June 2009
Volume 50, Issue 6
Free
Cornea  |   June 2009
Liposome Assay for Evaluating Ocular Toxicity of Surfactants
Author Affiliations
  • Yash Kapoor
    From the Department of Chemical Engineering, University of Florida, Gainesville, Florida.
  • Brett A. Howell
    From the Department of Chemical Engineering, University of Florida, Gainesville, Florida.
  • Anuj Chauhan
    From the Department of Chemical Engineering, University of Florida, Gainesville, Florida.
Investigative Ophthalmology & Visual Science June 2009, Vol.50, 2727-2735. doi:10.1167/iovs.08-2980
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Yash Kapoor, Brett A. Howell, Anuj Chauhan; Liposome Assay for Evaluating Ocular Toxicity of Surfactants. Invest. Ophthalmol. Vis. Sci. 2009;50(6):2727-2735. doi: 10.1167/iovs.08-2980.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

purpose. The ocular toxicity of various compounds is typically determined by the Draize eye test, which has been criticized in the past for its lack of reproducibility and the cruelty associated with harsh testing conditions for animals. In this study, a liposome-based assay was developed for estimating ocular toxicity of surfactants.

methods. The release of calcein dye from liposomes induced by interactions with surfactants was studied and correlated to Draize eye scores. First, the liposome assay was conducted for various surfactants at identical concentrations and correlated to the Draize scores. Next, mechanistic and geometric considerations were used to determine the appropriate surfactant concentration that should be used in the liposome assay.

results. Correlations between the percentage of dye released and the Draize scores were drastically improved after surfactant concentrations were chosen based on CMC/200, where CMC is the critical micelle concentration of the surfactants. With this choice of surfactant concentration, excellent correlations were obtained with Draize scores from three separate sources and two different ocular surfactant loadings (Pearson = 0.99, 0.82, 0.78, and 0.74; Spearman = 0.94, 0.79, 0.79, and 0.85). Subsequently, the ocular toxicities of six nonionic surfactants, Brij 700, -58, -56, -78, -97, and -98 were shown to be minimal based on the proposed correlations.

conclusions. The modified liposome assay developed in this study could be used in conjunction with other in vitro assays to obtain initial estimates for ocular toxicities and thus minimize the need for the Draize test.

