Effect of smoking on the diagnostic performance of computational fluid dynamics-derived CT-derived fractional flow reserve: a cross-sectional study

Introduction

Myocardial ischemia results from an imbalance between coronary blood flow and myocardial demand, primarily manifesting as angina pectoris (1,2). Coronary artery disease (CAD) and coronary microvascular dysfunction (CMD) are major causes of myocardial ischemia, with CAD presenting as epicardial coronary artery stenosis and CMD as dysfunction of the downstream coronary microvascular network (3). Invasive measurement of fractional flow reserve (FFR) via coronary angiography is currently the gold standard for assessing obstructive CAD, but its invasiveness and high cost limit its clinical application (4).

FFR can be estimated from the computational fluid dynamics (CFD) simulation of patient-specific blood flow in a three-dimensional (3D) model of coronary arteries reconstructed from clinical imaging data, most commonly computed tomography (CT), named CT-derived FFR (FFRct). FFRct presents the anatomical and functional severity of coronary artery stenosis and is consistent with measured FFR value, providing a non-invasive option in diagnosing CAD and myocardial ischemia (5).

However, the diagnostic value of FFRct is challenged by some factors. First, currently, FFRct is mainly applied in assessing obstructive coronary lesions. The accuracy of FFRct is lower in obstructive lesions where CMD can influence the hemodynamics (6-8). Notably, among smokers which is a particularly high-risk group for CMD, the covert CMD is not preemptively excluded as patients may never be explicitly diagnosed, therefore affecting the reliability of FFRct (9). Smoking can affect coronary microvascular function, leading to increased microvascular resistance which is a main manifestation of CMD (10). Previous studies have indicated that CMD can significantly impact coronary blood flow and FFR via “branch steal”, namely, more blood flow redirected to the non-CMD branches due to the increase resistance in a branch affected by CMD, where FFRct may over- or underestimate FFR in the non-CMD and CMD branches, and FFR may not indicate the hemodynamic severity of the stenosis (11). Therefore, the accuracy of FFRct and its diagnostic value in smokers warrant further investigation (12).

This study aimed to compare the diagnostic accuracy of FFRct in smokers vs. non-smokers, using invasive FFR as the reference standard. We also evaluated whether myocardial volume, used as a boundary condition in CFD simulations, differs between smoking groups and whether this affects FFRct performance. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2409/rc).

Methods

Study design and patient population

This is a sub-study of the HBFlows trial. The HBFlows (CT-based non-invasive FFR measurement system) study is a multi-center, prospective, observational cohort study designed to explore the diagnostic performance of FFRct in identifying obstructive CAD compared to invasive FFR. It prospectively screened patients with clinical suspicion of CAD and planned invasive coronary angiography (ICA) from six clinics in China (Zhongshan Hospital Fudan University, Hangzhou First People’s Hospital, The First Affiliated Hospital of Wenzhou Medical University, The Second Affiliated Hospital of Zhejiang University School of Medicine, The Second Affiliated Hospital of Nanchang University, and The Eighth Affiliated Hospital of Sun Yat-sen University). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Zhongshan Hospital Fudan University (No. 2019-139R). Written informed consent was provided by all enrolled patients. All participating local institutions were informed of and agreed to the study. All patients underwent coronary CT angiography (CCTA), FFRct, ICA, and invasive FFR within 45 days from 2020 to 2022. To investigate the relationship between coronary physiology and smoking, patients’ smoking status was strictly recorded in their lifestyle records. Smoking status was dichotomized into two groups according to standard epidemiological criteria, with a total of 298 participants included in the final analysis: (I) smokers (n=106): individuals who had smoked ≥100 cigarettes in their lifetime, comprising two subcategories: current smokers (n=69): currently smoking at the time of enrollment, with a median smoking duration of 28.5 years [interquartile range (IQR), 18.0–38.0 years] and median consumption of 18 cigarettes per day (IQR, 12–25 cigarettes per day); former smokers (n=37): had smoked ≥100 cigarettes in their lifetime but had quit smoking for ≥12 months before enrollment, with a median time since cessation of 9.0 years (IQR, 4.0–16.0 years) and median cumulative exposure of 22.3 pack-years (IQR, 14.1–35.7 pack-years). (II) Non-smokers (n=192): individuals who had smoked <100 cigarettes in their lifetime. The primary analysis was patient-based, with one target lesion per patient included for diagnostic performance evaluation. For patients with multiple epicardial lesions, the target vessel was defined as the vessel with the highest degree of stenosis on CCTA; in cases of equal stenosis severity, the left anterior descending artery was selected as the target vessel, consistent with prior FFRct validation studies (11,13). The inclusion criteria comprised individuals with at least one lesion in the epicardial coronary arteries, with a CCTA diameter ≥2.0 mm and lumen diameter stenosis of 30–90%. The exclusion criteria were as follows: (I) prior coronary artery revascularization; (II) acute coronary syndrome or history of myocardial infarction; (III) heart failure; (IV) renal insufficiency (glomerular filtration rate <45 mL/kg/
1.73 m²); (V) implanted cardiac devices; (VI) poor-quality CCTA images due to tachyarrhythmias; (VII) allergy to iodine contrast agents; (VIII) non-assessable CCTA images for coronary anatomical modeling; (IX) pregnancy; and (X) failure of ICA and FFR procedures. The predefined primary endpoint of the study was the difference in FFRct diagnostic accuracy (using invasive FFR ≤0.80 as the reference standard) between smokers and non-smokers. Key secondary endpoints included the following: (I) difference in mean myocardial volume between groups; (II) difference in FFRct and invasive FFR values between groups; (III) difference in FFRct measurement error (absolute difference between FFRct and invasive FFR) between groups; and (IV) difference in area under the receiver operating characteristic (ROC) curve (AUC) between groups.

