1 Introduction

Osteoarthritis (OA) is a common, chronic, and debilitating condition affecting the synovial joint [1]. Substantial limitations with existing treatments have driven intensive research intended to identify new therapies using a ‘treat-to-target’ approach [2,3,4]. One emerging treat-to-target option is known as geniculate artery embolization (GAE) [5,6,7]. GAE was developed for patients who are resistant to conservative therapies and not yet indicated for a knee replacement [6]. GAE is a transcatheter based clinical intervention (ca. 80 min) that targets abnormal angiogenesis [6] which facilitates the growth of new unmyelinated sensory nerves inducing pain signalling and sensitization [4, 8]. Based on this mechanistic understanding, and the treat-to-target nature of GAE, it has been clinically demonstrated to provide substantial pain relief for patients [5, 7].

GAE requires embolic microspheres to generate vessel occlusion. However, at this point in time, there exists no microspheres designed to meet the demands of this clinical intervention. The ideal embolic microspheres for GAE should possess several features including, in the first instance, the ability to degrade in situ. It has been determined that degradable microspheres (i) ameliorate clinical risks by ensuring no long-term presence in the body, (ii) facilitate vessel recanalization post treatment(s), and (iii) allow flexibility for repeat embolization procedures if necessary [9,10,11]. Furthermore, patients have an overwhelming preference for technologies which are eliminated from the body on completion of such treatments. Coupled with degradation, it is considered beneficial to have microspheres that exhibit certain imageability characteristics. For example, it may be beneficial to have microspheres with intrinsic x-ray attenuation characteristics (for intraprocedural imaging), while also exhibiting appropriate MRI relaxation characteristics (so as not to confound follow-up imaging on MRI) [12]. The expanded selection criteria for an ideal GAE microsphere can be described as (i) ensuring no residual mass remains in situ after ca. 72 h, with (ii) a particle size distribution (PSD) of 100–300 μm, (iii) a density of < 2.4 g/cm3, and (iv) preferably a suspension time of ≥ 2 min. These design inputs were gathered through one-on-one interviews with users and leading experts in the field via two mechanisms; firstly, through independent third-party interviews conducted by TL Health (PA, USA) and via direct interaction of the researchers with users at relevant international conferences (including the European Conference on Interventional Oncology, Global Embolization Symposium and Technologies, and Cardiovascular and Interventional Radiological Society of Europe).

The purpose of this work is to characterize the dissolution behaviour and imageability of 16 borate glasses, and to identify a preferred composition which best addresses the ideal design features of an embolic microsphere for GAE. Additional data relating to the systematic evaluation of the physical and chemical characteristics of this glass system are available elsewhere in the literature [13].

2 Materials and Methods

Sixteen glass compositions (including five replicates for statistical purposes) were established (Table 1) using Design-Expert Software (Version 12.0.9) based on an I-optimal approach [13]. Each composition was synthesized using a melt quench technique and subjected to baseline characterization (density, x-ray diffraction (XRD), 11B MAS NMR, and differential scanning calorimetry (DSC)) as described elsewhere [13]. Glasses were synthesized with analytical grade boric anhydride (99.98% trace metals basis and CAS number 1303–86-2), strontium carbonate (≥ 99.9% trace metals basis and CAS number 1633–05-2), and potassium carbonate (≥ 99% trace metals basis and CAS number 584–08-7) (Sigma-Aldrich, USA). Each composition was melted in a 100 mL Pt crucible (XRF Scientific, Montreal) and were subsequently placed at room temperature and subjected to a ramp rate of 25 ℃/min to 600 ℃ for 60 min. Thereafter, the ramp rate was increased to 20 ℃/min until 1200 ℃ was achieved as the final dwell temperature and held for 60 min [13]. Samples were then stored under desiccated conditions prior to experimentation.

Table 1 BKSA compositions (with five replicate compositions 1 and 13, 4 and 8, 5 and 7, 6 and 12, and 9 and 10) by mol% as characterized in Manchester et al. [13]
Full size table

Dissolution experiments for each glass were conducted in two stages. Firstly, glasses were subjected to screening, using sterile saline as the dissolution media, to identify chemistries with desirable mass loss characteristics. Individual Corning® 15 ml centrifuge tubes were weighed in triplicate with 10 mL saline (0.9% Sodium chloride, Baxter, Illinois, USA). 0.1 g of each glass composition (100–300 μm) was placed in falcon tubes (n = 3 per composition for each media) and then in a shaking incubator (120 RPM and 37 °C for 2 h) then immediately centrifuged (3.0 RCF/ 4.4 RPM) for 15 min. The supernatant was decanted from the pellet and the pellet (in the original test tube) was dried (50 °C for 96 h) then weighed to determine the average residual mass. The preferred glass (BKSA16) was subjected to residual mass experiments in physiologically representative media using a mass: volume ratio of 0.1 g (glass) to 10 mL of 10% bovine calf serum (BCS)/90% Dulbecco’s Modified Eagle Medium (DMEM) (Sigma Aldrich, USA). Residual mass was determined after 72 h using the same conditions as described for the screening stage.

