Predictors of one-year clinically driven revascularization following endovascular treatment of isolated atherosclerotic popliteal artery lesions

Abstract

Isolated atherosclerotic popliteal artery lesions (IAPAL) commonly require interventional treatment. This study aimed to develop a prediction model using procedure-related variables for one-year clinically driven target lesion revascularization (CDTLR) events after intervention. Clinical data were retrospectively collected from 217 patients who underwent endovascular treatment for isolated atherosclerotic popliteal artery lesions between 2017 and 2022. Based on inclusion and exclusion criteria, all patients were randomly divided into training and testing sets at a ratio of 7:3. In the training set, LASSO regression, logistic regression, and random forest were used to identify the most significant variables for outcome events. These variables were then incorporated into a multivariate logistic regression model. The prediction model was visualized using a nomogram and validated using training and testing sets. The final nomogram consisted of three independent predictors: body weight, drug-coating balloon angioplasty, and post-procedural outflow score. The regression equation was: Y = 3.65–0.0645×weight − 1.04×(DCB = use) − 1.21×(post-procedural outflow score = 2) − 0.465×(post-procedural outflow score = 3). The prediction model demonstrated C-indices of 0.756 and 0.689 in the training and validation cohorts, respectively. Calibration curves showed satisfactory agreement in both cohorts. The prediction model incorporating body weight, drug-coating balloon angioplasty, and post-procedural outflow score may assist in predicting one-year clinically driven target lesion revascularization in patients with isolated atherosclerotic popliteal artery lesions, providing valuable information for individualized treatment strategies.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-17283-9.

Introduction

Peripheral arterial disease (PAD) is a chronic progressive vascular disease that severely threatens cardiovascular health. The primary clinical characteristics of PAD are progressive stenosis or occlusion of the lower limb arterial system, resulting in intermittent claudication, chronic ischemic pain, and lower limb functional impairment^1,2. The global prevalence of PAD is approximately 1.52%³, with an increasing trend year by year. Some patients may be asymptomatic in the early stage of the disease, but as the condition progresses, they will face serious complications, such as chronic lower limb ischemia, ulcer formation, and amputation risk. Although over 50% of patients are asymptomatic, 3%-4%⁴ may ultimately require amputation, and the economic and health burden on patients cannot be ignored.

Isolated atherosclerotic popliteal artery lesions (IAPAL) are relatively common in patients with peripheral arterial occlusive disease⁵. However, due to knee joint flexion⁶biomechanical stress⁷and relatively small arterial diameter, the popliteal artery has long been considered a challenging vessel for vascular reconstruction⁸. More problematically, popliteal artery occlusive lesions are often accompanied by severe vascular calcification⁹.

Currently, endovascular treatment has become a common therapeutic approach for lower limb arterial diseases, including popliteal artery interventions, primarily involving stents¹⁰drug-coating balloons¹¹and atherectomy devices¹². Due to the inherent risk of stent fracture¹³the popliteal artery has been considered a non-stent zone. To overcome stent-related issues, researchers have proposed drug-coating balloon angioplasty (DCB) and atherectomy devices as alternative solutions. These innovative technologies not only reduce stent-related complications but also better protect vascular endothelial function and vascular structural integrity.

Clinical Driven Target Lesion Revascularization (CDTLR) is a critical indicator for evaluating the long-term effectiveness of endovascular treatment. For patients with popliteal artery occlusive disease, predicting and managing CDTLR remains a clinical challenge. The current literature has relatively limited research on the long-term prognosis of popliteal artery endovascular treatment, highlighting the urgent necessity of constructing precise predictive models.

In the endovascular treatment of isolated atherosclerotic popliteal artery lesions, predicting and evaluating surgical-related variables is of critical importance. These variables constitute a complex system for clinical decision-making and prognosis assessment, and their significance is self-evident.

Nomograms, as an advanced statistical prediction tool, can provide individualized risk assessment by integrating clinical and imaging indicators^14,15. This study aims to construct an innovative predictive model to comprehensively analyze the critical factors influencing 1-year Clinical Driven Target Lesion Revascularization (CDTLR) in patients with popliteal artery atherosclerotic disease. This approach will not only provide clinicians with a precise risk stratification tool but also promote the development of personalized treatment strategies, ultimately improving patient prognosis and quality of life.