The ocular toxicity of various compounds is typically determined by the Draize eye test. The test has been criticized for its lack of reproducibility and the cruelty associated with harsh testing conditions for animals. 1 2 Alternatives to this test have been proposed, and the suitability of these assays has been explored by attempting to correlate the data from these assays with the Draize scores. A good correlation would imply that the in vitro assay is suitable for predicting the Draize score and thus can be used for predicting ocular toxicity. A varied level of success has been obtained by each newly proposed method but a widely accepted model to assess toxicity in vivo has not yet been found. Vian et al. 3 explored three in vitro assays and showed relatively poor correlations with Draize scores. Matsukawa et al. 4 reported mixed results for the EYTEX test (Ropak Labs, Irvine, CA), clearly showing several weaknesses of the test. The use of red blood cells to measure ocular toxicity has been touted as quick, inexpensive, and effective. 5 The correlation between this test and the Draize eye score has been found to be poor, although the authors claim that the test could effectively verify the toxicity of chemicals with Draize scores greater than 50. 6 Okahata and Ebato 7 used a lipid-coated quartz microbalance to correlate partition coefficients of surfactants between lipid bilayers and distilled water with Draize scores and found excellent correlations. Perhaps the most successful type of test developed thus far has been those using cell cultures to assess the cell permeability of the compounds in question. Cottin and Zanvit 8 reported successful correlation results from such a test. Some researchers have used tissue engineering to mimic the corneal epithelium, and these mimics can also be used effectively for exploring the toxicity of various compounds. 9 10 11 Although some success has been obtained with these assays, a simple, quantitative, and robust test to replace in vivo testing is still elusive. 12 In fact, it is believed that “replacing animal ocular surface testing will require an array of in vitro procedures.” 1  
In another in vitro method of assessing ocular toxicity, liposomes are used to mimic the corneal membranes and dye leakage is measured from the core of the liposomes on exposure to the test substance that is being examined for toxicity. The advantages of using liposome leakage to assess ocular toxicity include low cost and the ability to evaluate many compounds rapidly. In addition, the test is quantitative, is likely not to vary significantly across users and laboratories, and requires relatively simple and inexpensive equipment and supplies. The test is based on the idea that the permeation of a test substance through lipid bilayers is the root cause of ocular toxicity, with toxicity being caused by the leakage of cellular components, which increases substantially on binding of the test substance to the bilayer. The liposome-based assay is designed so that the lipid composition of the bilayers imitates the composition of corneal epithelial cells. The test measures the leakage of fluorescent dye from the liposome core on interaction with a test substance. It is noted that even though we can choose the lipid composition to mimic the cornea, there will still be significant differences between the liposomes and the corneal epithelium due to lack of proteins in liposomes, which can lead to differences between the in vitro and in vivo response. The maximum score of the Draize eye test is 110, with 80 of 110 points coming from the corneal damage, suggesting that the assessment of corneal toxicity should be the main focus of an in vitro alternative. This fact first inspired researchers to test liposomes as a possible means of assessing the ocular toxicity of surfactants. 13 14 Since that time, a few others have examined liposome leakage as well. 7 15 Several studies performed with the liposome assay have focused on surfactant toxicity, and reasonable correlations to in vivo data were obtained in some cases, although a few gross outliers were typically present. 
In this article, we propose that the lack of good correlation in some studies between the liposome-based assay and the Draize test is due to neglect of mechanistic issues, and that a better correlation can be obtained by designing the liposome assay with consideration of mechanistic factors. Specifically, while attempting to correlate results from the Draize test to an increase in liposome permeability on exposure to surfactants, most researchers evaluated the liposome permeability at a fixed surfactant concentration or at a concentration that induced 50% dye leakage, and the concentrations were significantly below the concentration used in the Draize test. Herein, we first show that the liposome permeability in the presence of surfactants at a fixed surfactant concentration does not correlate well with the Draize scores. We then show that the correlations are significantly improved when the chosen surfactant concentration in the liposome assay is CMC/200, where CMC is the critical micelle concentration, which varies significantly across the surfactants explored in this study. We also show that the rationale for this choice of surfactant concentration in the liposome assay is based on mechanistic considerations. Finally, we use the liposome assay that we developed to determine the ocular toxicity of several Brij surfactants for which available ocular toxicity data were very limited or nonexistent. 
Materials and Methods
Methanol, chloroform, Dulbecco’s phosphate-buffered saline (PBS) without calcium chloride and magnesium chloride, Sephadex G-50 (fine), cholesterol (CH), sodium dodecyl sulfate (SDS), polyoxyethylene sorbitan monolaurate (Tween 20), polyoxyethylene sorbitan monooleate (Tween 80), hexadecyltrimethylammonium bromide (CTAB), cetyl pyridinium chloride (CPC), benzalkonium chloride (BKC), Brij 56, -98, -76, -78, -97, and -700 were purchased from Sigma Aldrich (St. Louis, MO). Glass Whatman GF/B microfiber filters, calcein dye (fluorexon), polyoxyethylene sorbitan monopalmitate (Tween 40), myristyltrimethylammonium bromide (MTAB), and Triton X-100 were purchased from Fisher Scientific (Pittsburgh, PA). Octadecyltrimethylammonium bromide (OTAB) was purchased from K & K Laboratories (Carlsbad, CA). The lipids 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), dissolved in chloroform; 1,2-dimyristoyl-sn-glycero-3-[phospho-rac-(1-glycerol)] (sodium salt; DMPG), in powder form; and 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC), in powder form, as well as a kit for liposome preparation (Mini-Extruder), were purchased from Avanti Polar Lipids, Inc. (Alabaster, AL). 
Liposome Preparation for Calcein Leakage Studies
Liposomes composed of a molar ratio of 8:6:1.5:1.5 of DMPC:DOPE:DMPG:CH encapsulating an aqueous 100 mM calcein dye solution were prepared via a combination of mixing, sonication, and extrusion. The ratio of lipids was chosen based on the composition of the corneal epithelium. 15 Lipids were combined in their respective molar ratios and then dissolved in a 9:1 mixture (by volume) of chloroform/methanol such that a 20-mg/mL concentration of lipids was obtained. The organic solvent was then evaporated under a stream of nitrogen. After an even and uniformly dried lipid film was obtained, the dried lipid layer was hydrated with 100 mM calcein dye dissolved in PBS such that the lipid concentration was 25 mg lipid/mL. The lipid suspension was then mixed in a vortex mixer (Fisher Scientific) for 1 to 2 minutes, followed by bath sonication for 20 minutes. The vesicles were gently stirred overnight at 30°C for approximately 20 hours, due to the increase in entrapped aqueous volume of liposomes with increased stir times. 16 After stirring, the liposome solution was extruded through a 100-nm membrane 15 times. To remove excess dye from the bulk, the liposome solution was passed through a mini-column of Sephadex G-50 (fine), with the centrifugation method used to ensure that a large portion of the lipids added to the Sephadex bed were recovered. 17 The resulting liposome solution was diluted by a factor of 1001 based on the observation that the calcein release at the subsequent concentration fell in the linear detection regimen. 
Liposome Leakage Studies
Surfactant-induced calcein leakage from liposomes was used to mimic surfactant permeation of corneal epithelial cells in vivo. The 100-mM calcein solution inside the liposomes was in the quenched state, as is the case with highly concentrated calcein and carboxyfluorescein, so that only the diluted dye that leaked into the bulk medium gave a signal. 17 The baseline fluorescence of the liposome solution before leakage was first measured with a fluorometer (Quantech Digital Filter; Thermo-Fisher Scientific, Waltham, MA) with excitation and emission filters at 490 and 515 nm, respectively. A concentrated surfactant solution was then added such that the final surfactant concentration was either 1 μg/mL or the critical micelle concentration (CMC) of the surfactant divided by 200 (discussed later). The liposome solution was kept at room temperature in the intervals between fluorescent measurements. To compute the percent release, the following formula was used:  
\[\%\ \mathrm{Release}{=}\ \frac{F_{t}{-}F_{0}}{F_{\mathrm{total}}{-}F_{0}}{\times}100\]
where F t was the fluorescence measurement at time t (10 minutes), F 0 was the fluorescence at time 0, and F total was the total calcein released, which was determined by breaking the liposomes with 100 μL of 20% (vol/vol) Triton X-100. Corrections were made to account for the dilution on addition of the surfactant and Triton X-100 solutions. All release experiments were performed at least twice. 
Draize Scores
Draize scores for 10 of the surfactants tested (SDS, Tween 20, Tween 40, Tween 80, Triton X-100, MTAB, CTAB, OTAB, CPC, and BKC) were obtained from three independently published studies. 4 18 19 Kennah et al. 18 published 24-hour average Draize scores for multiple concentrations of surfactants, enabling two correlations from their publication. Their scores for concentrations of 10% and 1% (vol%) were used. For SDS, those authors reported Draize scores for 30%, 15%, and 3% surfactant concentrations, and their data were fitted to a calibration curve to compute the corresponding Draize scores at 10% and 1%. Tachon et al. 19 reported the maximum Draize score after 1 hour or 24 hours at 10% wt/vol surfactant concentrations. Finally, Matsukawa et al. 4 reported both the maximum average scores and the 24-hour average scores for 10% solutions; the latter was used for consistency. Table 1shows the surfactants used for the in vitro/in vivo correlations and their corresponding Draize scores from each publication. Note that not all surfactants were studied in each published work. 
Data Analysis
Draize scores and percentage of calcein release from liposomes after 10 minutes were correlated by using Pearson’s correlation coefficient (r p) and Spearman’s rank correlation coefficient (r s). Both coefficients were computed for comparisons with either in vitro method, as both Pearson’s 3 4 6 7 8 13 18 20 and Spearman’s 3 6 8 19 have been reported for correlations between in vitro methods and Draize data. The prediction equations calculated from Draize data at 10% from Kennah et al., 18 Tachon et al., 19 and Matsukawa et al. 4 were compared with 95% confidence intervals used for mean Draize scores as a function of percentage calcein leakage (JMP software; SAS, Cary, NC). The predicted Draize scores from the correlations were used to assign each surfactant to one of three categories based on the following scale: mild/moderate (0–25), irritant (>25–50), and severe (>50–110). Several different scales for toxicity based on Draize data have been used in the literature. 3 4 8 18 The scale used in the present study was based on the Texaco single-digit toxicity classification system, 21 although we have modified the scale so that only three classifications of toxicity are specified. The toxicity classifications based on actual Draize data and predicted Draize scores from liposome leakage were then compared, to determine the qualitative rate of correspondence. 
Results
Draize Score/Leakage Correlations at a Constant Test Concentration
In some of the prior studies, researchers measured the leakage of dyes from liposomes after exposure to surfactants at a fixed concentration and then attempted to correlate the leakage with the Draize score. In this section, we report results from a similar study in which the bulk concentration of surfactant in the liposome solution was fixed at 1 μg/mL, which was below the CMC value of all the surfactants tested. It is a common belief that the Draize eye score, if reported from different laboratories, can have statistically significant differences. 2 It is thus perhaps unreasonable to average Draize scores for a given surfactant from studies conducted in different laboratories. To avoid the impact of these differences in our study, we chose three publications with Draize eye scores available for most of the surfactants that we used in our study (Table 1)and we report three separate correlations for the liposome assay and the Draize scores from the three studies. The confidence intervals of the three separate correlations are later compared to determine whether there are statistical differences between the predictions based on the three correlations. In Figure 1 , we plotted the Draize scores from different studies versus the percentage dye leakage after 10 minutes at fixed surfactant concentrations. We used Draize scores obtained at two different concentrations from the publication of Kennah et al. 18 to obtain the two correlations. The correlation coefficients (Pearson and Spearman) for the fits at fixed concentrations are reported in Table 2 . As expected, the correlation coefficients for all cases are not very promising, and there are a few gross outliers. 
Draize Score/Leakage Correlations at Adjusted Concentrations
As shown in the previous section, the liposome assay does not correlate well with the Draize scores for surfactants if the surfactant concentration in the liposome leakage studies is kept fixed. In this section, we report the results for leakage studies from liposomes with the surfactant concentration kept at CMC/200. Specifically, we correlate the Draize data with the percentage dye leakage from liposomes after 10 minutes when the concentration of the surfactant introduced in the solution was adjusted to CMC/200. Table 3shows the CMCs taken from the literature 22 23 and the concentrations used for the test. Figure 2shows the correlations between the reported Draize scores and the percentage dye leakage after 10 minutes. Consequently, the following correlation equations for various sets of Draize scores obtained via different in vivo concentrations and from different groups were produced: 
Kennah et al., 18 concentration 1% vol/vol:  
\[\mathrm{Draize\ score}{=}12.57\ {\cdot}\ \mathrm{ln}(\%\ \mathrm{Leakage}){+}3.76\]
 
