Development of artificial intelligence tools for invasive Doppler-based coronary microvascular assessment.

Aims: Coronary flow reserve (CFR) assessment has proven clinical utility, but Doppler-based methods are sensitive to noise and operator bias, limiting their clinical applicability. The objective of the study is to expand the adoption of invasive Doppler CFR, through the development of artificial intelligence (AI) algorithms to automatically quantify coronary Doppler quality and track flow velocity. Methods and results: A neural network was trained on images extracted from coronary Doppler flow recordings to score signal quality and derive values for coronary flow velocity and CFR. The outputs were independently validated against expert consensus. Artificial intelligence successfully quantified Doppler signal quality, with high agreement with expert consensus (Spearman's rho: 0.94), and within individual experts. Artificial intelligence automatically tracked flow velocity with superior numerical agreement against experts, when compared with the current console algorithm [AI flow vs. expert flow bias -1.68 cm/s, 95% confidence interval (CI) -2.13 to -1.23 cm/s, P < 0.001 with limits of agreement (LOA) -4.03 to 0.68 cm/s; console flow vs. expert flow bias -2.63 cm/s, 95% CI -3.74 to -1.52, P < 0.001, 95% LOA -8.45 to -3.19 cm/s]. Artificial intelligence yielded more precise CFR values [median absolute difference (MAD) against expert CFR: 4.0% for AI and 7.4% for console]. Artificial intelligence tracked lower-quality Doppler signals with lower variability (MAD against expert CFR 8.3% for AI and 16.7% for console). Conclusion: An AI-based system, trained by experts and independently validated, could assign a quality score to Doppler traces and derive coronary flow velocity and CFR. By making Doppler CFR more automated, precise, and operator-independent, AI could expand the clinical applicability of coronary microvascular assessment.

Phasic flow patterns of right versus left coronary arteries in patients undergoing clinical physiological assessment.

BACKGROUND: Coronary blood flow in humans is known to be predominantly diastolic. Small studies in animals and humans suggest that this is less pronounced or even reversed in the right coronary artery (RCA). AIMS: This study aimed to characterise the phasic patterns of coronary flow in the left versus right coronary arteries of patients undergoing invasive physiological assessment. METHODS: We analysed data from the Iberian-Dutch-English Collaborators (IDEAL) study. A total of 482 simultaneous pressure and flow measurements from 301 patients were included in our analysis. RESULTS: On average, coronary flow was higher in diastole both at rest and during hyperaemia in both the RCA and LCA (mean diastolic-to-systolic velocity ratio [DSVR] was, respectively, 1.85±0.70, 1.76±0.58, 1.53±0.34 and 1.58±0.43 for LCArest, LCAhyp, RCArest and RCAhyp, p<0.001 for between-vessel comparisons). Although the type of RCA dominance affected the DSVR magnitude (RCAdom=1.55±0.35, RCAco-dom=1.40±0.27, RCAnon-dom=1.35; standard deviation not reported as n=3), systolic flow was very rarely predominant (DSVR was greater than or equal to 1.00 in 472/482 cases [97.9%] overall), with equal prevalence in the LCA. Stenosis severity or microvascular dysfunction had a negligible impact on DSVR in both the RCA and LCA (DSVR x hyperaemic stenosis resistance R2 =0.018, p=0.03 and DSVR x coronary flow reserve R2 <0.001, p=0.98). CONCLUSIONS: In patients with coronary artery disease undergoing physiological assessment, diastolic flow predominance is seen in both left and right coronary arteries. Clinical interpretation of coronary physiological data should therefore not differ between the left and the right coronary systems.

Immediate impact of coronary artery bypass graft surgery on regional myocardial perfusion: Results from the Collaborative Pilot Study to Determine the Correlation Between Intraoperative Observations Using Spy Near-Infrared Imaging and Cardiac Catheterization Laboratory Physiological Assessment of Lesion Severity.

