A Sorption Study between Ophthalmic Drugs and Multi Dose Eyedroppers in Simulated Use Conditions

Introduction

Ophthalmic solutions are currently the main galenic form used for local drug delivery to the eye. Many ocular diseases (glaucoma, eye infections, keratitis,…) require their administration once or multiple times a day, exposing the eye to contact with the substances present in the formulation. Most of the time, if the ophthalmic solution is presented in a multidose conditioning, the excipients will include a preservative to maintain microbiological stability.

Several studies have shown preservatives to be toxic for the eye, inducing problems such as allergic reactions, dry eye symptoms and subconjunctival fibrosis (1, 2, 3), and therefore encourage alternatives to withdraw them from eye drop formulation. Two possibilities are then available for manufacturers: unidose eyedroppers or multidose eyedroppers containing a system that prevents microbiological contaminations. Both have their advantages and drawbacks. Unidose eyedroppers are more expensive to produce, have a bigger environmental impact and their small size may be an issue for people having prehension problems. On the other hand, one of the risks of contamination preventing devices in multidose eyedroppers could be drug sorption with the systems designed to preserve sterility, like filter membranes or others. One commercially available preservative free multidose eyedropper (Novelia^®, marketed by Nemera^®) contains silicone parts isolating the solution from the outside, by using silicone’s properties of gas permeation to allow pressure equilibrium with drop release and properties of shape memory to act as a valve. Silicone is a material known for being at risk of interactions with certain drugs it comes in contact with, be it wanted or not, for example proteins, ciprofloxacin and prostaglandins (4, 5, 6). Such a device is therefore especially at risk of developing interactions with the ophthalmic drugs it could be put into contact with.

Cyclosporin, Latanoprost, Rifamycin, Timolol and Norfloxacin are 5 active pharmaceutical ingredients (API) commonly used for the treatment of various ophthalmic affections, with an administration schedule varying from once to multiples times a day, for a treatment duration of up to several days, months or even years for chronic diseases. They are also greatly different in terms of structure, molecular mass and lipophilicity, and some of them are known to be at risk for adsorption in certain matters. The aim of our work was to evaluate in clinical use conditions the potential risk of sorption of commercialized formulations of the 5 previously cited APIs with the Novelia^® system, in comparison with an eyedropper made exclusively of low density polyethylene (LDPE) and thus without a sterility conserving system, used as control.

Materials and methods

Samples preparation

Standards, reagents, drugs and eyedroppers

Acetonitrile, methanol, ammonium carbonate, sodium acetate, hydrochloric acid, phosphoric acid and all of the reference substances were purchased from Sigma-Aldrich laboratories (Saint-Quentin-Falavier, France). Materials for cyclosporine preparation (including cyclosporine powder, batch IE1400287A, exiry date 04/19) were purchased from Inresa (Bartenheim, France) and from Cooperation Pharmaceutique Française (Melun, France). Rifamycin Chibret^® (batch ST025, expiry date 01/20), Timabak^® (timolol, 2.5 mg.ml⁻¹: batch 2G33, expiry date 01/19; 5 mg.ml⁻¹: batch 9F06, expiry date 11/18) and Chibroxine^® (norfloxacin, batch 5D40/A, expiry date 10/18) were purchased from Thea Laboratories (Clermont-Ferrand, France), and Xalatan^® (latanoprost, batch R65646, expiry date 08/18), was purchased from Pfizer (Paris, France). Simple low density polyethylene 10 ml eyedroppers (further referred as LDPE eyedroppers) were purchased from CAT Laboratories (Lorris, France) and Novelia^® 10 ml eyedroppers (further referred as LDPE-Si eyedroppers, represented in Figure 1) were obtained from Nemera Laboratories (La Verpillère, France), both in sterile packages (ethylene oxide sterilization mode for LDPE eyedroppers and gamma radiations sterilization mode for LDPE-Si eyedroppers). It is to be noted that the rifamycin formulation we studied is presented in sealed glass flask with a polyvinylchloride adaptable end piece and the timolol formulation we studied is presented in LDPE eyedroppers including a polyethersulfone membrane (further referred as LDPE-PES eyedroppers).

Figure 1:

Schematic representation of Novelia^® cap.

Reconditioning and storage

Cyclosporine micellar solutions were prepared according to the method described by Chennell et al (7). For those and for all solutions to be tested in a different eyedropper from the commercial one, reconditioning into the tested eyedroppers was realized under the laminar air flow of an ISO 4.8 microbiological safety cabinet to prevent microbial contamination. All units were thereafter stored at 25 °C±2 °C in a constant climate chamber with 60 % residual humidity except for cyclosporine units that were stored at 5 °C±3 °C in a fridge.

