Continuous Versus Bolus Thermodilution‐Derived Coronary Flow Reserve and Microvascular Resistance Reserve and Their Association With Angina and Quality of Life in Patients With Angina and Nonobstructive Coronaries: A Head‐to‐Head Comparison

Abstract

Background

Coronary flow reserve (CFR) and microvascular resistance reserve (MRR) are physiological parameters to assess coronary microvascular dysfunction. CFR and MRR can be assessed using bolus or continuous thermodilution, and the correlation between these methods has not been clarified. Furthermore, their association with angina and quality of life is unknown.

Methods and Results

In total, 246 consecutive patients with angina and nonobstructive coronary arteries from the multicenter Netherlands Registry of Invasive Coronary Vasomotor Function Testing (NL‐CFT) were investigated. The 36‐item Short Form Health Survey Quality of Life and Seattle Angina questionnaires were completed by 153 patients before the invasive measurements. CFR and MRR were measured consecutively with bolus and continuous thermodilution. Mean continuous thermodilution‐derived coronary flow reserve (CFR_abs) was significantly lower than mean bolus thermodilution‐derived coronary flow reserve (CFR_bolus) (2.6±1.0 versus 3.5±1.8; P<0.001), with a modest correlation (ρ=0.305; P<0.001). Mean continuous thermodilution‐derived microvascular resistance reserve (MRR_abs) was also significantly lower than mean bolus thermodilution‐derived MRR (MRR_bolus) (3.1±1.1 versus 4.2±2.5; P<0.001), with a weak correlation (ρ=0.280; P<0.001). CFR_bolus and MRR_bolus showed no correlation with any of the angina and quality of life domains, whereas CFR_abs and MRR_abs showed a significant correlation with physical limitation (P=0.005, P=0.009, respectively) and health (P=0.026, P=0.012). In a subanalysis in patients in whom spasm was excluded, the correlation further improved (MRR_abs versus physical limitation: ρ=0.363; P=0.041, MRR_abs versus physical health: ρ=0.482; P=0.004). No association with angina frequency and stability was found.

Conclusions

Absolute flow measurements using continuous thermodilution to calculate CFR_abs and MRR_abs weakly correlate with, and are lower than, the surrogates CFR_bolus and MRR_bolus. Absolute flow parameters showed a relationship with physical complaints. No relationship with angina frequency and stability was found.

Nonstandard Abbreviations and Acronyms

ANOCA

angina and no obstructive coronary artery disease

CCA

classification agreement

CFR

coronary flow reserve

CFR_abs

continuous thermodilution‐derived coronary flow reserve

CFR_bolus

bolus thermodilution‐derived coronary flow reserve

CFT

coronary function test

CMD

coronary microvascular dysfunction

IMR

index of microvascular resistance

MRR

microvascular resistance reserve

MRR_abs

continuous thermodilution‐derived microvascular resistance reserve

MRR_bolus

bolus thermodilution‐derived microvascular resistance reserve

Q

absolute coronary flow (in mL/min)

T _mn

mean transit time

Clinical Perspective.

What Is New?

• Absolute flow measurements using continuous thermodilution to calculate continuous thermodilution‐derived coronary flow reserve and continuous thermodilution‐derived microvascular resistance reserve correlate with, but are lower than, the surrogates bolus thermodilution‐derived coronary flow reserve and bolus thermodilution‐derived microvascular resistance reserve.

• For the first time a relationship between intracoronary physiological parameters and physical limitation and health in angina and no obstructive coronary artery disease was found: continuous thermodilution‐derived coronary flow reserve and continuous thermodilution‐derived microvascular resistance reserve correlate with Seattle Angina Questionnaire and Short Form Health Survey physical complaint scores, whereas bolus thermodilution‐derived coronary flow reserve and bolus thermodilution‐derived microvascular resistance reserve.

What Are the Clinical Implications?

• After being proven feasible, safe, and highly reproducible, the continuous thermodilution method to assess microvascular function now correlates with physical symptoms and might therefore be a superior measurement reflecting the disease status of patients with coronary microvascular dysfunction compared with bolus thermodilution.

Coronary microvascular dysfunction (CMD) is a highly prevalent condition in patients with chest pain. ¹ Until recently, assessment of CMD relied on 2 parameters: coronary flow reserve (CFR) and microvascular resistance (index of microvascular resistance [IMR]/hyperemic microvascular resistance). ² Both can be measured during invasive coronary angiography with the use of adenosine to induce maximal hyperemia and minimal resistance. ³ However, current methods to determine these parameters have several shortcomings and are only surrogates for true coronary blood flow, which withholds widespread adoption. ⁴

This has led to the development of the less operator‐dependent method continuous thermodilution, which enables direct quantification of true (absolute) coronary blood flow (Q) and microvascular resistance (absolute coronary resistance in Woods units). ⁵ These indices correlate well with absolute flow derived from positron emission tomography (PET) ⁶ and are associated with severity of angina. ⁷ However, only hyperemic state measurements were validated, precluding the measurement of indices such as CFR. This makes interpatient comparisons difficult.