Materials and methods

Study design and participants

Figure 1 shows the flow chart of our study. This retrospective study was conducted using the electronic medical record system of a large medical center, collecting patients who underwent endovascular treatment at Xuanwu Hospital between August 2017 and August 2022. Inclusion criteria were: (1) newly developed atherosclerotic lesions in the popliteal artery; (2) patients with Rutherford grades 2–5; (3) digital subtraction angiography (DSA) showing stenosis exceeding 70% or complete occlusion. Exclusion criteria included: (1) ipsilateral superficial femoral artery lesions; (2) recent stroke (within 6 weeks) or myocardial infarction (within 6 months); (3) previous failed endovascular treatment or open surgery; (4) allergy to heparin or low molecular weight heparin; (5) severe coagulation disorders or contraindications to antiplatelet therapy; (6) poor overall health status (such as severe heart or kidney failure) rendering surgery intolerable; (7) absence of distal outflow through dorsalis pedis or plantar arteries post-surgery; (8) adjacent aneurysms or acute thrombosis. Finally, a total of 217 patients were included. All patients were randomly allocated to training and validation sets at a ratio of 7:3.

Fig. 1.

Flow chart illustrating patient selection, randomization, and allocation to training and internal test cohorts.

Study parameters

We collected baseline medical information, including age, sex, height, weight, body mass index (BMI), comorbidities (hypertension, diabetes, cardio-cerebrovascular disease, hyperlipidemia, kidney diseases), smoking status, and patient Rutherford category. Pre-procedural and post-procedural clinical data were also recorded, encompassing vascular calcification, vascular lesion length, cumulative P1, P2, and P3 involvement, endovascular recanalization strategies, and devices used (atherectomy devices, drug-coating balloon angioplasty, stent implantation), ankle-foot outflow tract grading, and pre- and post-procedural outflow scores.

The endpoint event was target lesion revascularization (TLR) occurring within 1 year. In this study, clinically driven TLR was defined as any reintervention for the target lesion prompted by recurrent symptoms (such as worsening claudication or ischemic pain) together with objective evidence of restenosis or occlusion, as demonstrated by imaging (duplex ultrasound, CTA, MRA, or angiography showing ≥ 50% diameter reduction or occlusion). A drop in ABPI was considered supportive but not mandatory. This definition is consistent with major clinical trials and international guidelines.

Specific characteristics are defined as follows:

1. (1) Vascular Calcification: Using the Peripheral Artery Calcification Scoring System (PACSS), combined with high-intensity fluoroscopy and digital subtraction angiography (DSA) imaging, the calcification degree in the target lesion was analyzed and graded from 0 to 4. Calcification is generally considered severe at a score of 2 or above.

2. (2) Pre- and post-procedural outflow score: The popliteal artery outflow tract includes the anterior tibial, posterior tibial, and peroneal arteries. We counted the number of patent vessels.

3. (3) Ankle-Below Artery Grading: Grade 0: Target artery enters the foot through the ankle with a complete foot arch; Grade 1: Target artery enters the foot through the ankle, but with an incomplete or severely diseased foot arch; Grade 2: No target artery enters the foot through the ankle. (Potentially adaptable to the Global Limb Anatomic Staging System)

4. (4) Restenosis: Restenosis was defined as a diameter reduction of 70% based on Doppler ultrasound criteria (significant peak systolic velocity (PSV) ratio of 2.0).

5. (5) Atherectomy Devices: Atherectomy procedures primarily refer to the use of the Turbo-Hawk plaque excision system (EV3, USA) or laser ablation (Spectranetics Corporation, USA).

