, whether AF is independently associated with myocardial perfusion in patients referred to diagnostic imaging due to suspected obstructive CAD remains unknown. Also, when investigating the association between AF and myocardial perfusion, previous studies have not utilized the severity and extent of coronary atherosclerosis as confounders.
We studied whether AF is independently associated with impaired myocardial perfusion in patients referred to noninvasive coronary imaging due to symptoms suggesting the presence of obstructive CAD. To investigate this possible independent association, we compared hyperemic MBF in patients with AF and in comprehensively matched patients with SR.
Methods
Subjects
We conducted a retrospective, registry-based study of patients who underwent CCTA and consecutive [15O]H2O PET MPI at Kuopio University Hospital, Finland, from 2017 to 2023. Each patient was clinically referred to the imaging due to symptoms suggesting obstructive CAD. The initial study population comprised 971 patients. Our exclusion criteria encompassed technically unsuccessful CCTA or PET MPI, prior myocardial infarction or coronary revascularization, heart failure, cardiomyopathy, at least moderate valvular heart disease, pacemaker, or any cardiac rhythm other than AF or SR during PET MPI. After applying the exclusion criteria, our final study population included 874 patients. Of these, 31 had ongoing AF, and 843 had SR during PET MPI. Eventually, we used propensity score matching (see the section Statistical methods for details) to match the 31 AF patients with 120 SR patients and included those 151 patients in this study’s final analyses. Fig 1 presents a flow chart for patient selection.
Fig 1. Flow chart for patient selection.
AF = atrial fibrillation, CCTA = coronary computed tomography angiography, PET = positron emission tomography, SR = sinus rhythm.
For this research, we secured written authorization from the registrar (the Wellbeing Services County of North Savo) to utilize their data. In Finland, scientific research based only on registry data requires no approval of an ethics committee or informed consent from patients. Therefore, before this study, we received an exemption from ethical approval from the ethics committee of the University of Eastern Finland (description number 16/2023). Data collection took place between February 2024 and September 2024, during which the authors had access to information that could identify individual patients of the study. After the data were collected and analyzed, they were anonymized so that no individual patients could be identified.
Clinical data
Clinical data were collected from electronic patient records and a questionnaire completed by the patients before CCTA and PET MPI. The data included age, sex, body mass index, medication, angina symptoms, comorbidities, and traditional risk factors for CAD (hypertension, dyslipidemia, diabetes, smoking history, and family history of CAD). Blood test results within three months of PET MPI were collected when available. Ongoing cardiac rhythm was determined based on 12-lead electrocardiography obtained during PET MPI. The type of AF was classified based on electronic patient records.
Imaging protocol
At Kuopio University Hospital, CCTA serves as the primary noninvasive modality for diagnosing CAD in patients with a low-to-moderate pretest probability. If CCTA is suggestive of obstructive CAD (≥ 50% luminal stenosis in any epicardial coronary artery) or if obstructive CAD cannot be excluded due to suboptimal image quality, patients are referred to PET MPI. Before PET MPI, the patients are instructed to abstain from smoking for at least 4 hours, caffeine for at least 12 hours, and alcohol for at least 48 hours. Also, dipyridamole is discontinued for a minimum of 48 hours before PET MPI. Most patients undergo CCTA and PET MPI consecutively on the same day. Rarely due to scheduling issues or inadequate patient preparation, CCTA and PET MPI are performed on separate days.
Coronary computed tomography angiography
For this current study, three different CT scanners by Siemens (Siemens Healthineers, Erlangen, Germany) were used for CCTA: Somatom Definition Flash, Somatom Force, and Somatom Edge. If the patient’s heart rate was over 65 beats/min, they were administered up to 20 mg of intravenous metoprolol. Immediately before imaging, the patients also received 1.25 mg of sublingual isosorbide dinitrate.
CCTA imaging began with the acquisition of unenhanced CT images, from which the coronary artery calcium score was determined according to Agatston et al. [19]. Next, contrast-enhanced CCTA scans were performed using prospective electrocardiograph-gated acquisition whenever possible. From CCTA images, we determined findings from the left main (LM), left anterior descending (LAD), left circumflex (LCX), and right (RCA) coronary arteries. A status was categorized for each artery: 0 = vessel not interpretable, 1 = no visible atherosclerosis, 2 = calcified or non-calcified plaque(s) with luminal stenosis < 50%, 3 = calcified plaque(s) with luminal stenosis ≥ 50%, 4 = non-calcified or partially calcified plaque(s) with luminal stenosis ≥ 50%, and 5 = total occlusion. This grading was adapted from the 2022 Coronary Artery Disease-Reporting and Data System [20]. A total CCTA status of 0–5 was formed based on the highest status from any individual coronary artery. For example, if a patient’s status for RCA, LM, and LAD was 2, and the status for LCX was 4, the total CCTA status was 4.
