- Email: [email protected]
Characterized and predictable rabbit uveitis model for antiinflammatory drug screening
Characterized and Predictable Rabbit Uveitis Model for Antiinflammatory Drug Screening
LAURAJAMIESON,DONNA MECKOLL-BRINCK, AND NANCY KELLER
A model has been developed for the screening of antiinflammatory ophthalmic drugs in rabbits. This simple and rapid method is reproducible and uses fewer animals proteins,
than
do some
then
other
challenged
methods.
intravitreally
Rabbits to induce
are sensitized a uveitis.
to bovine
The
serum
basis of the in-
flammatory response is shown to be due primarily to an immunologic mechanism. At 24 hr postchallenge, animals are sorted into treatment groups of approximately equal “titer” as defined by the slit lamp examination iris score which measures the magnitude of the immunologic response. Topical ocular treatment (control or drug) is then initiated and continued for a total of four days. Iris inflammation is evaluated by daily slit lamp exams. Results indicate that this model has statistically narrower frequency distribution of iris ratings than another published method, and is faster than other equally accurate and precise methods that depend upon blood antibody titer to sort animals. Iris inflammation; Key Words: Antibody titre; Animal model
Aqueous humor protein; Aqueous humor PMNs;
INTRODUCTION
A simple rabbit eye iris inflammation model has been developed for the study of the antiinflammatory activity of ophthalmic drugs. For screening antiinflammatory compounds in various formulations, a method that is relatively short in duration, yet reproducible from experiment to experiment, is most desirable. Our model is rapid to prepare and use, and it has been found to be reproducible over the last few years, with 11 experiments done to evaluate several different antiinflammatory compounds and drug delivery modes. This article presents the general mechanism responsible for the induced inflammation, an example of a typical experiment, and a comparison of experimental precision of our model with that of a previously published model. METHODS New Zealand White rabbits are sensitized to antigen-bovine serum proteins (lyophilized bovine serum, ICN Pharmaceuticals)-by three subcutaneous injecFrom InSite Vision (L.J., N.K.), Alameda, California, USA, and Boehringer Mannheim, Tutzing, Federal Republic of Germany. Address reprint requests to: Laura Jamieson, InSite Vision, 965 Atlantic Avenue, Alameda, CA 94501, USA. Received June 9,1988; accepted March 3,1989. 329 Journalof
Pharmacological
Methods
0 1989 Elsevier Science Publishing
21, 329-338 (1989) Co., Inc., 655 Avenue of the Americas, New York, NY 10010
330
1. lamieson et al.
tions given at 2-day intervals. Each sensitizing injection is 12.5 mg of antigen per kg of body weight. Adult animals in the weight range of 2.2-3.6 kg are used. The antigen (5 mg/mL) is freshly prepared in normal saline before use. Three days after the last sensitizing antigen dose, one eye of each rabbit is challenged by an intravitreal injection of 0.25 mg of antigen into the central portion of the vitreous. Prior to any intravitreal injection, the eyes are anesthetized by topical application of 2 drops of 0.5% tetracaine hydrochloride. Only one eye of each animal is used to avoid the confusion of any consensual effects. There were no overt signs of discomfort as the inflammatory response developed in sensitized and challenged eyes during the 96-hr postchallenge study period. Iris inflammation is measured by slit lamp examination according to the scale given below: O-no response l-slight congestion of iris vessels 2-mild congestion of iris vessels 3-moderate congestion of iris vessels and mild iris edema 4-marked congestion and dilation of iris vessels and moderate iris edema Slit lamp examination is first performed then daily for the succeeding 3 days.