Statistical analysis

Statistical analysis was performed using R software (version 4.4.1). For categorical variables, frequencies and percentages were used for description, with the Pearson chi-square test and Fisher’s exact test employed for inter-group comparisons. For continuous variables, an independent Student’s t-test or Mann-Whitney U test was applied based on the results of normality assessment. Non-normally distributed data were presented as median (interquartile range). In this study, three different statistical methods—LASSO regression, logistic regression, and random forest—were used for variable selection and to evaluate their impact on classification model performance. A predictive model for target vessel revascularization was established in the training cohort, and a nomogram was plotted. The nomogram’s total score was the sum of points assigned to each risk factor, with higher scores indicating greater risk. The performance of the nomogram was assessed using receiver operating characteristic (ROC) curves and calibration curves, with the area under the ROC curve (AUC) ranging from 0.5 (no discriminative ability) to 1 (perfect discrimination). Decision curve analysis (DCA) was employed to determine the net benefit threshold of the prediction. Results with p-values < 0.05 were considered statistically significant.

Results

Baseline characteristics

The basic characteristics of the study population are presented in Tables 1 and 2. A total of 217 patients met the inclusion criteria and were enrolled in this study, with 34 patients experiencing target lesion revascularization (TLR) within 12 months, and 183 patients without TLR, resulting in a TLR rate of 15.6%. The data were randomly divided into training and internal test cohorts at a 7:3 ratio, with 151 patients in the training cohort and 66 patients in the test cohort. In the training cohort, 23 patients experienced TLR, with a TLR rate of 15%; in the test cohort, 11 patients experienced TLR, with a TLR rate of 17%. The median ages were 70 and 68 years in the training and internal test cohorts, respectively, with no statistically significant difference (p = 0.21). There were no significant differences between the two groups in the prevalence of hypertension, diabetes, cardio-cerebrovascular disease, and kidney diseases (dialysis status) (all p > 0.05). Regarding gender composition, the proportions of males were 58% and 70% in the training and internal test cohorts, respectively (p = 0.093). In terms of smoking status, 28% were smokers and 72% were non-smokers in the training cohort, while 41% were smokers and 59% were non-smokers in the internal test cohort (p = 0.057). The median lesion length was 80 mm in the training cohort and 55 mm in the internal test cohort, with a statistically significant difference (p = 0.012). The utilization rates of arterial plaque excision were 69% and 52% in the training and internal test cohorts, respectively (p = 0.014). In summary, although the two groups were relatively comparable in age and gender distribution, there were differences in clinical characteristics such as lesion length and arterial plaque excision.

Table 1.

Preoperative baseline data of the training set and test set.

Characteristic	N	Internal test cohort, N = 661	Training cohort, N = 1511	P-value
TLR	217	11 (17%)	23 (15%)	0.792
Male	217	46 (70%)	87 (58%)	0.0932
Age (years)	217	68.0 (62.0, 76.0)	70.0 (64.0, 78.0)	0.213
Height (cm)	217	165.5 (162.0, 173.0)	167.0(160.0, 171.0)	0.533
Weight (kg)	217	69.0 (60.0, 75.)	70.0 (62.0, 75.0)	0.863
BMI (kg/m2)	217			0.354
18.5 ~ 25		41 (62%)	76 (50%)
25 ~ 30		21 (32%)	65 (43%)
>30		4 (6.1%)	8 (5.3%)
<18.5		0 (0%)	2 (1.3%)
Rutherford category	217			0.582
3		45 (68%)	95 (63%)
4		11 (17%)	24 (16%)
5		10 (15%)	32 (21%)
Diabetes	217	44 (67%)	105 (70%)	0.682
Hypertension	217	47 (71%)	113 (75%)	0.582
Smoke	217	27 (41%)	42 (28%)	0.0572
Cardio-cerebrovascular disease	217	22 (33%)	61 (40%)	0.322
Dialysis	217	2 (3.0%)	12 (7.9%)	0.244
Hyperlipidemia	217	29 (44%)	86 (57%)	0.0772

¹n (%), M (P25, P75), ²Pearson’s Chi-squared test, ³Wilcoxon rank sum test, ⁴Fisher’s exact test. TLR: Target lesion revascularization.

Table 2.

Surgical data of the training set and test set.

