Incidence, Clinical Factors, and Endotype Characterization of Coronary Microvascular Dysfunction in Patients With Non-Obstructive Coronary Artery Disease Undergoing 13N-Ammonia Positron Emission Tomography

Shiro Miura; Atsutaka Okizaki; Osamu Manabe; Hiraku Kumamaru; Chihoko Miyazaki; Takehiro Yamashita

doi:10.1253/circj.CJ-25-0212

Abstract

Background: Coronary microvascular dysfunction (CMD) is characterized by impaired myocardial flow reserve (MFR) in patients with non-obstructive coronary artery disease (CAD). The real-world incidence of CMD, risk factors for decreased MFR, and characteristics of the two CMD endotypes (classical and endogenous) in this population remain unclear.

Methods and Results: After screening 1,313 patients with suspected or known CAD who underwent ¹³N-ammonia positron emission tomography (PET), 345 with non-obstructive CAD were retrospectively enrolled in the study. Stress/resting myocardial blood flow (MBF) was quantified using ¹³N-ammonia PET. PET-assessed CMD (impaired MFR <2.0) was diagnosed in 60 (17%) patients. Independent predictors of decreased MFR included older age, female sex, anemia, and hypertension; however, these factors accounted for only 32% of the observed variability in MFR. Symptomatic status was not an independent predictor of decreased MFR. Patients with classical CMD (resting MBF <1.3 mL/min/g) had higher summed stress scores and stress/resting coronary vascular resistance, whereas patients with endogenous CMD (resting MBF ≥1.3 mL/min/g) showed female dominance, higher stress MBF, and a higher resting rate-pressure product.

Conclusions: Among patients with non-obstructive CAD, approximately 1 in 5 had PET-assessed CMD. Age, female sex, anemia, and hypertension were independent clinical factors associated with impaired MFR, which explained a limited portion of its variability. Further examination of unidentified or unmeasured factors is warranted.

Central Figure

Non-obstructive coronary artery disease (CAD), diagnosed via invasive coronary angiography (CAG), has a prevalence of 20–50%.¹^–³ Patients with non-obstructive CAD often experience refractory angina symptoms, persistent anxiety, limited exercise tolerance, and impaired quality of life; these conditions, alongside increased cardiovascular morbidity and mortality,¹ necessitate repeated healthcare utilization.⁴ Although non-obstructive CAD is receiving increased attention, the underlying pathophysiology remains unclear, likely resulting in missed diagnoses in clinical practice. Both vasomotion abnormalities and coronary microvascular dysfunction (CMD) contribute to the onset of angina and myocardial ischemia in emerging non-obstructive CAD.² However, previous studies¹ on CMD in non-obstructive CAD were relatively small and performed in specific populations (e.g., women),⁵ had invasive CMD assessments performed on patients referred for coronary angiography, and were potentially affected by methodological and technical factors.¹^,⁴ Comprehensive invasive assessments of both endothelial and non-endothelial mechanisms are considered the gold standard for the diagnosis of CMD.⁶ However, because of limited data, the incidence of and clinical risk factors for CMD in the general population remain unclear.

Myocardial perfusion positron emission tomography (PET) is an effective option for assessing CMD by the non-invasive measurement of stress and resting myocardial blood flow (MBF), as well as for the detection of patients with impaired myocardial flow reserve (MFR).⁷^,⁸ Myocardial perfusion PET provides essential diagnostic and prognostic information to guide treatments that could improve symptoms and outcomes.⁸ This highlights the need to identify population-based clinical risk factors for CMD to determine who would benefit from non-invasive or invasive CMD assessment. Further, the importance of standardizing the categorization of PET-assessed-CMD into classical (predominantly stress MBF alterations) and endogenous (predominantly resting MBF alterations) types to facilitate diagnosis and treatment of microvascular angina has been emphasized.⁸ However, it remains unclear whether these endotypes have different clinical features.

In this study we investigated the prevalence of PET-assessed CMD in a clinical population and clarified correlations between a decrease in MFR and significant clinical factors (including age, sex, symptoms, conventional cardiovascular risk factors, and PET-related parameters). We also determined the characteristics of the two distinct CMD endotypes identified on the basis of PET flow data.

Methods

Study Design and Participants

In this single-center retrospective study, we screened 1,313 consecutive patients who had undergone myocardial perfusion PET/computed tomography (CT) for a clinical indication of myocardial ischemia and CMD between November 2016 and April 2024 at Sapporo Kojinkai Memorial Hospital (Figure 1). We initially excluded 536 patients with prior documented myocardial infarction, coronary revascularization (percutaneous coronary intervention or coronary artery bypass grafting), known cardiomyopathy, previous hospitalization for heart failure, left ventricular ejection fraction <30%, congenital heart disease, significant valvular heart disease, and missing data. Among them, 683 underwent coronary assessments with coronary CT angiography (CCTA), invasive CAG, or both, within 6 months of the PET scan. These patients were divided into 2 groups according to the presence of obstructive or non-obstructive CAD (Figure 1), with those diagnosed with non-obstructive CAD (n=345) included in the study.

Figure 1.

Patient enrolment and disposition flowchart. CAD, coronary artery disease; CMD, coronary microvascular dysfunction; MBF, myocardial blood flow; MFR, myocardial flow reserve; PET, positron emission tomography.