CCTA acquisition and image analysis

CCTA was performed by using single- or dual-source CT scanners with four CT scanner models across six clinical centers: 320-row detector CT scanner (Aquilion One Vision; Canon Medical Systems Corporation, Tochigi, Japan) in 87 patients (29.2%), 256-row wide-detector CT scanner (Revolution HD; GE Healthcare, Chicago, IL, USA) in 102 patients (34.2%), 128-slice multidetector CT scanner (Definition AS; Siemens Healthineers, Erlangen, Germany) in 70 patients (23.5%), and third-generation 192-slice multidetector dual-source CT scanner (SOMATOM Force; Siemens Healthineers) in 39 patients (13.1%). There was no significant difference in scanner type distribution between smokers and non-smokers (P=0.62). The CCTA acquisition followed the guidelines set by the Society of Cardiovascular Computed Tomography (SCCT). Prior to the scan, sublingual nitroglycerin (0.1 mg per dose, nitroglycerin spray) was administered 3–5 minutes before the procedure.

Non-contrast CT images were obtained at a 70% R-R interval using prospective electrocardiogram (ECG) gating with parameters set at 120 kV, slice thickness of 3 mm, and no iterative reconstruction. Coronary artery calcium (CAC) scores were calculated using the Agatston method.

CCTA images were acquired using either prospective or retrospective ECG-triggered techniques at 35–75% of the R-R interval. The scanning parameters were as follows: tube voltage ranged from 70 to 120 kV depending on body mass index (BMI); tube current was 340 mA for dual-source CT (CARE Dose 4D, Siemens) and 350–700 mA for revolution CT (smart mA); field of view was 20 cm × 20 cm; reconstruction thickness was 0.625 or 0.75 mm. All scanners employed iterative reconstruction techniques (flash: SAFIRE; force: ADMIRE; revolution CT: ASiR).

In the CT core laboratory, a blind assessment of epicardial coronary artery stenosis with a diameter ≥2 mm was conducted using a 17-segment coronary artery model. CCTA images were visualized through axial and multiplanar reconstructions. Stenosis of the target vessels was visually categorized into groups of 30–49%, 50–69%, and 70–90%. Coronary artery lesions causing ≥50% lumen stenosis were defined as obstructive CAD.

ICA and FFR measurements

Two cardiologists with over 10 years of experience performed ICA and FFR measurements according to the guidelines provided by the American College of Cardiology (ACA). To ensure comprehensive visualization and analysis of the vessels, two or more optimized projection angles were selected for each major coronary artery.