However, with the recent validation of resting measurements when using continuous thermodilution by low infusion of saline (10 mL/min), it has become possible to assess both absolute CFR and a new index of microvascular function: microvascular resistance reserve (MRR). ⁸ , ⁹ MRR is the ratio of true resting microvascular resistance (not confounded by epicardial disease) and hyperemic microvascular resistance. ⁹

To date, these continuous thermodilution‐derived parameters have not been compared with the invasive standard of assessing microvascular function using bolus thermodilution. Furthermore, no large studies are available on the relationship of angina symptoms and quality of life (QoL) with bolus or continuous thermodilution‐derived CFR and MRR. Because angina is the limiting complaint for these patients, insights into the association between symptoms and the physiological parameters of CMD will help to achieve optimal treatment and evaluation of treatment success.

Therefore, the aim of this exploratory study was first to investigate the relationship between these novel parameters of absolute CFR and MRR as compared with bolus thermodilution‐derived surrogate indices. Second, we explored the relationship of both the absolute and surrogate values of CFR and MRR with angina and QoL.

METHODS

Study Population

This study was conducted within the Netherlands Registry of Invasive Coronary Vasomotor Function Testing (NL‐CFT) of patients with angina and nonobstructive coronary artery disease (ANOCA) undergoing clinically indicated coronary function testing (CFT) for suspected coronary vasomotor disease. ¹⁰ In total, 246 consecutive patients who underwent comprehensive coronary physiology measurements, including continuous thermodilution resting flow measurements at the Radboud University Medical Center in Nijmegen, Catharina Hospital Eindhoven and Maasstad Hospital, Rotterdam, were included.

Written informed consent was obtained in all patients and local institutional review board approval requirement was waived, because this study is a registry and not a clinical trial. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Study Design

Patients participating in the NL‐CFT were asked to fill in the Seattle Angina Questionnaire (SAQ) and Short Form Health Survey (SF‐36) questionnaire just before the CFT.

The SAQ consists of 5 domains: physical limitation, angina stability, angina frequency, treatment satisfaction, and QoL, and the SF‐36 comprises 36 questions covering 8 QoL dimensions related to physical functions and well‐being. ¹¹ , ¹²

All patients underwent continuous thermodilution measurements as part of the CFT. The CFT was performed according to a standardized protocol as described previously by Konst et al ¹³ In short, patients were instructed to withhold all vasoactive medication for 24 to 48 hours before the procedure. After excluding obstructive coronary artery disease, a stepwise dose escalating acetylcholine provocation test was performed to evaluate epicardial and or microvascular coronary spasm. Next, an intracoronary bolus of at least 300 μg of nitroglycerin was administered and repeated in case of persisting complaints after 1 minute. Subsequently, a pressure–temperature guidewire (PressureWireX Guidewire; Abbott, Abbott Park, IL) was advanced into the distal third part of the left anterior descending artery. Adequate pressure and temperature equalization were ensured. At least 2 minutes after nitroglycerin infusion, to ensure resting state, the following measurements were systematically performed by protocol:

1. Bolus thermodilution technique (surrogate values). At least 3 consecutive thermodilution curves were obtained by brisk injection of 3 mL of saline at room temperature into the coronary artery. Mean transit times (T _mn) of the distal temperature decrease were acquired at rest and maximal hyperemia, induced by peripheral intravenous infusion of adenosine (140 to 180 µg/kg per min). Thermodilution curves were considered adequate when a unimodal shape without distortion and with <20% variation between 3 measurements was observed. Measurements with poor quality curves were excluded and repeated.

2. Continuous thermodilution technique (absolute values). ¹⁴ After waiting at least 2 minutes to ensure that the hyperemic effect of the bolus thermodilution method disappeared, a dedicated rapid exchange coronary infusion microcatheter, Rayflow (Hexacath, France), was positioned in the proximal segment of the vessel. The pressure–temperature sensor was located at the same position as with bolus thermodilution, 3 to 6 cm distal from the tip of the RayFlow catheter. Then, low infusion rate of saline (10 mL/min) at room temperature was started, using an automated infusion system. The infusion rate was lowered in case signs of hyperemia (dropping of the resting P _d/P _a value). After obtaining a steady temperature state (T), the pressure/temperature wire was gently pulled back to the tip of the infusion catheter to assess the saline infusate temperature. To acquire hyperemic Q and R, the same protocol was repeated with increased saline infusion (20 mL/min). ¹⁵

Calculations

Surrogate and Absolute Flow

The bolus thermodilution‐derived surrogate of flow is calculated using the inverse of the mean transit time of the saline bolus: flow ~1/T _mn.

