BioMimics 3D Stent in Femoropopliteal Lesions: 3-Year Outcomes with Propensity Matching for Drug-Coated Balloons

Background: Through its helical centreline geometry, the BioMimics 3D vascular stent system is designed for the mobile femoropopliteal region, aiming to improve long-term patency and the risk of stent fractures. Methods: MIMICS 3D is a prospective, European, multi-centre, observational registry to evaluate the BioMimics 3D stent in a real-world population through 3 years. A propensity-matched comparison was performed to investigate the effect of the additional use of drug-coated balloons (DCB). Results: The MIMICS 3D registry enrolled 507 patients (518 lesion, length 125.9 ± 91.0 mm). At 3 years, the overall survival was 85.2%, freedom from major amputation 98.5%, freedom from clinically driven target lesion revascularisation 78.0%, and primary patency 70.2%. The propensity-matched cohort included 195 patients in each cohort. At 3-year follow-up, there was no statistically significant difference in clinical outcomes, such as overall survival (87.9% in the DCB vs. 85.1% in the no DCB group), freedom from major amputation (99.4% vs. 97.2%), clinically driven TLR (76.4% vs. 80.3%), and primary patency (68.5% vs. 74.4%). Conclusion: The MIMICS 3D registry showed good 3-year outcomes of the BioMimics 3D stent in femoropopliteal lesions, demonstrating the safety and performance of this device under real-world conditions, whether used alone or in combination with a DCB.


Introduction
Successful procedural outcomes and long-term durability are the main goals in the interventional treatment of peripheral artery disease (PAD), especially in the femoropopliteal lesions. Endovascular treatment options have substantially improved over the last decades, but there is still room for enhancements, such as improvement in long-term patency and durability of stents in this highly mobile region [1].
The BioMimics 3D vascular stent system (Veryan Medical Ltd., Horsham, UK) features a new-generation self-expanding nitinol stent, with a 3D helical centreline. It is designed to resist forces such as compression, bending and torsion of the artery, and aims to improve patency rates compared to other bare metal stents by mimicking natural structures [2,3]. In the MIMICS randomised controlled trial, the BioMimics 3D stent was superior to straight  (2) and transition sections (1). The last 3 crowns at the stent ends define the transition sections, where the length of the crown is increased, slightly reducing the radial force imposed on the vessel, thus mitigating a step-change in the profile at the stent ends. Image provided by Veryan Medical.

Endpoints and Definitions
Endpoints at 36 months are death, target limb amputation, clinically driven target lesion revascularisation (CD-TLR), primary patency (site-assessed, defined as peak systolic velocity ratio <2.4 by duplex ultrasound), and functional and clinical outcomes, including ankle-brachial index (ABI) and Rutherford class.

Statistical Analysis
The analysis is based on the intention-to-treat population. Continuous variables are expressed as the mean ± standard deviations, categorical variables are expressed as fre- Figure 1. Unconstrained BioMimics 3D stent. The BioMimics 3D stent has a helical centreline (2) and transition sections (1). The last 3 crowns at the stent ends define the transition sections, where the length of the crown is increased, slightly reducing the radial force imposed on the vessel, thus mitigating a step-change in the profile at the stent ends. Image provided by Veryan Medical.

Endpoints and Definitions
Endpoints at 36 months are death, target limb amputation, clinically driven target lesion revascularisation (CD-TLR), primary patency (site-assessed, defined as peak systolic velocity ratio <2.4 by duplex ultrasound), and functional and clinical outcomes, including ankle-brachial index (ABI) and Rutherford class.