Myocardial perfusion imaging
PET MPI was performed with [15O]H2O using either Biograph mCT or Biograph Vision PET/CT scanner (Siemens Healthineers, Erlangen, Germany). Imaging was performed during pharmacologically induced hyperemia, achieved with a continuous intravenous infusion of adenosine at a rate of 140 µg/kg/min over 6.5 minutes. During adenosine infusion, an intravenous bolus of [15O]H2O with a weight adjusted activity of 500–800 MBq was injected, and dynamic images were acquired over 5 minutes. A low-dose CT was used for attenuation correction. Since our standard clinical protocol included only hyperemic imaging, rest images were unavailable for most patients and were excluded from this study. Blood pressure (BP) was measured at rest and immediately after the [15O]H2O bolus with an automated sphygmomanometer device (SunTech Tango M2, SunTech Medical Inc., Morrisville, North Carolina, USA). A 12-lead electrocardiogram was monitored using the GE Case (General Electric Company, Boston, Massachusetts, USA). Rate pressure product (RPP) was calculated as systolic BP multiplied by heart rate (HR).
PET images acquired with Biograph mCT were reconstructed with two iterations and 21 subsets, with a 3.18 mm x 3.18 mm pixel size. For images acquired with Biograph Vision, image reconstruction was performed with eight iterations, five subsets, and a pixel size of 2.83 mm x 2.83 mm. A matrix size of 128 x 128 and Gaussian 6 mm post-filtering were used for all images regardless of the PET/CT scanner.
Hyperemic MBF was quantified with Carimas software (Turku PET Centre, Turku, Finland) using a single-tissue compartment model [21,22]. MBF was quantified using the standard 17-segment model [23], with measurements recorded for each segment in mL/g/min. MBF was also assessed as a global value and for the standardized perfusion territories of LAD, LCX, and RCA. MBF was considered impaired when any individual segment (excluding basal septal segments 2 and 3) had a value of < 2.3 mL/g/min [24].
Invasive coronary angiography and revascularization
Based on the noninvasive imaging protocol and patient-specific evaluation, attending cardiologists considered whether to refer individual patients to ICA. Additionally, if significant stenoses were present in ICA, they decided whether to perform revascularization procedures.
Statistical methods
We formed two groups for statistical analyses: patients who, during PET MPI, had ongoing AF and those who had ongoing SR. To assess the possible independent association between AF and myocardial perfusion, we formed the groups using propensity score matching. Matching was based on six criteria: age, sex, body mass index, number of traditional risk factors for CAD (0–5), calcium score, and total CCTA status [24–26].
We performed power calculations to assess the required number of patients in each group. The estimated effect size was based on the study by Madsen et al [18]. In their study, the mean difference in hyperemic MBF between AF and SR patients was 0.70 mL/g/min in patients without CAD and 0.42 mL/g/min in patients with CAD. Because our patient population included those with and without obstructive CAD, we employed an effect size of 0.56 mL/g/min, which represents an average between these two patient groups. According to the power calculations, 28 AF patients and 111 SR patients would be needed to achieve a statistical power greater than 80% with a two-tailed α-level less than 0.05. Therefore, propensity score matching was performed using a 1:4 ratio, i.e., each AF patient was matched with four SR patients. Matching was performed with R 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria) using a logistic regression model. AF and SR patients were matched according to the closest propensity score (within ± 0.2 standard deviations).
Apart from propensity score matching, statistical analyses were conducted using IBM SPSS Statistics version 29 (IBM, Armonk, New York, USA). We used the Shapiro-Wilk test to assess the normality of distribution for continuous variables. The results were presented as mean ± standard deviation for normally distributed data, median (interquartile range, IQR) for nonnormally distributed data, or n (%) for categorical data. Finnish Social and Health Data Permit Authority restricts the presentation of data in registry studies. Thus, we were unable to present categorical data with n < 3 in exact numbers.
To compare continuous variables between groups, we used the Student’s T-test for normally distributed data and the Mann-Whitney U-test for nonnormally distributed data. Depending on the expected cell count, either the chi-squared test or Fisher’s exact test was used to compare nominal data between groups. Statistical significance was set at p-value < 0.05.
To further assess the possible independent association between AF and hyperemic MBF, we performed multivariate general linear model analyses with three models. In each model, the presence of AF was a fixed factor. In the first model, the covariates included each variable used in propensity score matching (i.e., age, sex, body mass index, number of traditional risk factors for CAD [0–5], calcium score, and total CCTA status). In the second model, the covariates included those of the first model as well as hemoglobin and the use of beta-blockers (i.e., patient characteristics that differed statistically significantly between the patients with AF and SR and were statistically significantly associated with global hyperemic MBF). In the third model, the covariates included those of the second model as well as diastolic blood pressure after [15O]H2O bolus, glomerular filtration rate (GFR), and the use of diuretics (i.e., variables that differed statistically significantly between AF and SR patients, but which were not statistically significantly associated with global hyperemic MBF). In the multivariate general linear model analyses, we observed no unequal variances (i.e., p-value in Levene’s test was consistently ≥ 0.05) or nonnormally distributed residuals. Also, we found no significant multicollinearity between the covariates, i.e., the variance inflation factor was consistently < 2.5 [27,28].