We ascertained the incidence of CMD and identified the clinical factors and PET-related parameters that were associated with a decrease in MFR. Non-obstructive CAD was defined as no significant stenosis in an epicardial coronary artery with a diameter ≥2 mm (<50% coronary diameter stenosis or >0.80 fractional flow reserve).⁹ CMD was defined as the presence of abnormal endothelial-independent coronary microvascular function (PET-derived global MFR <2.0).³^,¹⁰^–¹² We also examined and compared the characteristics between the two distinct CMD endotypes (classical and endogenous).⁸^,¹¹ To this end, patients with CMD were classified as classical or endogenous CMD based on resting MBF (<1.3 and ≥1.3 mL/min/g, respectively).¹¹

This study was approved by the Institutional Review Board of Sapporo Kojinkai Memorial Hospital (Approval no. 2019-7). Although all patients provided written informed consent before undergoing CAG, CCTA, and PET examinations, the requirement for specific consent for study participation was waived due to the retrospective nature of the study. This study complied with all the IRB requirements based on the Declaration of Helsinki and the ethical principles of the Belmont Report.

Clinical Data Definitions

Demographic data, risk factors, and laboratory results at the time of PET examination were obtained from medical records. Following the conventional classification,⁴ symptomatic status was classified as typical angina, atypical angina, and asymptomatic. Details of traditional cardiovascular risk factors, including hypertension (prior clinical diagnosis or the use of antihypertensive medications), diabetes (prior clinical diagnosis or the use of antidiabetes medications, including insulin), dyslipidemia (prior clinical diagnosis or the use of lipid-lowering medications), chronic kidney disease (defined as an estimated glomerular filtration rate <60 mL/min/1.73 m² [calculated using the Modification of Diet in Renal Disease Study for the Japanese population] or a urinary albumin–creatinine ratio >30 mg/g),¹³ smoking status, and body mass index (BMI; underweight <18.5 kg/m², normal weight 18.5–24.9 kg/m², and overweight or obese ≥25.0 kg/m²), were recorded. Biochemical parameters included fasting blood glucose, HbA1c, serum total cholesterol, low-density lipoprotein, high-density lipoprotein, and triglyceride levels. Anemia was diagnosed using World Health Organization criteria (hemoglobin level <13 and <12 g/dL in men and women, respectively).¹⁴

PET Imaging

Patients underwent PET using a whole-body PET/CT scanner (Biograph mCT Flow 64-4R PET/CT system; Siemens Healthcare GmbH, Erlangen, Germany) with ¹³N-ammonia as the myocardial perfusion tracer at rest and during pharmacological stress, as described previously.¹⁵ Briefly, a single-day ¹³N-ammonia PET/CT scan was performed at rest and during hyperemia induced by ATP infusion for 5 min at a rate of 160 μg/kg/min; 3 MBq/kg ¹³N-ammonia was injected over a period of 30 s, starting 3 min after the ATP infusion. In the 24 h before the scans, patients were instructed to abstain from caffeine- and methylxanthine-containing substances and drugs. Resting scans were performed with 3 MBq/kg ¹³N-ammonia for 30 s, 1 h after the stress scan. Electrocardiograms were monitored continuously throughout each flow measurement, and heart rate and blood pressure were measured at 1-min intervals.

PET images were analyzed quantitatively using the Syngo MBF software package (Siemens Healthcare). Global MFR was calculated by dividing stress MBF by resting MBF. Using a standard 17-segment model and a 5-point grading system, semiquantitative assessments of myocardial perfusion (segmental perfusion scores) were performed by 2 experienced radiologists (C.M. and M.N.). The summed stress score (SSS), summed resting score, and summed difference score (SDS) were assessed. To correct the resting MBF for baseline work, we calculated the resting rate–pressure product (RPP) as heart rate×systolic blood pressure at rest, the corrected resting MBF as resting MBF×resting RPP/8,500, and the corrected MFR as stress MBF divided by corrected resting MBF.¹⁶ Stress/resting coronary vascular resistance (CVR; stress/resting systolic blood pressure divided by stress/resting MBF) was calculated separately to evaluate the microvascular burden.¹⁶ Based on the PET myocardial perfusion images, stress-induced myocardial ischemia was independently evaluated by 2 experienced radiologists (C.M., M.N.). Disagreements, if any, on the presence of myocardial ischemia were resolved by discussion with a third experienced radiologist (A.O.).