Kennah et al., concentration 10% vol/vol:  
\[\mathrm{Draize\ score}{=}16.19\ {\cdot}\ \mathrm{ln}(\%\ \mathrm{Leakage}){+}34.83\]
 
Matsukawa et al., 4 concentration 10% wt/vol:  
\[\mathrm{Draize\ score}{=}18.18\ {\cdot}\ \mathrm{ln}(\%\ \mathrm{Leakage}){+}21.64\]
 
Tachon et al., 19 concentration 10% wt/vol:  
\[\mathrm{Draize\ score}{=}13.75\ {\cdot}\ \mathrm{ln}(\%\ \mathrm{Leakage}){+}22.35\]
 
Both Spearman and Pearson correlation coefficients in Table 2show a significant improvement over the values obtained by comparing the dye release at fixed surfactant concentrations. Because the eventual goal of this assay is to categorize surfactants into groups based on toxicity, it is useful to determine the rate of correspondence between the qualitative classifications assigned according to Draize data and liposome leakage (see data analysis). Clearly, the number of surfactants correctly identified into one of three irritancy categories has improved in most cases (Table 2) . These observations are confirmation that the liposome assay is a good predictor of the Draize scores and consequently ocular toxicity, when the surfactant concentration in the assay is chosen to be CMC/200. 
We also evaluated the 95% confidence intervals for the Draize score predictions as a function of percentage of dye leakage from liposomes after 10 minutes for 10% surfactant solutions. The confidence intervals were evaluated for the three correlations given in equations 2to 3 4 , and the results are shown in Figure 3 . The three confidence intervals overlap, which is encouraging, as it suggests that any of the correlations could be used to predict Draize scores. 
Discussion
Mechanism of Surfactant Toxicity
Surfactants have been shown to cause toxicity by penetrating the epithelial cell membrane, causing damage to the ocular tissue. Jester et al. 24 suggested that the degree of ocular irritation caused by surfactants depends on the initial area and depth of injury, suggesting that the area and volume both determine the toxicity levels of surfactants. Tachon et al. 19 pointed out that damaged cells release lysosomal enzymes, histamine, and inflammatory mediators and suggested that the ability to permeate cells was a major sign of toxicity. Okahata and Ebato 7 claimed that eye irritancy is due to the penetration of surfactant molecules into the lipid bilayer as they correlated Draize scores to the partition coefficients of surfactants. They did not find a significant correlation between the hydrophilic–lipophilic balance (HLB) and the Draize score and concluded that mere lipophilicity arguments cannot determine the irritancy of a surfactant. Other properties such as electrostatic interactions, steric effects, and charge density significantly contribute to the interaction between the surfactant and the corneal epithelium. The liposomes used in our study have been designed to mimic the lipid composition of the corneal epithelium. 15 Based on this, an increase in the permeability of liposomes in the presence of surfactants should directly correlate with eye irritancy. 
The mechanism of surfactant toxicity is depicted in Figure 4 . In some ocular formulations, surfactants are introduced at concentrations above their CMCs, and the free concentration in the tears is the CMC, whereas the remaining surfactant molecules form micelles. The adsorption of surfactants to the lipid bilayers of the ocular epithelia occurs mainly by penetration of the surfactant monomers into the bilayers. This point was specifically addressed by Okahata and Ebato, 7 who measured the uptake of surfactants onto a lipid-coated quartz microbalance and showed good correlation with Draize scores. Absorption became saturated for nonionic and cationic surfactants above the CMC, pointing to a lack of direct micelle interaction with the membrane. Furthermore, this point was supported by the work of Hall-Manning et al., 25 who found skin irritancy to be related to the CMC, rather than the absolute concentration tested. Thus, only the free surfactant, which is present at the CMC, should interact with the corneal epithelium. As the surfactant molecule starts to penetrate the epithelial bilayer, micelles on the ocular surface break, and the concentration of free surfactant on the ocular surface is constantly maintained at the CMC until all the surfactant micelles have either drained or have been broken. As the concentration of surfactant introduced on the ocular surface increases, the total time for drainage of the surfactant from the ocular surface will also vary, leading to an increase in toxicity. Once the concentration of the surfactant introduced exceeds a critical value where the time of drainage is longer than the time at which the toxicity is assessed, ocular toxicity should saturate. These points are supported by the Draize data given by Kennah et al. 18 and Matsukawa et al., 4 which show increases in Draize scores well above the CMCs of surfactants and then subsequent leveling of the scores. Thus, residence time is a key factor to consider when comparing one Draize score to another. This point is quantitatively addressed in a later section. While most ocular formulations and Draize test solutions contain surfactant at concentrations above the CMC, the liposome assays are typically conducted at concentrations below the CMC. This leads to a contradiction that is evident by the presence of outliers in the correlations between the Draize scores and the percentage leakage from the liposomes in 10 minutes shown in Figure 1and in similar studies published previously. We propose that this inconsistency could be eliminated by conducting the liposome assay for various surfactants at concentrations that are a fixed fraction of the CMC. In the following discussion, we present a rationale for choosing this fraction (i.e., the ratio of the CMC that we propose should be used in the liposome assay). 
The degree of surfactant penetration inside the ocular surface should also be proportional to the surface area available for the surfactant to diffuse into the epithelium. Similarly, when liposomes are used as an alternative, the amount of surfactant penetrating the lipid bilayer should be directly proportional to the surface area of the liposomes. This clearly suggests that experiments have to be designed with differences between liposomes and the corneal epithelium taken into consideration. To illustrate the importance of differences between liposome and corneal geometry, it is instructive to consider a mass balance on substances such as lysosomal enzymes, histamine, and inflammatory mediators that begin to leak from inside the corneal cells due to the toxic effects of surfactant penetration into the bilayer of the corneal epithelium. The mass balance yields  
\[V_{\mathrm{Cornea}}\ \frac{dC_{S}}{dt}\ {=}\ {-}K_{\mathrm{Perm}}A_{\mathrm{Cornea}}C_{\mathrm{S}}\]
where V Cornea is the cellular volume of the corneal epithelium, K Perm is the permeability of the corneal epithelium to the species that leaks out, A Cornea is the corneal area available for penetration, and C S is the concentration of the species of interest inside the corneal cells. Similarly, a mass balance on a test component such as a dye present inside the liposomes can be obtained by  
\[V_{\mathrm{Liposome}}\ \frac{dC_{\mathrm{Dye}}}{dt}\ {=}\ {-}K_{\mathrm{Perm,Lipo}}\ A_{\mathrm{Liposome}}C_{\mathrm{Dye}}\]
where V Liposome is the volume of the liposomes, K Perm,Lipo is the permeability of the liposomes, A Liposome is the liposome surface area available for penetration, and C Dy e is the concentration of the species that leaks out. The presence of surfactants is manifested in increased permeabilities for both the cornea and the liposomes. Based on these equations, the time scale for the leakage of the molecules is KA/V. The surface area-to-volume (A/V) ratio is much larger for liposomes due to their small size. Thus, if the surfactant concentration for the liposome assay is chosen to be the same as that in the Draize test, or even the CMC, the time scale for dye leakage will be extremely small, and so the percentage leakage will be very large unless the measurements are done at extremely short intervals. Since short-time measurements are prone to artifacts due to issues such as mixing, it is more appropriate to ensure that the time scale for the leakage in the liposomes is comparable to that in the corneal cells. Since the time scale is KA/V, the higher values of A/V can be compensated for by a lower K for the liposome assay. The permeability K is related to the amount of surfactant that binds to the liposomes, and so the value of K in the liposome assay can be controlled by controlling the surfactant concentration in the assay. Based on these arguments, the following equation can be used to evaluate the effective concentration that should be tested in the liposome assay to correctly predict the irritancy of surfactants in vivo  
\[C_{\mathrm{TEST}}{=}\ \frac{A_{\mathrm{Cornea}}V_{\mathrm{Liposome}}}{V_{\mathrm{Cornea}}A_{\mathrm{Liposome}}}\ \mathrm{CMC}\]
 
