Fluorescence Quantification of Silicone Oil Release upon Contact with Liquid Therapeutic Formulations

Abstract

Prefilled syringes are valuable drug delivery systems, offering convenience and precision dosing. Among the critical factors influencing their performance is the stability of the silicone oil layer, which acts as a lubricant, guaranteeing the gliding properties of the plunger. The silicone oil, if it comes in contact with therapeutic formulations, can be subject to drug–container interactions, potentially leading to silicone oil release into the solution, thereby altering the gliding properties of the syringe and leading to unwanted particle formation, compromising drug efficacy and safety. Different measurement techniques, such as visual inspection, dynamic light scattering and spectroscopic analysis, are used to assess silicone oil layer stability in prefilled syringes. However, a quantitative, rapid and low-volume screening method to rapidly evaluate container compatibility for therapeutic formulations is not available. Here, we present a multi-well-based screening protocol allowing users to quantify, through fluorescence, the silicone oil released into a solution upon contact with liquid formulations. Fluorescently labeled uniform silicone oil layers of the desired thickness are deposited in glass-bottom wells and exposed to typical formulations, containing surfactants and monoclonal antibodies. The release of silicon oil as a function of contact time is quantified using fluorescence calibration. Beyond its use as a screening tool to evaluate drug–container compatibility, our protocol can contribute to the fundamental understanding of the factors and mechanisms influencing silicone oil layer stability and, furthermore, to the optimization of drug delivery systems.

1. Introduction

Pharmaceutical companies are increasingly using prefilled syringes (PFSs) as an alternative to vial packaging for the delivery of monoclonal antibodies (mAbs) in parenteral formulations. The major advantages of using PFS packaging are, among others, to simplify the drug administration by the caregiver, reduce the risk of microbial contamination, provide a reduced injection cost with a lower dead volume, and preserve the drug products until delivered safely for injection [1,2].

To ensure the efficient gliding of the stopper during drug injection and to attain optimal PFS functionality, commonly, polydimethylsiloxane (PDMS) medical-grade silicone oil is used as a lubricant to coat both the barrel and stopper surfaces.

Several papers describe the implications of the silicone oil–water interface for drug aggregation and particle formation in biological formulations [1,3,4,5,6]. Therapeutic proteins’ exposure to such interfaces can lead to their adsorption, structural perturbation and aggregation [7,8,9,10,11]. To minimize the effects of the silicone oil–liquid interface, surfactants are often included in the formulation. Surfactants are molecules that protect biologics against interactions with hydrophobic interfaces and are commonly used in mAb formulations to reduce protein adsorption and aggregation in siliconized PFSs [10,12,13,14]. However, although surfactants are effective in preserving the stability of biologics, these amphiphilic molecules can also favor interactions between the drug product components and the silicone oil. Indeed, in combination with other variables (such as pH, buffer, tonicity agents, etc.), the presence of surfactants can promote silicone oil layer degradation via a complex set of mechanisms described as “de-lub”, “wet”, and “wash” processes [15,16,17]. These surfactant-induced changes in the silicone oil layer can trigger changes in its lubricant properties. Moreover, drug component interactions with silicone coatings may lead to subvisible particle formation in the solution. Together, the loss of lubrication and the increase in subvisible particles challenge PFSs’ performance [3,4,16,18,19,20].

From this perspective, a thorough analysis of the interaction potential between the formulation composition and the silicone coating is pivotal. The present study was designed to propose a screening protocol to identify the risk of degradation of the silicone coating by exposure to pertinent formulation components. Fluorescent silicone layers with desired thicknesses were deposited onto glass surfaces in multi-well plates, mimicking PFS silicone coatings. The impact of surfactants and monoclonal antibodies (mAbs) on the stability of the silicone layer was addressed using a calibrated quantitative fluorescence readout of the silicone released into solution upon contact with formulations.