Objectives: Coronary artery bypass grafting (CABG) is performed using anatomic guidance. Data connecting the physiologic significance of the coronary vessel stenosis to the acute physiologic response to grafting are lacking. The Collaborative Pilot Study to Determine the Correlation Between Intraoperative Observations Using Spy Near-Infrared Imaging and Cardiac Catheterization Laboratory Physiological Assessment of Lesion Severity study is the first to compare preintervention coronary physiology with the acute regional myocardial perfusion change (RMP-QC) at CABG in a per-graft analysis. Methods: Non-emergent patients undergoing diagnostic catheterization suitable for multivessel CABG were enrolled. Synergy between Percutaneous Coronary Intervention with Taxus score, fractional flow reserve (FFR), instantaneous wave free ratio (iFR), and quantitative coronary angiography was documented in 75 epicardial coronary arteries, with 62 angiographically intermediate and 13 severe stenoses. At CABG, near-infrared fluorescence analysis quantified the relative change (post- vs pregrafting, termed RMP-QC) in the grafted vessel's perfusion territory. Scatter plots were constructed for RMP-QC versus quantitative coronary angiography and RMP-QC versus FFR/iFR. Exact quadrant randomization test for randomness was used. Results: There was no relationship between RMP-QC and quantitative coronary angiography percent diameter stenosis, whether all study vessels were included (P = .949) or vessels with core-lab quantitative coronary angiography only (P = .922). A significant nonrandom association between RMP-QC and FFR (P = .025), as well as between RMP-QC and iFR (P = .008), was documented. These associations remained when excluding vessels with assigned FFR and iFR values (P = .0092 and P = .0006 for FFR and iFR, respectively). Conclusions: The Collaborative Pilot Study to Determine the Correlation Between Intraoperative Observations Using Spy Near-Infrared Imaging and Cardiac Catheterization Laboratory Physiological Assessment of Lesion Severity study demonstrates there is no association between angiographic coronary stenosis severity and the acute perfusion change after grafting; there is an association between functional stenosis severity and absolute increase in regional myocardial perfusion after CABG.

Aerosolised fluorescein can quantify FFP mask faceseal leakage: a cost-effective adaptation to the existing point of care fit test.

BACKGROUND: A qualitative fit test using bitter-tasting aerosols is the commonest way to determine filtering face-piece (FFP) mask leakage. This taste test is subjective and biased by placebo. We propose a cheap, quantitative modification of the taste test by measuring the amount of fluorescein stained filter paper behind the mask using image analysis. METHODS: A bitter-tasting fluorescein solution was aerosolised during mask fit tests, with filter paper placed on masks' inner surfaces. Participants reported whether they could taste bitterness to determine taste test 'pass' or 'fail' results. Filter paper photographs were digitally analysed to quantify total fluorescence (TF). RESULTS: Fifty-six healthcare professionals were fit tested; 32 (57%) 'passed' the taste test. TF between the taste test 'pass' and 'fail' groups was significantly different (p<0.001). A cut-off (TF = 5.0 × 106 units) was determined at precision (78%) and recall (84%), resulting in 5/56 participants (9%) reclassified from 'pass' to 'fail' by the fluorescein test. Seven out of 56 (12%) reclassified from 'fail' to 'pass'. CONCLUSION: Fluorescein is detectable and sensitive at identifying FFP mask leaks. These low-cost adaptations can enhance exiting fit testing to determine 'pass' and 'fail' groups, protecting those who 'passed' the taste test but have high fluorescein leak, and reassuring those who 'failed' the taste test despite having little fluorescein leak.

Individual Lesion-Level Meta-Analysis Comparing Various Doses of Intracoronary Bolus Injection of Adenosine With Intravenous Administration of Adenosine for Fractional Flow Reserve Assessment.

