Tiny Vessels, Big Verdicts: Microvascular Resistance Can Flip Heart Stenosis Diagnoses

When cardiologists weigh whether a narrowed coronary artery deserves a stent, they lean on two physiological yardsticks: fractional flow reserve, or FFR, which measures the pressure drop across a blockage during maximal vessel dilation, and coronary flow reserve, or CFR, which tracks how much blood flow can rise from rest to peak exertion. In a perfect world, the two indices would tell the same story. In reality, they disagree in roughly 30 percent of cases, leaving clinicians to puzzle over plaques that look severe on one measure and benign on the other. A new simulation study published in the Annals of Biomedical Engineering argues that a frequently overlooked player—the resistance of the heart’s smallest vessels—can single-handedly push identical blockages across the clinical decision lines that determine treatment.

The research, led by Tej Jolly, Arnav Garcha, and Noelia Grande Gutiérrez of Carnegie Mellon University, tackled a stubborn methodological problem. Patient studies have long hinted that the health of the coronary microvasculature, the dense network of arterioles and capillaries downstream of the major arteries, influences both FFR and CFR. But in real patients, microvascular status is tangled up with anatomical variation: every artery has a different lesion length, severity, and branching pattern, making it nearly impossible to isolate what the microcirculation alone is doing. Computational studies have tried to fill the gap, yet most either fix the downstream resistance at a single value or sample only a narrow range of microvascular states, and patient-specific models confound geometry with physiology.

To break that confounding, the team built a multiscale modeling framework that couples three-dimensional computational fluid dynamics of the epicardial coronary arteries with a lumped-parameter network representing the microvascular bed. The three-dimensional component solves the incompressible Navier–Stokes equations over unstructured tetrahedral meshes averaging 3.3 million elements, with boundary-layer refinement near the walls and local refinement around stenoses. Blood was treated as a Newtonian fluid, a reasonable approximation given that simulated shear rates exceeded 100 per second, and flow was assumed laminar, with Reynolds numbers averaging around 200 and peaking at 967 under the most severe stenosis and highest flow. Transient simulations ran for four cardiac cycles at a heart rate of 70 beats per minute, with periodic convergence defined as less than one percent change in outlet pressure and flow between cycles.

The anatomical foundation was equally deliberate. The researchers generated a baseline left coronary artery model using population-mean values for branch diameters, bifurcation angles, and segment lengths drawn from a computed tomography angiography study of 300 adults with zero coronary calcium scores. Onto this fixed skeleton they imposed nine plaque configurations, varying lesion location between the left anterior descending artery and the left circumflex, lesion length between a focal 1.2 centimeters and a diffuse 2.5 centimeters, and stenosis severity across zero, 45, and 60 percent diameter reduction. The 45 percent severity was chosen after exploratory sweeps showed that FFR most often crossed its clinical threshold of 0.8 between 40 and 50 percent narrowing, precisely the borderline zone where diagnostic uncertainty is greatest. Stenoses were modeled as asymmetric constrictions to mimic the eccentric plaque morphology commonly seen in atherosclerosis.

Each of the nine geometries was then run under four hyperemic microvascular resistance states, produced by scaling the microvascular resistor in the lumped network to represent total coronary resistance equal to 24, 43, 62, and 81 percent of its resting value. The baseline of 24 percent reflects the maximum resistance reduction achievable with intravenous adenosine, the hyperemic agent used in clinical FFR measurement, while the elevated states span progressively impaired vasodilation. The resulting hyperemic microvascular resistance values ranged from 1.3 to 8.1 millimeters of mercury per centimeter per second, bracketing the clinical dysfunction threshold of 2.5 and matching pathologically observed ranges. Nine additional resting simulations brought the total to 45, and the simulated cases were verified against a retrospective observational cohort of 299 patients, occupying the same joint distributions of stenotic burden, microvascular resistance, and index values as the clinical data.