Absolute coronary flow (Q) (in mL/min) was calculated, using the following formula: ¹⁶

$$Q=Q_i\bullet \frac{T_i}{T}\bullet 1.08$$

where Q _i is the infusion rate of saline by the infusion pump (10 mL/min at rest and 20 mL/min at hyperemia), T _i is the temperature of the infused saline when it exits the infusion catheter, and T is the temperature of the mixture of blood and saline in the distal part of the coronary artery during steady‐state infusion, both as a difference to blood temperature before infusion. The correction factor 1.08 is necessary to compensate for the difference in specific heat between saline and blood. ¹⁷

Coronary Flow Reserve

The bolus thermodilution‐derived CFR (CFR_bolus) was determined by dividing the resting T _mn by the hyperemic T _mn.

The continuous thermodilution‐derived CFR (CFR_abs) was determined by dividing the average absolute coronary flow (Q) at rest by the average absolute flow during hyperemia.

For CFR, a cutoff value of 2.0 was used, with lower values indicating abnormal flow reserve as defined by current consensus documents. ² Furthermore, an additional analysis was performed using a cutoff of 2.5 for CFR_bolus, as proposed by Demir et al. ¹⁸

Microvascular Resistance Reserve

The MRR was calculated according to the theoretical framework by de Bruyne et al, which demonstrates that the MRR is independent of epicardial disease and independent of variation of driving pressure between resting and hyperemic measurements. ⁹

$$\mathrm{MRR}=\frac{Q_{\max }}{Q_{\text{rest}}}·\frac{P_{\mathrm{a},\mathrm{hyp}}}{P_{\mathrm{d},\mathrm{hyp}}}·\frac{P_{\mathrm{a},\text{rest}}}{P_{\mathrm{a},\mathrm{hyp}}}$$

where the first term is classical CFR, the second term compensates for epicardial disease, and the third term for hemodynamic variations. The latter equation can also be rewritten as:

$$\mathrm{MRR}=\frac{\mathrm{CFR}}{\mathrm{FFR}}·\frac{{\mathrm{P}}_{\mathrm{a},\text{rest}}}{{\mathrm{P}}_{\mathrm{a},\mathrm{hyp}}}$$

FFR indicates fractional flow reserve; and P _a, aortic pressure.

MRR can be calculated for both the bolus thermodilution method (MRR_bolus) and the continuous thermodilution method (MRR_abs). However, due to the infusion of adenosine that is needed for the assessment of MRR_bolus, a difference of 10% to 30% in P _a,rest and P _a,hyp is observed, whereas intracoronary infusion of saline, needed for MRR_abs, is not expected to lead to difference in aortic pressure during rest and hyperemia. ⁹

For MRR, we used the cutoff of 2.5 to determine abnormal versus normal, as was previously described by our study group. ¹⁹

All calculations were performed by dedicated software (Coroventis Coroflow, Uppsala, Sweden).

Statistical Analysis

Continuous variables are presented as mean ± SD or as median (interquartile range). Categorical variables are presented as frequency (percentage). Comparisons were analyzed using the Student's t test (for approximately normal distribution) and the Mann–Whitney U test (for nonnormal distribution). The correlation between indexes was estimated by calculating the Pearson correlation coefficient (r), and agreement between indexes was assessed using the classification agreement (CCA), Cohen's kappa coefficient, and Bland–Altman plots of the relative differences.

The SAQ and SF‐36 scores were transformed to a score of 0 to 100, where higher scores indicate better function. Both questionnaires were averaged to provide an overall metric of angina and QoL: the SAQ into the SAQ summary score and the SF‐36 in a physical health and a mental health component (Table S1). ¹¹ , ¹² Poor health status was defined as an SAQ summary score below 50. SAQ Angina Frequency scores of 0 to 30 points translate to having daily angina, 31 to 60 points to having weekly angina, 61 to 99 points to having monthly angina, and 100 points to having no angina. ¹¹ Based on the study by Beatty et al, ²⁰ a 5‐unit difference in SAQ health limitation was considered clinically relevant.

The following analyses were performed:

1. Comparison of bolus versus continuous thermodilution resting and hyperemic flow.

2. Comparison of bolus versus continuous derived CFR.

3. Comparison of bolus versus continuous derived MRR.

4. Bolus versus continuous thermodilution‐derived CFR and MRR for the different SAQ domains and the SF‐36 physical health and mental health.