2. Materials and Methods

2.1. Chemicals

DOW CORNING^® DC360 medical fluid is a silicone oil with a viscosity of 1000 cSt, a density of 0.97 g·cm⁻³ and a refractive index of 1.4. The silicone oil labeling was done with aminopropyl-terminated polydimethylsiloxane 900–1100 cSt (GELEST, Morrisville, PA, USA, DMS-A31), which is an amino-terminated silicone oil with a similar viscosity to DC360 and similar density and refractive index (0.98 g·cm⁻³ and 1.4, respectively), and BODIPY^TM TMR-X NHS ES (INVITROGEN, Eugene, OR, USA, D6117) diluted in dimethyl sulfoxide. The excitation and emission wavelengths of BODIPY are 544 and 570 nm, respectively. Ethanolamine was purchased from Sigma Aldrich (Saint-Quentin-Fallavier, France, E9508), and Tertbutanol (TBA) from ROTH (Lauterbourg, France, 4323.1).

A 10 mM Histidine buffer (Sigma Aldrich, H3911) at pH 6 containing 2% (w/v) Trehalose (ROTH, 5151) was prepared (named hereafter buffer) and used for all of the preparations. Nonionic surfactant solutions were prepared to evaluate the silicone oil release in solution. Polysorbate 80 (Sigma Aldrich, P1754) and Polysorbate 20 (Sigma Aldrich, P1379) were diluted at 100, 200, and 400 ppm, and Poloxamer 188 (Corning™, Manassas, VA, USA, 61-161-RM) at 400, 800 and 10,000 ppm in buffer.

Trastuzumab, a humanized immunoglobulin G isotype 1 (IgG1), was formulated in buffer by Promise Proteomics (Grenoble, France) at 5 g·L⁻¹ with an averaged molecular mass of 145,529 Da. Trastuzumab was prepared at 1 g·L⁻¹ in buffer, and PS80, PS20 or P188 were added at indicated concentrations.

2.2. Silicone Oil Fluorescent Labeling

A total of 80 µL of 10 mM BODIPY, diluted in dimethyl sulfoxide, was mixed with 1 mL of aminopropyl-terminated silicone oil in 4 mL of TBA for 3 h at ambient temperature under rotative agitation. The solution was transferred to a crystallizer, and ethanolamine was then added to block fluorescent label excess. The oil–water emulsion was transferred into Eppendorf tubes, and TBA was evaporated under vacuum during 1.5 h at 37 °C using a Vacufuge plus Vacuum concentrator (Eppendorf AG, Hamburg, Germany). The aqueous phase was removed by pipetting. A second ethanolamine wash was performed, followed by three water washing steps to remove potential ethanolamine traces. The residual silicone oil was diluted in 20 mL of TBA and left to dry in the Vacufuge. The silicone oil labeling rate was 0.03%, estimated thanks to the concentration of Bodipy measured using the absorbance of the labeled oil and considering a molar ratio of 2:1, respectively, for the Bodipy and the aminopropyl-terminated silicone oil. Controls of the fluorescence spectra of Bodipy were performed before and after the labeling to confirm the preservation of the fluorescence properties of Bodipy, once grafted to the silicone oil (See Appendix A Figure A1).

2.3. Reflective Interference Contrast Microscopy (RICM)

RICM [21] pictures (200 by 150 µm) were taken with an Olympus IX71 inverted microscope, which was equipped with a motorized stage (Märzhauser Wetzlar, Wetzlar, Germany), an Olympus DP30BW camera, a 20× magnification air objective (OLYMPUS, Rungis, France, UPLanFI, numerical aperture 0.5, working distance 1.8 mm) and a CoolLED PE-300lite lamp (CoolLED, Andover, UK). The illumination is controlled by a shutter that can be manually or automatically actuated via an RS232 port. A reflector cube containing a polarizer, a semi-reflecting plate (set at 45° to the optical axis) and a cross polarizer were selected to illuminate the sample. A filter (blue CHROMA, ET436/20×, 436 ± 10 nm) was placed between the lamp and the reflector cube to select the emission peak of the lamp to study the silicone oil layers. The Image Pro Plus 5 software (Media Cybernetics, Rockville, MD, USA) controls the microscope, the camera, the shutter, the image acquisition and the stage in Visual Basic language customizable thanks to the library given by the supplier. The stage was calibrated using the software Win-commander 4. The picture size is 1360 by 1024 px and has 256 gray levels (8 bits).