Results

Evaluation of drug sorption

The results of the drug sorption evaluations are presented hereafter.

Cyclosporine

The evolution of cyclosporine concentrations in the delivered drops is presented in Figure 2.

Figure 2:

Evolution of cyclosporine concentrations (mean ± 95 % confidence interval, n ≥ 3) in drops through time, at 1 mg.ml⁻¹ (A), 10 mg.ml⁻¹ (B) and 20 mg.ml⁻¹ (C) in low-density polyethylene only eyedroppers (blue cross) and low-density polyethylene with silicone membrane eyedroppers (red circle). Purple dash lines=90 % and 110 % of reference concentration (RC).

The RC for the 1 mg.ml⁻¹ concentration was of 1.06 mg.ml⁻¹. Mean concentrations in delivered drops didn’t vary by more than 3.5 % and 7.9 % from RC throughout the study, for respectively tested LDPE and LDPE-Si eyedroppers. The mean concentration for the 48h drop of LDPE-Si eyedroppers was 1.01±0.07 mg.ml⁻¹ (95.6±7.0 %). Measured concentration of residual solution was of 1.07±0.00 mg.ml⁻¹ (100.8±0.4 %) and 1.06±0.00 mg.ml⁻¹ (100.0±0.1 %) for respectively LDPE and LDPE-Si eyedroppers.

The RC for the 10 mg.ml⁻¹ concentration was of 10.72 mg.ml⁻¹. Concentrations in delivered drops didn’t vary by more than 5.3 % and 6.3 % from RC throughout the study, for respectively tested LDPE and LDPE-Si eyedroppers. Measured concentration of residual solution was of 10.38±0.22 mg.ml⁻¹ (96.8±2.0 %) and 10.47±0.30 mg.ml⁻¹ (97.6±2.8 %) for respectively LDPE and LDPE-Si eyedroppers.

The RC for the 20 mg.ml⁻¹ concentration was of 21.17 mg.ml⁻¹. Mean concentrations in delivered drops didn’t vary by more than 5.6 % and 8.6 % from RC throughout the study, for respectively tested LDPE and LDPE-Si eyedroppers. The mean concentration for the 24h drop of LDPE-Si eyedroppers was 19.34±1.14 mg.ml⁻¹ (91.4±5.4 %). Measured concentration of residual solution was of 20.81±0.35 mg.ml⁻¹ (98.3±1.6 %) and 19.72±0.67 mg.ml⁻¹ (93.1±3.1 %) for respectively LDPE and LDPE-Si eyedroppers.

Rifamycin

The evolution of ryfamycin concentrations in the delivered drops is presented in Figure 3.

Figure 3:

Evolution of rifamycin concentrations (mean ± 95 % confidence interval, n ≥ 3) in drops through time in low-density polyethylene only eyedroppers (blue cross), low-density polyethylene with silicone membrane eyedroppers (red circle) and glass and polyvinylchloride eyedroppers (green diamond). Purple dash lines=90 % and 110 % of reference concentration (RC).

The RC was of 10,392 UI.ml⁻¹. Concentrations in delivered drops varied from initially measured concentration by 8.7 % for glass eyedroppers, by 6.9 % for LDPE eyedroppers, and by 11.2 % for LDPE-Si eyedroppers. Measured concentration of residual solution was of 9573±128 UI.ml⁻¹ (92.1±1.2 %), 9532±425 UI.ml⁻¹ (93.0±1.4 %), and 9274±262 UI.ml⁻¹ (89.2±2.5 %) for respectively glass, LDPE and LDPE-Si eyedroppers.

Latanoprost

The evolution of latanoprost concentrations in delivered drops is presented in Figure 4.

Figure 4:

Evolution of latanoprost concentrations (mean ± 95 % confidence interval, n ≥ 3) in drops through time in low-density polyethylene only eyedroppers (blue cross) and low-density polyethylene with silicone membrane eyedroppers (red circle). Purple dash lines=90 % and 110 % of reference concentration (RC).

The RC was of 50.04 µg.ml⁻¹. Concentrations in delivered drops from LDPE eyedroppers didn’t vary by more than 7.8 % from RC throughout the study. Measured concentration of residual solution in these eyedroppers was 49.81±0.44 µg.ml⁻¹ (99.6±0.9 %). Concentrations in delivered drops from LDPE-Si eyedroppers decreased by a maximum of 76.4 % at 24h then stabilizing around 15 µg.ml⁻¹ (65 % loss from RCs). Measured concentration of residual solution in these eyedroppers was of 29.77±0.69 µg.ml⁻¹ (59.5±1.4 %).

