The Predictive Value of Quantitative Flow Ratio and Its Derived Angiographic Microvascular Resistance for Restenosis Following Drug‐Coated Balloon Angioplasty

ABSTRACT

Background

Initially, drug‐coated balloon (DCB) angioplasty was primarily employed for in‐stent restenosis (ISR) treatment. Over time, its indications have broadened to include de novo small‐vessel lesions and bifurcation lesions. However, there is a lack of effective strategies to reduce restenosis rates post‐DCB angioplasty.

Aims

Exploring the predictive value of quantitative flow ratio (QFR) and its derived angiographic microvascular resistance (AMR) for vascular restenosis following DCB angioplasty.

Methods

This study enrolled 108 patients who received DCB angioplasty during the period between February 2021 and October 2024. All patients underwent follow‐up coronary angiography at 1 year post‐procedure. Based on angiographic findings, patients were divided into a restenosis group (n = 38) and a non‐restenosis group (n = 70). The study compared preoperative parameters, surgical data, and postoperative variables.

Results

The restenosis group demonstrated a significantly higher prevalence of diabetes (p < 0.05), greater history of smoking (p < 0.05), lower postoperative QFR values (p < 0.05), and a higher proportion of patients with AMR values exceeding 2.5 (p < 0.05) compared to the non‐restenosis group. Multivariate logistic regression analysis identified postoperative QFR and high AMR values as independent predictors of restenosis after DCB therapy. ROC curve analysis demonstrated that the AUC for postoperative QFR in predicting restenosis was 0.727 (95% CI: 0.518–0.936, p < 0.05), which improved to 0.903 (95% CI: 0.782–1, p < 0.01) when combined with high AMR values.

Conclusions

Coronary angiography‐derived QFR and AMR are closely associated with vascular restenosis in patients treated with DCB. Routine postoperative measurement of QFR and AMR may enhance risk prediction for restenosis following DCB angioplasty.

1. Introduction

Coronary atherosclerotic heart disease (CHD) is a cardiac condition caused by lipid deposits forming plaques on the inner walls of coronary arteries, leading to vascular narrowing or occlusion. This results in myocardial ischemia, hypoxia, or even necrosis. As a common disease threatening the health of both urban and rural residents in China, epidemiological data shows the prevalence of cardiovascular diseases continues to rise in China [1], with CHD emerging as one of the leading causes of morbidity and mortality [2]. The current treatment options for CHD include medical conservative therapy, coronary artery bypass grafting (CABG), and percutaneous coronary intervention (PCI). Compared to CABG, PCI offers advantages such as minimal invasiveness and faster recovery, making it the predominant treatment modality today. Among PCI techniques, drug‐eluting stents (DES) and DCB are widely used. With the growing adoption of the “stent‐less intervention” concept, DCB applications have expanded significantly. Initially, DCB angioplasty was primarily employed for in‐stent restenosis (ISR) treatment. Over time, its indications have broadened to include de novo small‐vessel lesions, bifurcation lesions, and large‐vessel native lesions, with efficacy validated by multiple studies [3, 4]. However, current research predominantly focuses on stent‐related restenosis, while studies on DCB‐induced restenosis remain limited. There is a lack of effective strategies to reduce restenosis rates post‐DCB angioplasty.

Studies have demonstrated that anatomical stenosis assessed by coronary angiography (CAG) does not fully correlate with functional myocardial ischemia, while a inverse relationship exists between post‐interventional fractional flow reserve (FFR) and adverse event risks [5, 6]. Additionally, coronary microcirculation plays a crucial role in regulating myocardial blood flow and metabolism [7], and is currently primarily assessed through the Index of Microcirculatory Resistance (IMR). However, both methods are invasive procedures associated with high material costs and expenses. They require adenosine‐induced maximal vasodilation, making the detection process time‐consuming and clinically inconvenient.

Based on this, quantitative flow ratio (QFR) emerged as a technique that reconstructs three‐dimensional coronary vessels from angiography images to calculate FFR. Its advantages—short analysis time, retrospective applicability, and elimination of disposable consumables—have garnered widespread acclaim. In recent years, with the increasing adoption of QFR, multiple studies have demonstrated its strong diagnostic consistency with invasive FFR [8, 9, 10]. Furthermore, Angiography‐derived Microcirculatory Resistance (AMR), a QFR‐derived parameter, has shown excellent correlation with pressure‐wire‐based IMR in clinical trials. These findings position AMR as a feasible, non‐invasive alternative to IMR for evaluating coronary microvascular disease (CMD) [11]. Therefore, this study collected clinical data from patients undergoing DCB angioplasty, utilizing the Pulse Medical AngioPlus QFR Measurement System (produced by Imaging Technologies Shanghai Co. Ltd.) to assess QFR and AMR. Baseline characteristics, intraoperative PCI details, and postoperative follow‐up data were systematically recorded to identify factors influencing post‐DCB restenosis. By evaluating coronary hemodynamic function (via QFR) and microcirculatory performance (via AMR), the study aims to: 1. Analyze the predictive value of QFR and AMR for restenosis risk after DCB treatment; 2. Provide evidence‐based insights to optimize clinical decision‐making protocols for DCB‐treated patients.