Equation 8implicitly assumes that permeability is linearly related to the surfactant concentration and also that the appropriate concentration that controls binding in the Draize test is not the total concentration but the surfactant monomer concentration, which equals the CMC. The average surface area-to-volume ratio for liposomes and the cornea were estimated as 63,132 and 345, respectively. These ratios are based on a surface area of 62 Å 2 occupied by a single lipid, 17 26 27 a computed liposome volume 28 based on a liposome radius of 55 nm, 29 a bilayer thickness of 35 to 40 Å, 30 and cell data taken from Maric et al. 31 and Farinas and Verkman. 32 The resulting surface area-to-volume ratios for liposomes and the cornea, respectively, is ∼183. From this value, we can evaluate the test concentration for different surfactant systems based on their CMC’s. Since the surface area-to-volume ratio is a rough calculation and the reported CMC values for surfactants can vary, we rounded the correction factor up to 200. Thus, we divided each CMC value by 200 to obtain the most physiologically relevant test concentration possible for our liposomal system. This surface area to volume ratio correction has not been made in previous reports; it is crucial to ensuring that the two systems allow for true a comparison. 7 13 14 15 It is important to point out that our correction factor of 200 depends heavily on our liposomes’ having mean diameters of ∼110 nm, and new correction factors must be computed when working with liposomes of dramatically different sizes. 29  
Comparison of Liposome Assay with Other In Vitro Assays
Matsukawa et al. 4 used the EYTEX test as an in vitro model to predict ocular toxicity. They argued that since protein denaturation is one of the most important factors in determining the extent of ocular irritation, it could be used as an alternative. The overall correlation coefficient between Draize scores and the EYTEX test (Ropak Labs, Irvine, CA) was reported as 0.313, which is poor compared to the values in this study (Table 4) . Also, with their technique, they were unable to predict toxicity for cationic surfactants, whereas with the liposome leakage technique we can accurately predict the irritancy of cationic surfactants. Vian et al. 3 compared three different in vitro techniques to determine the ocular toxicity of various surfactants. They used neutral red uptake assay, the MTT tetrazolium salt assay, and the total protein content assay for correlating in vivo Draize data and concluded that among the three techniques, the result of the neutral red uptake assay was best correlated with the Draize score. Tachon et al. 19 used cell mortality and inhibition of cell growth as assays to assess surfactant toxicity. They obtained reasonable agreement with the Draize test score and concluded that penetration of surfactant in the cell lines was responsible for cell damage. Cottin and Zanvit 8 correlated toxicity to cell leakage using a fluorescent dye. They measured the amount of surfactant needed to induce 20% leakage and related it to Draize scores via a nonlinear relationship. They found strong correlations between in vivo and in vitro assays and suggested that this method could be another addition for an in vitro alternative to the Draize eye test. Kennah et al. 18 have suggested in the past that there is a need for another in vivo alternative to the Draize test due to poor reproducibility and the subjective assessment which differs from one researcher to other. They sought to accomplish this by measuring the corneal thickness of rabbit eyes before and after exposure. They found a linear relationship between their test and the Draize score with a reasonable correlation. All the relevant correlations, including our work, are compared in Table 4 . As can be seen from the results, liposome leakage studies have the potential to be used as one of the in vitro alternatives for early predictions of the irritancy of surfactants and possibly other substances as well. 
Prediction of Ocular Toxicity for Nonionic Surfactants
Surfactants have been actively explored in literature as potential permeability enhancers to increase the permeability of some common ocular drugs. 20 Recently, it was proposed that some nonionic surfactants could be incorporated inside commercial contact lenses to attenuate drug release. 33 We predicted the ocular irritancy levels of similar surfactants by performing identical experiments as before, where the test concentration of surfactant was adjusted according to equation 8 . Leakage from liposomes was evaluated as discussed in the previous section and Draize scores were predicted from the correlations (equations 2-5) 3 4 5obtained by fitting Draize scores from various sources. In Table 5 , Draize score predictions for six nonionic surfactants are presented. Since all the predictions were made for a surfactant concentration of 10% or 1% (wt/vol) on the ocular surface, the predicted Draize scores should correspond to the same tested concentrations. As expected, the predicted Draize scores are higher for the higher concentration with Brij 78, -700, -56, and -58 showing negligible toxicity for the 1% concentration correlation. 
The mechanism of toxicity should be similar for a particular class of surfactants. For nonionic surfactants, the determining factor in toxicity should be the hydrophobic interaction of the surfactant with the lipid bilayer. Hydrophobic interaction of stearyl and cetyl chains should be stronger than that of oleyl chains because of the double bond of the oleyl chain, which makes oleyl surfactants more hydrophilic and suggests that Brij 78 (C18H37(OCH2OCH2)20OH), Brij 700 (C18H37(OCH2OCH2)100OH), Brij 56 (C16H33(OCH2OCH2)10OH), and Brij 58 (C16H33(OCH2OCH2)20OH) should be more toxic than Brij 97 (C18H35(OCH2OCH2)10OH) and Brij 98 (C18H35(OCH2OCH2)20OH). On the other hand, the CMCs of Brij 97 and -98 are much higher than that of the other surfactants (Table 3) . Thus, when present at concentrations higher than the CMC, the number of monomers for Brij 98 and -97 surfactants would be much larger than the other surfactants, leading to more toxicity. This explains the higher Draize score for Brij 97 and -98 compared with other surfactants at both 10% and 1% surfactant loading. Because no Draize data are available for these systems, we can at best speculate that the oleyl series of surfactants should be more toxic than the stearyl and cetyl groups of surfactants if administered on the ocular surface above their respective CMCs, due solely to the larger number of monomers available for epithelial cell penetration for the oleyl series. 
Draize scores for Brij 78 have been reported previously, and the authors found that the Draize score at 1% wt/vol surfactant loading was 2. 20 This value is in agreement with the value predicted by the correlations developed in the present study, as shown in Table 5for this surfactant at similar concentrations. 
Model for Micelle Depletion from the Ocular Surface
As explained in the Mechanism of Surfactant Toxicity section, several studies suggest that only the surfactant monomer directly interacts with the ocular epithelia. Since the concentration of the monomers is fixed at CMC irrespective of the total surfactant loading in the eye drops instilled in the tear film, one may conclude that the Draize scores should be independent of the total surfactant concentration. Experiments, however, have shown that as the concentration of the test material on the ocular surface is increased to concentrations above CMC, the corresponding Draize scores also increases, whereas at really higher concentrations, there is a saturation of Draize scores. 18 This apparent paradox can be resolved by noting that the concentration of the surfactant in the eyes decreases over time due to tear drainage and transport through the epithelia. The residence time of the surfactant in the tear film increases with increasing loading, and thus even though the surfactant binding rates to the epithelia saturate at the CMC, the total amount taken up by the epithelia continue to increase with increasing surfactant concentration, because of an increase in the time for which the epithelia are exposed to the surfactant. This issue can be explained more clearly by performing a mass balance on the surfactant and tears on the ocular surface. The mass balances yield  
\[\frac{d(VC_{\mathrm{Micelle}}{+}VCMC)}{dt}\ {=}\ {-}q_{\mathrm{Drainage}}(C_{\mathrm{Micelle}}{+}\mathrm{CMC}){-}k_{\mathrm{Cornea}}A_{\mathrm{Cornea}}\mathrm{CMC}{-}k_{\mathrm{Conj}}A_{\mathrm{Conj}}\mathrm{CMC}\]
and the tear balance on the ocular surface can be given by  
\[\frac{dV}{dt}{=}q_{\mathrm{Secretion}}{-}q_{\mathrm{Drainage}}{-}q_{\mathrm{Conjunctiva}}{-}q_{\mathrm{Evaporation}}\]
where C Micelle is the concentration of the surfactant monomer present as micelles on the ocular surface; A Cornea is the area of the cornea; A Conj is the area of the conjunctiva; k Cornea is the permeability of the surfactant in the cornea; k Conj is the permeability of surfactant in the conjunctiva; q Secretion is the rate of tear production; q Drainage is the rate of tear drainage; q Conjunctiva is the rate of tear penetration inside the conjunctiva; q Evaporation is the rate of tear evaporation from the ocular surface; and V is the volume of tears on the ocular surface. 
These coupled equations show that the residence time of the surfactant micelles on the ocular surface should be a function of surfactant concentration. 
It is instructive to consider an extreme case in which the dominant mechanism for surfactant loss from the tear film is canalicular drainage. This is a likely scenario at surfactant concentrations much above CMC. In this case, the time for the surfactant concentration in the tear film to go from the initial concentration C i to CMC is equal to V/q ln C i /CMC. Thus, a higher initial C i leads to a longer duration in which the surfactant concentration in the tear volume is larger than the CMC, and the longer time leads to a larger influx of the surfactant into the cornea, causing higher toxicity. This clearly indicates that the Draize eye score, which should be indicative of the amount of surfactant monomer diffusing inside the corneal epithelium, should increase with increasing surfactant concentrations due to increased residence times. 
Conclusions
Damage to the corneal epithelium can be attributed to the disruption of membrane fluidity due to the penetration of external agents such as surfactants, and the subsequent release of lysosomal enzymes, histamine, and inflammatory mediators. Interactions between surfactants and the corneal surface are governed by their respective CMCs, as micelles are not expected to interact with lipid bilayers. We took this mechanism into account to develop a liposome-based assay for evaluating the ocular toxicity of surfactants. The assay is based on measuring percentage leakage of a dye from the core of the liposomes in a certain period after exposure of liposomes to surfactants at concentration of CMC/200. The specific choice of the concentration is based on mechanistic considerations related to the assumption of negligible interaction between the micelles and the epithelia and on geometric considerations. The dye leakage from the liposomes does not correlate well with the Draize scores if the surfactant concentrations are fixed at any value, but the correlations are significantly better when the surfactant concentrations are chosen to be CMC/200. Although there is good correlation between dye leakage and Draize scores for various surfactants, lack of a complete eye model and failure to mimic the complexity of the corneal epithelium may lead to errors in the proposed approach. The correlations obtained in our study for the liposome assay are better or comparable to other in vitro tests reported in the literature. Thus, the liposome-based assay alone or in conjunction with other in vitro assays can be used to evaluate the initial toxicity of various surfactants and thus minimize the need for the Draize test. In addition to developing the modified liposome-based assay, we also focused on using the assay to predict the ocular toxicity of six nonionic surfactants for which ocular toxicity data were scarce or nonexistent. We propose that Brij 78, -700, -56, and -58 are mildly to moderately comfortable when placed in the eye at concentrations of 10% (wt/vol), whereas Brij 97 and -98 are irritating at similar concentrations. At 1% (wt/vol), all the surfactants examined are likely to be in the mild to moderate category, causing little to no discomfort. 
 