Each composition was imaged on a clinical CT scanner Somatom Definition AS + (Siemens Healthineers, Erlangen, Germany) at 70 and 120 kVp using 400 mAs, pitch = 0.5, and 1 mm slice thickness [12]. Both full and half strength Omnipaque 350 (Iohexol; 350 mg of non-ionic iodine per mL, GE Healthcare Ireland, Cork, Ireland) were used as controls. Radiopacity measurements was reported as Hounsfield Units (HU) ± standard deviation (SD) [12].

MRI susceptibility was established for each composition by dispersing the materials (100–300 µm) in a non-aqueous gel made with 1% Evonik Intelimer IPA 13–1 NG polymer in peanut oil (2, 4, 6, 8, 10% w/w, n = 5). The gels were then loaded in 5 mm tubes and exposed to magnetic stirring, horizontal rotation, and heating before cooling on ice to solidify the gel. Each composition was measured at room temperature using Agilent 3 T preclinical MRI to obtain R2 (CPMG) MRI relaxometry measurements. A linear regression analysis and extrapolation to 100% volume fraction was used to provide values for the particles [12].

Irregular BKSA16 particles (100–300 µm), were subjected to flame spheroidization using a propane oxygen flame to transform the morphology of the glass frit to a microsphere (described elsewhere [14]). After processing the glass microspheres were packaged dry clean containers and sterilized using gamma irradiation (average dose 30 kGy, Nordion, Ontario, Canada).

Scanning Electron Microscopy (SEM) was conducted on BKSA16 microspheres (BKSA16M). A Hitachi S4700 Cold Field Emission scanning electron microscope (with an Oxford Analytical 80 mm SDD detector for energy dispersive x-ray spectroscopy) was used to image the microspheres with a condenser lens 1 set to 3, an emission current of 20 mA, and accelerating voltage of 10 kV. Prior to imaging the specimens, they were placed on aluminum stubs and coated with gold–palladium (20 nm) to prevent product contamination [9].

Following spheroidization of BKSA16 from irregular particles (BKSA16I) to microspheres (BKSA16M), the samples were recharacterized (density, XRD, 11B MAS NMR, and DSC) as previously described in Manchester et al. [13]. The data pre- [13] and post-spheroidization was compared to determine the impact of spheroidization on the material.

The suspension time of BKSA16 microspheres was qualitatively determined using accepted methods from the relevant clinical literature [15]. Succinctly, 0.1 g of glass (100–300 μm) placed into sealed 15 mL tubes with 10 mL of Omnipaque 350 (n = 3). Once added, a timer was started, and the tubes were inverted (10x) to create a suspension. Subsequently, the falcon tubes were held upright, placed against a black background and monitored until the suspension fell below 2/3 of the total volume [15]. The time was then averaged for three runs to determine the average suspension time.

3 Results

Average residual mass at 2 h (n = 3) was assessed in saline is presented in Fig. 1A. From these data a cubic model with statistical significance (p-value < 0.0001) was established (Table 2). The average residual mass at 2 h ranged from 7 to 108% (where a value > 100% represents a residual mass greater than the initial mass used). The ranking of factors that facilitate an increase in the final residual mass are B > K*Sr > Sr > K > B*Sr > B*K, where B*K and B*Sr result in a decrease in the final residual mass.