Prior to FFR assessment, adenosine was administered intravenously (140–180 µg/kg/min) to induce hyperemia. The pressure guidewire with a sensor tip was then advanced to 2 cm distal to the site of stenosis. The position of the sensor was recorded to ensure that the FFRct values were obtained at the same location. The guidewire was then gradually withdrawn, and the FFR value was automatically displayed on the monitor. An FFR value of ≤0.80 was considered hemodynamically significant, indicating that the stenosis has a significant impact on myocardial blood flow.

Measurement of left ventricular volume

Myocardial volume measurement was performed by two independent experienced analysts (with >3 years of cardiac CT segmentation experience) who were blinded to patient smoking status and clinical outcomes. Segmentation disagreements were resolved by consensus with a senior cardiac radiologist (with >10 years of experience). Myocardial volume was defined as the total volume of the left ventricular myocardial wall (excluding the left ventricular cavity), measured from end-diastolic CCTA images. The measurement was not indexed to body surface area in the primary analysis, as we aimed to evaluate the raw volume used directly in the CFD boundary condition calculation. Using Mimics software (Materialise NV, Leuven, Belgium), we first preprocessed the imported CCTA images, including filtering and noise reduction, to enhance image quality. Next, we employed the software’s semi-automatic or manual segmentation tools to delineate the endocardial and epicardial boundaries of the left ventricle layer by layer on the CCTA images. During this process, we carefully adjusted thresholds and used contouring tools to ensure accurate segmentation at each layer. After completing the segmentation across all layers, we used Mimics’ 3D reconstruction feature to generate a 3D model of the left ventricle. Finally, we measured the volume of the segmented left ventricle model using the software’s volume calculation tool, providing an accurate measurement of the left ventricular volume. The entire process was meticulous and thorough, ensuring the accuracy of the measurement results.

FFRct calculation

The FFRct calculation software system, developed and provided by Hangzhou Artery Co., Ltd. (Hangzhou, China), employs the principles of CFD and is conducted by core research personnel in a blinded manner. The process includes the following steps: (I) constructing a 3D anatomical model of the coronary arteries under simulated maximum hyperemia; (II) defining the centerline and boundary of the lumen; and (III) performing the FFRct calculation (Figure 1). The FFRct value for the target vessel is recorded at the position during the FFR assessment. An FFRct value of ≤0.80 is identified as a significant flow-limiting lesion. The coronary artery geometries were reconstructed from medical images using Mimics. During segmentation, all coronary branches with a diameter larger than 0.8 mm were explicitly segmented and included in the computational model, whereas smaller vessels were excluded due to imaging resolution limits. The FFRct software used in this study (ArteryFrac™, Hangzhou Artery Co., Ltd.) has been previously validated in one large multicenter prospective study (13), showing a per-patient diagnostic accuracy of 89% against invasive FFR. The same algorithm version was used for all patients in this study, with no smoking-specific adjustments to the core calculation model. The myocardial volume adjustment used in boundary condition setting is a standard pre-specified component of the algorithm, applied uniformly to all patients regardless of smoking status. The processing failure rate for FFRct calculation in this study was 1.0% (3/301 patients), with a median processing time of 11.7 minutes per case, consistent with clinical workflow requirements.

Figure 1 FFRct calculation workflow. 3D, three-dimensional; CTA, computed tomography angiography; FFR, fractional flow reserve; FFRct, computed tomography-derived fractional flow reserve.

The calculation assumes that blood flow follows Newtonian fluid dynamics (with constant viscosity) and is modeled using time-dependent incompressible Navier-Stokes equations with specific boundary conditions. The Navier-Stokes equations consist of two fundamental equations describing blood flow, velocity, and pressure: the conservation of momentum and the conservation of mass. A no-slip boundary condition is applied on the vessel walls (assuming zero velocity). A patient-specific time-dependent inflow rate is used as the inlet boundary condition. The flow or velocity at the aortic inlet is estimated from the patient’s individual myocardial volume using the scaling law (14). At the outlet boundary, an impedance boundary condition circuit model is employed, with parameters derived from the personalized myocardial volume.