at 24 hr after the intravitreal
challenge,
RESULTS The basis of the iris inflammation response was determined by measuring: I) the effects of an intravitreal injection itself, because of an injection of only the saline vehicle into vitreous of unsensitized animals; 2) the influence of any contaminants in the antigen, such as endotoxin, following injection of the bovine serum proteins into vitreous of unsensitized animals; and 3) the effects because of an immunogenic response, subsequent to injection of the antigen into the vitreous of sensitized animals. Figure 1 shows the iris inflammation ratings in the sensitized and unsensitized rabbits. The effect of the intravitreal injection procedure itself on iris inflammation is negligible, as shown by the minimal response in the iris of the saline control group that has never been exposed to antigen. Bovine serum proteins injected into the vitreous of unsensitized rabbits provoked a slight to mild inflammation, probably because of some endotoxin contamination (Bito, 1974). In contrast to the results in unsensitized rabbits injected with either saline or with antigen, the response in the sensitized animals was in or near the moderate range (a score of 3) for the first 3 days with only some decrease on day 4. The three groups of data in Figure 1 were compared using the nonparametric Kruskal-Wallis test (chi-square approximation) with a Bernoulli correction for multiple comparisons so that a difference was considered statistically significant at p < 0.0125 (i.e., an alpha of 0.05 divided by 4 comparisons). There was a statistically significant difference among the groups at every time point (days 1, 2, 3, and 4).
Antiinflammatory
Drug Screening Model
sensitized and 3
: 2
8 +! F” a+I
unsensitized, intravitreal injection of bovine serum proteins
2; 4g S
1
(/, a -
Y unsensitized, intravitreal $ saline injection l 0 0
1
DAYS AFTER FIGURE 1. saline (n = proteins (n proteins as
2
INTRAVITREAL
3
4
INJECTION
Iris inflammation in (A) unsensitized animals given an intravitreal injection of 17), (B) in unsensitized animals given an intravitreal injection of bovine serum = 17), and (C) in sensitized animals given an intravitreal injection of bovine serum challenge (n = 20).
The saline control group had consistently lower scores than the groups injected with antigen. The sensitized group had consistently higher scores than the unsensitized groups at each time point. On day 4 postintravitreal injection of antigen, aqueous humor is drawn for analysis immediately after euthanization of the animals by T-61 Euthanasia Solution. The leukocyte protein concentration (Lowry et af., 1951) and the polymorphonuclear (PMN) count are determined in the sample from each injected eye. These data are given on Table 1 with the data from normal untreated animals for comparison. The process of injection by itself provokes a small rise in protein in the aqueous, a response not unexpected in rabbits (Bito, 1984). A significantly greater but still smafl
331
TABLE 1 Protein and PMNs in Aqueous Humor of Rabbit Eyes 4 Days After lntravitreal Injections PROTEINCONC. (MC/o.1 ML) EXPERIMENTAL CONDITIONS Normal, untreated Unsensitized, saline injected into vitreous Unsensitized, bovine serum proteins injected into vitreous Sensitized, intravitreal challenge with bovine serum proteins
0
1 MEAN
PMN COUNT (PERMM~) MEAN
(SEM)
N
MEAN
(SEW
0.032
(0.004)
0
0.092
(0.050)
1
(1)
17
0.447
(0.088)
143
(30)
17
1.79
(0.27)
1250
(187)
20
2 ANTIBODY IN SERUM
3
4
-
5
15
6
TITRE (,vg IgM/ml) (1:32 dilution)
FIGURE 2. Iris inflammation scores on day 4 postchallenge versus antibody titers in serum from blood drawn 6 hr after the third and final antigen injection. The number of animals (one eye of each animal) at each iris score is shown on the graph in parentheses.