RICM can be used to determine the distance between two media with different refractive indices, here the glass bottom of the 96-well plate and the silicone oil deposited on it (Appendix A Figure A2). First, a monochromatic incident beam (I₀) passes through the glass (refractive index n₁ =1.5), where the light is partially reflected (I₁) and partially transmitted through the silicone oil layer above (refractive index n₂ =1.4). Similarly, once the transmitted light meets the silicone oil–air interface, a part of the beam is again reflected (I₂) and transmitted with a phase shift φ = π. The camera acquires the resulting intensity from the interference between the two reflected beams (I₁ and I₂) called I. This intensity is expressed for quasi-normal incidence as in Equation (1).

$$I=I_1+I_2+2\sqrt{I_1{I}_2}\cos 4\pi \left({\displaystyle \frac{n1 t+n2 h}{\lambda }}\right)+\pi$$

(1)

where

• λ is the wavelength of the monochromatic beam,
• $t$ is the thickness of the glass,
• h is the height of the silicone oil layer,
• $I_1=r_{01}^2{I}_0$ and $I_2=\left(1-r_{01}^2\right)r_{20}^2{I}_0$.

r is the Fresnel reflection coefficient, $r= {\displaystyle \frac{n_i-n_j}{n_i+n_j}}$, and i, j = (0, 1, 2) are the propagation media, in this case air, glass and silicone oil. Most of the light is transmitted at the oil–glass and glass–air interfaces (Appendix A Figure A2).

The interference fringes obtained in the RICM images are the direct observation of the variation in the distance between the two interfaces (glass–silicone oil and silicone oil–air). The distance between two fringes can be expressed as ${\displaystyle \frac{\lambda }{2n_2}}$.

λ is the wavelength, and n₂ is the refractive index of the silicone oil.

The height variation between two given positions can be given by h as in Equation (2).

$$h=N_{fringe}k-1$$

(2)

where N_fringe is the number of fringes between two selected positions, and $k={\displaystyle \frac{2\pi n_2}{\lambda }}$, with n₂ being the refractive index of the silicone oil and λ being the wavelength of the monochromatic beam.

2.4. Episcopic Fluorescence

A Zeiss Axio-Observer 7 inverted microscope equipped with an Axiocam 506 MonoD camera (Zeiss, Oberkochen, Germany) and a 20× magnification air objective with a 0.8 numerical aperture and a working distance of 0.55 mm (ZEISS Plan apochromat) was used for fluorescence microscopy. The episcopic illumination is made with a solid-state source Colibri 7 with the spectral filter DS red (λ_ex = 545 ± 25 nm; λ_em = 605 ± 70 nm). The microscope is equipped with a motorized stage controlled using the Zen software vs 2.3. The picture has 256 gray levels (8 bits).

2.5. Confocal Microscopy

A LEICA SP8 confocal microscope (Leica, Wetzlar, Germany) with a DMI600 XY motorized stage was used to determine the height of the silicone oil layers. Images were taken with an HC PL FLUOTAR 40xDRY objective, numerical aperture 0.6, in fluorescence contrast mode with a DPSS 561 nm source at 100 mW. A photomultiplier tube (PMT) detector was used to detect the transmitted light. The acquired image size was 512 × 512 px with a resolution of 176 px per nm, and the pinhole size was 1 airy unit. The microscope was controlled with the LAS X Life science microscope software.

Confocal microscopy was used to measure the total height of the silicone oil layers. The resulting image was processed using the ImageJ (vs 1.44p) 3D viewer program, which allows us to draw the xz profile of the layer. The fluorescence was plotted as a function of the z-stacks, as represented in Figure 1.