2. Methods

2.1. Study Population

The study investigated 136 patients (136 lesions) exclusively treated with DCBs for the first time between February 2021 and October 2024. After excluding 28 patients (28 lesions) who were lost to follow‐up angiography or had poor‐quality angiographic images, 108 patients (108 lesions) who underwent DCB angioplasty and received follow‐up CAG at our hospital 1 year post‐procedure were included (Figure 1). The cohort comprised 82 males (76%) and 26 females (24%), with a mean age of 65 years. Based on angiographic findings, patients were divided into a restenosis group (n = 38) and a non‐restenosis group (n = 70). Among them, 22 patients underwent PCI again due to acute coronary syndrome (ACS) occurring during follow‐up or severe stenosis observed during repeat coro‐nary angiography. Vascular restenosis is defined as a stenosis exceeding 50% diameter stenosis at the lesion site at 1 year post‐PCI, compared with the post‐procedural lumen measurement immediately after the initial PCI procedure.

Figure 1.

Flow chart of the study.

Inclusion Criteria: 1. Patients meeting clinical indications for DCB angioplasty for coronary artery disease (CAD); 2. Either of the following within the specified timeframe: Rehospitalization for PCI due to chest pain within 1 year post‐procedure; Asymptomatic follow‐up CAG (CAG) at 1 year post‐procedure. Exclusion Criteria: 1. Poor‐quality CAG images (e.g., inadequate visualization of lesions for quantitative analysis); 2. Contraindications to anticoagulation/antiplatelet therapy; Severe comorbidities including hepatic/renal dysfunction, malignancy, autoimmune diseases, hematologic disorders, or other severe systemic conditions.

2.2. Surgical Procedure

All patients underwent CAG and DCB angioplasty post‐admission, with procedures performed by the same multidisciplinary team via transradial access using ioversol as the contrast agent, followed by tailored dual antiplatelet therapy (DAPT) (aspirin combined with clopidogrel or ticagrelor) and continuous electrocardiographic monitoring as standard postoperative care based on individual clinical status and risk stratification.

2.3. Data Collection

Through the hospital electronic medical record (EMR) management system, the following patient‐related data are collected: ①Demographic information: Age, gender, smoking history, hypertension, and history of diabetes mellitus; ②Pre‐PCI Indicators: Complete Blood Count (CBC), High‐Density Lipoprotein (HDL), Low‐Density Lipoprotein (LDL), Apolipoprotein A (ApoA), Apolipoprotein B (ApoB), Lipoprotein(a) [Lp(a)], Glycated Hemoglobin (HbA1c), Left Ventricular Ejection Fraction (LVEF); ③Surgical Data: Postoperative AMR, Postoperative QFR, Target vessel, Postoperative residual stenosis; ④Postoperative Follow‐up: Restenosis status of the diseased vessel.

The QFR and AMR values are measured using BoDong Medical Imaging Technology (Shanghai) Co. Ltd.'s Angioplus Quantitative Flow Fraction Measurement System: ①Angiographic Imaging Acquisition: Select appropriate angiographic views to obtain complete and clear vascular imaging; ②AI‐Powered Analysis: The system employs artificial intelligence algorithms to automatically compute QFR and AMR values within 30 s.

This study has been approved by the Ethics Committee of Ningbo Medical Center Li Huili Hospital (No. 2025‐126).

2.4. Statistical Methods

Categorical data are expressed as counts (%) and analyzed using the Chi‐square (χ²) test, while normally distributed continuous variables are reported as mean ± standard deviation and non‐normally distributed variables as median and interquartile range (IQR). The normality of data distribution was assessed via the Kolmogorov‐Smirnov test, with independent samples t‐tests applied for normally distributed continuous variables and Mann−Whitney U tests for non‐normal distributions. Logistic regression analysis was performed to identify independent influencing factors of restenosis after (DCB) angioplasty. The diagnostic performance was evaluated by plotting the Receiver Operating Characteristic (ROC) curve and calculating the Area Under the Curve (AUC). A two‐tailed p < 0.05 was considered as statistically significant. All data statistics were performed using the SPSS, version 26.0 (SPSS Inc., IBM).