Antiinflammato~
Drug Screening Model
increase in aqueous humor protein is seen in unsensitized rabbits injected intravitreally with the bovine serum proteins. The highest protein concentration is found in the aqueous of the sensitized, challenged rabbits. The PMN count is normal in unsensitized animals injected intravitreally with saline, although a rise is observed in unsensitized animals injected intravitreally with bovine serum proteins. However, the mean PMN count in sensitized and challenged rabbits is about nine times that in the unsensitized animals injected intravitreally with the bovine serum proteins. The results-iris score, aqueous humor protein concentration, and aqueous humor PMN measurements-all indicated that the major portion of the inflammation was due to an immunologic mechanism. Assay of IgM and IgG was done to help confirm the presence of an immunologic stimulus in the inflammatory response. The relationship between serum IgM titer (by EL&A) at 6 hr after the last sensitizing antigen dose and the iris inflammation on day 4 of the study (Figure 2) shows that the IgM titer is a good predictor of a sustained inflammato~ response and confirms the predominantly immunogenic basis of the response. No IgM is detected in sera from unsensitized animals. No IgG was found in the sera from the sensitized rabbits. This is the expected result given that the antibody is measured in sera collected 6 hr after the last sensitizing antigen dose and before challenge. IgM is formed early in a reaction to sensitization, and IgG later. The results of a typical experiment done in this model to measure steroid antiinflammatory activity are given in Figure 3 and Table 2. Iris inflammation in the saline group is sustained as well or better than that in other models in which drug activity is measured (Baldwin et al., 1970; Bonnet et al., 1976; Kulkarni et al., 1981). The iris inflammation, aqueous protein, and PMN count in the saline group are all comparable to those values shown for the sensitized and challenged animals in previous experiments (Figure 1, Table 1). The effects of the dexamethasone treatment, begun after the slit lamp examination l-day postchallenge, are readily apparent. The nonparametric Wilcoxon rank sums test with a criterion of significance of p < 0.0125 was used to analyze the data in Figure 3. By 96 hr, there was a statistically significant difference between dexamethasone and saline efficacy in the reduction of inflammation. TABLE 2 Protein and PMNs in Aqueous Humor 4 Rays Postchallenge in a Typical Experiment in Sensitized Rabbits PMN COUNT (PERMM~)
PROTEINCONC. (MC/o.1 Mt.) MEAN
(GM)
MEAN
Saline (0.9% NaCI)
1.98
co.44
1048
(93
Dexamethasone (0.1% suspension)
0.49
(0.15)
378
(101)
TREATMENT
WM)
N 12 12
One drop (0.05 mL) of saline or dexamethasone suspension applied topically eight times per day at hourly intervals (9 A.M.-~ P.M.) for 4 days starting 24 hr postchallenge, after slit lamp examinations.
333
334
L. Jamieson et al.
4
3
1
0 0
1
DAYS AFTER
2 INTRAVITREAL
3
4
CHALLENGE
FIGURE 3. Distribution of iris scores in a typical experiment. There were 12 animals in each group; saline (0, dexamethasone (0). One eye of each animal treated with saline or O.I% dexamethasone at hourly intervals for a total of eight treatments each day, one 0.05.mL drop in each dose. Iris scores are given for 24,48, 72, and 96 hr after antigen challenge.
DISCUSSION
The basis for the initial methods used in our laboratory, was the model described by Baldwin et al. (1970). This model requires only 1 wk to adequately sensitize the animals to the antigen, but treatment is initiated before challenge so that there is no opportunity to measure the iris score in nondrug treated eyes. Thus, in this method (Baldwin et al., 1970), animals are assigned to treatment groups (saline control or drug) without any knowledge of the animal’s ability to respond to an antigen challenge. The omission of some assay of the animal’s immunologic state
AntiinfIammato~
Drug Screening Model
when forming control and treatment groups results in a great variation and a wide range of iris scores within groups at the 24-hr slit lamp examination. The iris score variation in the absence of any drug treatment is thought to reflect the variation in antibody titer within the groups. Rather than making antibody titer a routine procedure in our final method, a faster way of assessing immune status was chosen. Animals were sorted into groups of approximately equal “titer,” as defined by the iris score from a slit lamp examination 24 hr after challenge and prior to starting any treatment. The timing of the critical events in the two methods is diagrammed in Figure 4. To compare the precision of the initial and final methods, frequency plots were made of the 24-hr and 96-hr cumulative ratings (range O-16) from the series of experiments done by the initial method and those done by the final method (Fig. 5a,b). The cumulative 96-hr score (summation of the 24-, 48-, 72-, and 96-hr ratings) is a measure of how the eye responded to either drug or placebo treatment during the entire study period. The hypothesis was that the initial method was less precise than the final method. The hypothesis would be proven if: 1) the 24-hr data from
INITIAL METHOD
si 0
24
40
72
86
TIME (hr)
FINAL METHOD s A
A
0
A
24
48
72
96
A
I 1 0.
TIME
120
144
168
(hr)
FIGURE 4. A schematic representation of the timing of the events in the initial and in the final methods. The treatment period is 7 hr during each day; eight topical instillations of placebo or drug are given during the treatment period. In the initial method, animals are sorted into groups by body weight, then treatment is begun and proceeds for 4 hr (5 doses) prior to the intravitreally administered antigen challenge and the first slit lamp examination of the iris. This initiation of treatment before challenge prophylactically treats the animals’ eyes prior to challenge so that the full effect of the challenge is unlikely to be expressed. In the final method, the slit lamp examination and drug or placebo treatment is not begun until 24 hr after antigen challenge.