Results
Clinical characteristics
Thirty-one AF patients were matched with 120 SR patients. Two of the AF patients could be matched with only two SR patients within the specified propensity score range (± 0.2 standard deviations). The patients’ clinical characteristics are shown in Table 1. Most AF patients had persistent or permanent AF. Among SR patients, 16 (13%) had previously documented AF paroxysms but had ongoing SR during PET MPI. No statistically significant differences existed between the groups in terms of age, sex, or body mass index. Also, no differences between the groups were observed in the five traditional risk factors for CAD. The use of an angiotensin-converting enzyme (ACE) inhibitor or an angiotensin II receptor (AR) blocker, a calcium channel blocker, and cholesterol medication were similar between the AF and SR patients. Beta-blocker and diuretic medications were more common in AF patients than in SR patients. Each of the three blood test results differed between the groups: GFR was lower, whereas hemoglobin and N-terminal pro-B-type natriuretic peptide (NT-proBNP) were higher in AF patients than in SR patients.
Table 1. Patient characteristics.
| AF patients (n = 31) |
SR patients (n = 120) |
p-value | |
|---|---|---|---|
| General characteristics | |||
| Age (years) | 72 (68–74) | 72 (68–76) | 0.541 |
| Females | 8 (26%) | 36 (30%) | 0.647 |
| Body mass index (kg/m2) | 30.1 (26.9–33.9) | 29.4 (25.1–33.5) | 0.632 |
| Atrial fibrillation | |||
| Paroxysmal | 5 (16%) | 16 (13%) | 0.771 |
| Persistent or permanent | 26 (84%) | – | – |
| Symptom | |||
| Typical angina | 3 (10%) | 22 (18%) | 0.248 |
| Atypical angina | 7 (23%) | 20 (17%) | 0.444 |
| Dyspnea | 14 (45%) | 42 (35%) | 0.296 |
| Other | 7 (23%) | 36 (30%) | 0.981 |
| Risk factor for CAD | |||
| Hypertension | 28 (90%) | 96 (80%) | 0.181 |
| Dyslipidemia | 23 (74%) | 98 (82%) | 0.353 |
| Diabetes (type 1 or type 2) | 6 (19%) | 22 (18%) | 0.896 |
| Ever-smoker | 10 (32%) | 39 (33%) | 0.980 |
| Family history of CAD | 17 (55%) | 72 (60%) | 0.603 |
| Comorbidity | |||
| Obstructive sleep apnea | 5 (16%) | 22 (18%) | 0.775 |
| Systemic cancer treatment or irradiation of the chest | < 3 | 6 (5%) | – |
| Chronic obstructive pulmonary disease or interstitial lung disease | 3 (10%) | 6 (5%) | 0.391 |
| Blood test results | |||
| GFR (mL/min/1.73 m2) | 70 (60–79) | 81 (68–89) | 0.016 |
| Hemoglobin (g/L) | 151 ± 15 (n = 25) | 143 ± 12 (n = 98) | 0.009 |
| NT-proBNP (ng/L) | 843 (466–1,653) (n = 14) |
88 (45–336) (n = 40) | < 0.001 |
| Medication | |||
| ACE inhibitor/AR blocker | 22 (71%) | 74 (62%) | 0.337 |
| Beta-blocker | 23 (74%) | 47 (39%) | < 0.001 |
| Calcium channel blocker | 9 (29%) | 32 (27%) | 0.792 |
| Diuretic | 13 (42%) | 24 (20%) | 0.011 |
| Cholesterol medication | 17 (55%) | 85 (71%) | 0.090 |
| Long-acting nitrate | < 3 | 8 (7%) | – |
| Insulin | < 3 | < 3 | – |
| Other diabetes medication | 4 (13%) | 19 (16%) | 0.787 |
Values are mean ± standard deviation, median (interquartile range), or n (%). ACE = angiotensin-converting enzyme, AF = atrial fibrillation, AR = angiotensin II receptor, CAD = coronary artery disease, GFR = glomerular filtration rate, NT-proBNP = N-terminal pro-B-type natriuretic peptide, SR = sinus rhythm.
Coronary computed tomography angiography
Table 2 presents the CCTA results. Premedication with beta-blockers was equally common between AF and SR patients. Calcium score and total CCTA status showed no difference between the groups. Also, no between-group differences were observed in the status of any individual epicardial coronary artery.
Table 2. CCTA results.
| AF patients (n = 31) |
SR patients (n = 120) |
p-value | |
|---|---|---|---|
| Premedication with a beta-blocker | 21 (68%) | 85 (71%) | 0.877 |
| Calcium score | 465 (80–1,079) | 274 (113–792) | 0.579 |
| Total CCTA status ≥ 3 | 27 (87%) | 110 (92%) | 0.487 |
| LM or proximal LAD status ≥ 3 | 21 (68%) | 73 (61%) | 0.479 |
| Distal LAD status ≥ 3 | 17 (55%) | 66 (55%) | 0.987 |
| LCX status ≥ 3 | 10 (32%) | 51 (43%) | 0.300 |
| RCA status ≥ 3 | 14 (45%) | 54 (45%) | 0.987 |
Values are median (interquartile range) or n (%). AF = atrial fibrillation, CCTA = coronary computed tomography angiography, LAD = left anterior descending coronary artery, LCX = left circumflex coronary artery, LM = left main coronary artery, RCA = right coronary artery, SR = sinus rhythm.
Myocardial perfusion imaging
The hemodynamic parameters during PET MPI are shown in Table 3. At rest, systolic BP was similar between the groups, while diastolic BP, HR, and RPP were higher in AF patients than in SR patients. After the [15O]H2O bolus, systolic BP, HR, and RPP were similar between the groups, whereas compared to SR patients, those with AF had higher diastolic BP.