Statistical Analysis

Baseline characteristics are reported as proportions (%) for categorical variables and as the median with interquartile range (IQR) for continuous variables. Fisher’s exact test and the Wilcoxon rank-sum test were used to assess the significance of differences in categorical and continuous variables, respectively. The incidence of each of the 3 CMD categories (i.e., no CMD, classical CMD, and endogenous CMD) was determined in the overall population based on symptomatic statuses (typical angina, atypical angina, and asymptomatic). The significance of differences between groups was determined using Chi-squared tests and analysis of variance (ANOVA) for categorical and continuous variables. Spearman’s correlation was used to describe sex-stratified associations between MFR and key continuous variables (age, BMI, hemoglobin, SSS, and resting RPP).²^,³^,⁵^,¹¹ The sex-stratified effects of resting RPP on MFR, stress MBF, and resting MBF were also ascertained. Multivariable linear regression analyses were used to identify predictors of reduced MFR. Model 1 included 10 predefined baseline clinical factors (age, sex, symptoms, hypertension, diabetes, dyslipidemia, chronic kidney disease, smoking status, obesity, and anemia). Model 2 included PET-derived parameters (SSS and resting RPP) in addition to all variables in Model 1. Linearity, variance homogeneity, and Gaussian distribution of the residuals were graphically assessed, with no significant multicollinearity identified using variance inflation factors. All tests were 2-sided, with statistical significance set at P<0.05. All statistical analyses were performed using Stata BE 18.0 (StataCorp, College Station, TX, USA) and R version 4.3.1 (R Project for Statistical Computing, Vienna, Austria).

Results

Baseline Participant Characteristics

The 683 patients who underwent ¹³N-ammonia myocardial perfusion PET and anatomical coronary assessments using CCTA or invasive CAG for suspected or known CAD were eligible for inclusion in this study (Figure 1). Of these, the 345 patients with evidence of non-obstructive CAD were included in the final study population; PET-assessed CMD was present in 60 (17%) patients. The baseline characteristics of patients with and without CMD are presented in Table 1. Overall, 196 (57%) patients were female, and the median age was 71 years (IQR 61–76 years). In all, overweight or obesity was present in 127 (37%) patients, and 180 (52%) presented with typical angina. CCTA and CAG were performed on 282 (82%) and 158 (46%) patients, respectively.

Table 1.

Baseline Characteristics in All Patients and in Those With and Without CMD Separately

	Overall (n=345)	CMD (n=60)	No CMD (n=285)	P value
Demographic characteristics
Age (years)	71 [61–76]	75 [73–81]	69 [60–75]	<0.001
Female sex	196 (57)	44 (73)	152 (53)	0.006
Body mass index (kg/m²)	23.9 [21.7–26.4]	22.9 [20.8–24.9]	24.3 [21.8–26.5]	0.03
Normal weight	204 (59)	41 (68)	163 (57)	0.06
Underweight	14 (4)	4 (7)	10 (4)
Overweight or obese	127 (37)	15 (25)	112 (39)
Symptomatic status
Typical angina	180 (52)	44 (73)	136 (48)	0.002
Atypical angina	125 (36)	12 (20)	113 (40)
Asymptomatic	40 (12)	4 (7)	36 (13)
Medical history
Hypertension	207 (60)	44 (73)	163 (57)	0.02
Dyslipidemia	173 (50)	33 (55)	140 (49)	0.48
Diabetes	83 (24)	16 (27)	67 (24)	0.62
Chronic kidney disease	245 (71)	36 (60)	209 (73)	0.04
Anemia	40 (12)	15 (25)	25 (9)	0.001
Atrial fibrillation	38 (11)	8 (13)	30 (11)	0.50
Peripheral vascular disease	17 (5)	5 (8)	12 (4)	0.19
Chronic lung disease	32 (9)	10 (17)	22 (8)	0.05
Current smoker	50 (15)	8 (13)	42 (15)	1.0
Laboratory values
Hemoglobin (g/dL)	13.6 [13.0–14.0]	13.0 [12.0–14.0]	14.0 [13.0–14.0]	<0.001
eGFR (pg/mL)	67.1 [58.5–76.5]	62.5 [52.4–71.4]	68.2 [59.8–77.0]	0.008
LDL-C (mg/dL)	108 [90–129]	100 [87–120]	110 [91–130]	0.02
HDL-C (mg/dL)	56 [46–68]	55 [47–63]	56 [46–69]	0.37
Triglycerides (mg/dL)	111 [83–169]	95 [80–125]	117 [85–182]	0.003
Fasting glucose (mg/dL)	112 [101–130]	111 [103–129]	112 [100–131]	0.88
HbA1c (%)	6.0 [5.6–6.3]	6.0 [5.6–6.3]	6.0 [5.6–6.3]	0.60
Coronary assessment
CCTA	282 (82)	45 (75)	237 (83)	0.14
Invasive coronary angiography	158 (46)	38 (63)	120 (42)	0.004
Deferred by fractional flow reserve	74 (21)	21 (35)	53 (19)	0.009
Hemodynamic parameters during PET
Resting heart rate (beats/min)	68 [60–78]	70 [61–85]	68 [60–77]	0.11
Stress heart rate (beats/min)	80 [70–89]	79 [68–91]	80 [70–88]	0.70
Resting SBP (mmHg)	131 [118–146]	138 [120–158]	128 [116–141]	0.007
Stress SBP (mmHg)	114 [103–129]	111 [99–131]	115 [104–128]	0.26
Resting DBP (mmHg)	71 [64–80]	74 [64–83]	71 [64–79]	0.08
Stress DBP (mmHg)	62 [56–69]	59 [52–66]	62 [57–69]	0.006
Resting RPP (mmHg · beats/min)	8,820 [7,425–10,656]	10,171 [7,481–12,156]	8,733 [7,425–10,268]	0.01
Stress RPP (mmHg · beats/min)	9,282 [7,800–10,830]	9,294 [6,660–10,686]	9,282 [7,957–10,830]	0.31
PET imaging parameters
Resting EDV (mL)	80 [66–98]	78 [62–92]	80 [66–99]	0.22
Resting LVEF (%)	71 [64–77]	71 [62–79]	71 [64–77]	0.91
Summed stress score	4 [2–7]	5 [3–8]	4 [2–6]	0.03
Summed resting score	2 [0–3]	2.00 [1.00–5.00]	2 [0–3]	0.002
Summed difference score	2 [1–4]	2 [1–3]	2 [1–4]	0.72
Resting MBF (mL/min/g)	0.95 [0.79–1.14]	1.15 [0.95–1.43]	0.91 [0.78–1.09]	<0.001
Corrected resting MBF^A (mL/min/g)	0.92 [0.77–1.09]	0.97 [0.79–1.23]	0.92 [0.76–1.08]	0.06
Stress MBF (mL/min/g)	2.60 [2.17–3.00]	1.90 [1.45–2.40]	2.69 [2.31–3.09]	<0.001
MFR	2.72 [2.20–3.31]	1.69 [1.46–1.86]	2.87 [2.46–3.43]	<0.001
Corrected MFR^A	2.79 [2.31–3.45]	1.84 [1.46–2.40]	2.93 [2.49–3.52]	<0.001
Resting CVR^B (mmHg/(mL/min/g))	138 [114–159]	118 [102–147]	140 [117–160]	<0.001
Stress CVR^B (mmHg/(mL/min/g))	46 [36–55]	61 [46–82]	44 [36–53]	<0.001
Stress-induced myocardial ischemia	62 (18)	12 (20)	50 (18)	0.71