Statistical Analysis
The analysis is based on the intention-to-treat population. Continuous variables are expressed as the mean ± standard deviations, categorical variables are expressed as frequencies. All calculations are based on the data available. The p-values were calculated using a Fisher's exact test for categorical variables, Student's t-test (means) for continuous variables and log-rank test for comparing Kaplan-Meier estimates. A p-value < 0.05 was considered significant.
A subgroup analysis of lesions treated with DCB was pre-specified in the clinical investigation plan. In addition, a propensity score model was used, that was built using a logistic regression to model the propensity to use a DCB as a function of the covariates, as detailed in Supplemental Table S1 (n = 15 variables). Missing data in the covariates were imputed using a single imputation via Monte Carlo Markov Chain methodology. The model fit was assessed via the c-statistic and the propensity scores before and after matching were plotted to examine the overlap between the groups. A 1:1 matched cohort was constructed using the GREEDY match algorithm. The analysis was performed using SAS version 9.4 (SAS institute, Cary, NC, USA).

Results
Baseline data have been described previously [2]. In brief, patients were 70.1 ± 10.0 years old on average, 65.5% were male, 68.0% smokers, and 36.9% had diabetes (Table 1). Lesions had a reference vessel diameter of 5.5 ± 0.7 mm and were 125.9 ± 91.0 mm long. The total stented length was 131.2 ± 80.1 mm (Tables 2 and 3).    There was no systematic x-ray assessment to assess stent fractures; nevertheless, fou stent fractures were reported (one of which was not confirmed by the clinical events com mittee).  There was no systematic X-ray assessment to assess stent fractures; nevertheless, four stent fractures were reported (one of which was not confirmed by the clinical events committee).

Propensity-Matched Cohorts
The propensity-matched cohorts consisted of 195 patients each. Both cohorts were well matched, as shown in Supplemental Figure S1, and Tables 1 and 2, except for a lower calcification grade rate (grade 4) in the DCB group (7.9% vs. 19.3%, p = 0.001).
The total stent length was 128.3 ± 80.9 mm in the DCB group vs. 132.2 ± 79.2 mm in the no-DCB group, p = 0.262 for lesions lengths of 123.6 ± 93.7 mm and 124.7 ± 86.6 mm, p = 0.402, respectively (Tables 2 and 3). In the unmatched cohort, the total stented length was 135.0 ± 85.5 mm vs. 127.7 ± 74.6 mm, p = 0.872, and the lesion length was 137.6 ± 99.8 mm vs. 115.0 ± 80.7 mm, p = 0.068, respectively. Spot-stenting, which we defined as stented length <80% of lesion length, was performed in 15.9% of lesions in the propensity-matched DCB cohort and 11.3% in the no-DCB cohort, p = 0.237, and for 15.3% vs. 11.1%, p = 0.239, in the unmatched cohort. At 3 years, 153 patients in the DCB and 145 patients in the no-DCB group completed their follow-up visit, and 22 and 26 patients had died, respectively. There was no statistically significant difference in clinical outcomes. Three-year survival (day 1095) was 87.9% and 85.1% in the DCB and no-DCB propensity-matched cohorts, freedom from major amputation was 99.4% and 97.2%, CD-TLR was 76.4%, and 80.3% and primary patency 68.5% and 74.4%, respectively (Table 4, Figure 3).