Table 3. Hemodynamics during PET MPI.
| AF patients (n = 31) |
SR patients (n = 120) |
p-value | |
|---|---|---|---|
| Systolic BP at rest (mmHg) | 129 (117–141) | 133 (119–142) | 0.812 |
| Diastolic BP at rest (mmHg) | 79 (72–84) | 65 (57–75) | < 0.001 |
| HR at rest (beats per min) | 74 (63–82) | 60 (54–66) | < 0.001 |
| RPP at rest | 9,840 (8,125–12,000) | 7,680 (6,760–9,135) | < 0.001 |
| Systolic BP after [15O]H2O bolus (mmHg) | 124 (114–145) | 126 (113–142) | 0.959 |
| Diastolic BP after [15O]H2O bolus (mmHg) | 70 (65–80) | 60 (52–66) | < 0.001 |
| HR after [15O]H2O bolus (beats per min) | 73 (64–88) | 74 (68–86) | 0.778 |
| RPP after [15O]H2O bolus | 9,513 (7,995–11,431) | 9,522 (7,884–11,397) | 0.833 |
Values are median (interquartile range). AF = atrial fibrillation, BP = blood pressure, HR = heart rate, MPI = myocardial perfusion imaging, PET = positron emission tomography, RPP = rate pressure product, SR = sinus rhythm.
Fig 2 presents the results of PET MPI. Global hyperemic MBF was lower in AF patients than in SR patients (median: 1.9 mL/g/min, IQR: 1.4–2.6 mL/g/min vs. 3.0 mL/g/min, IQR: 2.4–3.6 mL/g/min, p < 0.001). Also, hyperemic MBF in each standardized perfusion territory was lower in AF patients (p < 0.001 for each territory). Hyperemic MBF was impaired (< 2.3 mL/g/min in ≥ 1 segment) more often in AF patients (n = 25/31, 81%) than in SR patients (n = 50/120, 42%, p < 0.001). In the multivariate analyses, AF was independently associated with reduced hyperemic MBF in all three models (p < 0.001).
Fig 2. Hyperemic MBF results.
The boxes indicate median and interquartile ranges. The ends of whiskers indicate 5 and 95 percentiles. AF = atrial fibrillation, SR = sinus rhythm, LAD = left anterior descending coronary artery, LCX = left circumflex coronary artery, MBF = myocardial blood flow, RCA = right coronary artery.
Invasive coronary angiography and revascularization
Of the 25 AF patients with impaired hyperemic MBF, 18 (72%) underwent ICA. Of the 50 SR patients with MBF impairment, ICA was performed on 35 (70%). The referral rate to ICA was similar between the groups (p = 0.858). However, the revascularization frequency was significantly lower in AF patients compared to SR patients: revascularization was performed on 6/18 (33%) AF patients versus 22/35 (63%) SR patients (p = 0.041). The remaining 12/18 (67%) AF patients and 13/35 (37%) SR patients had no significant stenoses in ICA and underwent no revascularization.
Among the AF patients referred to ICA, we found no statistically significant difference in global hyperemic MBF between those revascularized and those not (1.4 ± 0.4 mL/g/min vs. 1.7 ± 0.6 mL/g/min, p = 0.070). Also, compared to revascularized AF patients, those not revascularized showed no difference in the lowest hyperemic MBF of any individual segment (mean 1.0 ± 0.3 mL/g/min vs. 1.3 ± 0.5 mL/g/min, p = 0.071).
Discussion
Our results suggest that in symptomatic patients with suspected obstructive CAD, AF is independently associated with impaired myocardial perfusion. Myocardial perfusion was assessed using the reference standard method for MBF quantification, [15O]H2O PET MPI. The conclusion of the independent association leans against the group comparison based on detailed propensity score matching as well as the multivariate analyses with three comprehensive models.
Prior studies have demonstrated that, compared to patients with SR, those with AF have lower myocardial perfusion. In the landmark study by Range et al., AF patients had not only lower hyperemic MBF but also lower resting MBF and MBF under cold-pressor testing when compared with matched SR patients [13]. In addition, Wijesurendra et al., using magnetic resonance imaging, and Takafuji et al., using CT imaging to measure hyperemic MBF, reported similar results [15,17]. Our findings regarding hyperemic MBF are in line with these previous studies. To the best of our knowledge, our study is the first to include CCTA findings as matching criteria, however. Considering the confounding effect of coronary atherosclerosis is critical, as CAD and AF are interdependent, and our patients underwent coronary imaging specifically due to suspicion of obstructive CAD. Our results consistently demonstrate that AF is independently associated with reduced hyperemic MBF in this patient population.