The headline finding is strikingly clean: for geometrically identical lesions, raising microvascular resistance increased FFR and decreased CFR. Across all 36 hyperemic simulations, the correlation between microvascular resistance and FFR was strong and positive, while the correlation with CFR was very strong and negative. The mechanism follows directly from the physics. Higher downstream resistance throttles hyperemic flow, and since the pressure drop across a stenosis scales with flow, a throttled flow shrinks the trans-stenotic pressure gradient, making FFR appear more reassuring even as the tissue receives less blood. Meanwhile, the ratio of hyperemic to resting distal velocity collapses toward unity, dragging CFR downward. In short, coronary flow in these models was governed primarily by the microcirculation rather than by the epicardial blockage itself, with flow bands overlapping broadly across lesions of different severities when stratified by stenosis.

The threshold analysis carries the most clinical weight. Because treatment decisions hinge on binary cutoffs—FFR of 0.8 and CFR of 2—the team fitted regression models within narrow bands around those values. Near the FFR threshold, the standardized coefficients for microvascular resistance and hyperemic stenosis resistance were comparable in magnitude and both highly significant, with models explaining over 90 percent of the variance. This means that near-threshold FFR readings reflect both the epicardial lesion and the microvascular state, so a shift across the cutoff cannot be uniquely attributed to plaque severity. Near the CFR threshold, by contrast, microvascular resistance dominated, consistent with CFR’s role as the gold standard for microvascular function. The response was also nonlinear: the transition from healthy to the first impaired microvascular state produced far larger changes in flow and FFR than transitions deeper into dysfunction, with sensitivity to microvascular resistance dropping by roughly 77 percent for FFR and 71 percent for CFR as both indices approached a plateau once hyperemic flow was already severely limited.

Perhaps most provocatively, the simulations reproduced both modes of FFR–CFR discordance in identical anatomies purely by varying microvascular resistance. The same plaque could land in the preserved-FFR, reduced-CFR quadrant or the reduced-FFR, preserved-CFR quadrant depending on the microvascular state alone, and preserved FFR with reduced CFR was consistently associated with higher microvascular resistance. Notably, lesion length showed no association with the direction of discordance in a Fisher’s exact test, suggesting that the focal-versus-diffuse distinction, often invoked to explain index disagreements, may be an inadequate surrogate for physiological state. The study also found that elevated microvascular resistance expanded the arterial wall area exposed to low wall shear stress at the LAD–LCx bifurcation, an atherogenic condition, and shrank areas of high shear stress linked to cap thinning and plaque rupture, with the bifurcation effects persisting even after controlling for bulk flow rate.

The implications reach from the catheterization lab to the computational pipeline. Clinically, lesions with FFR at or below 0.50 remain significant regardless of microvascular status, but when microvascular resistance is elevated, an FFR above 0.8 can coexist with a pathologically low CFR, and severe microvascular disease can push CFR below threshold even with no stenosis at all and an FFR near 1. In such cases, flow-oriented evidence from perfusion imaging or CFR itself should carry more weight. Conversely, when microvascular resistance is low and flow runs high, pressure-based indices can be penalized, and flow-normalized metrics such as hyperemic stenosis resistance become especially informative. For the growing field of CT-derived FFR, the message is that distal boundary conditions are not mere numerical settings—they encode microvascular health, and uncertainty in them propagates directly into predicted indices. The authors argue that hyperemic microvascular resistance should be reported routinely alongside FFR and CFR, both in the clinic and in computational studies, so that geometric effects are not mistaken for physiological ones and so that cross-study comparisons become reproducible. Future work will extend the framework to patient-specific coronary trees, compliant walls, branch-specific and time-varying microvascular parameterizations, and prospective validation against simultaneous invasive pressure-flow measurements and perfusion imaging.

Subject of Research: The effect of hyperemic microvascular resistance on coronary hemodynamic diagnostic indices such as FFR and CFR, investigated through multiscale computational fluid dynamics simulation.

Article Title: Impact of Microvascular Resistance on Coronary Hemodynamic Diagnostic Indices: A Multiscale Simulation Study

Article References: Impact of Microvascular Resistance on Coronary Hemodynamic Diagnostic Indices: A Multiscale Simulation Study. (n.d.). https://doi.org/10.1007/s10439-026-04378-1

Image Credits: AI Generated

DOI: 10.1007/s10439-026-04378-1

Keywords: coronary artery disease, fractional flow reserve, coronary flow reserve, microvascular resistance, computational fluid dynamics, coronary hemodynamics, lumped-parameter network, stenosis, wall shear stress, revascularization, CT-FFR, cardiovascular modeling