Subgroup analyses of the fourth item were performed in patients with either proven absence of epicardial vasospasm or proven absence of both epicardial and microvascular spasm, to preclude the influence of complaints by spasm. ²

A 2‐sided P value <0.05 was considered statistically significant. Analyses were performed with SPSS and GraphPad Prism 7.

RESULTS

Clinical and Procedural Characteristics

A total of 246 patients underwent a CFT between February 2020 and May 2022. Mean age was 59±9 years and 204 (83%) were female. In 151 patients a noninvasive ischemia detection test was available, of which 20% were positive for ischemia. Furthermore, of the total cohort, 19% had a history of myocardial infarction, 50% had a history of hypertension, 45% had dyslipidemia, and 11% had diabetes. Nonobstructive coronary artery disease was confirmed, as displayed by a mean fractional flow reserve of 0.90±0.05. However, 7 patients had a fractional flow reserve < 0.80 based on diffuse disease, of whom 4 had a discordant normal relative flow reserve >0.90.

All patients were referred because of chest complaints, either classified as angina or shortness of breath, with a poor health status in 43% of the patients. In total, 153 patients filled in the SAQ and SF‐36, of whom 15% had a prior positive ischemia detection test.

Further baseline characteristics are presented in Table 1 and Table S2.

Table 1.

Patient Characteristics

	Total cohort N=246
Age, y	58.6	±8.7
Female sex	209	(85%)
At least weekly angina (SAQ frequency <66)		(61%)
Poor health status (SAQ summary score <50)		(43%)
Relevant medical history
History of myocardial infarction	44	(18%)
History of percutaneous coronary intervention	30	(12%)
Previous angiography	169	(71%)
Previous coronary computed tomography	84	(36%)
Noninvasive detection test performed/available	151	(62%)
Ischemia on noninvasive detection test	29	(12%)
Autoimmune disorder	31	(13%)
Asthma	37	(15%)
Chronic obstructive pulmonary disease, Global Initiative for Chronic Obstructive Lung Disease III to IV	2	(0.8%)
Glomerular filtration rate <30	1	(0.4%)
Creatinine, μmol/L	65	±14
Left ventricular ejection fraction	58	±5
Cardiovascular risk factors
Hypertension	123	(50%)
Dyslipidemia	111	(46%)
Diabetes	28	(12%)
Current smoker	17	(7%)
Former smoker	82	(34%)
Premature coronary artery disease in first‐degree relative	119	(49%)
Cerebrovascular accident/transicent ischemic attack	22	(9%)
Body mass index	27.4	±4.6
Medication use
Aspirin	71	(29%)
Statins	97	(39%)
Nitrates	58	(24%)
Betablockers	62	(25%)
Calcium channel blocker	136	(55%)
Angiotensin‐converting enzyme inhibitor/angiotensin receptor blocker	86	(35%)

SAQ indicates Seattle Angina Questionnaire.

Comparison of Bolus Versus Continuous Thermodilution Resting and Hyperemic Flow

As displayed in Table 2, mean bolus thermodilution‐derived surrogate of resting flow (1/T _mn) was 1.6±1.0 1/s and mean hyperemic surrogate of flow was 4.7±2.6 1/s. Mean continuous thermodilution‐derived absolute resting Q was 85±34 mL/min and mean hyperemic Q was 210±80 mL/min. There was no significant correlation between bolus versus continuous thermodilution resting‐ and hyperemic flow measurements (Table 2).

Table 2.

Procedural Results

Bolus thermodilution		Continuous thermodilution		Correlation (P value)
1/Tmn rest	1.6±1.0	Resting absolute coronary flow (mL/min)	85±35	0.109	0.09
1/Tmn hyp	4.7±2.5	Hyperemic absolute coronary flow (mL/min)	212±82	0.107	0.10
FFR	0.90±0.05	FFR	0.89±0.09	0.419*	0.001*
Index of microvascular resistance	22.2±11.9	Absolute resistancehyp (WU)	450±156	0.181*	0.05*
CFR	3.45±1.76	CFR	2.63±0.95	0.305*	0.001*
MRR	4.2±2.5	MRR	3.1±1.1	0.280*	0.001*

Correlations are presented as Pearson's correlation coefficient. CFR indicates coronary flow reserve; FFR, fractional flow reserve; MRR, microvascular resistance reserve; T _mn, mean transit time; and WU, Wood units.

^*Represents a signficant correlation coëfficient.