To solve the continuous Navier-Stokes equations using a computer, we first discretize the equations to obtain a finite-dimensional system. The spatial discretization of the Navier-Stokes equations is performed using a stable finite element method, and time discretization is achieved using a fully implicit second-order backward differentiation formula. The nonlinear system at each time step is solved using the Newton-Krylov-Schwarz (NKS) method, which is a highly scalable parallel solver for nonlinear partial differential equations, applied to complex problems such as compressible Euler equations, partial differential equation (PDE)-constrained optimization problems, and blood flow simulations in pulmonary arteries.

Statistical analysis

The normality of quantitative data was assessed using the Kolmogorov-Smirnov test. Continuous variables following a normal distribution (age, BMI, heart rate, blood pressure) are presented as mean ± standard deviation (SD), whereas non-normally distributed variables (myocardial volume, FFR, FFRct) are reported as median and IQR, as determined by the Kolmogorov-Smirnov test. Pearson’s correlation analysis was used to evaluate the correlation between FFRct and FFR, and differences were visualized using Bland-Altman plots. The diagnostic performance of FFRct and CCTA was assessed using accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) with invasive FFR as the reference standard. Comparisons of diagnostic performance between FFRct and CCTA were conducted using Chi-squared tests, Fisher’s exact tests, or McNemar’s tests. Differences between FFRct and FFR groups were evaluated using Chi-squared tests or Mann-Whitney U tests. DeLong’s test was used to compare AUCs between smokers and non-smokers. Differences in sensitivity, specificity, and accuracy between groups were compared using z-tests for proportions. A P value of <0.05 was considered statistically significant. There were no missing data for the primary variables (smoking status, myocardial volume, CT-FFR, invasive FFR) in the final analysis cohort. Statistical analyses were performed using commercial software [SPSS 23.0 (IBM Corp., Armonk, NY, USA), MedCalc 15.8 (MedCalc Software, Ostend, Belgium)].

Results

Baseline characteristics

A total of 402 cases were screened for this clinical trial. Of these, 84 were not eligible, and 318 were successfully enrolled. Nine cases dropped out during the trial. Among the 309 evaluable cases, 3 could not have their images assessed due to imaging issues, and 8 deviated from the protocol for various reasons. Ultimately, 298 cases completed the trial as per the protocol, including 192 non-smokers and 106 smokers. A flowchart is shown in Figure 2. Among these, 106 were smokers (mean age 63.8±8.9 years) and 192 were non-smokers (mean age 62.8±9.2 years). Patient characteristics and angiographic findings are summarized in Tables 1,2. There were no statistically significant differences between the two groups in terms of sex, age, BMI, heart rate, blood pressure, prevalence of diabetes, dyslipidemia, hypertension, or treatment modalities (Table 1). Additionally, quantitative coronary angiography (QCA) and echocardiographic analyses revealed no significant differences between the two groups with respect to lesion diameter, left ventricular physiological parameters, or FFR values (Table 2).

Figure 2 Study flow diagram. CAC, coronary artery calcium; CTA, computed tomography angiography; FFR, fractional flow reserve.

Differences in myocardial volume between non-smokers and smokers

Figure 3 illustrates the distribution of myocardial volume among patients. The average myocardial volume in the non-smokers group was 157.89 cm³ (IQR, 133.16–191.26 cm³), whereas that in the smokers group was 193.36 cm³ (IQR, 166.15–222.89 cm³). The difference between the two groups was statistically significant (Z=−6.249, P<0.0001).

Figure 3 The difference in myocardial volume between non-smokers and smokers.

FFR and FFRct among non-smokers and smokers

Figure 4A shows the distribution of invasive measured FFR. The FFR values obtained from ICA for the non-smokers group and smokers group were 0.85 (IQR, 0.79–0.90) and 0.85 (IQR, 0.78–0.90), respectively, with no statistically significant difference (Z=−0.010, P=0.496). Figure 4B presents the distribution of FFR derived from CFD. The FFRct values for the non-smokers group and smokers group were 0.86 (IQR, 0.77–0.93) and 0.86 (IQR, 0.77–0.92), respectively, with no statistically significant difference (Z=−0.086, P=0.466).

Figure 4 FFR and FFRct among non-smokers and smokers. FFR, fractional flow reserve; FFRct, computed tomography-derived fractional flow reserve.

Differences between non-smokers and smokers in FFR and FFRct

Bland-Altman analysis revealed that the level of difference between measured FFR and FFRct was similar in both the non-smokers and smokers groups (Figure 5). Linear regression analysis demonstrated a strong linear relationship between FFR and FFRct in both the non-smokers and smokers groups (Figure 6).