Characteristic	N	Internal test cohort, N = 661	Training cohort, N = 1511	P-value
Atherectomy	217	34 (52%)	104 (69%)	0.0142
Stent implantation	217	18 (27%)	58 (38%)	0.112
Drug-coating balloon angioplasty	217	44 (67%)	90 (60%)	0.322
Serve calcification	217	27 (41%)	55 (36%)	0.532
P1	217	43 (65%)	105 (70%)	0.522
P2	217	19 (29%)	56 (37%)	0.242
P3	217	26 (39%)	71 (47%)	0.302
Pre-procedural outflow score	217			0.392
1		18 (27%)	55 (36%)
2		34 (52%)	65 (43%)
3		14 (21%)	31 (21%)
Post-procedural outflow score	217			0.342
1		24 (36%)	67 (44%)
2		31 (47%)	55 (36%)
3		11 (17%)	29 (19%)
Lesion Length	217	55.0 (30.0, 90.0)	80.0 (40.0, 120.0)	0.0123
Occlusion Lesion	217	30 (45%)	85 (56%)	0.142
Ankle-Below Artery Grading	217			0.834
0		21 (32%)	50 (33%)
1		44 (67%)	100 (66%)
2		1 (1.5%)	1 (0.7%)

¹n (%), M (P25, P75), ²Pearson’s Chi-squared test, ³Wilcoxon rank sum test, ⁴Fisher’s exact test. Severe calcification: PACSS ≥ 2.

Predictive model

Candidate predictive factors included: age, sex, height, weight, BMI classification (< 18.5, 18.5–25.0, 25.0–30.0, > 30.0), Rutherford staging, hypertension, diabetes, cardio-cerebrovascular disease, kidney diseases (dialysis status), hyperlipidemia, smoking, pre- procedural outflow score, post- procedural outflow score, lesion length, lesion type (stenosis or occlusion), calcification degree, atherectomy devices, drug-coating balloons, stent implantation, and ankle-below artery grading. These factors were initially included in the original model and subsequently underwent variable selection through LASSO regression analysis, logistic regression, and random forest in the training cohort to assess their impact on classification model performance.

The cross-validation error plot of the LASSO regression model (Figure S1) shows that the most simplified and regularized model contains 5 variables at the lambda.min. Logistic regression selects important features by calculating the impact of each variable and assessing its contribution to the target variable. As an ensemble learning method, random forest predicts results using 500 decision trees and evaluates variable contributions by calculating feature importance in tree splitting. This method effectively handles non-linear relationships between features and provides an intuitive assessment of feature importance.

To evaluate model performance, we used 10-fold cross-validation to optimize model parameters and select the best model. By comparing the accuracy of LASSO regression, logistic regression, and random forest models on the test dataset, we determined their performance in the classification task. Figure 2 displays the visualization of important variables in each model, intuitively presenting the impact of different methods on feature selection. LASSO regression reduces redundant variables through regularization, logistic regression highlights the influence of core variables, and random forest reveals complex interactions between variables.

Fig. 2.

Comparison of variable importance rankings identified by LASSO regression, logistic regression, and random forest methods.

The study followed rigorous statistical principles for variable selection, screening predictive factors through multiple methods. Considering 34 positive events, we referred to both the conventional 10 events per variable (EPV) rule¹⁶ and recent recommendations for higher EPV thresholds to ensure model stability and reduce the risk of overfitting¹⁷. To further justify the number of predictors, we used the R package pmsampsize to estimate the minimum sample size and events per predictor parameter (EPP) required for model development. Based on our study parameters (anticipated Cox-Snell R² = 0.1058, C-statistic = 0.75, outcome prevalence = 0.156, and shrinkage factor = 0.9), the minimum sample size required for developing a new model with one predictor is 203, with at least 32 events and an EPP of 31.67. Our actual sample size (n = 217, 34 events) meets these requirements, indicating that our model is unlikely to be overfitted and has adequate statistical power for the number of predictors included. Therefore, we ultimately included 3 core variables in the final logistic model (Table 3): drug-coating balloons, body weight, and post-procedural outflow score.

Table 3.

Information on the results of multi-factor logistic regression analysis.