BACKGROUND: Intravenous infusion of adenosine is considered standard practice for fractional flow reserve (FFR) assessment but is associated with adverse side-effects and is time-consuming. Intracoronary bolus injection of adenosine is better tolerated by patients, cheaper, and less time-consuming. However, current literature remains fragmented and modestly sized regarding the equivalence of intracoronary versus intravenous adenosine. We aim to investigate the relationship between intracoronary adenosine and intravenous adenosine to determine FFR. METHODS: We performed a lesion-level meta-analysis to compare intracoronary adenosine with intravenous adenosine (140 µg/kg per minute) for FFR assessment. The search was conducted in accordance to the Preferred Reporting for Systematic Reviews and Meta-Analysis statement. Lesion-level data were obtained by contacting the respective authors or by digitization of scatterplots using custom-made software. Intracoronary adenosine dose was defined as; low: <40 µg, intermediate: 40 to 99 µg, and high: ≥100 µg. RESULTS: We collected 1972 FFR measurements (1413 lesions) comparing intracoronary with intravenous adenosine from 16 studies. There was a strong correlation (correlation coefficient =0.915; P<0.001) between intracoronary-FFR and intravenous-FFR. Mean FFR was 0.81±0.11 for intracoronary adenosine and 0.81±0.11 for intravenous adenosine (P<0.001). We documented a nonclinically relevant mean difference of 0.006 (limits of agreement: -0.066 to 0.078) between the methods. When stratified by the intracoronary adenosine dose, mean differences between intracoronary and intravenous-FFR amounted to 0.004, 0.011, or 0.000 FFR units for low-dose, intermediate-dose, and high-dose intracoronary adenosine, respectively. CONCLUSIONS: The present study documents clinically irrelevant differences in FFR values obtained with intracoronary versus intravenous adenosine. Intracoronary adenosine hence confers a practical and patient-friendly alternative for intravenous adenosine for FFR assessment.

Sex Differences in Instantaneous Wave-Free Ratio or Fractional Flow Reserve-Guided Revascularization Strategy.

OBJECTIVES: This study sought to evaluate sex differences in procedural characteristics and clinical outcomes of instantaneous wave-free ratio (iFR)- and fractional flow reserve (FFR)-guided revascularization strategies. BACKGROUND: An iFR-guided strategy has shown a lower revascularization rate than an FFR-guided strategy, without differences in clinical outcomes. METHODS: This is a post hoc analysis of the DEFINE-FLAIR (Functional Lesion Assessment of Intermediate stenosis to guide Revascularization) study, in which 601 women and 1,891 men were randomized to iFR- or FFR-guided strategy. The primary endpoint was 1-year major adverse cardiac events (MACE), a composite of all-cause death, nonfatal myocardial infarction, or unplanned revascularization. RESULTS: Among the entire population, women had a lower number of functionally significant lesions per patient (0.31 ± 0.51 vs. 0.43 ± 0.59; p < 0.001) and less frequently underwent revascularization than men (42.1% vs. 53.1%; p < 0.001). There was no difference in mean iFR value according to sex (0.91 ± 0.09 vs. 0.91 ± 0.10; p = 0.442). However, the mean FFR value was lower in men than in women (0.83 ± 0.09 vs. 0.85 ± 0.10; p = 0.001). In men, an FFR-guided strategy was associated with a higher rate of revascularization than an iFR-guided strategy (57.1% vs. 49.3%; p = 0.001), but this difference was not observed in women (41.4% vs. 42.6%; p = 0.757). There was no difference in MACE rates between iFR- and FFR-guided strategies in both women (5.4% vs. 5.6%, adjusted hazard ratio: 1.10; 95% confidence interval: 0.50 to 2.43; p = 0.805) and men (6.6% vs. 7.0%, adjusted hazard ratio: 0.98; 95% confidence interval: 0.66 to 1.46; p = 0.919). CONCLUSIONS: An FFR-guided strategy was associated with a higher rate of revascularization than iFR-guided strategy in men, but not in women. However, iFR- and FFR-guided strategies showed comparable clinical outcomes, regardless of sex. (Functional Lesion Assessment of Intermediate Stenosis to guide Revascularization [DEFINE-FLAIR]; NCT02053038).

Clinical Events After Deferral of LAD Revascularization Following Physiological Coronary Assessment.