Table 1.
 
Draize Scores Used for In Vitro/In Vivo Correlations
Table 1.
 
Draize Scores Used for In Vitro/In Vivo Correlations
Surfactant Tachon et al. 19 Draize Scores (10%) Matsukawa et al. 4 Draize Scores (10%) Kennah et al. 18 Draize Scores (10%) Kennah et al. 18 Draize Scores (1%)
SDS 37.34 14.7 40.33* 5.83*
Tween 20 5.67 0.01, † 1 0.01
Tween 80 3.83 0.01
Triton × 100 40.33 59 2
CPC, ‡ 52.67 93 84 36
Tween 40 1.5
MTAB 42.66
CTAB 44 39
BKC 78 98 56
OCTAB 56.3
Figure 1.
 
Draize scores versus liposome leakage after 10 minutes induced by surfactants at concentrations of 1 μg/mL and logarithmic correlations for Draize scores from (a) Kennah et al. 18 evaluated at 10% (vol/vol) with r p = 0.38, r s = 0.49; (b) Kennah et al. 18 evaluated at 1% (vol/vol) with r p = 0.74, r s = 0.59; (c) Tachon et al. 19 evaluated at 10% (wt/vol) with r p = 0.26, r s = 0.43; and (d) Matsukawa et al. 4 evaluated at 10% (wt/vol) with r p = 0.74, r s = 0.63.
Figure 1.
 
Draize scores versus liposome leakage after 10 minutes induced by surfactants at concentrations of 1 μg/mL and logarithmic correlations for Draize scores from (a) Kennah et al. 18 evaluated at 10% (vol/vol) with r p = 0.38, r s = 0.49; (b) Kennah et al. 18 evaluated at 1% (vol/vol) with r p = 0.74, r s = 0.59; (c) Tachon et al. 19 evaluated at 10% (wt/vol) with r p = 0.26, r s = 0.43; and (d) Matsukawa et al. 4 evaluated at 10% (wt/vol) with r p = 0.74, r s = 0.63.
Table 2.
 