Fig. 1
figure 1

Contour plots of responses for A average residual mass at 2 h in sterile saline, B radiopacity at 70 kVp. C CT radiopacity at 120 kVp, and D MRI R2 slope. The x-axes are labelled as ‘A’: mol% B2O3, ‘B’: mol% K2O, and ‘C’: mol% SrO

Full size image
Table 2 Model equations and model statistics for each response per Fig. 1 and data for density, XRD (% crystallinity), 11B MAS NMR (B[3]: B[4]), and DSC (Tg onset and inflection) pre- [13] and post-spheroidization of BKSA16
Full size table

CT radiopacity (n = 5) at 70 kVp (Fig. 1B) produced a statistically significant (p-value < 0.0001) reduced quadratic model (Table 2). CT Radiopacity (70 kVp) ranged from 212 to 5249 HU. Analysis of the coefficients at 70 kVp (Table 2) resulted in the following ranking of factors that lead to an increase in radiopacity: Sr > K > B > B*Sr > K*Sr. CT radiopacity (n = 5) at 120 kVp (Fig. 1C) produced a statistically significant (p-value < 0.0001) reduced cubic model (Table 2). CT Radiopacity (120 kVp) ranged from 161 to 3600 HU. Analysis of the coefficients at 120 kVp (Fig. 1C) ranked the factors driving an increase in radiopacity as: Sr > K*Sr > K > B > B*Sr > B*K. Notably, interactions B*K and B*Sr resulted in a reduction in CT radiopacity.

Analysis of the R2 slope (n = 2) for MRI (Fig. 1D) provided a statistically significant (p-value 0.0152) reduced cubic model (Table 2) with values ranging from 1.298 to 21.12. Analysis of the coefficients for the R2 slope (Table 2) resulted in the ranking of factors for increased slope (i.e., higher relaxivity) as follows: B > K*Sr > Sr > K > B*K > B*Sr (where B*K and B*Sr decrease slope).

BKSA16 was processed into microspheres and morphology was confirmed via SEM imaging (Fig. 2A). Imaging confirmed the absence of irregular particles, surface contamination or other adverse findings. For completeness, and to account for any changes in the chemical and physical properties arising from microspheres processing, the irregular particles (BKSA16I) and microspheres (BKSA16M) had their basic characteristics (methods described elsewhere [13]) compared (Table 2).

Fig. 2
figure 2

A SEM image of BKSA16 microspheres (100x), B residual mass in saline at 2 h for BKSA16I and BKSA16M, C residual mass in 10% BSC/90% DMEM at 72 h for BKSA16I and BKSA16M, and D suspension times for BKSA16I and BKSA16M

Full size image

BKSA16I and BKSA16M were compared based on residual mass in saline after 2 h (Fig. 2B). BKSA16I achieved a residual mass of 84.25 ± 5.04% and BKSA16M achieved a residual mass of 88.44 ± 4.1% after 2 h. Average residual mass (n = 3) for the irregular particles (BKSA16I) and microspheres (BKSA16M) were also evaluated in 10% BCS/90% DMEM at 72 h (Fig. 2C). Notably, BKSA16 had residual mass at 72 h in both forms (irregular particles and microspheres). Finally, the average suspension time (n = 3) for BKSA16I was determined to be 2.01 min (run 1: 2.05 min, run 2: 1.20 min, and run 3: 1.58 min). The average suspension time for BKSA16M was determined to be 1.15 min (run 1: 1.11 min, run 2: 1.18 min, and run 3: 1.17 min).

4 Discussion

Average residual mass in saline (Fig. 1A) lead to the elimination of seven compositions (BKSA1, 3, 4, 8, 13, 14, and 15) as they achieved ca. < 50% residual mass at 2 h. This decision was informed from previous work where mass loss in vitro was a significant underestimate of in vivo dissolution rates [9, 12]. As such these compositions have a low probability of clinical success based on dissolution requirements. It was determined that radiopacity sufficient for clinical use can be achieved in glasses with high mol% SrO (Fig. 1B and C) such as, BKSA6, 7, 9, 10, 11, and 12. These compositions achieve radiopacity comparable to microspheres deemed sufficient for clinical imaging in the literature. For example, Negussie et al., and Ashrafi et al., have established imageable particles as having attenuation values of > 3972 HU and 4718 HU respectively [16, 17]. Additionally, Contour™ particles in half strength contrast have reported attenuation values of ca. 2500 HU [18]. Similarly, BKSA6, 7, 9, 10, 11, and 12 exhibit CT radiopacity values in the range 3563–5249 HU at 70 kVp. SrO was found to be the largest driver of increased CT radiopacity and responsible for facilitating a larger residual mass following dissolution experiment (Table 2). As such control of this critical constituent is an important consideration in materials design. Imageability on MRI was also assessed (Fig. 1D) and the compositions were shown to induce minimal R2 contrast changes and as such, these particles will not confound follow-up R2-weighted MRI scans in a clinical setting as required [12].