Unless indicated otherwise, data are presented as the median [interquartile range] or n (%). ^ACorrected resting MBF was calculated as (resting MBF×8,500)/resting RPP, where RPP is the rate-pressure product; corrected MFR was calculated as (stress MBF/corrected resting MBF). ^BStress/resting CVR was calculated by dividing stress/resting mean arterial pressure by MFR. CCTA, coronary computed tomographic angiography; CMD, coronary microvascular dysfunction; CVR, coronary vascular resistance; DBP, diastolic blood pressure; EDV, end-diastolic volume; eGFR, estimated glomerular filtration rate; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; LVEF, left ventricular ejection fraction; MBF, myocardial blood flow; MFR, myocardial flow reserve; PET, positron emission tomography; SBP, systolic blood pressure.

PET analysis revealed a median resting ejection fraction of 71% (IQR 64–77%), a median SSS of 4 points (IQR 2–7 points), and a median resting RPP of 8,820 mmHg · beats/min (IQR 7,425–10,656 mmHg·beats/min). Median stress MBF, resting MBF, MFR, and corrected MFR were 2.60 mL/min/g (IQR 2.17–3.00 mL/min/g), 0.95 mL/min/g (IQR 0.79–1.14 mL/min/g), 2.72 (IQR 2.20–3.31), and 2.79 (IQR 2.31–3.45), respectively. Stress-induced myocardial ischemia was observed in 62 (18%) patients. In contrast to patients without CMD, those with CMD were older (P<0.001), more often female (P<0.006), and had higher rates of hypertension (P=0.021) and anemia (P=0.001). Patients with CMD also had higher resting RPP (P=0.01), resting MBF (P<0.001), and stress CVR (P<0.001); however, the corrected resting MBF was similar between the 2 groups (P=0.06). Stress MBF (P<0.001), MFR (P<0.001), and resting CVR (P<0.001) were significantly lower in patients with than without CMD. However, stress-induced myocardial ischemia was comparable between the 2 groups (P=0.71).

Clinical Factors to Predict MFR

Potential determinants of a decrease in MFR were evaluated from conventional clinical risk factors and PET-related indices in all patients (Table 2). In univariate analysis, age, sex (female), hypertension, dyslipidemia, chronic kidney disease, anemia, current smoking, SSS, and resting RPP were significantly associated with a decrease in MFR. However, there was no association between a decrease in MFR and obesity, symptoms, and diabetes. Multivariable analysis showed that age, sex, hypertension, and anemia remained statistically significant independent predictors of MFR (Model 1), with R²=0.32 (adjusted R²=0.30) for MFR. When PET-related variables such as SSS and resting RPP were included in Model 2, SSS and resting RPP were also found to be independently related to MFR (R²=0.37; adjusted R²=0.35).

Table 2.