335
336
1. Jamieson et al.
r
1
SALINE
MAXIDEX
35 FINAL
INITIAL
INITIAL :N = 33)
(N = 32) 30
25
20
15
10
5
0
12
3
4
24 HOUR RATINGS FIGURE 5a. The frequency distribution of iris ratings within each group at the 24.hr slit lamp examination.
the drug-treated groups assigned by the initial method showed a wider distribution of inflammation ratings than the final method or 2) cumulative 24-96-hr ratings would allow a clear distinction between the efficacy of saline and that of dexamethasone. In Figure 5a, there was significant difference (x-square = 10.62, degrees of freedom = 3, p < 0.01) between the ratings in the initial method Maxidex group and the final method Maxidex group. The difference was due to ratings of 0, 1, 2, 3, and 4 in the initial method group, but ratings over a narrower band in the final method group, i.e., 2-4. In Figure Sb, the cumulative-24-96-hr-ratings for control and drug treated groups are shown. There is no significant difference between saline groups. However, the initial method Maxidex group has a significantly wider distribution of ratings than does the final method Maxidex group (x-square = 7.74, degrees of freedom = 2, p < 0.02).
Antiinflammatory
r
Drug Screening Model
SALINE
MAXIDEX
50 FINAL (N = 32
INITIAL (N = 33)
1
FINAL (N = 56)
INITIAL (N = 48)
40
30
20
10
ll ill o-4
5-9
10-14
o-4
5-9
96 HOUR
o-4
10-14
CUMULATIVE
FIGURE 5b. The frequency distribution ratings) in each group.
5-9
IO-14
RATINGS
of ranges of cumulative 96-hr iris rating (24-!l6-hr
In the model described in this article (“final method”), the sensitivity and accuracy of the measurement of antiinflammatory drug activity is enhanced by a pretreatment sorting of animals into groups according to their 24hr postchallenge slit lamp examination scores. This approach establishes a mean initial level of inflammation and distribution of iris scores that are about the same in each group. Drug treatment may then be started in fairly homogeneous groups of animals. Other models use antibody titer (Bonnet et al., 1976) or response to a previous challenge (Kulkarni et al., 1981) to sort animals into groups before treatment. Both of these techniques are effective but result in a more lengthy or complex experiment compared to the method reported in this article. The model described here has been used to determine the relative efficacy of dexamethasone given topically and intravitreally in a simple saline solution or in a polymer vehicle to sustain drug release in the eye (Keller et al., 1986). The model was sufficiently sensitive to differentiate between different drug doses and the effects of the two vehicles used in intravitreal drug administration. The authors are grateful to Dr. J&Lien Li for help with the ELISA method and interpretation of the antibody data. The authors also wish to express their gratitude to Dr. Priscilla Crevert for the nonparametric statistical analyses.
337
338
1. Jamieson et al.
REFERENCES Baldwin HA, Borgmann AR (1970) An improved immunogenic uveitis test in rabbits for evaluation of antiinflammatory compounds. froc Sot Exptl Biol Med 133:1326-1330.
Bito LZ (1984) Species differences in the responses of the eye to irritation and trauma: A hypothesis of divergence in ocular defense mechanisms, and the choice of experimental animals for eye research. Exp Eye Res 39:807-829. Bito LZ (1974) The effects of experimental uveitis on anterior uveal prostaglandin transport and aqueous humor composition. invest Ophthalmol13:959-966.
Bonnet P, Faure JP, LeDouarec JC (1976) Standardization of an experimental immune uveitis in the
rabbit for topical testing of drugs. Mod
frobl
Ophthalmol16:285-304.
Keller N, Jamieson L, Olejnik 0, Archibald R, Ehtesham S, Meckoll D (1986) Efficacy of intravitreal dexamethasone in experimental uveitis: Effect of drug delivery in a biodegradable polymer. Invest Ophthalmol27:248.
Kulkarni PS, Bhattacherjee P, Eakins KE, Srinivasan BD (1981) Anti-inflammatory effects of betamethasone phosphate, dexamethasone phosphate and indomethacin on rabbit ocular inflammation induced by bovine serum albumin. Curr Eye Res 1143-47. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ (1951) Protein measurement with the Folin phenol reagent. / Biol Chem 193:265-275.