Timolol

The evolution of timolol concentrations in delivered drops is presented in Figure 5.

Figure 5:

Evolution of timolol concentrations (mean ± 95 % confidence interval, n ≥ 3) in drops through time, at 2.5 mg/ml (A) and 5 mg/ml (B) in low-density polyethylene only eyedroppers (blue cross), low-density polyethylene with silicone membrane eyedroppers (red circle) and low-density polyethylene with polyethersulfone membrane eyedroppers (orange triangle). Purple dash lines=90 % and 110 % of reference concentration (RC).

The RC for the 2.5 mg.ml⁻¹ concentration was of 2.47 mg.ml⁻¹. Concentrations in delivered drops didn’t vary by more than 3.6 %, 1.0 % and 2.1 % from RC throughout the study, for respectively tested LDPE-PES, LDPE and LDPE-Si eyedroppers. Measured concentration of residual solution was of 2.45±0.06 mg.ml⁻¹ (99.1±2.6 %), 2.46±0.09 mg.ml⁻¹ (99.6±3.8 %), and 2.47±0.03 mg.ml⁻¹ (100.1±1.2 %) for respectively LDPE-PES, LDPE and LDPE-Si eyedroppers.

The RC for the 5 mg.ml⁻¹ concentration was of 4.97 mg.ml⁻¹. Concentrations in delivered drops didn’t vary by more than 3.4 % and 4.1 % from RC throughout the study, for respectively tested LDPE and LDPE-Si eyedroppers. Concentrations in delivered drops varied from initially measured concentration by 10.1 % for LDPE-PES eyedroppers as an increase at the 24h point, but thereafter follow the same scheme as the other eyedroppers. Measured concentration of residual solution was of 4.98±0.12 mg.ml⁻¹ (100.1±2.5 %), 5.02±0.06 mg.ml⁻¹ (100.9±1.1 %), and 5.00±0.15 mg.ml⁻¹ (100.6±3.0 %) for respectively LDPE-PES, LDPE and LDPE-Si eyedroppers.

Norfloxacin

The evolution of norfloxacin concentrations in delivered drops is presented in Figure 6.

Figure 6:

Evolution of norfloxacin concentrations (mean ± 95 % confidence interval, n ≥ 3) in drops through time in low-density polyethylene only eyedroppers (blue cross) and low-density polyethylene with silicone membrane eyedroppers (red circle). Purple dash lines=90 % and 110 % of reference concentration (RC).

The RC was 3.10 mg.ml⁻¹. Concentrations in delivered drops didn’t vary by more than 4.1 % and 3.6 % from RC throughout the study, for respectively tested LDPE and LDPE-Si eyedroppers. Measured concentration of residual solution was of 3.10±0.04 mg.ml⁻¹ (99.9±1.2 %), and 3.08±0.03 mg.ml⁻¹ (99.4±1.0 %) for respectively LDPE and LDPE-Si eyedroppers.

Discussion

The objective of this study was to conduct a preliminary investigation of potential sorption phenomena between several commercialized ophthalmic solutions and LDPE-Si eyedropper in comparison to LDPE eyedroppers. The results are in favour of an absence of sorption between LDPE-Si and timolol and norfloxacin (no loss of concentration throughout the study). For rifamycin solutions, a difference that could be assimilated to a drug sorption was noted, but the variation from RC was significant beyond a 7-days conservation (versus the 15-days recommendation in glass bottles). Results for cyclosporine (micellar formulation) at a concentration of 1 mg.ml⁻¹ showed a non-significant decrease in concentrations when compared to RC (7.9 %), but the difference was significant when compared to obtained concentrations from LDPE eyedroppers. The results for the latanoprost formulation showed a massive decrease (up to 76 % loss) in latanoprost concentrations in delivered drops from LDPE-Si eyedroppers, clearly indicating the existence of a specific interaction with LDPE-Si. For the LDPE eyedroppers, the variations in concentration of the APIs remained insignificant throughout the study, for all tested formulations, when compared to the corresponding RC.