Considering the data from our previous work [13] and in conjunction with dissolution and imageability findings, BKSA16 was selected as it best addressed the GAE microsphere requirements, and was further evaluated. While it has low CT radiopacity, BKSA16 also has a relatively low concentration of SrO (5.69 mol%) and B2O3 (73.91 mol%) which facilitates potential increases in the clinical unit dose (from 44 to 55.91 mg (Shymka 2021)) without raising toxicological concerns (BKSA16 margin of safety (where > 1 indicates low toxicological risk) at 44 mg dose: boron = 1.27 and strontium = 1.67 (Shymka 2021)). Due to the novelty of the procedure, there currently exists no cleared/approved products with a defined dose and as a result the dosing for this indication remains ambiguous. Nevertheless, on the basis of the existing clinical data which confirm doses within the range of ca. 0.2 mL [6, 7], it can be estimated that a probable dose for GAE will be in the range of 44–55 mg (Shymka 2021). For the purposes of this work, a 44 mg dose was established based on the unit dose administered (0.2 ml Embozene particles) in clinical studies [7], a particle size of 100 µm, and a density of 2.3 g/cm3 (Shymka 2021). Additionally, BKSA16 has a density of ca. 2.3 g/cm3 (comparable to a range of commercially available contrast media) which will support adequate suspension, reduced sedimentation/aggregation, and ease of delivery [19]. Finally, BKSA16 has sufficient mass remaining (84.25% at 2 h) which should be adequate to support residual mass > 2 h in vivo and thereby, reduce the risk of ineffective occlusion [20]. To overcome the potential issues of low intrinsic CT radiopacity, these particles may be delivered via traditional TAE methods using half strength contrast to safely and effectively image the particles in a clinical setting and as such address this design requirement [21].

Following the selection of BKSA16, the irregular particles were processed into microspheres (Fig. 2A) and characterized (Table 2). The properties pre/post spheroidization were absent of major differences. As such, the properties of BKSA16 are stable, and these particles can be made into microspheres substantial without alteration of the glass properties. The residual mass in saline at 2 h was also compared between the BKSA16 irregular particles (BKSA16I) and microspheres (BKSA16M) (Fig. 2B). No major differences in dissolution occurred when comparing the two forms and both forms maintained a residual mass at 2 h which will facilitate effective occlusion of the vasculature. The dissolution of BKSA16I and BKSA16M was further assessed in 10% BCS/90% DMEM to provide insight into their dissolution behaviour in vivo and both forms were found to have a residual mass of 14.28 and 26.99% at 72 h, respectively (Fig. 2C). The dissolution of borate networks has become an area of interest in the literature, and while our understanding is nascent in this regard, it has been suggested that hydrolysis and hydration drive dissolution of such networks in aqueous media [22]. However, benchtop models have been shown to significantly overestimate mass loss rates in comparison to animal models (e.g. 100% dissolution ca. 48 h and ca. 24 h, respectively) [9, 12]. Thus, it may be appropriate to suggest 100% dissolution will likely occur in vivo in < 72 h meeting the requirements for a microsphere indicated for GAE, however, further animal studies are required to fully characterize the dissolution behaviour of BKSA16 in situ. Lastly, the suspension times of BKSA16I and BKSA16M were assessed and compared to microspheres currently used in TAE procedures (ca. 2 min) [17, 23]. BKSA16I achieved a consistent suspension time with commercially available microspheres used in TAE’s procedures and BKSA16M settled out of solution quicker (ca. 1.20 min). However, as stated in Johnson et al. [24] a shorter suspension time of ca. 30 s is still considered sufficient for clinical use, can support the delivery of the particles (with similar handling when compared to drug loaded DC Bead), and is disregarded as a “clinical concern when sufficient soluble contrast media is used” [24]. As such, BKSA16 is able to sustain a suspension time that is compatible for use clinically. While the data are preliminary and further studies are required, BKSA16 remains a promising candidate for use as an embolic microsphere indicated for GAE.

5 Conclusions

Glasses based on B2O3, K2O and SrO were systematically evaluated as potential degradable embolic materials for GAE using a statistical design methodology. Modulation of critical properties across a range of clinical variables were assessed (e.g., dissolution, CT-radiopacity, MRI R2 relaxivity, and suspension times). Based on the data, one exemplary composition (BKSA16) was processed into spherical form, and no substantive changes to critical properties were observed as a result of processing. BKSA16 is a promising candidate for use as a degradable embolic microsphere indicated for GAE and warrants further development and investigation.