LAURAJAMIESON,DONNA MECKOLL-BRINCK, AND NANCY KELLER
A model has been developed for the screening of antiinflammatory ophthalmic drugs in rabbits. This simple and rapid method is reproducible and uses fewer animals proteins,
than
do some
then
other
challenged
methods.
intravitreally
Rabbits to induce
are sensitized a uveitis.
to bovine
The
serum
basis of the in-
flammatory response is shown to be due primarily to an immunologic mechanism. At 24 hr postchallenge, animals are sorted into treatment groups of approximately equal “titer” as defined by the slit lamp examination iris score which measures the magnitude of the immunologic response. Topical ocular treatment (control or drug) is then initiated and continued for a total of four days. Iris inflammation is evaluated by daily slit lamp exams. Results indicate that this model has statistically narrower frequency distribution of iris ratings than another published method, and is faster than other equally accurate and precise methods that depend upon blood antibody titer to sort animals. Iris inflammation; Key Words: Antibody titre; Animal model
Aqueous humor protein; Aqueous humor PMNs;
INTRODUCTION
A simple rabbit eye iris inflammation model has been developed for the study of the antiinflammatory activity of ophthalmic drugs. For screening antiinflammatory compounds in various formulations, a method that is relatively short in duration, yet reproducible from experiment to experiment, is most desirable. Our model is rapid to prepare and use, and it has been found to be reproducible over the last few years, with 11 experiments done to evaluate several different antiinflammatory compounds and drug delivery modes. This article presents the general mechanism responsible for the induced inflammation, an example of a typical experiment, and a comparison of experimental precision of our model with that of a previously published model. METHODS New Zealand White rabbits are sensitized to antigen-bovine serum proteins (lyophilized bovine serum, ICN Pharmaceuticals)-by three subcutaneous injecFrom InSite Vision (L.J., N.K.), Alameda, California, USA, and Boehringer Mannheim, Tutzing, Federal Republic of Germany. Address reprint requests to: Laura Jamieson, InSite Vision, 965 Atlantic Avenue, Alameda, CA 94501, USA. Received June 9,1988; accepted March 3,1989. 329 Journalof
Pharmacological
Methods
0 1989 Elsevier Science Publishing
21, 329-338 (1989) Co., Inc., 655 Avenue of the Americas, New York, NY 10010
330
1. lamieson et al.
tions given at 2-day intervals. Each sensitizing injection is 12.5 mg of antigen per kg of body weight. Adult animals in the weight range of 2.2-3.6 kg are used. The antigen (5 mg/mL) is freshly prepared in normal saline before use. Three days after the last sensitizing antigen dose, one eye of each rabbit is challenged by an intravitreal injection of 0.25 mg of antigen into the central portion of the vitreous. Prior to any intravitreal injection, the eyes are anesthetized by topical application of 2 drops of 0.5% tetracaine hydrochloride. Only one eye of each animal is used to avoid the confusion of any consensual effects. There were no overt signs of discomfort as the inflammatory response developed in sensitized and challenged eyes during the 96-hr postchallenge study period. Iris inflammation is measured by slit lamp examination according to the scale given below: O-no response l-slight congestion of iris vessels 2-mild congestion of iris vessels 3-moderate congestion of iris vessels and mild iris edema 4-marked congestion and dilation of iris vessels and moderate iris edema Slit lamp examination is first performed then daily for the succeeding 3 days.
at 24 hr after the intravitreal
challenge,
RESULTS The basis of the iris inflammation response was determined by measuring: I) the effects of an intravitreal injection itself, because of an injection of only the saline vehicle into vitreous of unsensitized animals; 2) the influence of any contaminants in the antigen, such as endotoxin, following injection of the bovine serum proteins into vitreous of unsensitized animals; and 3) the effects because of an immunogenic response, subsequent to injection of the antigen into the vitreous of sensitized animals. Figure 1 shows the iris inflammation ratings in the sensitized and unsensitized rabbits. The effect of the intravitreal injection procedure itself on iris inflammation is negligible, as shown by the minimal response in the iris of the saline control group that has never been exposed to antigen. Bovine serum proteins injected into the vitreous of unsensitized rabbits provoked a slight to mild inflammation, probably because of some endotoxin contamination (Bito, 1974). In contrast to the results in unsensitized rabbits injected with either saline or with antigen, the response in the sensitized animals was in or near the moderate range (a score of 3) for the first 3 days with only some decrease on day 4. The three groups of data in Figure 1 were compared using the nonparametric Kruskal-Wallis test (chi-square approximation) with a Bernoulli correction for multiple comparisons so that a difference was considered statistically significant at p < 0.0125 (i.e., an alpha of 0.05 divided by 4 comparisons). There was a statistically significant difference among the groups at every time point (days 1, 2, 3, and 4).