The height (h, Figure) is calculated according to Equation (3).

$$h=\left(Z_2-Z_1\right)-{2R}_z$$

(3)

where h is the total height (µm), Z1 and Z2 (µm) are the coordinates at ${\displaystyle \frac{(F_{max}-F_{min})}{2}}$, and R_z (µm) is the z resolution of the microscope. R_z is determined as in Equation (4).

$$R_z={\displaystyle \frac{1.4n_{air}\lambda }{{NA}^2}}$$

(4)

where n_air = 1, λ = 0.561 µm, NA = 0.6 µm and Rz = 2.18 µm.

2.6. Silicone Oil Quantification in the Supernatant

Siliconized wells were filled with 200 µL of sample solution of interest, i.e., buffer, buffer + surfactant, buffer + mAb, and buffer + surfactant + mAb. The wells were incubated for 8 h at 37 °C with agitation at 1200 rpm (Heidolph, Schwabach, Germany, Titramax 100), and samples were analyzed at indicated times.

At each time point, 100 µL was collected from the wells and put in a black polystyrene well. 100 µL of TBA was added to homogenize the suspension prior to reading the fluorescence (λ_ex = 544 nm; λ_em = 570 nm; bandwidth = 10 nm) on a Tecan M1000 spectrophotometer (Tecan, Männedorf, Switzerland). A minimum of four sample replicates per plate in three independently siliconized plates were used for each experiment.

3. Results and Discussion

3.1. Silicone Oil Layer Deposition, Topography and Thickness

The labeled silicone oil was mixed at 1% with Dow Corning ^® 360 medical fluid (Dow Coning, Midland, MI, USA) in TBA for 3 h at ambient temperature under rotative agitation. This mixture was then used to deposit defined amounts (12 g·L⁻¹, 6 g·L⁻¹, 1.2 g·L⁻¹ and 0.06 g·L⁻¹) in a volume of 50 µL into glass-bottom microplate wells (CELLVIS Gerasdorf, Austria, P96-0-N). Microplates were put in an oven with a solvent extraction for 1 h at 50 °C until complete TBA evaporation. The obtained silicone oil layers contained 600, 300, 60 and 3 µg of silicone, respectively.

The 3D fluorescence picture reconstructed from fluorescence images of a typical 600 µg silicone oil layer is presented in Figure 2a. Note that the image reconstruction does not cover the entire well bottom surface, represented by a black circle in Figure 2a.

It shows that the silicone oil fully covers the surface of the well with a heterogenous distribution. Intensely fluorescent, randomly distributed spots corresponding potentially to dust particles or entrapped solvent crystals can be seen. Three different positions were defined for further analysis (Figure 2): A, in the center, B, the middle, and C, at the edge of the well, respectively. A and C show lower fluorescence intensities than B, as illustrated in the fluorescence profile (Figure 2a). To estimate the height difference between A, B and C, RICM images were acquired (example in Figure 2b). As explained in Materials and Methods, counting the fringes obtained through RICM allows us to estimate the height (h) between defined positions thanks to Equation (2), relating the wavelength and h. For each couple of positions (A–B and B–C), we calculated h (

Table 1. Height differences (h) measured in 600 µg silicone oil layers between positions A and B and positions B and C (Figure 2c) using RICM interference fringes. Three independent silicone oil layers and four images for each layer were used (n = 12).

Position	A–B	B–C
h (µm)	4.71 ± 0.79	1.65 ± 0.25

).

These data allow us to propose a topography profile for the deposited silicone oil layer (Figure 2c), which can be described from the center to the edge of the well as a valley in A followed by an elevated plateau in B, which is 4.71 µm higher than A, and ending in a second valley in C, which is 1.65 µm lower than B, before reaching the edge of the well. Silicone oil layers obtained with different deposited masses show decreasing fluorescence intensity maxima as the layers become thinner. Also, thinner silicone oil layers (120 and 60 µg) show a more gradual increase in the thickness of the oil layer from the center A towards the edge of the well without a marked plateau B. It is likely that these topographies are the result of the evaporation process of the solvent. The shape of the meniscus of the silicone solution indeed depends on the physical and chemical properties of the solid, liquid and air phases and results in complicated liquid convection movements near the walls of a well [22].

