**3. Discussion**

This study was designed to evaluate the use of keratose as a novel excipient for peripheral applications and, specifically, to determine the feasibility of the keratose excipient to retain paclitaxel under peripheral vascular mechanical environments. This was accomplished by developing a novel vascular-simulating *ex vivo* flow system and testing in a clinically relevant pre-clinical model. Furthermore, the vascular biological response to the keratose excipient was also investigated in the pre-clinical model. The *ex vivo* model arterial drug concentration results demonstrated that keratose significantly improves the retention of paclitaxel as compared to a commercially available DCB. Histomorphometric results of rabbit arteries treated by keratose demonstrated the safety and efficacy of the excipient in the delivery of paclitaxel. Overall, these results demonstrate the potential of the keratose as a DCB excipient for peripheral applications.

Drug-coated balloons are the next-generation treatment for PAD. Approved in the US since late 2014, DCB represented a shift in the approach to treating peripheral artery disease. While the DES provides a scaffold for long-term drug release, DCBs are limited in the time they can interact with the target lesion (~30 s to 2 min). Therefore, a major goal of any excipient is to support the retention of the therapeutic agen<sup>t</sup> to the arterial wall surface, even under vascular deformation conditions. In two recent studies, the embolization of release particulates from all currently FDA-approved DCB coatings was investigated [27,28]. Twenty-eight days post-delivery, their results also demonstrated evidence of distal embolization, including embolic crystalline material, in downstream tissue. Remarkably, pharmacokinetic analysis of the distal tissue showed similar or higher levels of paclitaxel concentration as compared to the arterial treatment site, in particular for the IN.PACT DCB. These results indicate the mobility of the DCB coating following deployment, although, to date, no studies have directly investigated the impact of vascular deformation on DCB performance.

The vascular-mimicking *ex vivo* system, to our knowledge, is the first system that can evaluate the acute drug-loading of arteries treated by endovascular devices under pulsatile and vascular deformation conditions using explanted pig arteries. Our testing of the KOS coating was performed under vascular deformation conditions of 10% artery shortening, 15◦/cm twisting at a frequency of 0.05 Hz (3 cycles/min). These conditions were selected to replicate the human periphery motion of the femoral artery (shortening lengths of 7% and twisting at 11.5◦/cm) and the popliteal–tibial artery motion (shortening of 15% and twisting at 19.9◦/cm) [32–35]. The frequency of the peripheral movement will be 0.06 Hz (5184 cycles/day or 3.6 cycles/min), which is based upon the average steps per day of adults in the US [36]. Our results indicated that the KOS coating maintained paclitaxel tissue levels under physiological pulsatile and vascular motion conditions 3 days post-delivery.

To further evaluate the DCB coating, we fluorescently tagged (NHS-Fluorescein, Thermo Scientific) the KOS to visualize the presence of the coating acutely (1 h) and three days post-delivery in arteries undergoing vascular deformation. The presence of the KOS was confirmed by confocal microscopy (Figure 4). The mechanism by which this process occurs is not fully elucidated in these studies. In drug-release experiments with small molecule drugs such as ciprofloxacin from hydrogel (rather than coating) forms of KOS, we have demonstrated that the rate of drug release correlates with the degradation rate of the hydrogel material [30]. We note that this degradation process does not refer to the breaking of peptide (amide) bonds in the keratin, but rather the dissolution of the keratin hydrogels. This correlation between drug release and KOS dissolution (or degradation) suggested an interaction between keratin and the drug. In the case of ciprofloxacin, this was found to be associated

with electrostatic interactions. While the physiochemical characteristics of paclitaxel are different than ciprofloxacin, such interactions (or others, such as hydrophobic interactions) could be in play and are an area for further study.

**Figure 4.** Representative confocal images arterial segments following keratose delivery. Confocal microscopy confirmed the presence of the keratose at (**A**) 1 h and (**B**) 3 days under peripheral deformation conditions. (**C**) Negative control depicts the lack of tissue autofluorescence during confocal imaging.

This previous finding of an interaction between KOS and small molecule drugs is noteworthy due to the findings of paclitaxel retention in the vessel at 3 days with vascular motion compared to DCB (Figure 2) and the presence of (fluorescently labeled) KOS on the vascular walls (Figure 4). That is, it is possible that paclitaxel remains associated with the KOS in a manner not possible with other synthetic polymers (e.g., PEG) or other (e.g., urea) excipients due to the properties of keratin. In particular, KOS has been shown to contain RGD and other integrin-binding sequences which may allow it to bind to the vascular cells [37,38]. Thus, KOS may have a unique ability to associate with the lumen through integrin-binding with vascular cells while simultaneously retaining the paclitaxel through electrostatic or hydrophobic interactions.

While it is well-recognized that arterial repair after balloon injury occurs more rapidly in animals than in humans, animal models still hold a predictive value for the observation of biological effects that may be associated with drug delivery [39]. In this study, histopathologic evaluation of the KOS–paclitaxel DCB, along with uncoated balloons, KOS-only balloons and paclitaxel-only DCBs were performed in a rabbit ilio–femoral injury model, which has been shown to be an appropriate model for the evaluation of endovascular devices [40–43]. Overall, the morphometric results demonstrated minimal neointimal growth, as percent area stenoses were less than eleven percent for all groups at the 7-day time point. These results were expected as, in general, peripheral rabbit arteries appear to be resistant to the development of aggressive neointimal growth with mild balloon to artery ratio (1.1–1.2:1), especially with plain balloon angioplasty [39,43]. Furthermore, as expected, injury scores were mild, ranging from 0.50 to 1.13 in all groups. However, by histologic evaluation, the safety and effectiveness of the KOS–paclitaxel DCB was still evident, based on vascular remodeling and healing. Specifically, neointimal thickness was significantly reduced in the KOS–paclitaxel DCB treatment group (no coating: 0.10 ± 0.011 mm vs. KOS-only: 0.069 ± 0.022 mm vs. PXL-only: 0.066 ± 0.018 mm vs. KOS-PXL: 0.53 ± 0.003 mm, *p* = 0.005). Importantly, the endothelization score was significantly reduced in the KOS–paclitaxel treated arteries, indicative of drug retention (Table 1). Additionally, there was a trend towards a lower neointimal area and higher loss of smooth muscle cells (SMCs) in the KOS–paclitaxel DCB group as compared to all others, indicative of drug effect (no coating: 0.00 ± 0.00 vs. KOS-only: 0.00 ± 0.00 vs. PXL-only: 0.25 ± 0.0.50 vs. KOS-PXL: 1.00 ± 0.82, *p* = 0.081). Overall, the *in vivo* data demonstrate the safety of the keratose coating and a reduction in neointimal growth by the keratose–paclitaxel DCB.

While our results support the concept of a keratose coating to deliver anti-proliferative drugs to arterial segments, the study was limited to a healthy animal model and thus did not take into consideration diseased arteries, as observed in patients with PAD. For the *ex vivo* studies, further characteristic testing of paclitaxel delivery via the drug-coated balloon is warranted to quantify the amount of drug remaining on the balloon following delivery and to quantify circulating paclitaxel levels. We also recognize that human lesions are more complex and often include fibrosis, calcification, hemorrhage and, in most cases, require de-bulking using balloons and atherectomy devices, which may alter drug transfer and retention. While preclinical studies involving healthy arteries are the standard model to determine the arterial time drug concentration of cardiac and stent-based intervention devices, further improvement may be found with a KOS-paclitaxel coating in injury models.

### **4. Materials and Methods**