Comparison of Bolus Versus Continuous Derived CFR

CFR_abs displayed a smaller range than CFR_bolus (1.0–7.0 versus 0.6–9.9). Mean CFR_abs was significantly lower than CFR_bolus (2.6±1.0 versus 3.5±1.8; P<0.001) (Figure 1A). A weak correlation was found between CFR_bolus and CFR_abs (ρ=0.305; P<0.001) (Figure 1B). The corresponding Bland–Altman plot demonstrated a bias toward overestimation of CFR by CFR_bolus 0.9±3.4 (Figure 1C). The mean bias was not constant throughout the range of values, with the highest differences in the highest range of values.

Figure 1. Bolus thermodilution CFR and MRR versus continuous thermodilution CFR and MRR in patients with ANOCA.

A and D, Dots and whiskers on both sides indicate means and SDs, respectively. B and E, The x axis represents continuous thermodilution values, and the y axis represents bolus thermodilution values. Black lines indicate the regression lines of both CFRs (B) and MRRs (E). The Pearson correlation was reported with the CCA. C and F, The differences given were calculated as follow‐up values minus baseline values. Blue lines indicate within 2 SDs, black dotted stripes indicate bias. ANOCA indicates angina and no obstructive coronary artery disease; CCA, classification agreement; CFR, coronary flow reserve; and MRR, microvascular resistance reserve.

The CCA of CFR between bolus and continuous thermodilution was 68% when a CFR cutoff of 2.0 was used and 66% when 2.5 was used; however, the Cohen's kappa coefficient was higher for a bolus CFR cutoff of 2.5: Κ=0.086 (P=0.001) in comparison with 2.0: Κ=0.058 (P=0.035; Table 3).

Table 3.

Classification Agreement

		Continuous thermodilution		Cohen's kappa coefficient
Bolus Thermodilution		CFR>2.0	CFR<2.0
	CFR>2.0	143 (60%)	46 (19%)	Κ=0.058, P=0.035
	CFR≤2.0	31 (13%)	20 (8%)
	CFR>2.5	126 (52%)	33 (14%)	Κ=0.086, P=0.001
	CFR≤2.5	48 (20%)	33 (14%)
		MRR>2.5	MRR<2.5
	MRR>2.5	132 (56%)	47 (20%)	Κ=0.215, P < 0.001
	MRR≤2.5	29 (12%)	29 (12%)

CFR indicates coronary flow reserve; and MRR, microvascular resistance reserve. For bolus thermodilution CFR both a cutoff of 2.0 and 2.5 are presented. The agreement, as represented by Cohen's kappa coefficient, is the highest between MRR_bolus and MRR continuous.

Comparison of Bolus Versus Continuous Derived MRR

MRR_abs displayed a smaller scatter than MRR_bolus (0.7 to 22.4 versus 1.0 to 8.2). Furthermore, mean MRR_abs was also significantly lower than MRR_bolus (3.1±1.1 versus 4.2±2.5; P<0.001) (Figure 1D). A weak correlation was found between MRR_bolus and MRR_abs (ρ=0.276; P<0.001) (Figure 1E). The corresponding Bland–Altman plot demonstrated a bias towards overestimation of MRR by MRR_bolus 1.2±5.2 (Figure 1F). The CCA of MRR based on a cutoff of 2.5, between bolus and continuous thermodilution was 68%.

The correlationns between indexes did not improve in the patients with ischemia on noninvasive detection testing (CFR ρ=0.174, P=0.36, MRR ρ=0.217, P=0.25). Furthermore, the correlation between bolus thermodilution‐derived resistive reserve ratio versus MRR_abs was similar to MRR_bolus versus MRR_abs (Figure S1).

Bolus Versus Continuous Thermodilution‐derived CFR and MRR and Association with Angina and QoL

In 153 patients baseline SAQ or SF‐36 questionnaires were available. Summary scores of questionnaires were calculated; SAQ (52±15), SF‐36 physical health (34±8), and mean mental health (47±10; Table S1).

CFR and MRR derived from both methods did not correlate to angina frequency, stability, or the SAQ summary score (Table 4).

Table 4.

Correlation of CFR and MRR Derived From Both Bolus and Continuous Thermodilution With Self‐Reported Angina Frequency and Stability

	Total cohort (n=153)
	Angina frequency (SAQ)
MRRabs	ρ=0.019; P=0.81
MRRbolus	ρ=−0.006; P=0.93
CFRabs	ρ=0.037; P=0.65
CFRbolus	ρ=0.016; P=0.85

	Angina stability (SAQ)
MRRabs	ρ=0.088; P=0.28
MRRbolus	ρ=0.092; P=0.27
CFRabs	ρ=0.108; P=0.18
CFRbolus	ρ=−0.107; P=0.19

	SAQ summary score
MRRabs	ρ=0.128; P=0.125
MRRbolus	ρ=0.144; P=0.051
CFRabs	ρ=0.140; P=0.09
CFRbolus	ρ=−0.052; P=0.54

Correlations are presented as Pearson's correlation coefficient. CFR indicates coronary flow reserve; CFR_abs, continuous thermodilution‐derived CFR; CFR_bolus, bolus thermodilution‐derived CFR; MRR, microvascular resistance reserve; MRR_abs, continuous thermodilution‐derived MRR; MRR_bolus, bolus thermodilution‐derived MRR; and SAQ, Seattle Angina Questionnaire.