The pathophysiology between AF and myocardial perfusion is complex, and whether this arrhythmia directly impairs perfusion remains debatable. Our study aimed to investigate the presence of an independent association between AF and myocardial perfusion. Being cross-sectional, this study lacked data to reliably assess causal mechanisms that connect the two. The direct effect of AF on myocardial perfusion could be related to irregular HR, leading to shortened and irregular duration of diastole [12,29]. AF directly affecting myocardial perfusion was supported by the finding of Takafuji et al. since they demonstrated an improvement in AF patients’ hyperemic MBF after catheter ablation [17]. In contrast, Wijesurendra et al. reported that catheter ablation yielded no significant improvement in hyperemic MBF in AF patients [15]. This discrepancy indicates that AF patients also have an underlying condition affecting myocardial perfusion. One such factor could be endothelial dysfunction [12], associated with reduced hyperemic MBF and increasing the risk of subsequent AF [30,31]. Also, in patients with hypertrophic and dilated cardiomyopathy, the presence of AF has been associated with reduced hyperemic MBF, suggesting impaired coronary microvascular function [14,32]. Overall, AF and myocardial perfusion are most likely connected by several mechanisms.
In [15O]H2O PET MPI, absolute hyperemic MBF < 2.3 mL/g/min has been established as the optimal cutoff value for detecting functionally significant CAD [24]. Accordingly, impairment in absolute hyperemic MBF most strongly guides a referral to ICA [33]. Still, in our study, only a few AF patients with hyperemic MBF < 2.3 mL/g/min had significant stenoses in ICA and were evaluated to benefit from revascularization. Moreover, Madsen et al. reported that only a few AF patients, even those with severely impaired hyperemic MBF, underwent revascularization [18]. Based on the results of our study and those of Madsen et al., we propose that for guiding a referral to ICA, AF patients would benefit from more individualized MBF cutoff values. Also, in addition to absolute MBF evaluation, the accuracy of [15O]H2O PET MPI in AF patients might be enhanced by assessing relative perfusion. This approach is supported by Madsen et al.‘s findings, where they observed that nearly all patients who underwent revascularization demonstrated regional hypoperfusion when evaluated using relative flow reserve measurements [18]. However, regional hypoperfusion was not a major determinant for a referral to ICA. Unfortunately, our study included only a few AF patients who underwent ICA. Thus, it lacked statistical power to assess whether impaired relative perfusion predicts revascularization in AF patients. Further research is warranted to conclude whether adjusted cutoff values for hyperemic MBF would increase diagnostic accuracy and whether relative perfusion metrics have additional value in this patient group.
The patient groups showed differences in each three blood test results, that is, GFR, hemoglobin, and NT-proBNP. The possible confounding effect of GFR and hemoglobin on global hyperemic MBF was considered with the multivariate analyses. When both measurements were included in the multivariate models, AF remained independently associated with global hyperemic MBF. NT-proBNP was higher in AF patients compared to SR patients. However, given that the measurement was available in only 36% (54/151) of the included patients, we excluded it from the multivariate analyses. In AF patients, interpreting elevated NT-proBNP is challenging. Jones et al. recently showed that in 155,347 primary care patients, NT-proBNP was 724 (IQR 245–1,535) ng/L in patients with AF but without diagnosed heart failure and 111 (IQR 51–250) ng/L in patients without AF but with diagnosed heart failure [34]. These figures closely resemble those in our study and highlight how the presence of AF is associated with significantly elevated NT-proBNP levels even in the absence of heart failure. Furthermore, several studies have demonstrated that AF is independently associated with significantly elevated NT-proBNP, as Liu et al. showed in their meta-analysis [35]. Since patients with diagnosed heart failure and cardiomyopathies, as well as those with prior myocardial infarction were excluded from our study, we suggest that AF presence mostly explains the differing NT-proBNP levels. Still, we acknowledge that recent echocardiography results were unavailable for most patients. We were thereby unable to evaluate the echocardiography-based measures of cardiac structure and function.
CCTA and ICA showed some discrepancies. While most patients had suspected significant stenosis in CCTA, only 33% of AF patients and 63% of SR patients showed a significant stenosis in ICA and were subsequently revascularized. In line with the latest practical CCTA guideline, ≥ 50% stenosis was used as the cutoff value that prompted consecutive PET MPI [20]. Still, angiographic stenosis severity correlates poorly with the lesion’s functional significance even up to 90% stenosis [36]. Also, compared to ICA, the specificity of CCTA to detect functionally significant CAD is lower [4]. Based on PET MPI, the decision to refer a patient to ICA was guided by hyperemic MBF of < 2.3 mL/g/min even when AF was present. The lower specificity of CCTA compared to ICA, and the assumably suboptimal hyperemic MBF cutoff value for AF patients likely explain the significantly lower revascularization frequency of AF patients. This highlights the demand for more individualized MBF cutoff values even when applying a hybrid imaging approach.