Antiinflammatory
Drug Screening Model
sensitized and 3
: 2
8 +! F” a+I
unsensitized, intravitreal injection of bovine serum proteins
2; 4g S
1
(/, a -
Y unsensitized, intravitreal $ saline injection l 0 0
1
DAYS AFTER FIGURE 1. saline (n = proteins (n proteins as
2
INTRAVITREAL
3
4
INJECTION
Iris inflammation in (A) unsensitized animals given an intravitreal injection of 17), (B) in unsensitized animals given an intravitreal injection of bovine serum = 17), and (C) in sensitized animals given an intravitreal injection of bovine serum challenge (n = 20).
The saline control group had consistently lower scores than the groups injected with antigen. The sensitized group had consistently higher scores than the unsensitized groups at each time point. On day 4 postintravitreal injection of antigen, aqueous humor is drawn for analysis immediately after euthanization of the animals by T-61 Euthanasia Solution. The leukocyte protein concentration (Lowry et af., 1951) and the polymorphonuclear (PMN) count are determined in the sample from each injected eye. These data are given on Table 1 with the data from normal untreated animals for comparison. The process of injection by itself provokes a small rise in protein in the aqueous, a response not unexpected in rabbits (Bito, 1984). A significantly greater but still smafl
331
TABLE 1 Protein and PMNs in Aqueous Humor of Rabbit Eyes 4 Days After lntravitreal Injections PROTEINCONC. (MC/o.1 ML) EXPERIMENTAL CONDITIONS Normal, untreated Unsensitized, saline injected into vitreous Unsensitized, bovine serum proteins injected into vitreous Sensitized, intravitreal challenge with bovine serum proteins
0
1 MEAN
PMN COUNT (PERMM~) MEAN
(SEM)
N
MEAN
(SEW
0.032
(0.004)
0
0.092
(0.050)
1
(1)
17
0.447
(0.088)
143
(30)
17
1.79
(0.27)
1250
(187)
20
2 ANTIBODY IN SERUM
3
4
-
5
15
6
TITRE (,vg IgM/ml) (1:32 dilution)
FIGURE 2. Iris inflammation scores on day 4 postchallenge versus antibody titers in serum from blood drawn 6 hr after the third and final antigen injection. The number of animals (one eye of each animal) at each iris score is shown on the graph in parentheses.
Antiinflammato~
Drug Screening Model
increase in aqueous humor protein is seen in unsensitized rabbits injected intravitreally with the bovine serum proteins. The highest protein concentration is found in the aqueous of the sensitized, challenged rabbits. The PMN count is normal in unsensitized animals injected intravitreally with saline, although a rise is observed in unsensitized animals injected intravitreally with bovine serum proteins. However, the mean PMN count in sensitized and challenged rabbits is about nine times that in the unsensitized animals injected intravitreally with the bovine serum proteins. The results-iris score, aqueous humor protein concentration, and aqueous humor PMN measurements-all indicated that the major portion of the inflammation was due to an immunologic mechanism. Assay of IgM and IgG was done to help confirm the presence of an immunologic stimulus in the inflammatory response. The relationship between serum IgM titer (by EL&A) at 6 hr after the last sensitizing antigen dose and the iris inflammation on day 4 of the study (Figure 2) shows that the IgM titer is a good predictor of a sustained inflammato~ response and confirms the predominantly immunogenic basis of the response. No IgM is detected in sera from unsensitized animals. No IgG was found in the sera from the sensitized rabbits. This is the expected result given that the antibody is measured in sera collected 6 hr after the last sensitizing antigen dose and before challenge. IgM is formed early in a reaction to sensitization, and IgG later. The results of a typical experiment done in this model to measure steroid antiinflammatory activity are given in Figure 3 and Table 2. Iris inflammation in the saline group is sustained as well or better than that in other models in which drug activity is measured (Baldwin et al., 1970; Bonnet et al., 1976; Kulkarni et al., 1981). The iris inflammation, aqueous protein, and PMN count in the saline group are all comparable to those values shown for the sensitized and challenged animals in previous experiments (Figure 1, Table 1). The effects of the dexamethasone treatment, begun after the slit lamp examination l-day postchallenge, are readily apparent. The nonparametric Wilcoxon rank sums test with a criterion of significance of p < 0.0125 was used to analyze the data in Figure 3. By 96 hr, there was a statistically significant difference between dexamethasone and saline efficacy in the reduction of inflammation. TABLE 2 Protein and PMNs in Aqueous Humor 4 Rays Postchallenge in a Typical Experiment in Sensitized Rabbits PMN COUNT (PERMM~)