CFR_abs and MRR_abs did demonstrate a relationship with the physical health component of both questionnaires, whereas CFR_bolus and MRR_bolus did not (Table 5).

Table 5.

Correlation of CFR and MRR Derived from Both Bolus and Continuous Thermodilution With Self‐Reported Physical Health and Limitation

	Total cohort (n=153)	Epicardial spasm excluded (n=80)	Epicardial and microvascular spasm excluded (n=34)
	Physical health (SF‐36)
MRRabs	ρ=0.208*; P=0.012*	ρ=0.322*; P=0.004*	ρ=0.471*; P=0.005*
MRRbolus	ρ=0.015; P=0.85	ρ=0.180; P=0.12	ρ=0.282; P=0.11
CFRabs	ρ=0.183*; P=0.026*	ρ=0.273*; P=0.015*	ρ=0.433*; P=0.011*
CFRbolus	ρ=0.015; P=0.86	ρ=0.100; P=0.39	ρ=0.173; P=0.34

	Physical limitation (SAQ)
MRRabs	ρ=0.215*; P=0.009*	ρ=0.354*; P=0.002*	ρ=0.345*; P=0.049*
MRRbolus	ρ=0.033; P=0.69	ρ=0.016; P=0.89	ρ=−0.001; P=0.99
CFRabs	ρ=0.230*; P=0.005*	ρ=0.324*; P=0.004*	ρ=0.309; P=0.08
CFRbolus	ρ=−0.240; P=0.78	ρ=0.‐0.027; P=0.82	ρ=−0.140; P=0.45

Correlations are presented as Pearson's correlation coefficient. CFR indicates coronary flow reserve; CFR_abs, continuous thermodilution‐derived CFR; CFR_bolus, bolus thermodilution‐derived CFR; MRR, microvascular resistance reserve; MRR_abs, continuous thermodilution‐derived MRR; MRR_bolus, bolus thermodilution‐derived MRR; SAQ, Seattle Angina Questionnaire; and SF‐36, 36‐Item Short Form Health Survey.

^*indicates significant correlation.

In the total cohort, both parameters showed a weak but positive correlation with SAQ physical limitation (ρ=0.230; P=0.005 for CFR_abs, and ρ=0.215; P=0.009 for MRR_abs) and with SF‐36 physical health (ρ=0.183; P=0.026 for CFR_abs, and ρ=0.208; P=0.012 for MRR_abs).

When evaluating the 2 subgroups (no epicardial spasm and no epicardial and microvascular spasm combined), we found increasing strength of the correlation (Table 5). MRR_abs demonstrated the strongest correlation with physical limitation and physical health in patients without any spasm. These are patients with either normal CFT or isolated CMD (ρ=0.345; P=0.049, ρ=0.471; P=0.005, respectively; Figure 2).

Figure 2. Scatterplot and correlation of both MRR_bolus and MRR_abs with physical limitation and health in patients with ANOCA without coronary artery spasm.

In patients with ANOCA without spasm, MRR_abs correlates with both physical limitation (SAQ) and physical health (SF‐36), whereas MRR_bolus does not correlate with these measures of angina and quality of life. ANOCA indicates angina and no obstructive coronary artery disease; MRR_abs, continuous thermodilution‐derived microvascular resistance reserve; MRR_bolus, bolus thermodilution‐derived microvascular resistance reserve; SAQ, Seattle Angina Questionnaire; and SF‐36, Short Form Health Survey.

Moreover, when the MRR_abs‐cutoff of 2.5 was used, patients with abnormal MRR_abs had significantly lower values of physical health (63±20 versus 55±22, P=0.016) and limitation (35±8 versus 32±8, P=0.029), whereas no differences in physical health and limitation were observed for MRR_bolus, as displayed in Figure 3.

Figure 3. The association of MRR_bolus and MRR_abs with physical health and limitation.

A, No difference in the physical health domain of the SF‐36 quality of life questionnaire, stratified by abnormal or normal bolus thermodilution MRR. B, No difference in the physical limitation domain of the SAQ, stratified by abnormal or normal bolus thermodilution MRR. C, Significant difference in physical health domain of the SF‐36 quality of life questionnaire, stratified by abnormal or normal continuous thermodilution MRR. D, Significant difference in physical limitation domain of the SAQ, stratified by abnormal or normal continuous thermodilution MRR. MRR_abs indicates continuous thermodilution‐derived microvascular resistance reserve; MRR_bolus, bolus thermodilution‐derived microvascular resistance reserve; NS, not significant; SAQ, Seattle Angina Questionnaire; and SF‐36, Short Form Health Survey.