We acknowledge that this single-center study had limitations. Unfortunately, resting MBF was unavailable for most patients, since the standard PET MPI protocol at Kuopio University Hospital includes only hyperemic imaging. Therefore, we were also unable to determine myocardial flow reserve, i.e., the ratio of hyperemic MBF to resting MBF. The higher resting RPP of AF patients compared to SR patients indicates that in AF patients, the demand for resting MBF is higher, which would also affect myocardial flow reserve. Due to the lack myocardial flow reserve assessment, we could not evaluate the vasodilatory capacity of the patients. The number of AF patients was limited. However, our initial patient population comprised 874 patients, which allowed us to perform propensity score matching with a 1:4 ratio. After appropriate statistical power calculations, we included 151 patients in the final analyses. Females were rather underrepresented in this study, which may affect the generalizability of our findings. Some measurements may have suffered from lower reliability in AF patients compared to SR patients. For example, automated blood pressure measurement is generally less accurate during uneven cardiac rhythm [37]. In addition, ongoing AF may introduce some unreliability in MBF quantification. However, the determination of MBF was based on imaging in which the shortest frames were 5 seconds long. In 5-second frames, variation caused by beat-to-beat hemodynamic fluctuations is already significantly attenuated. Moreover, a time-activity curve—upon which the calculation was based—was fitted to data comprising multiple frames. In this overall process, the impact of variations caused by instantaneous beat-to-beat fluctuations was further reduced. Consequently, the potential source of error arising from the irregular heart rhythm of AF patients is unlikely to be significant. Additionally, we aimed to minimize the contribution of MBF quantification errors by excluding patients with technically unreliable PET MPI such as those who demonstrated significant motion artefacts during imaging. Finally, since this study was cross-sectional, we were unable to assess the possible causality between AF and myocardial perfusion.
Conclusions
In patients with symptoms suggestive of obstructive CAD, the presence of AF was independently associated with impaired hyperemic MBF. Clinicians should consider this association when interpreting PET MPI results as part of individualized clinical decision-making for this highly prevalent patient group. Further research is warranted to establish whether patients with AF would benefit from adjusted hyperemic MBF cutoff values to more accurately diagnose obstructive CAD.
Acknowledgments
We thank biostatistician Tuomas Selander for his assistance with propensity score matching and the multivariate analyses.
Data Availability
The data underlying this manuscript cannot be publicly shared since it contains sensitive health information and data access is subject to statutory restrictions and the terms of the relevant data permits. The researchers are not authorized to share individual-level data to third parties. Eligible researchers may apply for data access from the Wellbeing Services County of North Savo (contact email: kirjaamo@pshyvinvointialue.fi). All applications will be independently evaluated by the data controllers and the data permit authority.
Funding Statement
TPL received a grant from the Finnish Foundation for Cardiovascular Research (November 10, 2020). URL: https://www.sydantutkimussaatio.fi/. The funder played no role role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
References
- 1.Safiri S, Karamzad N, Singh K, Carson-Chahhoud K, Adams C, Nejadghaderi SA. Burden of ischemic heart disease and its attributable risk factors in 204 countries and territories, 1990–2019. Eur J Prev Cardiol. 2022;29:420–31. [DOI] [PubMed] [Google Scholar]
- 2.Gulati M, Levy PD, Mukherjee D, Amsterdam E, Bhatt DL, Birtcher KK, et al. 2021 AHA/ACC/ASE/CHEST/SAEM/SCCT/SCMR Guideline for the Evaluation and Diagnosis of Chest Pain: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2021;144(22):e368–454. doi: 10.1161/CIR.0000000000001029 [DOI] [PubMed] [Google Scholar]
- 3.Vrints C, Andreotti F, Koskinas KC, Rossello X, Adamo M, Ainslie J, et al. 2024 ESC Guidelines for the Management of Chronic Coronary Syndromes: Developed by the Task Force for the Management of Chronic Coronary Syndromes of the European Society of Cardiology (ESC) Endorsed by the European Association for Cardio-Thoracic Surgery (EACTS). Eur Heart J. 2024;45:3415–537. [DOI] [PubMed] [Google Scholar]
- 4.Knuuti J, Ballo H, Juarez-Orozco LE, Saraste A, Kolh P, Rutjes AWS, et al. The performance of non-invasive tests to rule-in and rule-out significant coronary artery stenosis in patients with stable angina: A meta-analysis focused on post-test disease probability. Eur Heart J. 2018;39(35):3322–30. doi: 10.1093/eurheartj/ehy267 [DOI] [PubMed] [Google Scholar]