Propensity-Matched Cohorts
The propensity-matched cohorts consisted of 195 patients each. Both cohorts were well matched, as shown in Supplemental Figure S1, and Tables 1 and 2, except for a lower calcification grade rate (grade 4) in the DCB group (7.9% versus 19.3%, p = 0.001).
The total stent length was 128.3 ± 80.9 mm in the DCB group versus 132.2 ± 79.2 mm in the no-DCB group, p = 0.262 for lesions lengths of 123.6 ± 93.7 mm and 124.7 ± 86.6 mm p = 0.402, respectively (Tables 2 and 3). In the unmatched cohort, the total stented length was 135.0 ± 85.5 mm versus 127.7 ± 74.6 mm, p = 0.872, and the lesion length was 137.6 ± 99.8 mm versus 115.0 ± 80.7 mm, p = 0.068, respectively. Spot-stenting, which we defined as stented length <80% of lesion length, was performed in 15.9% of lesions in the propensity-matched DCB cohort and 11.3% in the no-DCB cohort, p = 0.237, and for 15.3% versus 11.1%, p = 0.239, in the unmatched cohort. At 3 years, 153 patients in the DCB and 145 patients in the no-DCB group completed their follow-up visit, and 22 and 26 patients had died, respectively. There was no statistically significant difference in clinical outcomes. Three-year survival (day 1095) was 87.9% and 85.1% in the DCB and no-DCB propensity-matched cohorts, freedom from major amputation was 99.4% and 97.2%, CD-TLR was 76.4%, and 80.3% and primary patency 68.5% and 74.4%, respectively (Table 4, Figure 3). As detailed in Figure 2, the mean baseline Rutherford class was identical amongst the groups (3.2 ± 0.8 in the DCB group and 3.2 ± 1.2 in the no-DCB group), whereas at follow-up, the mean Rutherford class was significantly higher in the DCB group (1.3 ± 1.1 versus 0.8 ± 1.0, p = 0.0001, at 3 years). As detailed in Figure 2, the mean baseline Rutherford class was identical amongst the groups (3.2 ± 0.8 in the DCB group and 3.2 ± 1.2 in the no-DCB group), whereas at follow-up, the mean Rutherford class was significantly higher in the DCB group (1.3 ± 1.1 vs. 0.8 ± 1.0, p = 0.0001, at 3 years).
A summary of the unmatched DCB and no-DCB cohort data is provided in Supplemental Tables S2-S5, and a comparison of outcomes in patients that received a DCB before vs. after stent implantation is provided in Supplemental Table S6.

Discussion
The real-world MIMICS 3D registry showed that treating femoropopliteal lesions using the BioMimics 3D stent results in good 3-year outcomes, whether the stent is used in combination with a DCB or not.

3-Year Outcomes in the Overall Cohort
BioMimics 3D is a next-generation stent with a 3D helical centreline that imparts a gentle helical shape to the artery, which allows the stent to shorten, elongate and twist with the artery. Furthermore, the helical geometry induces swirling flow that increases wall shear stress, which has been shown to reduce intimal hyperplasia [5,6]. Moreover, the 3D helical geometry improves the stent's biomechanical compatibility which reduces vascular injury and the risk of stent fracture, both of which can be the consequence of using traditional straight stents in the mobile femoropopliteal arteries [7]. The stent demonstrated statistically significant better primary patency at 2 years compared to straight stents, and numerically higher freedom from TLR in a randomised controlled trial, where the curves diverged beyond 12 months in favour of the BioMimics 3D stent [4].
Likewise, the MIMICS 3D registry showed good outcomes in terms of TLR. Putting the data into perspective, whilst acknowledging that at this stage there are no direct RCTs comparing BioMimics 3D to drug-based devices, the MIMICS 3D TLR rate is similar to that seen in DES and DCB trials, whether in the overall, DCB or no-DCB cohort. In the overall cohort, with a mean lesion length of 125.9 ± 91.0 mm, freedom from CD-TLR at 3 years was slightly lower (78.0%) compared to 87.2% for the Zilver PTX DES (Cook Medical, Bloomington, Indiana) in the Zilver PTX randomized controlled trial, and 85.3% in the MAJESTIC trial using the Eluvia DES (Boston Scientific, Marlborough, MA), but in those trials, lesions were only approximately half as long as those in MIMICS 3D (lesion lengths of 66.4 mm and 70.8 mm, respectively) [8,9]. The Real-PTX randomised controlled trial had slightly longer lesions with 155.5 mm, and a lower freedom from TLR (68.9% for the Zilver PTX stent and 71.3% for DCBs) [10]. The In.Pact Global DCB registry had a similar lesion length (120.9 mm, provisional stenting in 21.2%), and reported 76.9% freedom from TLR at 3 years [11].