Figure 5 Correlation between FFR and FFRct. A good correlation was observed in both non-smokers (r=0.726, P<0.0001) and smokers (r=0.730, P<0.0001). FFR, fractional flow reserve; FFRct, computed tomography-derived fractional flow reserve.

Figure 6 Consistency between FFR and FFRct. Bland-Altman plots for non-smokers and smokers demonstrate good consistency. FFR, fractional flow reserve; FFRct, computed tomography-derived fractional flow reserve; SD, standard deviation.

Differences between FFR and FFRct

In this study, we compared the absolute and relative errors between non-smokers and smokers. The absolute error for the non-smoker group was 0.010 (range, −0.030 to 0.050), whereas that for the smoker group was 0.010 (range, −0.033 to 0.043). Statistical analysis revealed a Z value of −0.065 and a P value of 0.474, indicating that the difference in absolute error between the two groups was not statistically significant (P>0.05). Regarding relative error (%), the non-smoker group had a relative error of 1.14% (range, −4.08% to 6.06%), whereas the smoker group had a relative error of 1.25% (range, −3.58% to 5.34%). The statistical analysis yielded a Z value of −0.125 and a P value of 0.450, suggesting that the difference in relative error between the two groups was also not statistically significant (P>0.05).

Diagnostic performance of FFRct in non-smokers and smokers

Figure 7 summarizes the diagnostic performance of FFRct in non-smokers and smokers, including sensitivity, specificity, accuracy, PPV, and NPV. The sensitivity, specificity, accuracy, PPV, and NPV of FFRct for non-smokers were 90.00%, 88.64%, 89.06%, 78.26%, and 95.12%, respectively, and those for smokers were 87.50%, 89.19%, 88.68%, 77.78%, and 94.29%, respectively.

Figure 7 AUCs for FFRct-smokers and FFRct-non-smokers are shown. AUC, area under the receiver operating characteristic curve; FFRct, computed tomography-derived fractional flow reserve.

Table 3 summarizes the AUC for non-smokers and smokers {0.928 [95% confidence interval (CI): 0.888–0.969] and 0.919 (95% CI: 0.858–0.979)}, showing that the AUCs for the two groups are similar.

Discussion

This cross-sectional study of 298 patients with suspected CAD found that although smokers had significantly larger myocardial volumes than non-smokers, there were no significant differences in invasive FFR, FFRct, or FFRct diagnostic performance between groups. These findings indicate that smoking status does not degrade the diagnostic performance of FFRct when personalized myocardial volume calibration is used in CFD simulations. Furthermore, this finding of this study suggests that there is no significant association between smoking status and the diagnostic efficacy of FFRct in the current study cohort. This conclusion still needs further validation through multi-factor studies with larger sample sizes. This study comprised an observational subgroup analysis that only focuses on the impact of smoking status on FFRct performance. It does not explore the independent effects of other CAD risk factors, including hypertension, diabetes, and family history of cardiovascular disease. The influence of these factors on the diagnostic efficacy of FFRct remains to be verified in further studies.

We observed significantly larger myocardial volumes in smokers compared to non-smokers, consistent with prior reports of smoking-related myocardial remodeling (15). However, in the absence of direct CMD measurements [e.g., index of microcirculatory resistance (IMR), coronary flow reserve (CFR)], we cannot determine whether this volume difference is related to microvascular dysfunction. Future studies with direct CMD assessment are needed to explore potential mechanistic pathways (16). Although echocardiographic LVEF and LVEDD were similar between groups in our cohort, these parameters are less sensitive for detecting early diffuse myocardial remodeling compared to volumetric CT measurements. The increased myocardial volume observed in this study may be associated with these pathophysiological changes (17,18). However, despite the increased myocardial volume in smokers, their FFR and FFRct values showed no significant differences from non-smokers, which may suggest that FFR and FFRct are more sensitive to evaluating large vessel lesions but have limited ability to reflect microcirculatory changes. Consistent with related studies, although smoking acutely increases blood pressure, large-scale epidemiological studies have found that smokers have slightly lower blood pressure levels than non-smokers (19).