Correlation Comparisons for Draize Scores and Leakage Experiments Performed at Surfactant Concentrations of 1 μg/mL and CMC/200
Table 2.
 
Correlation Comparisons for Draize Scores and Leakage Experiments Performed at Surfactant Concentrations of 1 μg/mL and CMC/200
Source Pearson (r p) Spearman (r s) Rate of Correspondence*
Kennah et al. 18 , † (10%, 1 μg/mL) 0.38 0.49 50%
Kennah et al., ‡ (10%, CMC/200) 0.82 0.79 60%
Kennah et al., § (1%, 1 μg/mL) 0.74 0.59 100%
Kennah et al., ∥ (1%, CMC/200) 0.99 0.94 100%
Matsukawa et al. 4 , † (10%, 1 μg/mL) 0.74 0.63 71.4%
Matsukawa et al., ‡ (10%, CMC/200) 0.78 0.79 71.4%
Tachon et al. 19 , † (10%, 1 μg/mL) 0.26 0.43 62.5%
Tachon et al., ‡ (10%, CMC/200) 0.74 0.85 87.5%
Table 3.
 
CMC for Surfactants Studied and Subsequent Test Concentrations for Liposome Leakage
Table 3.
 
CMC for Surfactants Studied and Subsequent Test Concentrations for Liposome Leakage
Surfactant Molecular Weight (g/mole) CMC (mM) CMC (μg/mL) Test Concentration, CMC/200 (μg/mL) Source
Triton X-100 647 0.200 129.40 0.647 Roche Applied Science
SDS 288.38 8.200 2364.72 11.824 Rosen 22
Tween 20 1226 0.050 61.30 0.307 Hait and Moulik 23
Tween 40 1283.65 0.023 29.52 0.148 Hait and Moulik 23
Tween 80 1309.68 0.010 13.10 0.065 Hait and Moulik 23
BKC 340 8.800 2992.00 14.960 Rosen 22
OTAB 391.9 0.310 121.49 0.607 Rosen 22
CTAB 364.48 0.980 357.19 1.786 Rosen 22
MTAB 308 3.600 1108.80 5.544 Rosen 22
CPC 339.986 0.900 305.99 1.530 Rosen 22
Brij 58 1120 0.007 7.84 0.039 Hait and Moulik 23
Brij 56 682 0.002 1.36 0.007 Hait and Moulik 23
Brij 97 709 0.400 283.60 1.418 Hait and Moulik 23
Brij 98 1153.54 0.265 305.69 1.528 Hait and Moulik 23
Brij 78 1151.54 0.0057 6.56 0.033 Hait and Moulik 23
Brij 700 4670 0.020 93.40 0.467 Hait and Moulik 23
Figure 2.
 
Draize scores versus liposome leakage after 10 minutes induced by surfactants at CMC/200 and logarithmic correlations for Draize scores from (a) Kennah et al. 18 evaluated at 10% (vol/vol) with r p = 0.82, r s = 0.79; (b) Kennah et al. 18 evaluated at 1% (vol/vol) with r p = 0.99, r s = 0.94; (c) Tachon et al. 19 evaluated at 10% (wt/vol) with r p = 0.74, r s = 0.85; and (d) Matsukawa et al. 4 evaluated at 10% (wt/vol) with r p = 0.78, r s = 0.79.
Figure 2.
 
Draize scores versus liposome leakage after 10 minutes induced by surfactants at CMC/200 and logarithmic correlations for Draize scores from (a) Kennah et al. 18 evaluated at 10% (vol/vol) with r p = 0.82, r s = 0.79; (b) Kennah et al. 18 evaluated at 1% (vol/vol) with r p = 0.99, r s = 0.94; (c) Tachon et al. 19 evaluated at 10% (wt/vol) with r p = 0.74, r s = 0.85; and (d) Matsukawa et al. 4 evaluated at 10% (wt/vol) with r p = 0.78, r s = 0.79.
Figure 3.
 
The 95% confidence intervals for mean Draize scores at 10% ocular loading for surfactants based on logarithmic correlations from percent dye leakage from liposomes after 10 minutes at surfactant CMC/200. Kennah et al. 18 (thick line); Tachon et al. 19 (thin line); Matsukawa et al. 4 (dashed line).
Figure 3.
 
The 95% confidence intervals for mean Draize scores at 10% ocular loading for surfactants based on logarithmic correlations from percent dye leakage from liposomes after 10 minutes at surfactant CMC/200. Kennah et al. 18 (thick line); Tachon et al. 19 (thin line); Matsukawa et al. 4 (dashed line).
Figure 4.
 
Schematic of surfactant-induced toxicity on the corneal surface.
Figure 4.
 
Schematic of surfactant-induced toxicity on the corneal surface.
Table 4.
 
Correlation Comparisons between the Liposome Leakage Method of Assessing Toxicity and Other Published Methods
Table 4.
 
Correlation Comparisons between the Liposome Leakage Method of Assessing Toxicity and Other Published Methods
Method(s) Source Pearson Spearman
EYTEX™ Matsukawa et al. 4 0.2–0.4 N/A
Various cell assays Vian et al. 3 0.48–0.62 0.53–0.64
Corneal thickness Kennah et al. 18 0.86 N/A
Cell growth inhibition Tachon et al. 19 N/A 0.652–0.845
Cell leakage Cottin and Zanvit 8 0.94 0.92
Liposome leakage Current Study 0.74–0.99 0.79–0.94
Table 5.
 
Predicted Draize Scores for 10% and 1% Stock Solutions of Selected Brij Surfactants
Table 5.
 