Confocal microscopy imaging was used to measure the silicone oil layer thickness directly in the positions indicated on Figure 3a. The thickness of the layer as a function of the silicone oil mass deposited in the well was measured at position B (Figure 3a) using confocal fluorescence images (Figure 3c left) and is plotted in Figure 3b. These measurements show a linear dependency of the thickness of the layer with the deposited silicone oil mass.

Table 2. Thickness at positions B, C and D obtained from confocal fluorescence images (B, n = 3; C, n = 3 and D, n = 3 images) for 600 µg deposited silicone oil.

Positions	B	C	D
Thickness (µm)	5.98 ± 0.89	4.5 ± 1.22	79.8 ± 2.5

summarizes the silicone oil layer thicknesses measured at positions B, C and D through confocal microscopy in a well containing 600 µg silicone oil. The height difference measured between B and C is about 1.5 µm, which is in good agreement with the 1.65 ± 0.25 µm estimated from interference fringes on RICM images (Table 1).

3.2. Quantification of Silicone Oil Release in Solution

When silicone oil is present in an aqueous phase, it forms perfectly round droplets of different sizes due to the non-miscibility of the oil in water. Figure 4a shows an example of such a fluorescent silicone oil droplet. Direct fluorescence measurements of silicone droplet emulsions in water produce highly variable results, as the fluorescent oil droplets, free to move in solution, create very inhomogeneous samples. To increase homogeneity, the silicone oil droplets were dissolved (Figure 4b) by adding TBA to the solution (1:1 volume ratio TBA:aqueous solution), which is miscible with both the oil and water-based buffers.

The mass of silicone oil released in solution was calibrated by measuring the fluorescence of known silicone oil masses (0 to 5000 µg) of labeled silicone oil in 1:1 TBA:water mixtures (Figure 4c). The calibration curve allows us to correlate the measured fluorescence to a mass of silicone oil in solution. The fluorescence background signal, corresponding to a 1:1 TBA:water mixture without silicone oil, was measured at 11.92 ± 0.79 AU. This value was subtracted to all fluorescence values. The limit of detection (LOD) is calculated as 3 times the standard deviation divided by the slope of the calibration curve; LOD = 1.36 μg.

3.3. Silicone Oil Release upon Contact with Surfactants

The silicone oil release in solution was studied with formulations containing PS20, PS80 or P188. A total of 200 µL of the indicated concentrations of each surfactant in 10 mM Histidine, 2% Trehalose buffer was pipetted into siliconized 96-well plates and agitated (1200 rpm) for 0, 4 and 8 h at 37 °C. The plates contained 1 mg of fluorescently labeled silicone oil per well. Plates were sealed with sealing tape for 96-well plates (GREINER, Kremsmünster, Austria, Platesealer EASYSEAL™T transparent 79 × 135 mm). At each time point, 100 µL of the supernatant was placed in a well of a black polystyrene 96-well plate containing 100 µL of TBA. The black plate was agitated for 5 min at 900 rpm before the fluorescence was measured. The calibration curve (Figure 4c) allowed us to calculate the resulting silicone oil mass released by the surfactants into the solution.

Figure 5 shows the mass of silicone oil released into the solution as a function of PS20 (yellow), PS80 (violet) and P188 (orange) concentrations after 0, 4 and 8 h incubation. Concerning PS20, the initial value (0 h) shows that no silicone oil is released whatever the concentration of surfactant. At 4 h, the released silicone oil mass in solution increases significantly, with an average of 2.4 µg for 100 ppm, 8.5 µg for 200 ppm and 10.1 µg for 400 ppm of PS20. At 4 h, the results show a significant difference between 100 ppm and either 200 ppm or 400 ppm (p < 0.05%); however, the difference between 200 ppm and 400 ppm is not statistically significant (p > 0.05%). Interestingly, the released silicone oil mass increase is not proportional to the surfactant concentration and does not evolve significantly (p > 0.05%) between 4 and 8 h. The boxplot sizes highlight the large variability between replicates.