BACKGROUND: Physicians are not always comfortable deferring treatment of a stenosis in the left anterior descending (LAD) artery because of the perception that there is a high risk of major adverse cardiac events (MACE). The authors describe, using the DEFINE-FLAIR (Functional Lesion Assessment of Intermediate Stenosis to Guide Revascularisation) trial, MACE rates when LAD lesions are deferred, guided by physiological assessment using fractional flow reserve (FFR) or the instantaneous wave-free ratio (iFR). OBJECTIVES: The purpose of this study was to establish the safety of deferring treatment in the LAD using FFR or iFR within the DEFINE-FLAIR trial. METHODS: MACE rates at 1 year were compared between groups (iFR and FFR) in patients whose physiological assessment led to LAD lesions being deferred. MACE was defined as a composite of cardiovascular death, myocardial infarction (MI), and unplanned revascularization at 1 year. Patients, and staff performing follow-up, were blinded to whether the decision was made with FFR or iFR. Outcomes were adjusted for age and sex. RESULTS: A total of 872 patients had lesions deferred in the LAD (421 guided by FFR, 451 guided by iFR). The event rate with iFR was significantly lower than with FFR (2.44% vs. 5.26%; adjusted HR: 0.46; 95% confidence interval [CI]: 0.22 to 0.95; p = 0.04). This was driven by significantly lower unplanned revascularization with iFR and numerically lower MI (unplanned revascularization: 2.22% iFR vs. 4.99% FFR; adjusted HR: 0.44; 95% CI: 0.21 to 0.93; p = 0.03; MI: 0.44% iFR vs. 2.14% FFR; adjusted HR: 0.23; 95% CI: 0.05 to 1.07; p = 0.06). CONCLUSIONS: iFR-guided deferral appears to be safe for patients with LAD lesions. Patients in whom iFR-guided deferral was performed had statistically significantly lower event rates than those with FFR-guided deferral.

Influence of the amount of myocardium subtended to a coronary stenosis on the index of microcirculatory resistance. Implications for the invasive assessment of microcirculatory function in ischaemic heart disease.

AIMS: The index of microcirculatory resistance (IMR) is increasingly used to quantify microcirculatory function. However, in normal coronary arteries, resistance increases with the branching structure of the coronary tree, which suggests that IMR could be influenced by the amount of downstream myocardial mass (MM). We aimed to evaluate the influence of the amount of MM subtended to an intermediate stenosis on the IMR. METHODS AND RESULTS: IMR, fractional flow reserve and coronary flow reserve (CFR) were measured in 123 coronary arteries (102 patients) with intermediate stenosis. Jeopardised MM was estimated with the Myocardial Jeopardy Index (MJI). MM was inversely associated with IMR (R.

Coronary pressure and flow relationships in humans: phasic analysis of normal and pathological vessels and the implications for stenosis assessment: a report from the Iberian-Dutch-English (IDEAL) collaborators.

BACKGROUND: Our understanding of human coronary physiological behaviour is derived from animal models. We sought to describe physiological behaviour across a large collection of invasive pressure and flow velocity measurements, to provide a better understanding of the relationships between these physiological parameters and to evaluate the rationale for resting stenosis assessment. METHODS AND RESULTS: Five hundred and sixty-seven simultaneous intracoronary pressure and flow velocity assessments from 301 patients were analysed for coronary flow velocity, trans-stenotic pressure gradient (TG), and microvascular resistance (MVR). Measurements were made during baseline and hyperaemic conditions. The whole cardiac cycle and the diastolic wave-free period were assessed. Stenoses were assessed according to fractional flow reserve (FFR) and quantitative coronary angiography DS%. With progressive worsening of stenoses, from unobstructed angiographic normal vessels to those with FFR ≤ 0.50, hyperaemic flow falls significantly from 45 to 19 cm/s, Ptrend < 0.001 in a curvilinear pattern. Resting flow was unaffected by stenosis severity and was consistent across all strata of stenosis (Ptrend > 0.05 for all). Trans-stenotic pressure gradient rose with stenosis severity for both rest and hyperaemic measures (Ptrend < 0.001 for both). Microvascular resistance declines with stenosis severity under resting conditions (Ptrend < 0.001), but was unchanged at hyperaemia (2.3 ± 1.1 mmHg/cm/s; Ptrend = 0.19). CONCLUSIONS: With progressive stenosis severity, TG rises. However, while hyperaemic flow falls significantly, resting coronary flow is maintained by compensatory reduction of MVR, demonstrating coronary auto-regulation. These data support the translation of coronary physiological concepts derived from animals to patients with coronary artery disease and furthermore, suggest that resting pressure indices can be used to detect the haemodynamic significance of coronary artery stenoses.