Variable	β	S.E.	OR1	95% CI1	P-value
intercept	3.65	1.85			0.0482
Weight	-0.0645	0.0268	0.94	0.89, 0.99	0.016
DCB
Not use	—	—	—	—	—
Use	-1.04	0.484	0.35	0.13, 0.90	0.031
Pre-procedural outflow score
1	—	—	—	—	—
2	-1.21	0.569	0.30	0.09, 0.88	0.033
3	-0.465	0.619	0.63	0.17, 2.03	0.5

Using binary logistic regression to analyze TLR occurrence, the results showed: body weight (OR: 0.94, 95% CI: 0.89–0.99) indicates a slight decrease in TLR risk with each 1 kg increase; drug-coating balloons (OR: 0.35, 95% CI: 0.13–0.90) and post- procedural outflow scores (2 points OR: 0.30, 3 points OR: 0.63) significantly reduced TLR probability.

Based on binary logistic regression analysis, we established a prediction equation for TLR occurrence within 12 months of interventional treatment for popliteal artery atherosclerotic occlusion: Y = 3.65–0.0645 body weight (body weight unit: kg) − 1.04 (drug-coating balloon = use) − 1.21 (Post- procedural outflow score = 2; reference: 1) − 0.465 (Post- procedural outflow score = 3; reference: 1) and developed a simple and user-friendly nomogram (Fig. 3).

Fig. 3.

Nomogram for predicting the probability of 1-year clinically driven target lesion revascularization (CDTLR) in patients with popliteal artery atherosclerotic lesions.

The nomogram is primarily intended for use after the procedure, as it incorporates post-procedural outflow score to help clinicians assess the risk of TLR and guide postoperative management and follow-up strategies. However, since the model also includes preoperative variables such as body weight and the use of drug-coating balloons, it can provide some reference for preoperative risk stratification and intervention planning. Thus, the nomogram may assist clinicians in both postoperative individualized management and in informing preoperative discussions and decision-making.

ROC curves

Receiver Operating Characteristic (ROC) curves were employed to assess the discriminative ability of the model in both training and test sets. The ROC analysis revealed an Area Under the Curve (AUC) of 0.756 for the training set and 0.689 for the test set, as illustrated in Fig. 4. These results indicate that our predictive model shows moderate discrimination and acceptable clinical prediction accuracy.

Fig. 4.

ROC curve graph for prediction with the nomogram model. (A) Training cohort, (B) internal test cohort.

Calibration curves

The Calibration Curve is a crucial visualization tool for assessing the accuracy of model prediction probabilities. Figure 5 illustrates the predicted calibration of target lesion revascularization (TLR) probability within 12 months following interventional treatment for popliteal artery atherosclerotic occlusion.

Fig. 5.

Calibration curve graph for prediction with the nomogram model. (A) Training cohort, (B) internal test cohort.

In the training set, the model’s predicted probabilities closely aligned with actual results, with a Brier score of 11.7 (8.2–15.2), indicating good model calibration. For the internal test set, the calibration curve showed some deviations between predicted and observed probabilities in certain regions, as reflected by a Brier score of 13.0 (6.9–19.1). These deviations suggest that while the model demonstrates reasonable calibration overall, there are areas where prediction accuracy could be improved. We acknowledge this limitation and recommend that future studies further refine and validate the model in larger and more diverse cohorts. The similar performance metrics between the training and internal test sets suggest the absence of significant overfitting.

Decision curve analysis

Figure S2 presents the Decision Curve Analysis (DCA) of the nomogram. The model curve intersects with the reference line near the 35% risk threshold, indicating that the model demonstrates good decision-making value across different clinical risk levels. This confirms that our developed predictive model possesses significant practical utility and potential clinical benefits in actual medical practice.

Discussion

This study developed and validated a nomogram based on multivariate logistic regression analysis to predict the risk of vascular revascularization within one year for patients with popliteal artery atherosclerotic lesions. The results demonstrated that drug-coating balloon usage, body weight, and post-procedural outflow score are independent predictive factors influencing prognosis. These findings not only hold significant clinical implications but also provide novel insights for developing personalized treatment strategies.