PS80 does not release silicone oil into the solution at 0 h, as observed for PS20. However, after 4 h, the amounts of silicone oil released increase to 6.1 µg on average for 100 ppm and 7.7 µg for 200 ppm; meanwhile, 400 ppm of PS80 releases 12.1 µg of silicone oil into the solution. The results at 8 h remain constant for each concentration (p > 0.05%). The variability between samples (large box plot size) and the distribution of the data set around the median (in particular for 400 ppm PS80) does not allow us to conclude on a significant effect of the surfactant concentration. These results show that 40 μg (200 μL × 200 ppm) of polysorbates can emulsify approximatively 8 μg of silicone oil after 4 h of incubation, which is a mass ratio of 5:1 PS:silicone oil.

In contrast with polysorbates, P188 does not create a significant silicone oil release as a function of the surfactant concentration (400 ppm, 800 ppm and 10,000 ppm) or as a function of time, and the mass of silicone oil released into the solution does not exceed 5 µg on average.

Different behaviors between the two types of non-ionic surfactants (polysorbates and poloxamer) have been demonstrated in the literature, notably concerning the high propensity of polysorbates to emulsify silicone oil in biotherapeutic solutions [23,24]. On one hand, the emulsification created by PS80 and PS20 can be explained by the molecular structure of polysorbates: a diblock composed of a hydrophilic head (polyoxyethylene sorbitan) with a long hydrophobic tail (fatty acid esters). The main difference between the two PSs used in this study resides in the length of the hydrophobic tail, which decreases the Hydrophilic–Lipophilic Balance (HLB) of PS80 as it has a longer tail (15 compared to 16.7 for PS20). Nevertheless, the two polysorbates seem to act similarly on the silicone oil release into the solution. On the other hand, P188 is a triblock molecule, made of two hydrophilic chains (polyethylene oxide (PEO)) separated by a hydrophobic central part (polypropylene oxide (PPO)). The HLB of P188 has been calculated to be 29, which classifies it as more hydrophilic in comparison with PS80 and 20, potentially explaining its low impact on the silicone oil release.

3.4. Silicone Oil Release upon Contact with mAbs and Surfactants

We also analyzed the silicone oil release in solution upon contact with typical formulations containing monoclonal antibodies (mAbs) and surfactants. The presence of mAb in the buffer does not create a significant silicone oil release over time as the mass of silicone oil in solution is low and close to the mass released by the buffer alone (Figure 6 green). Similar conclusions can be drawn in the presence of Poloxamer 188 (Figure 6 orange). Indeed, in the presence of P188, except at 0 h, where the mass of silicone oil is slightly higher than for the other time points, the silicone oil release is close to zero, with and without mAb.

The addition of polysorbates, however, increases the silicone oil release into the solution significantly and in a similar way for 200 ppm PS20 and 200 ppm PS80. At 0 h, only the formulation of PS with mAb shows a significant amount of silicone oil in solution, which stays rather stable over 8 h. Without mAb, the silicone oil release sharply increases in the presence of PS between 0 h and 4 h and stays stable beyond. Moreover, the silicone oil release after 8 h is highly similar on average for the two formulations containing PS (without mAb, around 7 µg, and with mAb, around 5 µg). The statistical distribution of the data, given by the box plot representation, highlights a larger variability when both molecules are present compared to the surfactant or mAb alone.