BioMimics 3D in Combination with DCB
Nearly half of the lesions in MIMICS 3D were additionally treated with a DCB. This, along with the very good outcomes, warrants further investigations of the effect of additional DCB therapy.
There are several reasons to consider the combination of bare metal stents and DCB therapy. First, stenting is used as bailout therapy after DCB therapy. This is required in around one quarter of procedures increasing with increased lesion length or when occlusions are treated [10][11][12][13][14][15]. In the IN.PACT Global study, provisional stenting was performed in 36.4% of cases in lesions ≥ 150 mm [13].
Second, the combination of stent plus DCB may be used in the context of spot-stenting. Spot-stenting has the advantage of limiting axial stiffness of the artery, possibly preventing kinking and occlusion, as the vessel (in contrast to DES treatment) is only stented where needed, avoiding "full-metal jackets" [16,17]. With the limitation that we arbitrarily determined spot-stenting as the total stented length ≤ 80% of lesion length, 15% of DCB lesions were categorised as having been treated with spot-stenting.
Third, the combination of a stent with a DCB is attractive as it can address recoil post-DCB and reduce restenosis through the addition of a drug, without the irritation of the drug-releasing excipients. In our series, more than half of the lesions were treated with primary stent placement followed by DCB treatment, predominantly to treat complex lesions, combining the advantages of a stent and DCB. This may be particularly useful in long lesions [18].
If DCB therapy is combined with a stent, it makes sense to use a stent with good long-term outcomes that could extend the period of high patency beyond the initial 12 months when drug-eluting devices might lose their effectiveness [13]. BioMimics 3D might be particularly useful in this scenario as it has prolonged clinical benefits due to its helical centreline, mimicking natural arterial geometry and inducing swirling flow as discussed above. This theoretical advantage has been tested in preclinical bench tests, but also in vivo, when BioMimics 3D demonstrated superior patency rates compared to straight nitinol stents [4]. Moreover, the transition zones which have reduced radial force at the stent edges might reduce edge reactions in the highly mobile and dynamic femoropopliteal region [19]. The purpose of the reduction in radial force is to mitigate a step change in profile at the stent ends, which could otherwise lead to a flow disturbance.
Notably, in our series, CD-TLR and patency rates were good in all subgroups, without a significant difference between DCB and non-DCB-treated lesions, whether in the unmatched or propensity-matched analysis. As for the overall cohort, outcomes were comparable to contemporary DES and DCB therapies. A randomised controlled trial would be desired to allow a direct comparison between the different treatment modalities.

Limitations
Potential limitations include those inherent to observational real-world studies. MIM-ICS 3D is not randomised, hampering the comparison to other devices and leaving room for bias. Inherent to a registry, patients were treated according to the standard of care. This means that the use of a DCB was left to the discretion of the operator. Furthermore, no routine imaging follow-up was performed, which might have resulted in an underreporting of stent fractures or restenosis. No imaging core laboratory was used, therefore the description of lesion length and patency was left to the study centre. Furthermore, the cut-off value for spot-stenting was arbitrarily selected. Lastly, while the propensity matching aimed to reduce the impact of baseline characteristics, results might still be biased because propensity matching is not as effective as randomisation.

Conclusions
The MIMICS 3D registry showed satisfying outcomes in patients treated with the helical centreline BioMimics 3D nitinol stent, which are comparable to those of drug-eluting stents. There was no difference in terms of CD-TLR, whether the stent was used alone or in combination with a DCB. This means that suboptimal results after DCB treatment should not be accepted as the combination of DCB plus BioMimics 3D results in good outcomes, and that DCB, in addition to BioMimics 3D stent treatment, may be an alternative to DES therapy when drug treatment is deemed necessary. Ultimately, a randomised controlled trial is needed to allow for a comparison between the different treatment modalities.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/jcdd10030126/s1, Table S1: Logistic regression results for a drugcoated balloon vs. no drug-coated balloon (c-statistic = 0.68); Figure S1: Propensity score distribution in the full dataset prior to matching (left panel) and in the matched cohort (right panel); Table S2: Baseline characteristics (unmatched cohorts);  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data associated with the paper are not publicly available, but are available from the corresponding author on reasonable request.