3. Results

3.1. Preoperative Data

The study enrolled 108 patients, with 70 showing no restenosis and 38 developing restenosis upon follow‐up. Preoperative characteristics including age, sex, LVEF, hypertension history, baseline HbA1c, LDL, and HDL showed no significant differences between groups. However, smoking (28.6% vs. 47.5%, p < 0.05) and diabetes (20% vs. 47.5%, p < 0.05) were significantly associated with higher postoperative restenosis rates, as detailed in Table 1.

Table 1.

Preoperative data.

Characteristics	Recurrent restenosis No (n = 70)	Yes (n = 38)	p value
Age, year	62.8 ± 13.5	68.7 ± 10.4	0.101
Female	22 (31.4%)	4 (10.5%)	0.086
LVEF, %	62.5 [IQR: 52.5–68.3]	64.0 [IQR: 60–67]	0.905
Hypertension	42 (60%)	28 (73.7%)	0.315
Diabetes mellitus	14 (20%)	18 (47.4%)	0.035
Smoking	20 (28.6%)	18 (47.5%)	0.035
HbA1c, %	5.84 ± 0.67	6.69 ± 1.23	0.062
LDL‐C, mmol/L	2.33 ± 0.81	2.69 ± 0.96	0.148
HDL‐C, mmol/L	0.99 ± 0.18	1.01 ± 0.20	0.420
ApoA, mmol/L	1.13 ± 0.21	1.03 ± 0.16	0.744
ApoB, mmol/L	0.77 ± 0.23	0.88 ± 0.31	0.134
Lp(a), mmol/L	0.19 ± 0.18	0.17 ± 0.19	0.551
Hemoglobin, g/L	7.27 ± 1.28	7.34 ± 0.21	0.956

Abbreviations: ApoA = Apolipoprotein A, ApoB = Apolipoprotein B, HDL‐C = high density lipoprotein cholesterol, LDL‐C = low density lipoprotein cholesterol, Lp(a) = Lipoprotein(a).

3.2. Surgical Data

Among the 108 patients, the target vessels included the left anterior descending artery (LAD, including diagonal branches) in 56 cases, the left circumflex artery (LCX, including obtuse marginal branches) in 30 cases, and the right coronary artery (RCA) in 22 cases, with no statistically significant differences observed among these vascular distributions. Similarly, previous PCI history and residual stenosis after PCI showed no statistically significant differences between groups. Compared with the non‐restenosis group, the DCBs used in the restenosis group had smaller diameter, longer length, lower inflation pressure, and longer inflation time, though none of these differences were statistically significant. Immediate postoperative QFR and AMR measurements revealed that the non‐restenosis group exhibited significantly higher QFR values compared to the restenosis group (0.914 ± 0.067 vs. 0.860 ± 0.083, p < 0.05), demonstrating the prognostic value of QFR in evaluating DCB treatment efficacy. Previous studies have identified the correlation between IMR and coronary microcirculation status. The《2019 ESC Guidelines for the Diagnosis and Management of Chronic Coronary Syndromes》recommended IMR ≥ 25 as the diagnostic threshold for microcirculatory dysfunction [12]. Recent research further revealed that the optimal cutoff value of AMR for predicting IMR ≥ 25 was 2.5 mmHg·s/cm. Based on these findings, patients with postoperative AMR values > 2.5 mmHg·s/cm were categorized as the high‐AMR group [11]. While the postoperative AMR values in the restenosis group were numerically higher than those in the non‐restenosis group, the difference did not reach statistical significance. However, subgroup analysis demonstrated a significantly higher proportion of high‐AMR patients in the restenosis group compared to the non‐restenosis group (p < 0.05), as detailed in Table 2.

Table 2.

Surgical data.

Characteristics	Recurrent restenosis No (n = 70)	Yes (n = 38)	p value
Target vessel, n			0.472
LAD/DIA	32	54
LCX/OM	22	8
RCA	16	6
History of PCI, %	22.9	21.1	0.879
NSE balloon, n (%)	26 (37.14)	12 (31.58)	0.683
Cutting balloon, n (%)	8 (11.43)	2 (5.26)	0.455
High‐pressure balloon, n (%)	10 (14.29)	2 (5.26)	0.314
DCB brands, n (%)			0.186
SeQuent	18	4
Restore	52	34
DCB diameter, mm	2.621 ± 0.498	2.382 ± 0.385	0.091
DCB length, mm	23.11 ± 4.303	25.00 ± 4.410	0.187
DCB pressure, atm	8.80 ± 2.153	8.37 ± 2.033	0.688
DCB time, s	70.59 ± 15.896	70.79 ± 13.971	0.825
IVUS, n (%)	2 (2.86)	0 (0)	0.457
Residual stenosis, %	19.7 ± 10.7	21.1 ± 10.5	0.634
Postoperative QFR	0.912 ± 0.070	0.860 ± 0.083	0.017
AMR, mmHg*s/cm	2.64 ± 0.64	2.79 ± 0.53	0.203
High‐AMR, n	26	22	0.047