- 5.Driessen RS, Raijmakers PG, Stuijfzand WJ, Knaapen P. Myocardial perfusion imaging with PET. Int J Cardiovasc Imaging. 2017;33(7):1021–31. doi: 10.1007/s10554-017-1084-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Schindler TH, Fearon WF, Pelletier-Galarneau M, Ambrosio G, Sechtem U, Ruddy TD, et al. Myocardial perfusion PET for the detection and reporting of coronary microvascular dysfunction: A JACC: Cardiovascular imaging expert panel statement. JACC Cardiovasc Imaging. 2023;16(4):536–48. doi: 10.1016/j.jcmg.2022.12.015 [DOI] [PubMed] [Google Scholar]
- 7.Schnabel RB, Yin X, Gona P, Larson MG, Beiser AS, McManus DD, et al. 50 year trends in atrial fibrillation prevalence, incidence, risk factors, and mortality in the Framingham Heart Study: A cohort study. Lancet. 2015;386(9989):154–62. doi: 10.1016/S0140-6736(14)61774-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Benjamin EJ, Levy D, Vaziri SM, D’Agostino RB, Belanger AJ, Wolf PA. Independent risk factors for atrial fibrillation in a population-based cohort. The Framingham Heart Study. JAMA. 1994;271(11):840–4. doi: 10.1001/jama.1994.03510350050036 [DOI] [PubMed] [Google Scholar]
- 9.Wilson PW. Established risk factors and coronary artery disease: The Framingham Study. Am J Hypertens. 1994;7(7 Pt 2):7S-12S. doi: 10.1093/ajh/7.7.7s [DOI] [PubMed] [Google Scholar]
- 10.van den Bos EJ, Constantinescu AA, van Domburg RT, Akin S, Jordaens LJ, Kofflard MJM. Minor elevations in troponin I are associated with mortality and adverse cardiac events in patients with atrial fibrillation. Eur Heart J. 2011;32(5):611–7. doi: 10.1093/eurheartj/ehq491 [DOI] [PubMed] [Google Scholar]
- 11.Parwani AS, Boldt L-H, Huemer M, Wutzler A, Blaschke D, Rolf S, et al. Atrial fibrillation-induced cardiac troponin I release. Int J Cardiol. 2013;168(3):2734–7. doi: 10.1016/j.ijcard.2013.03.087 [DOI] [PubMed] [Google Scholar]
- 12.Wijesurendra RS, Casadei B. Atrial fibrillation: effects beyond the atrium? Cardiovasc Res. 2015;105(3):238–47. doi: 10.1093/cvr/cvv001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Range FT, Schäfers M, Acil T, Schäfers KP, Kies P, Paul M, et al. Impaired myocardial perfusion and perfusion reserve associated with increased coronary resistance in persistent idiopathic atrial fibrillation. Eur Heart J. 2007;28(18):2223–30. doi: 10.1093/eurheartj/ehm246 [DOI] [PubMed] [Google Scholar]
- 14.Range FT, Paul M, Schäfers KP, Acil T, Kies P, Hermann S, et al. Myocardial perfusion in nonischemic dilated cardiomyopathy with and without atrial fibrillation. J Nucl Med. 2009;50(3):390–6. doi: 10.2967/jnumed.108.055665 [DOI] [PubMed] [Google Scholar]
- 15.Wijesurendra RS, Liu A, Notaristefano F, Ntusi NAB, Karamitsos TD, Bashir Y, et al. Myocardial perfusion is impaired and relates to cardiac dysfunction in patients with atrial fibrillation both before and after successful catheter ablation. J Am Heart Assoc. 2018;7(15):e009218. doi: 10.1161/JAHA.118.009218 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Byrne C, Hasbak P, Kjær A, Thune JJ, Køber L. Myocardial perfusion during atrial fibrillation in patients with non-ischaemic systolic heart failure: A cross-sectional study using Rubidium-82 positron emission tomography/computed tomography. Eur Heart J Cardiovasc Imaging. 2019;20(2):233–40. doi: 10.1093/ehjci/jey089 [DOI] [PubMed] [Google Scholar]
- 17.Takafuji M, Kitagawa K, Nakamura S, Kokawa T, Kagawa Y, Fujita S, et al. Hyperemic myocardial blood flow in patients with atrial fibrillation before and after catheter ablation: A dynamic stress CT perfusion study. Physiol Rep. 2021;9(22):e15123. doi: 10.14814/phy2.15123 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Madsen S, Kjaerulff MLG, Ejlersen JA, Ranghøj Nielsen BR, Jakobsen L, Sörensen J, et al. Guiding early revascularization using [15O]H2O positron emission tomography myocardial perfusion imaging: Impact of atrial fibrillation. Eur Heart J Cardiovasc Imaging. 2024;25(7):958–67. doi: 10.1093/ehjci/jeae043 [DOI] [PubMed] [Google Scholar]
- 19.Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M Jr, Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol. 1990;15(4):827–32. doi: 10.1016/0735-1097(90)90282-t [DOI] [PubMed] [Google Scholar]
- 20.Cury RC, Leipsic J, Abbara S, Achenbach S, Berman D, Bittencourt M, et al. CAD-RADSTM 2.0 - 2022 Coronary Artery Disease-Reporting and Data System: An Expert Consensus Document of the Society of Cardiovascular Computed Tomography (SCCT), the American College of Cardiology (ACC), the American College of Radiology (ACR), and the North America Society of Cardiovascular Imaging (NASCI ). JACC Cardiovasc Imaging. 2022;15: 1974–001. [DOI] [PubMed] [Google Scholar]
- 21.Rainio O, Han C, Teuho J, Nesterov SV, Oikonen V, Piirola S, et al. Carimas: An extensive medical imaging data processing tool for research. J Digit Imaging. 2023;36: 1885–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Iida H, Kanno I, Takahashi A, Miura S, Murakami M, Takahashi K, et al. Measurement of absolute myocardial blood flow with H215O and dynamic positron-emission tomography. Strategy for quantification in relation to the partial-volume effect. Circulation. 1988;78(1):104–15. doi: 10.1161/01.cir.78.1.104 [DOI] [PubMed] [Google Scholar]