Univariable and Multivariable-Adjusted Associations of Clinical Factors and MFR

	Univariable analysis				Multivariable analysis: Model 1				Multivariable analysis: Model 2
	β	95% LCI	95% UCI	P value	β	95% LCI	95% UCI	P value	β	95% LCI	95% UCI	P value
Age (10-year increments)	−0.452	−0.383	−0.248	<0.0001	−0.332	−0.304	−0.159	<0.0001	−0.289	−0.273	−0.130	<0.0001
Female (vs. male)	−0.283	−0.643	−0.303	<0.0001	−0.262	−0.599	−0.274	<0.0001	−0.237	−0.558	−0.234	<0.0001
Obesity (vs. normal weight)
Underweight	0.014	−0.159	0.209	0.34	−0.059	−0.265	0.063	0.23	−0.050	−0.253	0.064	0.24
Overweight or obese	−0.052	−0.660	0.232	0.79	−0.013	−0.447	0.336	0.78	0.004	−0.358	0.400	0.91
Symptom (vs. asymptomatic)
Typical angina	0.098	−0.119	0.459	0.25	0.101	−0.090	0.437	0.20	0.076	−0.124	0.386	0.32
Atypical angina	−0.131	0.495	0.061	0.13	−0.028	−0.296	0.202	0.71	−0.038	−0.305	0.178	0.61
Hypertension	−0.219	−0.544	−0.195	<0.0001	−0.122	−0.374	−0.030	0.01	−0.085	−0.308	0.020	0.09
Dyslipidemia	−0.125	−0.381	−0.033	0.02	−0.031	−0.206	0.100	0.50	−0.032	−0.201	0.095	0.48
Diabetes	−0.019	−0.243	0.167	0.72	0.038	−0.105	0.256	0.41	0.065	−0.050	0.302	0.20
Chronic kidney disease	−0.179	−0.516	−0.136	0.001	−0.045	−0.253	0.088	0.34	−0.008	−0.182	0.153	0.86
Anemia	−0.218	−0.856	−0.270	<0.0001	−0.158	−0.643	−0.172	0.001	−0.152	−0.621	−0.164	0.001
Current smoker	0.115	0.024	0.519	0.03	−0.004	−0.231	0.212	0.93	−0.001	−0.216	0.211	0.98
Summed stress score	−0.202	−0.063	−0.020	<0.0001					−0.151	−0.050	−0.012	0.001
Resting RPP (per 100-mmHg·beats/min increase)	−0.347	−0.014	−0.007	<0.0001					−0.186	−0.008	−0.003	<0.0001

LCI, lower confidence interval; UCI, upper confidence interval. Other abbreviations as in Table 1.

Figure 2 shows sex-stratified scatterplots of MFR plotted against several parameters of interest (i.e., age, BMI, hemoglobin levels, SSS, and resting RPP), alongside associations among resting RPP, stress MBF, and resting MBF. MFR was significantly negatively correlated with age (r=−0.45, P<0.0001), SSS (r=−0.20, P<0.0001), and resting RPP (r=−0.35, P<0.0001), significantly positively correlated with hemoglobin levels (r=0.27, P<0.0001), and was not correlated with BMI (r=0.10, P=0.61). Resting MBF was significantly correlated with resting RPP (r=0.47, P<0.0001) and tended to be higher in women than in men at all RPP levels, whereas no such correlation or tendency was observed for stress MBF (r=0.06, P=0.23).

Figure 2.

Sex-stratified correlations between myocardial flow reserve (MFR) and (A) age, (B) body mass index (BMI), (C) hemoglobin level, (D) summed stress score, (E) resting rate–pressure product (RPP), and (H) resting myocardial blood flow (MBF). (F,G) Correlations by sex between resting RPP and stress MBF (F) and resting MBF (G). CI, confidence interval.

Subgroup Analysis of Classical vs. Endogenous CMD

Of the 60 patients with CMD, 37 (62%) and 23 (38%) had classical and endogenous CMD, respectively. Intergroup comparisons are presented in Table 3. The proportion of women was significantly higher among patients with endogenous than classical CMD (96% vs. 60%, respectively; P=0.002). No significant intergroup differences were found in symptomatic status and medical history. Regarding PET-related parameters, resting RPP (P<0.001), stress RPP (P<0.001), stress MBF (P<0.001), and corrected MFR (P=0.001) were significantly higher in patients with endogenous than classical CMD. In contrast, SSS (P<0.001), resting CVR (P<0.001), and stress CVR (P<0.001) were significantly higher in patients with classical CMD. Corrected resting MBF (P=0.14) and MFR (P=0.61) were comparable between the 2 groups.

Table 3.

Patient Characteristics and PET-Related Parameters According to CMD Endotype

	Classical CMD (n=37)	Endogenous CMD (n=23)	P value
Demographic characteristics
Age (years)	74 [72–80]	79 [75–84]	0.07
Female sex	22 (60)	22 (96)	0.002
Symptomatic status
Typical angina	29 (78)	15 (65)	0.34
Atypical angina	5 (14)	7 (30)
Asymptomatic	3 (8)	1 (4)
Medical history
Hypertension	25 (68)	19 (83)	0.24
Dyslipidemia	19 (51)	14 (61)	0.59
Diabetes	9 (24)	7 (30)	0.76
Chronic kidney disease	1 (3)	2 (9)	0.55
Anemia	6 (16)	9 (39)	0.06
Atrial fibrillation	5 (14)	3 (13)	1.0
Current smoker	5 (14)	3 (13)	1.0
Hemodynamic parameters during PET
Resting heart rate (beats/min)	65 [56–76]	83 [70–92]	<0.001
Stress heart rate (beats/min)	70 [64–84]	89 [78–98]	<0.001
Resting SBP (mmHg)	127 [117–147]	151 [136–165]	0.001
Stress SBP (mmHg)	105 [99–129]	126 [103–139]	0.09
Resting RPP (mmHg · beats/min)	8,052 [7,176–10,750]	12,640 [10,385–14,516]	<0.001
Stress RPP (mmHg · beats/min)	8,000 [6,466–9,750]	10,578 [9,426–12,449]	<0.001
PET imaging parameters
Resting EDV (mL)	80 [69–100]	75 [58–85]	0.16
Resting LVEF (%)	69 [61–77]	71 [64–84]	0.23
Summed stress score	7 [4–10]	3 [2–5]	<0.001
Summed rest score	4 [2–6]	2 [1–2]	0.001
Summed difference score	2 [1–4]	1 [0–2]	0.03
Resting MBF (mL/min/g)	0.98 [0.83–1.11]	1.50 [1.41–1.64]	<0.001
Corrected resting MBF^A (mL/min/g)	0.91 [0.73–1.21]	1.03 [0.87–1.26]	0.14
Stress MBF (mL/min/g)	1.55 [1.24–1.87]	2.56 [2.18–2.72]	<0.001
MFR	1.71 [1.46–1.87]	1.65 [1.48–1.80]	0.61
Corrected MFR^A	1.68 [1.19–2.01]	2.23 [1.81–2.78]	0.001
Resting CVR^B (mmHg/(mL/min/g))	140 [118–164]	102 [90–110]	<0.001
Stress CVR^B (mmHg/(mL/min/g))	73 [52–97]	48 [43–56]	<0.001
Stress-induced myocardial ischemia	11 (30)	1 (4)	0.02