Our findings are consistent with recent large-scale registries of popliteal endovascular treatment. The EMO-POP registry¹⁸ (651 patients) reported a 26-month freedom-from-TLR rate of 76.5%, while our 12-month TLR-free rate was 84.4%, similar to the K-POP registry’s 87.2%¹⁹ at 12 months with DCB. Differences in follow-up duration and patient characteristics may explain the variation. Overall, our results align with contemporary real-world outcomes and underscore the value of individualized risk assessment in this population.

Drug-coating balloons (DCB) were confirmed as a significant predictive factor in this study, consistent with previous research findings^11,18,20. Prior studies^18,21,22 have demonstrated that using debulking devices or DCB can reduce the one-year target lesion revascularization (TLR) rate in popliteal artery patients. In the era of rapidly evolving endovascular interventions, despite persistent high restenosis rates, DCB has emerged as a novel treatment strategy with unique advantages. DCB operates by releasing drugs (primarily paclitaxel²³ through microscopic pores on the balloon surface. During percutaneous transluminal angioplasty, the balloon’s folded state before expansion prevents premature drug loss, while expansion facilitates drug penetration into the arterial wall. Although some drugs are carried away by blood flow, most drugs remain in the local arterial wall, effectively inhibiting excessive endothelial cell proliferation²⁴ and reducing restenosis risk.

Atherectomy procedures^25,26 involve specialized devices for plaque and thrombus removal, increasing the effective intravascular diameter and reducing risks of elastic recoil and dissection, thus preventing intimal hyperplasia. Notably, while atherectomy is an important therapeutic approach^11,12,18it did not demonstrate significant predictive value in our model’s variable selection process. This suggests that atherectomy’s impact on one-year CDTLR might be less significant compared to other factors, aligning with previous studies such as Wardle et al.‘s systematic review, which found uncertain evidence regarding atherectomy’s impact on patency rates.

Our model, through rigorous statistical methods, was the first to incorporate DCB as a crucial predictive variable in the nomogram. This finding not only reflects current trends in vascular interventional treatment but also provides critical guidance for clinicians in developing personalized treatment strategies.

Weight was incorporated into the model as a continuous variable, offering advantages over previous research approaches. Traditional studies often categorized weight using specific cutoff points (such as overweight or obesity), potentially leading to information loss and reduced statistical power. While obesity is widely recognized as a cardiovascular disease risk factor^27,28in patients with existing cardiovascular disease, obesity may paradoxically serve as a protective factor for cardiovascular target lesion revascularization^29–31. According to Gruberg et al.³²patients with normal weight might face higher in-hospital complications and mortality risks during percutaneous coronary intervention (PCI). The relationship between weight and prognosis may operate through multiple mechanisms: First, weight correlates with vessel diameter, with lower-weight patients typically having relatively smaller vessel diameters³³. This not only increases surgical technical difficulty³⁴ but may also elevate restenosis risk^35–39 and potentially limit the size of usable devices. Second, extremely thin patients might respond poorly to surgery and subsequent revascularization due to insufficient bodily resources and reserves, increasing the likelihood of requiring repeated interventional treatment. Furthermore, abnormal weight is often associated with systemic inflammatory states^40–42potentially influencing post-vascular reconstruction prognosis. Our model preserved the continuous characteristics of weight, thereby enhancing predictive accuracy and enabling more precise quantification of weight variation’s impact on prognosis.

The observed association between higher body weight and reduced TLR risk in our model is counterintuitive and should be interpreted with caution. Although the “obesity paradox” has been reported in some cardiovascular contexts, the evidence in PAD remains limited and controversial. It is possible that body weight in our cohort acts as a surrogate for vessel diameter, nutritional status, or other comorbidities, rather than reflecting a true protective effect. Notably, BMI categories were not predictive in our analysis, which may indicate multicollinearity or residual confounding among related variables. Further research is needed to clarify the complex relationship between body weight, BMI, vessel size, and clinical outcomes in PAD populations.