PROTEINCONC. (MC/o.1 Mt.) MEAN
(GM)
MEAN
Saline (0.9% NaCI)
1.98
co.44
1048
(93
Dexamethasone (0.1% suspension)
0.49
(0.15)
378
(101)
TREATMENT
WM)
N 12 12
One drop (0.05 mL) of saline or dexamethasone suspension applied topically eight times per day at hourly intervals (9 A.M.-~ P.M.) for 4 days starting 24 hr postchallenge, after slit lamp examinations.
333
334
L. Jamieson et al.
4
3
1
0 0
1
DAYS AFTER
2 INTRAVITREAL
3
4
CHALLENGE
FIGURE 3. Distribution of iris scores in a typical experiment. There were 12 animals in each group; saline (0, dexamethasone (0). One eye of each animal treated with saline or O.I% dexamethasone at hourly intervals for a total of eight treatments each day, one 0.05.mL drop in each dose. Iris scores are given for 24,48, 72, and 96 hr after antigen challenge.
DISCUSSION
The basis for the initial methods used in our laboratory, was the model described by Baldwin et al. (1970). This model requires only 1 wk to adequately sensitize the animals to the antigen, but treatment is initiated before challenge so that there is no opportunity to measure the iris score in nondrug treated eyes. Thus, in this method (Baldwin et al., 1970), animals are assigned to treatment groups (saline control or drug) without any knowledge of the animal’s ability to respond to an antigen challenge. The omission of some assay of the animal’s immunologic state
AntiinfIammato~
Drug Screening Model
when forming control and treatment groups results in a great variation and a wide range of iris scores within groups at the 24-hr slit lamp examination. The iris score variation in the absence of any drug treatment is thought to reflect the variation in antibody titer within the groups. Rather than making antibody titer a routine procedure in our final method, a faster way of assessing immune status was chosen. Animals were sorted into groups of approximately equal “titer,” as defined by the iris score from a slit lamp examination 24 hr after challenge and prior to starting any treatment. The timing of the critical events in the two methods is diagrammed in Figure 4. To compare the precision of the initial and final methods, frequency plots were made of the 24-hr and 96-hr cumulative ratings (range O-16) from the series of experiments done by the initial method and those done by the final method (Fig. 5a,b). The cumulative 96-hr score (summation of the 24-, 48-, 72-, and 96-hr ratings) is a measure of how the eye responded to either drug or placebo treatment during the entire study period. The hypothesis was that the initial method was less precise than the final method. The hypothesis would be proven if: 1) the 24-hr data from
INITIAL METHOD
si 0
24
40
72
86
TIME (hr)
FINAL METHOD s A
A
0
A
24
48
72
96
A
I 1 0.
TIME
120
144
168
(hr)
FIGURE 4. A schematic representation of the timing of the events in the initial and in the final methods. The treatment period is 7 hr during each day; eight topical instillations of placebo or drug are given during the treatment period. In the initial method, animals are sorted into groups by body weight, then treatment is begun and proceeds for 4 hr (5 doses) prior to the intravitreally administered antigen challenge and the first slit lamp examination of the iris. This initiation of treatment before challenge prophylactically treats the animals’ eyes prior to challenge so that the full effect of the challenge is unlikely to be expressed. In the final method, the slit lamp examination and drug or placebo treatment is not begun until 24 hr after antigen challenge.