- 23.Cerqueira MD, Weissman NJ, Dilsizian V, Jacobs AK, Kaul S, Laskey WK, et al. Standardized myocardial segmentation and nomenclature for tomographic imaging of the heart. A statement for healthcare professionals from the Cardiac Imaging Committee of the Council on Clinical Cardiology of the American Heart Association. Circulation. 2002;105(4):539–42. doi: 10.1161/hc0402.102975 [DOI] [PubMed] [Google Scholar]
- 24.Danad I, Uusitalo V, Kero T, Saraste A, Raijmakers PG, Lammertsma AA, et al. Quantitative assessment of myocardial perfusion in the detection of significant coronary artery disease: Cutoff values and diagnostic accuracy of quantitative [(15)O]H2O PET imaging. J Am Coll Cardiol. 2014;64(14):1464–75. doi: 10.1016/j.jacc.2014.05.069 [DOI] [PubMed] [Google Scholar]
- 25.Bajaj NS, Osborne MT, Gupta A, Tavakkoli A, Bravo PE, Vita T, et al. Coronary microvascular dysfunction and cardiovascular risk in obese patients. J Am Coll Cardiol. 2018;72(7):707–17. doi: 10.1016/j.jacc.2018.05.049 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Winther S, Schmidt SE, Mayrhofer T, Bøtker HE, Hoffmann U, Douglas PS, et al. Incorporating coronary calcification into pre-test assessment of the likelihood of coronary artery disease. J Am Coll Cardiol. 2020;76(21):2421–32. doi: 10.1016/j.jacc.2020.09.585 [DOI] [PubMed] [Google Scholar]
- 27.Zuur AF, Ieno EN, Elphick CS. A protocol for data exploration to avoid common statistical problems. Methods Ecol Evol. 2009;1(1):3–14. doi: 10.1111/j.2041-210x.2009.00001.x [DOI] [Google Scholar]
- 28.Dormann CF, Elith J, Bacher S, Buchmann C, Carl G, Carré G, et al. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography. 2012;36(1):27–46. doi: 10.1111/j.1600-0587.2012.07348.x [DOI] [Google Scholar]
- 29.Goette A, Corradi D, Dobrev D, Aguinaga L, Cabrera J-A, Chugh SS, et al. Atrial cardiomyopathy revisited-evolution of a concept: a clinical consensus statement of the European Heart Rhythm Association (EHRA) of the ESC, the Heart Rhythm Society (HRS), the Asian Pacific Heart Rhythm Society (APHRS), and the Latin American Heart Rhythm Society (LAHRS). Europace. 2024;26(9):euae204. doi: 10.1093/europace/euae204 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Vartiainen N, Hartikainen JEK, Laitinen TM, Kuikka P-I, Mussalo H, Laitinen TP. Association between peripheral endothelial function and myocardial perfusion in patients with coronary artery disease. Eur Heart J Imaging Methods Pract. 2024;2(1):qyae010. doi: 10.1093/ehjimp/qyae010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Corban MT, Godo S, Burczak DR, Noseworthy PA, Toya T, Lewis BR, et al. Coronary endothelial dysfunction is associated with increased risk of incident atrial fibrillation. J Am Heart Assoc. 2020;9(8):e014850. doi: 10.1161/JAHA.119.014850 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Sciagrà R, Sotgia B, Olivotto I, Cecchi F, Nistri S, Camici PG, et al. Relationship between atrial fibrillation and blunted hyperemic myocardial blood flow in patients with hypertrophic cardiomyopathy. J Nucl Cardiol. 2009;16(1):92–6. doi: 10.1007/s12350-008-9005-5 [DOI] [PubMed] [Google Scholar]
- 33.Stenström I, Maaniitty T, Uusitalo V, Ukkonen H, Kajander S, Mäki M, et al. Absolute stress myocardial blood flow after coronary CT angiography guides referral to invasive angiography. JACC Cardiovasc Imaging. 2019;12(11 Pt 1):2266–7. doi: 10.1016/j.jcmg.2019.08.002 [DOI] [PubMed] [Google Scholar]
- 34.Jones NR, Taylor KS, Ordóñez-Mena JM, Goyder CR, Hobbs FDR, Taylor CJ. NT-proBNP testing for heart failure diagnosis in people with atrial fibrillation: A diagnostic accuracy study. PLoS Med. 2025;22(10):e1004550. doi: 10.1371/journal.pmed.1004550 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Liu Y, Xiao Y, Chen X, Zhang F. Association between plasma brain natriuretic peptide/N-terminal pro-brain natriuretic peptide levels and atrial fibrillation: Evidence from a meta-analysis. Chin Med J (Engl). 2014;127(15):2824–8. [PubMed] [Google Scholar]
- 36.Pal T, Fearon WF, De Bruyne B, Oldroyd KG, Leesar MA, Ver Lee PN. Angiographic versus functional severity of coronary artery stenoses in the FAME study. J Am Coll Cardiol. 2010;55:2816–21. [DOI] [PubMed] [Google Scholar]
- 37.Stergiou GS, Kyriakoulis KG, Stambolliu E, Destounis A, Karpettas N, Kalogeropoulos P, et al. Blood pressure measurement in atrial fibrillation: Review and meta-analysis of evidence on accuracy and clinical relevance. J Hypertens. 2019;37(12):2430–41. doi: 10.1097/HJH.0000000000002201 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data underlying this manuscript cannot be publicly shared since it contains sensitive health information and data access is subject to statutory restrictions and the terms of the relevant data permits. The researchers are not authorized to share individual-level data to third parties. Eligible researchers may apply for data access from the Wellbeing Services County of North Savo (contact email: kirjaamo@pshyvinvointialue.fi). All applications will be independently evaluated by the data controllers and the data permit authority.