The inclusion of post-operative outflow tract quantity underscores the significance of vascular anatomical factors in prognostic prediction. Adequate distal outflow⁴³ not only influences immediate surgical success but is also closely associated with long-term patency rates^39,44. Our study found that the lowest vascular reconstruction rate occurred when the post-operative outflow tract quantity was 2, while the highest reconstruction rate was observed when the outflow tract quantity was 1. This finding differs from some previous research results^5,45,46. Saratzis et al.¹⁸ suggested that target lesion revascularization (TLR) outcomes in popliteal artery occlusion are associated with poor pre-operative outflow. Xu et al.³⁷ and other studies^47,48 demonstrated that target vessel diameter and hemodynamic characteristics significantly impact revascularization outcomes, particularly after reconstructive surgery⁴⁹. Insufficient post-operative outflow may lead to tissue ischemia and potential complications, thereby increasing the need for reintervention. Interestingly, our model found that patients with two patent tibial vessels (score = 2) had a lower TLR risk than those with three (score = 3), a non-monotonic and biologically unexpected result. This may be due to small sample size, residual confounding, or clinical practice: patients with only one patent vessel are more likely to receive further intervention to open another outflow tract, while those with two or three are not. The post-procedural outflow score was defined as the number of patent tibial arteries (anterior tibial, posterior tibial, peroneal) with uninterrupted flow to the foot on final DSA, independently assessed by two radiologists. This limitation should be interpreted with caution.

In summary, low body weight may indirectly affect revascularization outcomes in patients with popliteal artery atherosclerotic occlusion by increasing the risk of post-operative complications, while post-procedural outflow directly influences blood flow and treatment effectiveness. Both factors play critical roles in predicting target lesion revascularization and should be given careful attention in clinical practice. These findings provide important guidance for pre-operative assessment and surgical planning, while also emphasizing the need to optimize outflow tract quantity during interventional treatments. To enhance clinical usability, we provide the raw regression formula for the nomogram in the Results section. Clinicians can manually input patient data into this formula to estimate individual risk.

The multivariate logistic regression method used to construct the predictive model offers several advantages: First, it can simultaneously assess the independent impacts of multiple predictive factors, effectively controlling potential confounding variables; second, the model can handle different types of predictive variables (continuous and categorical), enhancing predictive flexibility; third, visualization through a nomogram makes complex statistical predictions intuitive and user-friendly.

However, the study has some limitations: First, it is a single-center retrospective study with a relatively limited sample size, which may affect the model’s stability. Most importantly, the lack of external validation significantly limits the generalizability and clinical applicability of our model. Although the model performed well in internal validation, its performance in independent, external populations remains unknown. We strongly recommend that future studies conduct multicenter, prospective external validation to rigorously assess and further refine the model. Third, some potential predictive factors (such as hemodynamic parameters) could not be included due to data acquisition constraints, potentially compromising the comprehensiveness of the prediction.

Future research directions include: (1) external validation in multi-center, large-sample cohorts; (2) incorporating more potential predictive factors to further optimize model performance; (3) exploring the possibility of integrating imaging features and molecular biomarkers into the predictive model; (4) developing a mobile application based on this model to promote its clinical application.

Conclusion

The nomogram developed in this study provides a practical tool for prognostic prediction in patients with popliteal artery atherosclerotic lesions. The model integrates treatment methods, patient characteristics, and anatomical factors, helping clinicians perform risk stratification and individualized treatment decisions. Nevertheless, further validation and optimization of the model are needed to enhance, which will help enhance its clinical application value.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Z.D. performed formal analysis and wrote the original draft. L.G. contributed to conceptualization and data curation. Z.T. was responsible for methodology and project administration. S.C. performed formal analysis and methodology. X.G. contributed to project administration and software. C.Z. provided supervision and validation. J.G. contributed to supervision, formal analysis, methodology, project administration, and writing the original draft. Y.G. contributed to conceptualization, supervision, and writing – review & editing. All authors reviewed the manuscript.

Funding

The author(s) declare that no financial support was received for the research, authorship, and/or publication of this article.

Data availability

The raw data supporting the conclusions of this article will be made available upon reasonable request by contacting the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.

Ethics statement

The studies involving human participants were approved by Xuanwu Hospital of Capital Medical University. They were conducted in compliance with local legislation and institutional guidelines. Informed consent was obtained in writing from all participants prior to their involvement in this study.