No other significant differences in angina scales were found based on the cutoffs of CFR and MRR (Table S3). Furthermore, no association was found between CMD assessed by bolus or continuous thermodilution and angina frequency (Table S4).

DISCUSSION

Our study is the first head‐to‐head comparison of bolus versus continuous thermodilution assessment of CMD. Furthermore, this is the first large study evaluating the association of CFR and MRR to angina and QoL.

The most important findings are as follows (Figure 4).

Figure 4. Study summary and main findings.

ANOCA indicates angina and no obstructive coronary artery disease; CFR_abs, continuous thermodilution‐derived coronary flow reserve; CFR_bolus, bolus thermodilution‐derived coronary flow reserve; MRR_abs, continuous thermodilution‐derived microvascular resistance reserve; MRR_bolus, bolus thermodilution‐derived microvascular resistance reserve; NL‐CFT, Netherlands Registry of Invasive Coronary Vasomotor Function Testing; SAQ, Seattle Angina Questionnaire; and SF‐36, Short Form Health Survey.

First, both continuous thermodilution CFR_abs and MRR_abs show, at best, a moderate correlation with its surrogate parameters CFR_bolus and MRR_bolus derived by bolus thermodilution. Interestingly, when assessed with continuous thermodilution, CFR and MRR are significantly lower than when using bolus thermodilution. We found a CCA of 68% between bolus and continuous thermodilution for both CFR and MRR.

Second, CFR and MRR are not associated with angina frequency and stability; however, CFR_abs and MRR_abs are associated with physical health and limitation, where bolus thermodilution‐derived surrogate parameters are not.

Bolus Versus Continuous Thermodilution‐derived Coronary Flow

The correlation we found between bolus and continuous thermodilution measurements are in concordance with previously reported data. Both measurements of flow have been compared with the noninvasive gold standard PET, and with doppler‐derived CFR (CFR_doppler). ⁶ , ²¹ , ²² Correlation of CFR_abs versus CFR_bolus (ρ=0.31) was similar compared with the correlation of CFR_doppler versus CFR_bolus (ρ=0.36). CFR_abs versus CFR_doppler show a superior correlation (ρ=0.90), although this was measured simultaneously and not consecutively, so a higher correlation is expected.

There was also a higher correlation between CFR_bolus versus CFR_PET (ρ=0.55), but this was still lower than the strong correlation between hyperemic Q _cont and Q _PET (ρ=0.91). Data on the correlation of CFR_abs with CFR_PET are not available.

Of note, both studies revealed similar patterns: CFR_bolus correlates with and is higher than CFR_doppler, especially at the higher ranges. This pattern of overestimation of CFR by bolus thermodilution was also seen in our study (Figure S2). Also, when we correct for epicardial disease (with MRR), the pattern of overestimation by bolus thermodilution remains.

CFR_bolus, as mentioned before, has been shown to overestimate CFR, especially at higher values by 3 different modalities. Therefore, Demir et al ¹⁸ recently proposed to increase the cutoff value for bolus thermodilution‐derived CFR to 2.5 instead of 2.0. When applying this in our study, it led to a similar CCA of 66% compared with 68%, but a slightly higher Cohen's kappa coefficient indicating a slightly higher agreement (Table 3).

Despite the somewhat higher agreement between bolus and continuous MRR, the agreement would still be classified as low. This might reflect the difference in hyperemia inducement: pharmacological (adenosine) versus direct (continuous infusion of saline), where the “direct” method now seems to represent a more physiological hyperemic state, as expressed by the correlation with complaints.

However, this might also reflect higher accuracy and operator independency of the continuous thermodilution method to assess absolute flow. ²³ , ²⁴ , ²⁵ Validation in prognostic studies is warranted to draw definite conclusions.

To summarize, we found a relatively high amount of classification disagreement between bolus and continuous thermodilution measurements, of which the pathophysiological meaning is unknown.

Of future interest is the subclassification of functional (abnormal CFR, normal IMR) versus structural (both abnormal) CMD. Both have been shown to have a less favorable outcome in ANOCA compared with patients with a normal functioning microcirculation. ¹⁸ Therefore, identifying and classifying CMD as functional versus structural, preferably by relationship between complaints and CFT test results seems crucial to determine prognosis.