Cyclosporine is the most lipophilic molecule studied here (LogP=7.5). The formulation employed is a micellar solution containing, among other excipients, polyvinylpyrrolidone as wetting agent and polyethoxylated castor oil. It is important to note that that this formulation’s stability was previously internally assessed for 6 months at 4 °C (data not published). As silicone tends to hold back lipophilic substances (10), a difference between LDPE and LDPE-Si eyedroppers was expected. Indeed, for the lowest concentration studied (1 mg.ml⁻¹), the Kruskal-Wallis test result exhibits a significant difference between LDPE and LDPE-Si measured concentrations. Our hypotheses are absorption of already adsorbed molecules or a non-saturation of silicone exposed surface, in both cases allowing new molecules to adsorb through time. However, those concentrations tend to normalize at the end of the study. Mean concentrations never went under the 90 % limit, a sole confidence interval minimum was found at 0.938 mg.ml⁻¹ (88.6 % of the RC) for the 48 hour drop. The 10 mg.ml⁻¹ concentration drops and 20 mg.ml⁻¹ concentration drops didn’t exhibit any significant variation between the different eyedroppers. Those concentrations also stayed in the 90–110 % range all along the study, a sole lower limit of confidence interval on the third analytical time for 20 mg.ml⁻¹ LDPE-Si eyedroppers was found at 18.2 mg.ml⁻¹ (86.0 % of RC), which is consistent with a previously published study (7). The concentration is here 10 to 20 folds the previous one, so we can highly suspect a saturation phenomenon of adsorption sites, consuming few enough of the molecules so it doesn’t reflect on the drops’ concentration. Regarding the lipophilicity of cyclosporine, we were expecting to observe a larger decrease in drops concentration. Obviously, the RC plays a role, but this is far to be the only parameter of interest. Cyclosporine is also a large molecule (1202.6 g.mol⁻¹), depending on the mechanism of interaction with silicone, it may be too big to reach some of the interaction sites. The formulation contains lipophilic excipients that may compete with cyclosporine for those sites, decreasing the adsorption of cyclosporine. Eventually the 20 mg.ml⁻¹ solution is far more viscous than the other two, this may explain the bigger variability of this solution. Several studies have already been published about cyclosporine interacting with its container, mostly with polyurethane or silicone catheters, and they all found different extents of adsorption, even leaching of adsorbed molecules. Concentrations were of the same order to those studied here, but exposed surfaces were far larger than for the conditions tested in this study (11).

Rifamycin is also lipophilic (LogP=4.9), usually used as 10,000 UI.ml⁻¹ solutions. This drug is commercially presented in sealed glass flask with a PVC eyedropper to be adapted on the top of the flask after opening. Rifamycin is susceptible to oxidation and hydrolysis out of a narrow pH range, so we had to set pH of the mobile phase in this range to avoid degradation. According to the World Health Organization, the nominal concentration of the studied formulation corresponds to a mass concentration of about 11.3 mg.ml⁻¹, so we didn’t expect a large difference regarding the results previously obtained for 10 mg.ml⁻¹ cyclosporine. The main difference observed is for the last drop at 324h (14 days) when LDPE-Si drops concentration is significatively lower than for the two other eyedroppers, in addition to being lower than 90 % of RC limit, implying it to be a significant decrease. A degradation process may have interfered as the concentrations do decrease between the first and last drops for all eyedroppers, but as the chromatographic method used wasn’t validated as stability indicating, it cannot be ascertained. The rifamycin formulation that was studied contained several excipients, some precisely to prevent degradation (like ascorbic acid used for its anti-oxidant properties (12)), which may have an impact on rifamycin’s adsorption, either positively or negatively.

Latanoprost is almost as lipophilic as rifamycin (LogP=4.3), but this molecule is far smaller (432.6 g.mol⁻¹versus 697.8 g.mol⁻¹), and is usually used a lot more diluted, as 50 µg.ml⁻¹ solutions. The formulation studied here is an aqueous solution and has the lowest concentration solution studied. A decrease of concentrations can be observed for LDPE-Si bottles from the first drop onwards and is at its maximum for drop 2. Drop concentrations then stabilize around 15 µg.ml⁻¹ and never increased again throughout the rest of the study. Another noticeable point is the concentration measured in the residual solution in those bottles after emitting the last analyzed drop, found at 30 µg.ml⁻¹, while in LDPE bottles this concentration stayed at 50 µg.ml⁻¹. Both bottle types are made of LDPE, so the concentration of the solution in contact with it should behave the same way. It was checked afterwards that the concentration of the solution didn’t significantly decrease in LDPE-Si eyedroppers without the silicone piece (data not shown). The impact seems a lot more important than for cyclosporine and rifamycine, but the concentration studied is minimum 20 folds lower, so a comparison is hazardous with the present data only. A possible explanation may be that the short contact between the solution and the silicone pieces when turning the eyedropper over to emit a drop is enough to adsorb and possibly absorb a large quantity of latanoprost, as we made sure that eyedroppers stood still and up in between drop release. Latanoprost also exists in a lipophilic vehicle formulation that may behave differently from the one tested. Indeed, it has been shown that some excipients may prevent adsorption of latanoprost to its container (13, 14). A comparison with rifamycin at the same molar concentration would also be advantageous as it would show the impact of the variation of molecular weight nearly alone.