Low coronary microcirculatory resistance associated with profound hypotension during intravenous adenosine infusion: implications for the functional assessment of coronary stenoses.

BACKGROUND: Intravenous adenosine infusion produces coronary and systemic vasodilatation, generally leading to systemic hypotension. However, adenosine-induced hypotension during stable hyperemia is heterogeneous, and its relevance to coronary stenoses assessment with fractional flow reserve (FFR) remains largely unknown. METHODS AND RESULTS: FFR, coronary flow reserve, and index of microcirculatory resistance were measured in 93 stenosed arteries (79 patients). Clinical and intracoronary measurements were analyzed among tertiles of the percentage degree of adenosine-induced hypotension, defined as follows: %ΔP(a)=-[100-(hyperemic aortic pressure×100/baseline aortic pressure)]. Overall, %ΔP(a) was -13.6±12.0%. Body mass index was associated with %ΔP(a) (r=0.258; P=0.025) and obesity, an independent predictor of profound adenosine-induced hypotension (tertile 3 of %ΔP(a); odds ratio, 3.95 [95% confidence interval, 1.48-10.54]; P=0.006). %ΔP(a) was associated with index of microcirculatory resistance (ρ=0.311; P=0.002), coronary flow reserve (r=-0.246; P=0.017), and marginally with FFR (r=0.203; P=0.051). However, index of microcirculatory resistance (ß=0.003; P<0.001) and not %ΔP(a) (ß=-0.001; P=0.564) was a predictor of FFR. Compared with tertiles 1 and 2 of %ΔP(a) (n=62 [66.6%]), stenoses assessed during profound adenosine-induced hypotension (n=31 [33.3%]) had lower index of microcirculatory resistance (12.4 [8.6-22.7] versus 20 [15.8-35.5]; P=0.001) and FFR values (0.77±0.13 versus 0.83±0.12; P=0.021), as well as a nonsignificant increase in coronary flow reserve (2.5±1.1 versus 2.2±0.87; P=0.170). CONCLUSIONS: The modification of systemic blood pressure during intravenous adenosine infusion is related to hyperemic microcirculatory resistance in the heart. Profound adenosine-induced hypotension is associated with obesity, lower coronary microcirculatory resistance, and lower FFR values.

Hemodynamic response to intravenous adenosine and its effect on fractional flow reserve assessment: results of the Adenosine for the Functional Evaluation of Coronary Stenosis Severity (AFFECTS) study.

BACKGROUND: We studied the hemodynamic response to intravenous adenosine on calculation of fractional flow reserve (FFR). Intravenous adenosine is widely used to achieve conditions of stable hyperemia for measurement of FFR. However, intravenous adenosine affects both systemic and coronary vascular beds differentially. METHODS AND RESULTS: A total of 283 patients (310 coronary stenoses) underwent coronary angiography with FFR using intravenous adenosine 140 mcg/kg per minute via a central femoral vein. Offline analysis was performed to calculate aortic (Pa), distal intracoronary (Pd), and reservoir (Pr) pressure at baseline, peak, and stable hyperemia. Seven different hemodynamic patterns were observed according to Pa and Pd change at peak and stable hyperemia. The average time from baseline to stable hyperemia was 68.2±38.5 seconds, when both ΔPa and ΔPd were decreased (ΔPa, -10.2±10.5 mm Hg; ΔPd, -18.2±10.8 mm Hg; P<0.001 for both). The fall in Pa closely correlated with the reduction in peripheral Pr (ΔPr, -12.9±15.7 mm Hg; P<0.001; r=0.9; P<0.001). ΔPa and ΔPd were closely related under conditions of peak (r=0.75; P<0.001) and stable hyperemia (r=0.83; P<0.001). On average, 56% (10.2 mm Hg) of the reduction in Pd was because of fall in Pa. FFR lesion classification changed in 9% using an FFR threshold of ≤0.80 and 5.2% with FFR threshold <0.75 when comparing Pd/Pa at peak and stable hyperemia. CONCLUSIONS: Intravenous adenosine results in variable changes in systemic blood pressure, which can lead to alterations in FFR lesion classification. Attention is required to ensure FFR is measured under conditions of stable hyperemia, although the FFR value at this point may be numerically higher.