Our team’s 3D modeling based on CCTA data enables non-invasive estimation of microcirculatory resistance, and the negative correlation between myocardial volume and microvascular density provides biological plausibility for volume-resistance dynamic calibration. This study found no significant differences in absolute and relative errors between smokers and non-smokers, indicating high computational consistency of FFRct in both groups. These findings demonstrate that this method can accurately simulate hemodynamics. Specifically, at the technical level, this study quantifies myocardial volume through 3D myocardial reconstruction and constructs a dynamic 3-element Windkessel model to individually calibrate microcirculatory resistance, which offers significant advantages over traditional methods. Early studies mostly relied on literature values to set fixed resistance parameters (20) or converted invasively measured IMR into flow resistance values using proportional coefficients (21), neither of which can reflect individual anatomical differences. In recent years, notable progress has been made in optimizing boundary condition settings by incorporating morphometric scaling laws, which allow for more personalized and accurate modeling of vascular hemodynamics (22). In contrast, the advantage of 3D reconstruction used in this study is its ability to directly reflect morphological changes of the myocardium and correlate myocardial volume with flow resistance through mathematical modeling, thereby enabling more accurate assessment of microcirculatory status.

Our study extends these previous findings by demonstrating that even with smoking-related increases in myocardial volume, the personalized boundary condition calibration using measured myocardial volume effectively mitigates potential bias in FFRct calculation. This is a key methodological contribution, as most prior FFRct studies use population-averaged boundary conditions rather than patient-specific myocardial volume measurements (23). Our results suggest that the use of personalized myocardial volume calibration may improve the robustness of FFRct across patient subgroups with different myocardial morphological characteristics.

From a clinical perspective, our findings provide important evidence supporting the reliability of FFRct in smoking populations, who are at higher risk of CAD but often have comorbidities that may affect functional test performance. The consistent diagnostic performance across smoking groups suggests that FFRct can be used with confidence in clinical decision-making for smokers, without the need for smoking-specific adjustment factors.

The results of this study indicate that FFRct has similar diagnostic performance in smokers and non-smokers and can be used to evaluate CAD in smokers. Although smoking is a well-established risk factor for ischemic heart disease, its impact on invasive objective measures of microvascular physiology remains unclear, largely due to challenges in reliably obtaining invasive measurement data. Some evidence suggests that smokers may exhibit microcirculatory dysfunction, but this association is not definitively confirmed. Thus, in clinical practice, it may be reasonable to consider incorporating other microcirculatory assessment indicators (such as IMR or CFR) when evaluating myocardial blood flow status in smokers, though the added clinical value of such invasive assessments requires further validation. Notably, a recent study combining CFD with invasive measurements suggested that the association between smoking and absolute coronary microvascular resistance is negligible (24). Additionally, the non-invasiveness and high precision of FFRct make it suitable for large-scale screening and long-term follow-up, particularly in smoking populations, facilitating early detection of CAD and guiding treatment decisions.

The limitations of this study include the following: first, the relatively small sample size, especially the smoker group with only 106 cases, which may affect the generalizability of the results; second, the cross-sectional design cannot determine the causal relationship between smoking and increased myocardial volume. In the future, we will continue to expand the sample size and conduct longitudinal studies to clarify the long-term effects of smoking on myocardial volume and microcirculation; combine multimodal imaging techniques (such as cardiac magnetic resonance imaging) and biomarkers to further optimize the FFRct calculation model; and explore the molecular mechanisms of microcirculatory dysfunction in smokers to provide a theoretical basis for personalized treatment. Smoking exposure was categorized based on self-reported history, with no objective biomarker verification (e.g., cotinine levels), which introduces a risk of exposure misclassification. Although we have provided detailed smoking exposure parameters including pack-years, smoking duration, and time since cessation, residual confounding from unmeasured smoking-related factors cannot be fully excluded.

Conclusions

Despite significantly larger myocardial volumes in smokers, FFRct demonstrates consistent measurement accuracy and diagnostic performance in both smokers and non-smokers. These findings support the utility of FFRct for evaluating CAD in smoking populations. This study provides a new perspective for evaluating CAD in smokers, highlights the clinical application potential of the novel FFRct, and identifies clarifying the long-term impact of smoking on myocardial volume and microcirculation, and further optimizing the FFRct computational model based on these findings as key issues to be addressed in future research.