Timolol is lightly lipophilic (LogP=1.2) and is the smallest molecule we studied here (316.4 g.mol⁻¹). It is usually used as 2.5 mg.ml⁻¹ or 5 mg.ml⁻¹ solutions. The particularity here is that the aqueous formulation studied here is initially presented in LDPE eyedroppers including a filter and a polyethersulfone membrane that drops get in contact with when delivered. We initially studied it as a control, but the results we obtained weren’t completely regular. For both concentration, the first drop concentration is lightly lower than for both LDPE and LDPE-Si eyedroppers and normalize on the second drop, implying a potential saturation phenomenon of one or both of the filtration devices. Also for the 5 mg.ml⁻¹ concentration, we observed a peak concentration in the third drop inducing a potential clinical overdose, implying a potential leaching of previously adsorbed molecules. Both profiles were superimposable for each concentration, and no drop presented a concentration outside the 90–110 % range. Concentrations studied here are in the same range than cyclosporine or rifamycin, but timolol is far smaller and less lipophilic. Despite its molecular weight that could increase the risk of interactions, it seems that the low lipophilicity plays a more important role and prevents any significant interaction with the device.

Norfloxacin is the only truly hydrophilic molecule we studied (LogP=−1.0) with a molecular weight comparable to timolol (319.3 g.mol⁻¹). As for timolol, both profiles were superimposable and no drop presented a concentration outside the 90–110 % range. Norfloxacin and timolol are very close in term of molecular weight and of concentration of studied formulations. As we didn’t observe many variations for timolol and norfloxacin is even less lipophilic, we didn’t expect any more variation here.

The main phenomenon that we were expecting with LDPE-Si eyedroppers was adsorption and maybe absorption into the silicone pieces. LDPE is known to be mostly inert toward molecules in contact, even if it has been shown to interact with insulin in perfusion tubes (15). However, as the body of both eyedroppers is in LDPE, interactions with the main body of the eyedroppers would have happened for both types, be it adsorption or permeation. As plasticizers do not enter into the composition of those eyedroppers, no leaching from them was to be expected.

This study contains several biases. From one molecule to another, several parameters and characteristics varied at the same time: lipophilicity, shape and molecular weight of the molecule, concentration and viscosity of solutions, different excipients, and different rates of drop release. We chose lipophilicity as a scale because it is known to have an impact on the capacity of a molecule to adsorb on silicone (16). Yet, even if there is evidence in our study of its impact, as only the most lipophilic drugs where subject a loss of concentration, it is difficult to precisely attribute the cause of the observed variations to one or another of those parameters. The type of sterilization might also have an influence but was not taken into account in this study, as the eyedroppers were sterilized by ethylene oxide and gamma irradiation respectively for LDPE and LDPE-Si eyedroppers.

Finally, it is to be pointed out that all marketed drugs tested were at least 1 year ahead from their expiry date, and cyclosporine solutions were prepared the day before the study started. A repetition of this study could be performed nearer to the expiry date to get the worst case results.

Conclusion

This work is in favor of an absence of significant interaction between silicone and the studied timolol and norfloxacin formulations, so they may be used with LDPE-Si eyedroppers without incurring any active substance loss in delivered eyedrops. Rifamycin may be used with LDPE-Si eyedroppers, but for a smaller conservation time, further studies should be performed to assess the occurrence of drug sorptions. In the same way, cyclosporine solutions should be further studied before any definite conclusion is reached. Our study also gives strong indications that Latanoprost is an active substance at risk of sorption, and shouldn’t be used with LDPE-Si eyedroppers without a more thorough evaluation.

Abbreviations

API

Active pharmaceutical ingredient

EMA

European Medicine Agency

HPLC

High pressure liquid chromatography

ICH

International Conference of Harmonization

LDPE

Low density polyethylene

PES

Polyethersulfone

RC

Reference concentration

RSD

Relative standard deviation

Si

Silicone

UV

Ultraviolet

Vis

Visible