CFR and MRR and the Association With Angina and QoL

In this study, for the first time a relationship between different intracoronary physiological parameters and physical health in a large cohort of patients with ANOCA was found. Although adenosine‐derived CFR and IMR (with bolus thermodilution) are associated with adverse prognosis, they have never been linked directly to symptoms. ²⁶ , ²⁷ Only little information is available on the association of coronary blood flow and angina. Bairey Merz et al found a correlation between both the myocardial perfusion index and resting flow and angina symptoms, but no information on the correlation with CFR was reported. ²⁸ , ²⁹ It is therefore difficult to measure therapy effect based on repeated testing and complaints.

Recently, our group described the association between bolus and continuous thermodilution‐derived parameters and angina severity. ⁷ Konst et al showed that in a cohort of 84 patients with ANOCA, patients with high absolute microvascular resistance (R) or low absolute hyperemic blood flow (Q) more frequently report severe angina (Canadian Cardiovascular Society class 3–4) (about 80% versus 60%), whereas bolus thermodilution‐derived CFR and IMR did not show an association.

Q and R thus seem ideal for intrapatient evaluation in relation to complaints, but because they are dependent on patient‐specific factors and myocardial mass they are less ideal for interpatient evaluation. ³⁰ MRR and CFR are relative measures and are therefore ideal for interpatient evaluation. In a larger cohort of patients with fully endotyped ANOCA and using a more extensive method to assess symptoms (SAQ and SF‐36), we found a clear correlation of CFR_abs and MRR_abs with physical limitation and health, whereas bolus thermodilution‐derived parameters did not show a relation. This CFR_abs and MRR_abs correlation became stronger when possibly bias by complaints of epicardial or microvascular spasm was excluded, indicating a relationship specific for the microcirculation. Interestingly, there was no relationship between the frequency of angina and CMD, several factors interplay in the frequency of angina in this population. We hypothesize that patients with CMD primarily have complaints during exercise, due to relative hypoperfusion because of lack of vasodilatory reserve or increased microvascular resistance. To avoid complaints, patients avoid exercise, leading to fewer complaints but increased physical limitation.

Clinical Implications

The need for quantitative assessment of coronary microvascular function is increasingly recognized. This novel quantitative method also allows for greater diagnostic certainty, ²³ , ²⁵ as traditional methods to determine coronary microvascular dysfunction have shortcomings. We would propose to extend the use of the continuous thermodilution method in cardiovascular research and further investigate the relation to anginal complaints (not only in CMD). Patient‐reported outcome measures such as SAQ have been validated in several studies and have linked SAQ scores to important clinical outcomes of patients with coronary artery disease. ¹¹ , ³¹ Both CFR and MRR derived with continuous thermodilution seem to be correlated with these patient‐reported outcomes and are able to detect clinically relevant differences in physical limitation. They might therefore be superior measurements reflecting the disease status of the patients with CMD compared with CFR_bolus and IMR. For this reason we advocate the use of CFR_abs or MRR_abs in future studies using repeated measurements in patients with ANOCA. However, further validation on prognosis is warranted in order to fully replace other methods in determining microvascular function in clinical practice.

Limitations

First, men were underrepresented (17%) in this study, as they are in more ANOCA population studies. Second, no adjustment of covariates was involved in the analysis. However, both methods are measuring the same indices during the same procedure, possible confounders should act on both methods similarly and presumably play no role. Only technical differences, such as timing of the methods (first or second) might influence the results.

Third, SAQ and SF‐36 are designed and validated for patients with stable angina caused by obstructive coronary artery disease; however, these are the best tools available and are frequently used in nonobstructive disease to give an impression of angina severity and physical status beyond stable disease.

CONCLUSIONS

Absolute flow measurements using continuous thermodilution to calculate CFR_abs and MRR_abs weakly correlate with, and are lower than, the surrogates CFR_bolus and MRR_bolus by bolus thermodilution measurements. In addition, absolute flow and bolus thermodilution‐derived parameters did not correlate with angina and QoL. However, absolute flow parameters did show a relationship with physical health and limitation. Therefore, they might be superior measurements, better reflecting disease status of patients with CMD when compared with CFR_bolus and IMR.

Sources of Funding

None.

Disclosures

P. Damman has received consultancy fees and research grants from Philips, and research grants from Abbott. N. van Royen has received research grants from Philips, Biotronik, Medtronic, and Abbott and speaker fees from Microport, Rainmed, and Abbott. A. Leen received speaker fees from Amgen. V. Paradies has received a research grant from Abbott to the institution and speaker fees from Boston Scientific and Abbott. P. Smits has received institutional grants from Abbott Vascular, Microport, and SMT and speaker/consulting fees from Abbott, Abiomed, Opsense, Microport, and Terumo. The remaining authors have no disclosures to report.

Supporting information

Tables S1–S4

Figures S1–S2