Acknowledgments

We would like to thank Xincheng Li for his significant contributions to the data collection and two rounds of manuscript revision.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2409/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2409/dss

Funding: This work was supported by the Natural Science Foundation of Zhejiang Province (No. LQN25H180006) and the Key Research and Development Program of Zhejiang Province (No. 2025C02144).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2409/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Zhongshan Hospital Fudan University (No. 2019-139R). Written informed consent was obtained from all enrolled patients. All participating local institutions were informed of and agreed to the study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Henderson A. Coronary heart disease: overview. Lancet 1996;348:s1-4. [Crossref] [PubMed]
2. Dalen JE, Alpert JS, Goldberg RJ, Weinstein RS. The epidemic of the 20(th) century: coronary heart disease. Am J Med 2014;127:807-12. [Crossref] [PubMed]
3. Boden WE, Marzilli M, Crea F, Mancini GBJ, Weintraub WS, Taqueti VR, Pepine CJ, Escaned J, Al-Lamee R, Gowdak LHW, Berry C, Kaski JC. Chronic Myocardial Ischemic Syndromes Task Force. Evolving Management Paradigm for Stable Ischemic Heart Disease Patients: JACC Review Topic of the Week. J Am Coll Cardiol 2023;81:505-14. [Crossref] [PubMed]
4. Johnson NP, Gould KL, Di Carli MF, Taqueti VR. Invasive FFR and Noninvasive CFR in the Evaluation of Ischemia: What Is the Future? J Am Coll Cardiol 2016;67:2772-88. [Crossref] [PubMed]
5. Chen J, Wetzel LH, Pope KL, Meek LJ, Rosamond T, Walker CM. FFR(CT): Current Status. AJR Am J Roentgenol 2021;216:640-8. [Crossref] [PubMed]
6. Tang CX, Liu CY, Lu MJ, Schoepf UJ, Tesche C, Bayer RR 2nd, et al. CT FFR for Ischemia-Specific CAD With a New Computational Fluid Dynamics Algorithm: A Chinese Multicenter Study. JACC Cardiovasc Imaging 2020;13:980-90. [Crossref] [PubMed]
7. Cury RC, Leipsic J, Abbara S, Achenbach S, Berman D, Bittencourt M, et al. CAD-RADS™ 2.0 - 2022 Coronary Artery Disease-Reporting and Data System: An Expert Consensus Document of the Society of Cardiovascular Computed Tomography (SCCT), the American College of Cardiology (ACC), the American College of Radiology (ACR), and the North America Society of Cardiovascular Imaging (NASCI). J Cardiovasc Comput Tomogr 2022;16:536-57. [Crossref] [PubMed]
8. Driessen RS, Danad I, Stuijfzand WJ, Raijmakers PG, Schumacher SP, van Diemen PA, Leipsic JA, Knuuti J, Underwood SR, van de Ven PM, van Rossum AC, Taylor CA, Knaapen P. Comparison of Coronary Computed Tomography Angiography, Fractional Flow Reserve, and Perfusion Imaging for Ischemia Diagnosis. J Am Coll Cardiol 2019;73:161-73. [Crossref] [PubMed]
9. Han Y, Hu Y, Yu C, Guo Y, Pei P, Yang L, Chen Y, Du H, Sun D, Pang Y, Chen N, Clarke R, Chen J, Chen Z, Li L, Lv JChina Kadoorie Biobank Collaborative Group. Lifestyle, cardiometabolic disease, and multimorbidity in a prospective Chinese study. Eur Heart J 2021;42:3374-84. [Crossref] [PubMed]
10. Qiu Z, Wang Y, Liu Y, Zhou Z, Wang Z. Diagnostic value of angiography-derived index of microcirculatory resistance (AMR) for coronary microcirculatory dysfunction (CMD) and its prognostic significance in patients with chronic coronary syndromes in the smoking population. Medicine (Baltimore) 2024;103:e37022. [Crossref] [PubMed]