These observations could be reminiscent of the differences in molecular adsorption/desorption behavior that have been characterized for proteins and surfactants at the silicone oil–water interface. Both surfactants and mAbs are able to adsorb onto silicone oil. Polysorbates are efficient at preventing mAb adsorption on silicone oil when they are first in contact with the surface. This was studied using Quartz Crystal Microbalance with dissipation monitoring (QCM-d), notably by Zheng et al. [25], who used gold siliconized sensors to monitor the adsorption of a therapeutic protein (0.3 g·L⁻¹) in a 20 mM Histidine buffer containing 600 mM sucrose at pH7. PS80 was added at 0.05% (i.e., 500 ppm). The data showed that PS80 significantly reduced the mass of mAb adsorbed on the surface (550 ng·cm⁻² vs. 150 ng·cm⁻² without and with PS80, respectively). On the other hand, Poloxamer 188 was shown to be less efficient than PS80 in reducing mAb adsorption on siliconized SiO₂ sensors, illustrating the different behavior of both surfactant molecules [26]. Similarly, this was illustrated by Kannan et al. [27] for PS20 vs. P188. Adsorbed mAb on a silicone oil layer is not desorbed in the presence of polysorbates or poloxamer [27]. One could thus expect different adsorption mechanisms to dominate when a mixture of mAb and surfactant is adsorbed to the silicone than when the adsorption is dominated by a single species, either mAb alone or surfactant alone. As a consequence, the stability of the silicone oil layer might vary when exposed concomitantly or separately to different adsorbed species. Interestingly, regarding our results, the stability of the silicone oil layer seems to be most affected by the presence of polysorbates, in the presence or absence of mAbs, with released silicone oil masses in solution that are higher than in the presence of mAb alone.

In the case of PS, the mass of released silicone oil seems to decrease when mAb is present at the same time as the surfactant, but the variability of the data is high. A potential contribution underlying this observation could be the direct interaction between mAb and PS in solution, thereby reducing the impact of PS on silicone oil release. In addition to surfactant chemistry, the effective surfactant concentration available to interact with silicone oil is expected to depend on competitive interactions with therapeutic proteins in solution. Such competition between protein–surfactant binding in solution and surfactant adsorption at hydrophobic interfaces is concentration-dependent and formulation-specific, further supporting the need for formulation-relevant screening. In the literature, the release of silicone oil by polysorbates is well documented, notably by Gerhardt et al. [28], in a study on mAb storage in syringes. They have shown that PS20 released increasing quantities of silicone oil with time, but, in the presence of mAb, the amount of released silicone oil tended to decrease with time. Similar conclusions (lower release of silicone oil with mAb + PS80 than PS80 alone) were drawn by Jiao et al. [23] in their syringe study.

4. Conclusions

The deposition of fluorescently labeled silicone oil thin films has been developed on glass bottom microplate wells in order to establish a quantifiable fluorescence screening protocol for silicone oil release into solution upon contact with formulations. Silicone deposition through solvent evaporation gives reproducible thin films with a scalable thickness over the entire surface of the well. The release of silicone oil into a solution is quantified as a function of the contact time with formulations containing selected concentrations of PS20, PS80 and P188 in the presence or absence of mAb (Trastuzumab). The results demonstrate first that the silicone oil release is surfactant- and mAb-dependent, as has been reported before in syringe studies. We also show that 200 ppm PS80, a surfactant that is often used in therapeutic formulations, induces similar released silicone oil masses to 200 ppm PS20, which are significantly higher than the masses released by 800 ppm P188. This demonstrates a higher potential for polysorbates than poloxamer to affect the stability of silicone oil films.

This work is a valuable contribution for rapidly determining, with a low sample volume (200 µL) and high throughput, if a formulation is at risk or not when stored in the presence of silicone oil. Indeed, the release of silicone oil particles in solution affects both the therapeutic proteins and the performance of the syringe used for its storage or injection. In addition, the developed deposition method to obtain silicone oil thin films in 96-well plates could be combined with other analytical methods frequently used in container-compatibility studies. For example, it could be implemented as a first screening protocol for the stability of therapeutic proteins by SEC or DLS upon exposure to silicone oil layers, facilitating costly and long in-syringe stability studies. Moreover, the deposited silicone oil layer stability could be investigated mechanistically using desired thicknesses and comparing full and partial glass surface coverage. Finally, the presented method could also be implemented in the development of novel silicone oil layer treatments to improve lubrication and stability.