Figure 3 shows representative images and findings for patients with classical and endogenous CMD. Focusing on symptomatic status (Figure 4), approximately one-quarter (24%) of 180 patients with typical angina had CMD, with the incidence of classical CMD (16%) being twice that of endogenous CMD (8%). However, the overall percentage (24%) of patients with CMD decreased significantly to only 10% among patients with atypical angina symptoms and in asymptomatic patients.

Figure 3.

Representative images and data for 2 symptomatic patients with (A–C) classical and (D–F) endogenous coronary microvascular dysfunction (CMD). (A,D) Invasive coronary computed angiography shows no significant stenosis in the coronary artery in each patient. (B,E) ATP-induced ¹³N-ammonia positron emission tomography (PET)-derived quantification of stress myocardial blood flow (MBF; Left), resting MBF (Center) and, myocardial flow reserve (MFR; Right) in the 17-segment standard American Heart Association model in 2 patients with classical (B) and endogenous (E) CMD. (C,F) Stress and resting quantitative assessments of myocardial perfusion ¹³N-ammonia PET based on the 3 major coronary territories with PET-derived parameters in 2 patients with classical (C) and endogenous (F) CMD; global MFR was lower than 2.0 in both patients (1.70 and 1.87, respectively). In contrast, resting MBF was 0.92 mL/min/g (<1.3) and 1.52 mL/min/g (≥1.3) in the patients with classical and endogenous CMD, respectively. CVR, coronary vascular resistance; EF, ejection fraction; LAD, left anterior descending coronary artery; LCX, left circumflex coronary artery; RCA, right coronary artery; RPP, rate–pressure product; SDS, summed difference score; SRS, summed rest score; SSS, summed stress score.

Figure 4.

Prevalence of patients with classical and endogenous coronary microvascular dysfunction (CMD) overall and in subsets according to symptom status at initial positron emission tomography.

Discussion

This study investigated the incidence of PET-assessed CMD in patients with non-obstructive CAD, evaluated the clinical factors and PET-related parameters contributing to CMD, and characterized two distinct CMD endotypes (classical and endogenous) based on PET flow data. The major findings of the study are as follows. First, of 683 patients with suspected or known CAD, 345 (51%) were found to have non-obstructive CAD, of whom 60 (17%) had CMD, with classical CMD (62%) being more prevalent than endogenous CMD (38%). Second, among the clinical factors evaluated, older age, female sex, anemia, and hypertension were independent predictors of a reduction in MFR. Older age, female sex, and anemia, remained significant predictors of a reduction in MFR after the addition of PET-related parameters, namely higher SSS and greater resting RPP, to the model. Third, these clinical factors accounted for only 30% of the observed variability in MFR; furthermore, after the addition of PET-related parameters, the factors combined only accounted for less than 40% of the observed variability in MFR. Fourth, patients with classical CMD were more likely to have higher SSS and resting and stress CVR, whereas patients with endogenous CMD were more likely to be female and have considerably higher stress MBF and resting RPP. These findings imply that non-invasive CMD assessments using myocardial perfusion PET could potentially accurately diagnose CMD and its endotypes, and contribute to the exploration of potential mechanisms of CMD beyond invasive CMD assessments, leading to the development of specific therapies or effective prognostic stratifications.