11. Geng Y, Liu H, Wang X, Zhang J, Gong Y, Zheng D, Jiang J, Xia L. Effect of microcirculatory dysfunction on coronary hemodynamics: A pilot study based on computational fluid dynamics simulation. Comput Biol Med 2022;146:105583. [Crossref] [PubMed]
12. Al-Lamee R, Thompson D, Dehbi HM, Sen S, Tang K, Davies J, et al. Percutaneous coronary intervention in stable angina (ORBITA): a double-blind, randomised controlled trial. Lancet 2018;391:31-40. [Crossref] [PubMed]
13. Guo W, He W, Lu Y, Yin J, Shen L, Yang S, et al. CT-FFR by expanding coronary tree with Newton-Krylov-Schwarz method to solve the governing equations of CFD. Eur Heart J Imaging Methods Pract 2024;2:qyae106. [Crossref] [PubMed]
14. Choy JS, Kassab GS. Scaling of myocardial mass to flow and morphometry of coronary arteries. J Appl Physiol (1985) 2008;104:1281-6. [Crossref] [PubMed]
15. Holmes KR, Gulsin GS, Fairbairn TA, Hurwitz-Koweek L, Matsuo H, Nørgaard BL, et al. Impact of Smoking on Coronary Volume-to-Myocardial Mass Ratio: An ADVANCE Registry Substudy. Radiol Cardiothorac Imaging 2024;6:e220197. [Crossref] [PubMed]
16. Freund O, Shetrit A, Bar-Shai A, Zornitzki L, Frydman S, Banai A, Shamir RA, Ben-Shoshan J, Arbel Y, Banai S, Konigstein M. Smoking and Respiratory Diseases in Patients with Coronary Microvascular Dysfunction. Am J Med 2024;137:538-544.e1. [Crossref] [PubMed]
17. Rader F, Rashid M, Nguyen TT, Luong E, Kim A, Kim E, Elashoff R, Davoren K, Moy N, Nafeh F, Merz NB, Ebinger J, Hamburg N, Lindner J, Cheng S. E-Cigarette Use and Subclinical Cardiac Effects. Circ Res 2020;127:1566-7. [Crossref] [PubMed]
18. Burroughs Peña MS, Swett K, Kaplan RC, Perreira K, Daviglus M, Kansal MM, Cai J, Giachello AL, Gellman MD, Velazquez EJ, Rodriguez CJ. Childhood and adult exposure to secondhand tobacco smoke and cardiac structure and function: results from Echo-SOL. Open Heart 2018;5:e000831. [Crossref] [PubMed]
19. Omvik P. How smoking affects blood pressure. Blood Press 1996;5:71-7. [Crossref] [PubMed]
20. Andreini D, Mushtaq S, Conte E, Mei M, Nicoli F, Melotti E, Pompilio G, Pepi M, Bartorelli AL, Onuma Y, Serruys PW. The usefulness of cardiac CT integrated with FFRCT for planning myocardial revascularization in complex coronary artery disease: a lesson from SYNTAX studies. Cardiovasc Diagn Ther 2020;10:2036-47. [Crossref] [PubMed]
21. Fairbairn TA. Fractional Flow ReserveCT (FFRCT). Eur Heart J 2017;38:1867-9. [Crossref] [PubMed]
22. Taylor CA, Fonte TA, Min JK. Computational fluid dynamics applied to cardiac computed tomography for noninvasive quantification of fractional flow reserve: scientific basis. J Am Coll Cardiol 2013;61:2233-41. [Crossref] [PubMed]
23. Narimani-Javid R, Moradi M, Mahalleh M, Najafi-Vosough R, Arzhangzadeh A, Khalique O, Mojibian H, Kuno T, Mohsen A, Alam M, Shafiei S, Khansari N, Shaghaghi Z, Nozhat S, Hosseini K, Hosseini SK. Machine learning and computational fluid dynamics derived FFRCT demonstrate comparable diagnostic performance in patients with coronary artery disease; A Systematic Review and Meta-Analysis. J Cardiovasc Comput Tomogr 2025;19:232-46. [Crossref] [PubMed]
24. Taylor DJ, Aubiniere-Robb L, Gosling R, Newman T, Hose DR, Halliday I, Lawford PV, Narracott AJ, Gunn JP, Morris PD. Sex differences in coronary microvascular resistance measured by a computational fluid dynamics model. Front Cardiovasc Med 2023;10:1159160. [Crossref] [PubMed]