Predicted Draize Scores for 10% and 1% Stock Solutions of Selected Brij Surfactants
Surfactant Predicted Draize Score (10%)* Irritation Class (10%), † Predicted Draize Score (1%), ‡ Irritation Class (1%), †
Brij 97 36.9 ± 7.3 Irritant 11.0 Mild/Moderate
Brij 98 34.7 ± 7.2 Irritant 9.2 Mild/Moderate
Brij 78 22.1 ± 6.7 Mild/Moderate 0, § , ∥ Mild/Moderate
Brij 700 24.4 ± 6.8 Mild/Moderate 0.8 Mild/Moderate
Brij 56 21.6 ± 6.7 Mild/Moderate 0, § Mild/Moderate
Brij 58 26.4 ± 6.8 Mild/Moderate 2.5 Mild/Moderate
WilhelmusKR. The Draize Eye Test. Surv Ophthalmol. 2001.(456)493–515.
WeilCS, ScalaRA. Study of intra- and interlaboratory variability in the results of rabbit eye and skin irritation tests. Toxicol Appl Pharm. 1971;19:276–360. [CrossRef]
VianL, VincentJ, MaurinJ, FabreI, GirouxJ, CanoJP. Comparison of three in vitro cytotoxicity assays for estimating surfactant ocular irritation. Toxic In Vitro. 1995;9(2)185–190. [CrossRef]
MatsukawaK, MasudaK, KakishimaH, et al. Interlaboratory validation of the in vitro eye irritation tests for cosmetic ingredients. (11) EYTEX. Toxic In Vitro. 1999;13:209–217. [CrossRef]
VinardellMP, MitjansM. Alternative methods for eye and skin irritation tests: an overview. J Pharm Sci. 2008;97(1)46–59. [CrossRef] [PubMed]
OkamotoY, OhkoshiK, ItagakiH, et al. Interlaboratory validation of the in vitro eye irritation tests for cosmetic ingredients. (3) evaluation of the haemolysis test. Toxic In Vitro. 1999;13:115–124. [CrossRef]
OkahataY, EbatoH. Absorption behaviors of surfactant molecules on a lipid-coated quartz-crystal microbalance: an alternative to eye-irritant tests. Anal Chem. 1991;63:203–207. [CrossRef]
CottinM, ZanvitA. Fluorescein leakage test: a useful tool in ocular safety assessment. Toxic In Vitro. 1997;11:399–405. [CrossRef]
FancyM, MungerR, FahimA. Instrument for non-invasive photonic assessments of biological materials. Proc SPIE Int Soc Opt Eng. 2004;5578:278–289.
KlenklerBJ, GriffithM, BecerrilC, West-MaysJA, SheardownH. EGF-grafted PDMS surfaces in artificial cornea applications. Biomaterials. 2005;26:7286–7296. [CrossRef] [PubMed]
LiuW, MerrettK, GriffithM, et al. Recombinant human collagen for tissue engineered corneal substitutes. Biomaterials. 2008;29:1147–1158. [CrossRef] [PubMed]
CurrenRD, HarbellJW. In vitro alternatives for ocular irritation. Environ Health Persp. 1998;106(S2)485–492. [CrossRef]
KatoS, ItagakiH, ChiyodaI, et al. Liposomes as an in vitro model for predicting the eye irritancy of chemicals. Toxic In Vitro. 1988;2(2)125–130. [CrossRef]
SunamotoJ, IwamotoK, ImokawaG, TsuchiyaS. Possible eye-irritant test using polysaccharide-coated liposomes as a corneal epithelium model. Chem Pharm Bull. 1987;35(7)2958–2965. [CrossRef] [PubMed]
Saarinen-SavolainenP, JarvinenT, SuhonenP, UrttiA. Amphiphilic properties of pilocarpine prodrugs. Int J Pharm. 1996;133:171–178. [CrossRef]
SzokaF, PapahadjopoulosD. Comparative properties and methods of preparation of lipid vesicles (liposomes). Ann Rev Biophys Bioeng. 1980;9:467–508. [CrossRef]
ZuidamNJ, VruehR, CrommelinDJA. Characterization of liposomes.TorchilinVP WeissigV eds. Liposomes. 2003; 2nd ed. 31–78.Oxford University Press New York.
KennahHE, II, HignetS, LauxPE, DorkoJD, BarrowCS. An objective procedure for quantitating eye irritation based upon changes of corneal thickness. Fundam Appl Toxicol. 1989;12:258–268. [CrossRef] [PubMed]
TachonP, CotovioJ, DossouKG, PrunierasM. Assessment of surfactant cytotoxicity: comparison with the Draize eye test. Int J Cosmetic Sci. 1989;11:233–243. [CrossRef]
SaettoneMF, ChetoniP, CerbaiR, MazzantiG, BraghiroliL. Evaluation of ocular permeation enhancers: in vitro effects on corneal transport of four β-blockers, and in vitro/in vivo toxic activity. Int J Pharm. 1996;142:103–113. [CrossRef]
DeSousaDJ, RouseAA, SmolonWJ. Statistical consequences of reducing the number of rabbits utilized in eye irritation testing: data on 67 petrochemicals. Toxicol Appl Pharmacol. 1984;76:234–242. [CrossRef] [PubMed]
RosenMJ. Surfactants and Interfacial Phenomena. 2004;105–177.Wiley-Interscience Hoboken NJ.
HaitSK, MoulikSP. Determination of critical micelle concentration (CMC) of nonionic surfactants by donor-acceptor interaction with iodine and correlation of CMC with hydrophile-lipophile balance and other parameters of the surfactants. J Surfactants Deterg. 2001;4(3)303–309. [CrossRef]
JesterJV, LiHF, PetrollWM, et al. Area and depth of surfactant-induced corneal injury correlates with cell death. Invest Ophthalmol Vis Sci. 1998;39(6)922–936. [PubMed]
Hall-ManningTJ, HollandGH, RennieG, et al. Skin irritation potential of mixed surfactant systems. Food Chem Toxicol. 1998;36:233–238. [CrossRef] [PubMed]
KimJH, KimMW. Temperature effect on the transport dynamics of a small molecule through a liposome bilayer. Eur Phys J E. 2007;23:313–317. [CrossRef] [PubMed]
LisLJ, McAlisterM, FullerN, RandRP, ParsegianVA. Interactions between neutral phospholipid bilayer membranes. Biophys J. 1982;37:657–666. [PubMed]
NewRRC. Liposomes. 1990;137.Oxford University Press New York.
HowellB, ChauhanA. Amitriptyline overdose treatment by pegylated anionic liposomes. J Colloid Interf Sci. 2008;324:61–70. [CrossRef]
BalgavyP, DubnickovaM, KucerkaN, KiselevMA, YaradaikinSP, UhrikovaD. Bilayer thickness and lipid interface area in unilamellar extruded 1,2-diacylphosphatidylcholine liposomes: a small-angle neutron scattering study. Biochim Biophys Acta. 2001;1512:40–52. [CrossRef] [PubMed]
MaricK, WiesnerB, LorenzD, KlussmannE, BetzT, RosenthalW. Cell volume kinetics of adherent epithelial cells measured by laser scanning reflection microscopy: determination of water permeability changes of renal principal cells. Biophys J. 2001;80:1783–1790. [CrossRef] [PubMed]
FarinasJ, VerkmanAS. Cell volume and plasma membrane osmotic water permeability in epithelial cell layers measured by interferometry. Biophys J. 1996;71:3511–3522. [CrossRef] [PubMed]
KapoorY, ChauhanA. Drug and surfactant transport in cyclosporine A and Brij 98 laden p-HEMA hydrogels. J Colloid Interf Sci. 2008;322(2)624–633. [CrossRef]
Figure 1.
 
Draize scores versus liposome leakage after 10 minutes induced by surfactants at concentrations of 1 μg/mL and logarithmic correlations for Draize scores from (a) Kennah et al. 18 evaluated at 10% (vol/vol) with r p = 0.38, r s = 0.49; (b) Kennah et al. 18 evaluated at 1% (vol/vol) with r p = 0.74, r s = 0.59; (c) Tachon et al. 19 evaluated at 10% (wt/vol) with r p = 0.26, r s = 0.43; and (d) Matsukawa et al. 4 evaluated at 10% (wt/vol) with r p = 0.74, r s = 0.63.
Figure 1.
 
Draize scores versus liposome leakage after 10 minutes induced by surfactants at concentrations of 1 μg/mL and logarithmic correlations for Draize scores from (a) Kennah et al. 18 evaluated at 10% (vol/vol) with r p = 0.38, r s = 0.49; (b) Kennah et al. 18 evaluated at 1% (vol/vol) with r p = 0.74, r s = 0.59; (c) Tachon et al. 19 evaluated at 10% (wt/vol) with r p = 0.26, r s = 0.43; and (d) Matsukawa et al. 4 evaluated at 10% (wt/vol) with r p = 0.74, r s = 0.63.
Figure 2.
 