335
336
1. Jamieson et al.
r
1
SALINE
MAXIDEX
35 FINAL
INITIAL
INITIAL :N = 33)
(N = 32) 30
25
20
15
10
5
0
12
3
4
24 HOUR RATINGS FIGURE 5a. The frequency distribution of iris ratings within each group at the 24.hr slit lamp examination.
the drug-treated groups assigned by the initial method showed a wider distribution of inflammation ratings than the final method or 2) cumulative 24-96-hr ratings would allow a clear distinction between the efficacy of saline and that of dexamethasone. In Figure 5a, there was significant difference (x-square = 10.62, degrees of freedom = 3, p < 0.01) between the ratings in the initial method Maxidex group and the final method Maxidex group. The difference was due to ratings of 0, 1, 2, 3, and 4 in the initial method group, but ratings over a narrower band in the final method group, i.e., 2-4. In Figure Sb, the cumulative-24-96-hr-ratings for control and drug treated groups are shown. There is no significant difference between saline groups. However, the initial method Maxidex group has a significantly wider distribution of ratings than does the final method Maxidex group (x-square = 7.74, degrees of freedom = 2, p < 0.02).
Antiinflammatory
r
Drug Screening Model
SALINE
MAXIDEX
50 FINAL (N = 32
INITIAL (N = 33)
1
FINAL (N = 56)
INITIAL (N = 48)
40
30
20
10
ll ill o-4
5-9
10-14
o-4
5-9
96 HOUR
o-4
10-14
CUMULATIVE
FIGURE 5b. The frequency distribution ratings) in each group.
5-9
IO-14
RATINGS
of ranges of cumulative 96-hr iris rating (24-!l6-hr
In the model described in this article (“final method”), the sensitivity and accuracy of the measurement of antiinflammatory drug activity is enhanced by a pretreatment sorting of animals into groups according to their 24hr postchallenge slit lamp examination scores. This approach establishes a mean initial level of inflammation and distribution of iris scores that are about the same in each group. Drug treatment may then be started in fairly homogeneous groups of animals. Other models use antibody titer (Bonnet et al., 1976) or response to a previous challenge (Kulkarni et al., 1981) to sort animals into groups before treatment. Both of these techniques are effective but result in a more lengthy or complex experiment compared to the method reported in this article. The model described here has been used to determine the relative efficacy of dexamethasone given topically and intravitreally in a simple saline solution or in a polymer vehicle to sustain drug release in the eye (Keller et al., 1986). The model was sufficiently sensitive to differentiate between different drug doses and the effects of the two vehicles used in intravitreal drug administration. The authors are grateful to Dr. J&Lien Li for help with the ELISA method and interpretation of the antibody data. The authors also wish to express their gratitude to Dr. Priscilla Crevert for the nonparametric statistical analyses.
337
338
1. Jamieson et al.
REFERENCES Baldwin HA, Borgmann AR (1970) An improved immunogenic uveitis test in rabbits for evaluation of antiinflammatory compounds. froc Sot Exptl Biol Med 133:1326-1330.
Bito LZ (1984) Species differences in the responses of the eye to irritation and trauma: A hypothesis of divergence in ocular defense mechanisms, and the choice of experimental animals for eye research. Exp Eye Res 39:807-829. Bito LZ (1974) The effects of experimental uveitis on anterior uveal prostaglandin transport and aqueous humor composition. invest Ophthalmol13:959-966.
Bonnet P, Faure JP, LeDouarec JC (1976) Standardization of an experimental immune uveitis in the
rabbit for topical testing of drugs. Mod
frobl
Ophthalmol16:285-304.
Keller N, Jamieson L, Olejnik 0, Archibald R, Ehtesham S, Meckoll D (1986) Efficacy of intravitreal dexamethasone in experimental uveitis: Effect of drug delivery in a biodegradable polymer. Invest Ophthalmol27:248.
Kulkarni PS, Bhattacherjee P, Eakins KE, Srinivasan BD (1981) Anti-inflammatory effects of betamethasone phosphate, dexamethasone phosphate and indomethacin on rabbit ocular inflammation induced by bovine serum albumin. Curr Eye Res 1143-47. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ (1951) Protein measurement with the Folin phenol reagent. / Biol Chem 193:265-275.