The incidence of non-obstructive CAD (51%) in the present study was consistent with that reported in previous studies.¹^,⁴ The prevalence of CMD among patients with non-obstructive CAD has previously been reported to range from 26% to 59%,³ with prevalence varying depending on the testing modality.³ In the present study, PET-assessed CMD was detected in 17% of patients, which is slightly lower than that reported in a similar study using ¹³N-ammonia PET¹² in which the PET-assessed prevalence was 43%. This discrepancy is likely due to differences in patient demographics, referral criteria, and geographic variations. Patients in the previous study were younger, with a mean age of 51 years, and all presented with typical angina.¹² In the present study, most patients (73%) with CMD presented with typical angina. The incidence of CMD was highest among patients with typical angina than among those with atypical angina or asymptomatic patients. As discussed previously, CMD may play a significant role in the development of angina pectoris and microvascular angina.²^,¹⁰^,¹⁷ In contrast, recent studies³^,⁴ showed that symptom characteristics were neither sensitive nor specific, and may not reliably identify patients with CMD. This aligns with our finding in the multivariable analysis that symptoms were not significantly associated with a decrease in MFR. Therefore, CMD cannot be ruled out based solely on symptom characteristics and their severity.⁴

We explored potential clinical factors related to a decrease in MFR and found that older age, female sex, hypertension, and anemia were independent predictors of a decrease in MFR, whereas BMI (obesity) was not. Aging is associated with increased pulse pressure, and arterial remodeling leads to impaired MFR through increased arterial wall stiffness, media thickening, and lumen enlargement.¹⁸ A myocardial perfusion PET study⁷ in healthy subjects showed that resting MBF increased with age, mainly because of enhanced RPP and decreased stress MBF in patients aged ≥65 years; subsequently, the combination could lead to a significant decrease in MFR. Women are more likely than men to have angina and non-obstructive CAD with CMD.¹⁸^,¹⁹ In fact, in the present study, 73% of patients with CMD were female. A study using invasive thermodilution found that microvascular function was similar between the sexes and that, instead, regulation of resting coronary flow differed because women were found to have a shorter resting mean transit time, whereas hyperemic mean transit times were similar between the sexes.¹⁹ In the present study, women clearly had a higher resting MBF at all levels of resting RPP and MFR. These findings are consistent with those of previous reports using myocardial perfusion PET.⁷ Notably, female sex remained a significant factor after adjusting for resting RPP, suggesting that female sex could be related to a decrease in MFR beyond increased resting coronary flow. Women may be particularly susceptible to vasomotor disorders and CMD due to unique factors such as inflammation, mental stress, autonomic dysfunction, and neuroendocrine dysfunction, all of which contribute to endothelial dysfunction and increase the risk of CMD.¹⁹^,²⁰ Although it has been suggested that there may be differences in the pathophysiology leading to the development of CMD between men and women,²¹ these concepts are beyond the scope of the present study.

In this study, anemia was also identified as a potential contributor to reduced MFR, which has not been reported previously, except in patients with end-stage renal disease.²² As a potential mechanism, reduced hemoglobin may impair the oxygen-carrying capacity of blood going to the myocardium, leading to microvascular ischemia and adverse microvascular remodeling, and causing CMD.

Importantly, traditional cardiovascular risk factors, including hypertension, hyperlipidemia, older age, obesity, smoking, and diabetes, are associated with CMD.²^,⁷^,¹⁰ However, these factors are not always present in patients with CMD.¹⁸ In the WISE⁵ and iPOWER⁴ studies, cardiovascular risk factors accounted for <20% of the observed variability in flow reserve. In the present study, these clinical factors explained only 32% of the observed variance, suggesting that additional unidentified factors must play a key role in CMD.⁴

In the present study, resting RPP was positively correlated with resting MBF, which indicated a negative association between MFR and resting RPP. In general, MBF increases the myocardial workload to meet myocardial oxygen or metabolic demands under resting conditions. Thus, conditions that increase myocardial oxygen consumption, such as arterial hypertension, increased myocardial contractility, increased left ventricular wall stress, sympathetic activation, and tachycardia, lead to increases in resting MBF.⁸ Previous studies reported a close correlation between resting MBF and resting RPP.⁸^,⁹^,¹¹ Notably, resting RPP, which may be provoked by a patient’s nervousness and/or anxiety during PET scans, could represent an important confounding factor leading to an increase in resting MBF, as observed in the present study. Given the lack of established standardized methods for addressing this issue, it may be reasonable to correct resting MBF using resting RPP. In fact, the corrected resting MBF values in our cohort, both with and without CMD, were within the normal range. More importantly, we believe that the cardiac workload should be considered in invasive CMD assessments to ensure accurate comparisons of coronary circulatory function by correcting or normalizing the resting MBF with the resting RPP.

Positive stress test results in patients without evidence of obstructive CAD are typically considered indicative of CMD. However, our study contradicts this notion, because we found no difference in the incidence of stress-induced myocardial ischemia between patients with and without CMD, highlighting a low incidence of 20% in the CMD group. In addition, no differences were found in the SSS or summed difference scores. These observations are in line with a previous large-scale study showing no association between the results of stress testing and invasively assessed CMD.¹ These findings suggest that CMD can present ischemia even without positive stress testing, mainly because stress testing would be mostly based on the demonstration of regional rather than diffuse ischemia⁴ and myocardial ischemia due to CMD commonly may be distributed diffusely or homogeneously.