Improvement in coronary haemodynamics after percutaneous coronary intervention: assessment using instantaneous wave-free ratio.

OBJECTIVE: To determine whether the instantaneous wave-free ratio (iFR) can detect improvement in stenosis significance after percutaneous coronary intervention (PCI) and compare this with fractional flow reserve (FFR) and whole cycle Pd/Pa. DESIGN: A prospective observational study was undertaken in elective patients scheduled for PCI with FFR ≤ 0.80. Intracoronary pressures were measured at rest and during adenosine-mediated vasodilatation, before and after PCI. iFR, Pd/Pa and FFR values were calculated using the validated fully automated algorithms. SETTING: Coronary catheter laboratories in two UK centres and one in the USA. PATIENTS: 120 coronary stenoses in 112 patients were assessed. The mean age was 63 ± 10 years, while 84% were male; 39% smokers; 33% with diabetes. Mean diameter stenosis was 68 ± 16% by quantitative coronary angiography. RESULTS: Pre-PCI, mean FFR was 0.66 ± 0.14, mean iFR was 0.75 ± 0.21 and mean Pd/Pa 0.83 ± 0.16. PCI increased all indices significantly (FFR 0.89 ± 0.07, p<0.001; iFR 0.94 ± 0.05, p<0.001; Pd/Pa 0.96 ± 0.04, p<0.001). The change in iFR after intervention (0.20 ± 0.21) was similar to ΔFFR 0.22 ± 0.15 (p=0.25). ΔFFR and ΔiFR were significantly larger than resting ΔPd/Pa (0.13 ± 0.16, both p<0.001). Similar incremental changes occurred in patients with a higher prevalence of risk factors for microcirculatory disease such as diabetes and hypertension. CONCLUSIONS: iFR and FFR detect the changes in coronary haemodynamics elicited by PCI. FFR and iFR have a significantly larger dynamic range than resting Pd/Pa. iFR might be used to objectively document improvement in coronary haemodynamics following PCI in a similar manner to FFR.

Clinical validation of the resting pressure parameters in the assessment of functionally significant coronary stenosis; results of an independent, blinded comparison with fractional flow reserve.

BACKGROUND: The role of resting pressure parameters, i.e. instantaneous wave-free ratio (iFR), and resting distal coronary pressure/aortic pressure (Pd/Pa) in assessing functionally significant stenosis remains controversial. We sought to assess the diagnostic performance of iFR and resting whole-cycle Pd/Pa in Asian patients. METHODS: In this study, 238 consecutive lesions (no total occlusions) in which fractional flow reserve (FFR) was measured with both intravenous and intracoronary adenosine administration were included. Coded resting pressure data were sent to the core laboratory in which iFR was calculated in a blinded fashion. RESULTS: FFR and iFR had unimodal distributions and the correlation was r = 0.77 (95% confidence interval, 0.71 to 0.82). In a receiver-operating-characteristic curve analysis, iFR had an area under the curve (AUC) of 0.9 at FFR ≤ 0.80. The best cut-off value for iFR was 0.90 with a sensitivity, specificity, positive and negative predictive values, and diagnostic accuracy of 76%, 86%, 82% and 80%, and 82%, respectively. The resting whole-cycle Pd/Pa cut-off of 0.91 demonstrated a diagnostic accuracy of 82% (AUC 0.9). However, iFR had higher discriminatory power than the resting whole-cycle Pd/Pa. CONCLUSION: Both iFR and resting whole-cycle Pd/Pa showed good diagnostic performance to define the functionally significant stenosis in an independent Asian cohort distributed unimodally and without total occlusions. However, further validation is needed to explore the areas of disagreement between different physiologic parameters prior to adoption of resting pressure parameters into routine clinical practice.