Abbreviations: AMR = angiographic microvascular resistance, DIA = diagonal branches, IVUS = intravascular ultrasound, LAD = left anterior descending artery, LCX = left circumflex artery, OM = obtuse marginal branches, RCA = right coronary artery.

3.3. Multivariate Regression Analysis

Using the occurrence of restenosis after DCB procedure as the dependent variable, smoking history, diabetes history, LDL levels, QFR and postoperative high AMR were included in multivariate regression analysis. The results identified high QFR value as a protective factor, while postoperative high AMR and diabetes history emerged as independent risk factors for target vessel restenosis after DCB treatment. The adjusted odds ratios (ORs) were <0.001, 6.803, and 22.799, with 95% confidence intervals (CIs) of 0.000–0.096, 1.049–44.110, and 1.584–328.114, respectively, as detailed in Table 3.

Table 3.

Multivariate regression analysis.

Risk factors	β	SE	Wald	p	OR	95% CI
QFR	−25.190	11.659	4.668	0.031	<0.001	0.000−0.096
Diabetes history	1.917	0.054	4.041	0.044	6.803	1.049−44.110
High‐AMR	3.127	1.361	5.281	0.022	22.799	1.584−328.114

3.4. ROC Curve

ROC curve analysis revealed that the AUC of QFR in predicting restenosis after DCB angioplasty was 0.701, with an optimal cut‐off value of 0.875 (95% CI: 0.552–0.851, p < 0.05), indicating that QFR has moderate predictive value for 1‐year target vessel restenosis in patients undergoing DCB angioplasty. Furthermore, a combined model integrating postoperative QFR, high AMR, and diabetes history demonstrated improved predictive accuracy, achieving an AUC of 0.857 (95% CI: 0.730–0.984, p < 0.01). This highlights that combining QFR with high AMR and diabetes history significantly enhances the precision of predicting 1‐year target vessel restenosis events post‐DCB angioplasty. For detailed results, refer to Table 4 and Figure 2.

Table 4.

Receiver Operating Characteristic (ROC) curve.

	Cut‐off	Area	Sensitivity	Specificity	95% CI
QFR	0.875	0.701	0.579	0.781	0.552−0.851
Diabetes history		0.696			0.476−0.798
High AMR		0.661			0.489−0.833
Combined model		0.857	0.714	0.929	0.730−0.984

Figure 2.

Receiver Operating Characteristic (ROC) curve. [Color figure can be viewed at wileyonlinelibrary.com]

4. Discussion

With the increasing advocacy for “intervention without implantation,” DCBs hold broad clinical prospects. However, restenosis after angioplasty remains a critical concern. Despite satisfactory immediate post‐PCI outcomes and strict adherence to DAPT and lipid‐lowering regimens, some patients still exhibit target vessel restenosis during follow‐up CAG, necessitating repeat revascularization [13, 14]. Therefore, identifying risk factors and developing predictive models for restenosis are imperative.

The findings of this study revealed that the restenosis group had significantly higher proportions of smokers and diabetic patients compared to the non‐restenosis group (p < 0.05). Diabetes was confirmed as an independent risk factor for restenosis after DCB angioplasty, aligning with previous research. Inflammation plays a critical role in atherosclerosis. Tobacco smoke contains harmful components such as nicotine and polycyclic aromatic hydrocarbons, which activate inflammatory pathways and exacerbate endothelial injury [15, 16, 17], thereby promoting target vessel restenosis. For diabetic patients, chronic hyperglycemia disrupts metabolic homeostasis, induces endothelial dysfunction, and activates NADPH oxidase (particularly NOX1/4). This leads to excessive mitochondrial reactive oxygen species (ROS) production, triggering oxidative stress that further damages vascular integrity [18]. This study confirmed that diabetic patients had a 6.803‐fold higher risk of restenosis compared to non‐diabetics, highlighting diabetes as a critical predictor for post‐DCB restenosis.