Regarding CMD endotypes, classical CMD was more prevalent than endogenous CMD in the present study, which is comparable to the results of a previous study using ¹³N-ammonia PET.¹¹ Notably, the endogenous CMD group had a higher proportion of women; otherwise, no distinct difference was observed in baseline medical history and symptomatic status between the two CMD endotype groups. Impaired MFR in patients with endogenous CMD was primarily due to increased resting MBF and normal stress MBF. In contrast, impaired MFR in patients with classical CMD was due to decreased stress MBF and normal resting MBF. The higher incidence (30%) of myocardial ischemia and higher SSS in patients with classical CMD suggest that reduced stress MBF may be reflected by a diffuse plaque burden in epicardial vessels or a focal/regional burden of impaired microcirculation.⁸ In addition, stress and resting CVR were significantly higher in patients with classical CMD than in those with endogenous CMD. These proposed CMD classifications based on PET flows (classical and endogenous CMD) are conceptually close to two distinct entities of invasively assessed CMD (structural and functional CMD, respectively), although they are not equivalent because of methodological differences.⁶^,¹⁷^,²³ In functional CMD, low coronary flow reserve arises from increased resting coronary blood flow, indicating reduced resting microvascular resistance; in structural CMD, low coronary flow reserve arises from reduced hyperemic flow due to elevated hyperemic microvascular resistance.⁶^,¹⁰^,²⁴ In fact, in 86 patients with angina and non-obstructive CAD undergoing coronary pressure and flow measurement at rest, during exercise, and under the conditions of adenosine-mediated hyperemia, functional CMD was characterized by elevated resting flow linked to enhanced nitric oxide synthase activity.¹⁷ In contrast, patients with structural CMD had endothelial dysfunction, which led to diminished peak coronary blood flow augmentation and increased demand during exercise.¹⁷ These characterizations of CMD endotypes could help with our understanding of the underlying pathophysiology and improve treatments and clinical outcomes.

Study Limitations

This study has some limitations. First, this was a single-center retrospective observational study, which included a relatively large sample size; however, many patients were excluded because they did not meet the inclusion criteria. These factors may limit the generalizability of our findings. Second, patients referred for myocardial perfusion PET are generally symptomatic and have comorbidities, which may add to referral bias. However, this may reflect real-world practice rather than what is observed in randomized trials. In addition, at Sapporo Kojinkai Memorial Hospital, in which both CCTA and myocardial perfusion PET are available, myocardial perfusion PET is mainly indicated when the CCTA reveals complex/multiple vessel obstructive CAD, has inconclusive findings (i.e., severe calcified lesions, in-stent lesions, inadequate imaging quality), or reveals non-obstructive CAD, requiring evaluation of myocardial ischemia and/or CMD, based on guidelines.²⁵ The diagnostic approach for patients with stable CAD and CMD is standardized at Sapporo Kojinkai Memorial Hospital, reducing physician discretion in this context. Thus, we believe that the referral bias would minimally affect the results and that the findings reflect real-world data. Third, we evaluated one aspect of CMD: ATP-induced impairment of MFR, which is mainly caused by dysfunction of endothelium-independent vasodilation in the microcirculation. We did not assess endothelium-dependent epicardial dysfunction or microvascular spasm, which can be evaluated using invasive diagnostic procedures.⁴ Therefore, the incidence of CMD may have been underestimated in the present study. Fourth, our study population is not identical to patients with angina/ischemia and non-obstructive CAD. Symptomatic status or evidence of myocardial ischemia were not accounted for in the inclusion criteria, which may have affected the incidence of CMD in this study. Fifth, we could not make prognostic comparisons between patients with classical and endogenous CMD because of the small sample size in each group. A difference in prognosis may further enhance the clinical importance of diagnosing CMD endotypes, as discussed previously.²⁶ Finally, the lack of correlation between invasively assessed and PET-assessed CMD may limit the broad clinical relevance of our findings; further studies are required to develop more accurate diagnostic approaches, whether invasive or non-invasive, by focusing on comparisons among these diagnostic tools.

Conclusions

Among patients with non-obstructive CAD undergoing myocardial perfusion PET, approximately 1 in 5 had PET-assessed CMD. Of the clinical factors, age, female sex, hypertension, and anemia were independent predictors of a decrease in MFR. However, these factors only partially explained the observed variability in MFR, suggesting the presence of additional unknown contributors to MFR reductions. We highlighted that the resting RPP during PET scans is a key factor that considerably affects MBF. Among patients with lower stress MBF and higher stress CVR, classical CMD was prominent; in contrast, endogenous CMD was characterized by a dominance of female sex and higher resting RPP, resting MBF, and stress MBF. We believe that these CMD classifications (classical and endogenous CMD) are relevant to the underlying mechanisms of CMD by referencing invasively assessed structural and functional CMD, respectively. Our research would be highly beneficial in identifying patients with CMD by providing noble insights into whether patients should proceed to further invasive CMD assessments and highlights the unique pathophysiology of each CMD endotype.

Acknowledgments

The authors thank Editage (https://www.editage.com/) for English language editing a draft of this paper.

Sources of Funding

This study did not receive any specific funding.

Disclosures

None.

IRB Information

The study was approved by the Institutional Review Board of Sapporo Kojinkai Memorial Hospital (Approval no. 2019-7).

Data Availability

Deidentified participant data will be shared upon request directly to the corresponding author. Datasets and study protocols can be requested and will be shared upon approval from the IRB of Sapporo Kojinkai Memorial Hospital. The data will remain available until the end of March 2029. Any analyses on the data will be approved and data will be shared as Excel files via email.

References

責任著者(Corresponding author)

J-STAGEへの登録はこちら（無料）