Draize scores versus liposome leakage after 10 minutes induced by surfactants at CMC/200 and logarithmic correlations for Draize scores from (a) Kennah et al. 18 evaluated at 10% (vol/vol) with r p = 0.82, r s = 0.79; (b) Kennah et al. 18 evaluated at 1% (vol/vol) with r p = 0.99, r s = 0.94; (c) Tachon et al. 19 evaluated at 10% (wt/vol) with r p = 0.74, r s = 0.85; and (d) Matsukawa et al. 4 evaluated at 10% (wt/vol) with r p = 0.78, r s = 0.79.
Figure 2.
 
Draize scores versus liposome leakage after 10 minutes induced by surfactants at CMC/200 and logarithmic correlations for Draize scores from (a) Kennah et al. 18 evaluated at 10% (vol/vol) with r p = 0.82, r s = 0.79; (b) Kennah et al. 18 evaluated at 1% (vol/vol) with r p = 0.99, r s = 0.94; (c) Tachon et al. 19 evaluated at 10% (wt/vol) with r p = 0.74, r s = 0.85; and (d) Matsukawa et al. 4 evaluated at 10% (wt/vol) with r p = 0.78, r s = 0.79.
Figure 3.
 
The 95% confidence intervals for mean Draize scores at 10% ocular loading for surfactants based on logarithmic correlations from percent dye leakage from liposomes after 10 minutes at surfactant CMC/200. Kennah et al. 18 (thick line); Tachon et al. 19 (thin line); Matsukawa et al. 4 (dashed line).
Figure 3.
 
The 95% confidence intervals for mean Draize scores at 10% ocular loading for surfactants based on logarithmic correlations from percent dye leakage from liposomes after 10 minutes at surfactant CMC/200. Kennah et al. 18 (thick line); Tachon et al. 19 (thin line); Matsukawa et al. 4 (dashed line).
Figure 4.
 
Schematic of surfactant-induced toxicity on the corneal surface.
Figure 4.
 
Schematic of surfactant-induced toxicity on the corneal surface.
Table 1.
 
Draize Scores Used for In Vitro/In Vivo Correlations
Table 1.
 
Draize Scores Used for In Vitro/In Vivo Correlations
Surfactant Tachon et al. 19 Draize Scores (10%) Matsukawa et al. 4 Draize Scores (10%) Kennah et al. 18 Draize Scores (10%) Kennah et al. 18 Draize Scores (1%)
SDS 37.34 14.7 40.33* 5.83*
Tween 20 5.67 0.01, † 1 0.01
Tween 80 3.83 0.01
Triton × 100 40.33 59 2
CPC, ‡ 52.67 93 84 36
Tween 40 1.5
MTAB 42.66
CTAB 44 39
BKC 78 98 56
OCTAB 56.3
Table 2.
 
Correlation Comparisons for Draize Scores and Leakage Experiments Performed at Surfactant Concentrations of 1 μg/mL and CMC/200
Table 2.
 
Correlation Comparisons for Draize Scores and Leakage Experiments Performed at Surfactant Concentrations of 1 μg/mL and CMC/200
Source Pearson (r p) Spearman (r s) Rate of Correspondence*
Kennah et al. 18 , † (10%, 1 μg/mL) 0.38 0.49 50%
Kennah et al., ‡ (10%, CMC/200) 0.82 0.79 60%
Kennah et al., § (1%, 1 μg/mL) 0.74 0.59 100%
Kennah et al., ∥ (1%, CMC/200) 0.99 0.94 100%
Matsukawa et al. 4 , † (10%, 1 μg/mL) 0.74 0.63 71.4%
Matsukawa et al., ‡ (10%, CMC/200) 0.78 0.79 71.4%
Tachon et al. 19 , † (10%, 1 μg/mL) 0.26 0.43 62.5%
Tachon et al., ‡ (10%, CMC/200) 0.74 0.85 87.5%
Table 3.
 
CMC for Surfactants Studied and Subsequent Test Concentrations for Liposome Leakage
Table 3.
 
CMC for Surfactants Studied and Subsequent Test Concentrations for Liposome Leakage
Surfactant Molecular Weight (g/mole) CMC (mM) CMC (μg/mL) Test Concentration, CMC/200 (μg/mL) Source
Triton X-100 647 0.200 129.40 0.647 Roche Applied Science
SDS 288.38 8.200 2364.72 11.824 Rosen 22
Tween 20 1226 0.050 61.30 0.307 Hait and Moulik 23
Tween 40 1283.65 0.023 29.52 0.148 Hait and Moulik 23
Tween 80 1309.68 0.010 13.10 0.065 Hait and Moulik 23
BKC 340 8.800 2992.00 14.960 Rosen 22
OTAB 391.9 0.310 121.49 0.607 Rosen 22
CTAB 364.48 0.980 357.19 1.786 Rosen 22
MTAB 308 3.600 1108.80 5.544 Rosen 22
CPC 339.986 0.900 305.99 1.530 Rosen 22
Brij 58 1120 0.007 7.84 0.039 Hait and Moulik 23
Brij 56 682 0.002 1.36 0.007 Hait and Moulik 23
Brij 97 709 0.400 283.60 1.418 Hait and Moulik 23
Brij 98 1153.54 0.265 305.69 1.528 Hait and Moulik 23
Brij 78 1151.54 0.0057 6.56 0.033 Hait and Moulik 23
Brij 700 4670 0.020 93.40 0.467 Hait and Moulik 23
Table 4.
 
Correlation Comparisons between the Liposome Leakage Method of Assessing Toxicity and Other Published Methods
Table 4.
 
Correlation Comparisons between the Liposome Leakage Method of Assessing Toxicity and Other Published Methods
Method(s) Source Pearson Spearman
EYTEX™ Matsukawa et al. 4 0.2–0.4 N/A
Various cell assays Vian et al. 3 0.48–0.62 0.53–0.64
Corneal thickness Kennah et al. 18 0.86 N/A
Cell growth inhibition Tachon et al. 19 N/A 0.652–0.845
Cell leakage Cottin and Zanvit 8 0.94 0.92
Liposome leakage Current Study 0.74–0.99 0.79–0.94
Table 5.
 
Predicted Draize Scores for 10% and 1% Stock Solutions of Selected Brij Surfactants
Table 5.
 
Predicted Draize Scores for 10% and 1% Stock Solutions of Selected Brij Surfactants
Surfactant Predicted Draize Score (10%)* Irritation Class (10%), † Predicted Draize Score (1%), ‡ Irritation Class (1%), †
Brij 97 36.9 ± 7.3 Irritant 11.0 Mild/Moderate
Brij 98 34.7 ± 7.2 Irritant 9.2 Mild/Moderate
Brij 78 22.1 ± 6.7 Mild/Moderate 0, § , ∥ Mild/Moderate
Brij 700 24.4 ± 6.8 Mild/Moderate 0.8 Mild/Moderate
Brij 56 21.6 ± 6.7 Mild/Moderate 0, § Mild/Moderate
Brij 58 26.4 ± 6.8 Mild/Moderate 2.5 Mild/Moderate
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×