Previous studies have highlighted that dyslipidemia, particularly elevated levels of LDL and its oxidized derivatives (ox‐LDL), plays a critical role in promoting inflammatory responses. These lipoproteins are readily internalized by macrophages and monocytes, leading to foam cell formation and the development of fatty streaks—a pivotal step in atherosclerotic plaque progression. Oxidized LDL further exacerbates endothelial dysfunction by activating toll‐like receptors (TLRs) and triggering immune‐mediated inflammatory cascades, which destabilize plaques and amplify the risk of target vessel restenosis [19]. In this study, while the non‐restenosis group exhibited lower preoperative LDL levels compared to the restenosis group, the difference was not statistically significant (p > 0.05). This lack of be attributed to the limited sample size, which reduces statistical power to detect subtle differences. Additionally, post‐PCI lipid management strategies, such as statin therapy and dietary interventions, could have modulated LDL levels in both groups, potentially masking underlying associations. Variability in adherence to lipid‐lowering regimens post‐PCI might also explain the observed results.

Previous studies have demonstrated that post‐PCI IMR is strongly associated with adverse cardiovascular events in patients with stable CAD and ST‐segment elevation myocardial infarction (STEMI) [20, 21]. AMR exhibits a high correlation with IMR, reflecting similar pathophysiological mechanisms. In this study, although there was no significant difference in mean AMR values between the restenosis and non‐restenosis groups, the proportion of patients with elevated AMR was significantly higher in the restenosis group (p < 0.05). This suggests that coronary microcirculatory dysfunction (CMD) plays a critical role in target vessel restenosis following drug‐coated balloon (DCB) angioplasty. Mechanistically, PCI‐induced mechanical vascular injury activates platelets and inflammatory pathways [22, 23], leading to the release of pro‐inflammatory cytokines such as IL‐6 and TNF‐α, which exacerbate endothelial damage and directly promote neointimal hyperplasia. These cytokines also amplify oxidative stress by generating ROS, which increase capillary permeability and impair microvascular autoregulation, creating a vicious cycle of persistent inflammation and endothelial dysfunction. Furthermore, prior research has confirmed the close interplay between CMD and systemic inflammation, with elevated CRP levels serving as a validated predictor of restenosis and poor prognosis in post‐PCI patients [24, 25].

FFR remains the gold standard for assessing the physiological severity of coronary stenosis. Despite its longstanding clinical use, the proportion of PCI patients undergoing preoperative coronary physiological evaluation remains relatively low [26]. In contrast, QFR—a technique that analyzes and calculates coronary lesion functionality based on angiographic imaging—has demonstrated strong correlation with FFR in multiple studies [27]. Previous research found that vessels with post‐PCI QFR < 0.9 had significantly higher risks of composite endpoint events compared to those with QFR ≥ 0.9 (25% vs. 3.5%, p < 0.001) [28], and QFR effectively predicts vascular‐related clinical outcomes after DCB angioplasty [29]. These findings align with the current study, which revealed significant differences in QFR between non‐restenosis and restenosis groups. QFR emerged as an independent predictor of target vessel restenosis, with an ROC AUC of 0.701 (95% CI: 0.552–0.851 an optimal cutoff value of 0.875, indicating higher restenosis risk when post‐procedural QFR falls below this threshold. Furthermore, combining QFR with risk factors such as diabetes history and elevated AMR improved predictive accuracy, yielding an AUC of 0.857 (95% CI: 0.730–0.984), highlighting enhanced prognostic value for post‐DCB restenosis.

5. Conclusion

In summary, QFR and its derivative AMR, as non‐invasive tools based on angiographic imaging, are poised to significantly improve the adoption rate of coronary physiological evaluations in future PCI patients. The restenosis model developed in this study following DCB angioplasty provides a framework for early postoperative risk stratification, offering scientific guidance for preventive strategies and clinical decision‐making. When diabetic patients exhibit lower post‐DCB QFR (<0.875) and elevated AMR (>2.5) after DCB angioplasty, implementing intensified interventions (e.g., enhanced lipid control) or surveillance protocols can significantly reduce the risk of target vessel restenosis or enable its early detection, thereby preventing postprocedural myocardial infarction.

6. Limitations

This study is a single‐center retrospective analysis with a limited sample size, and given the scarcity of existing AMR research both domestically and internationally, the findings may carry potential biases. Compared with previous studies, the restenosis rate in this study was higher, which might be attributed to the generally severe vascular lesions, longer lesion lengths, thinner lumen diameters of the enrolled cases, as well as the lower utilization rates of adjunctive tools such as cutting balloons and IVUS. Additionally, the inability to perform QFR computations due to loss to follow‐up, suboptimal angiographic image quality, or vessel tortuosity may have impacted the final analysis. Further large‐scale prospective clinical studies are warranted to validate and expand upon these results.

Conflicts of Interest

The authors declare no conflicts of interest.