Clinicopathologic Spectrum of Primary Uveal Melanocytic Lesions in an Animal Model

Jacob Pe'er, MD, 1 Robert Folberg, MD, 1•2 Stephen J. Massicotte, MD, 1

Jeffrey Baron, PhD,3 Rita Parys-Van Ginderdeuren, MD, 1

Bridget Zimmerman, PhD,4 Margaret L. Meyer, 1 Hallie Worsey, BA1

Background: Currently, there are no animal models of primary uveal melanoma in an eye large enough to allow documentation of the clinical evolution of the lesion by either funduscopy or fundus photography.

Methods: The authors induced primary uveal melanocytic lesions in the eyes of Dutch (pigmented) rabbits using a two-stage carcinogenesis protocol involving initiation with 4 weekly topical applications of 1 0 ~-tl of a 1% solution of 7, 12-dimethyl­benz[a]anthracene (DMBA) in acetone (21 eyes) followed by 12 weekly topical appli­cations of a 1 0 ~-tl solution of either 0.25% or 0.5% croton oil in acetone. They also investigated the effect of initiation with DMBA without promotion and the effects of chronic topical exposure to acetone and proparacaine.

Results: Exposure to DMBA followed by promotion with croton oil in either con­centration was the most effective means of inducing clinically detectable fundus lesions. Histologically, a spectrum of melanocytic proliferations developed including benign nevi, nevi with varying grades of cytologic atypia, and clusters of confluent atypical mela­nocytes that may represent early melanomas. Although clinical regression of fundus lesions was noted in eight eyes after promotion had been stopped, five of these eyes showed unequivocal histologic evidence of a residual uveal melanocytic lesion. Chronic ocular irritation is capable of inducing cytologically benign subclinical uveal melanocytic proliferations.

Conclusions: The conventional classification of human uveal melanocytic lesions includes only nevi and melanomas, but a comparison of the results of this study with descriptions of human uveal melanocytic nevi suggests the existence of a spectrum of intermediate atypical precursor lesions in humans. Ophthalmology 1992;99:977-986

Originally received: October 24, 1991. Revision accepted: January 27, 1992. 1 Department of Ophthalmology, University oflowa, Iowa City. 2 Department of Pathology, University of Iowa, Iowa City. 3 Department of Pharmacology, University oflowa, Iowa City. 4 Division of Biostatistics, Department of Preventive Medicine, University of Iowa, Iowa City.

Dr. Pe'erwas on sabbatical leave from the Department of Ophthalmology, Hadassah-Hebrew University Medical School, Jerusalem, Israel.

Dr. Massicotte is currently affiliated with the Department of Ophthal­mology, Indiana University, Indianapolis.

Supported in part by United States Public Health Service grant EY07043 (Dr. Folberg), by the Center for Electron Microscopy Research, University of Iowa, Iowa City, Iowa, and by an unrestricted grant from Research to Prevent Blindness, Inc, New York, New York.

Reprint requests to Robert Folberg, MD, The University of Iowa, 233MRC, Iowa City, IA, 52242.

Compared with the volume of information describing the evolution of human primary cutaneous1

•2 and conjunc­

tival3-6 melanoma through various well-defined precursor lesions, information about the in situ histogenesis of pri­mary human choroidal melanoma is strikingly incom­plete. In his 1976 Jackson Memorial Lecture, Gass7 stated that "most choroidal nevi are not present at birth and . . . arise probably from very small cell rests as hyper­plastic lesions that exhibit growth primarily but not exclusively in late childhood and early adulthood." Of course, it is not possible to verify this hypothesis for pri­mary human uveal melanoma because, unlike the skin and conjunctiva which are accessible to biopsy for estab­lishing clinicopathologic correlations of early precursor stages, it is not feasible to obtain tissue from early cho­roidal precursor lesions except from the rare eyes studied clinically that are obtained subsequently at postmortem examination. 7

977

Ophthalmology Volume 99, Number 6, June 1992

If an animal model of primary uveal melanoma were available, it would be possible to study the emergence of the earliest clinically detectable precursor lesions and to describe the pathology of these precursors. Unfortunately, although animal models of uveal melanoma have been developed using the transplantation of tumors into animal eyes, animal models of primary uveal melanoma are not available for either studying the natural history of this cancer or assisting in the development and evaluation of diagnostic procedures and treatment protocols for pa­tients.8·9

Animal models of both primary cutaneous mela­noma10-12 and conjunctival primary acquired melanosis13

have been induced by the chronic topical application of the chemical carcinogen, 7,12-dimethylbenz[a]anthracene (DMBA). Each of these models mimic their human coun­terparts both clinically and histologically. Recently, we demonstrated that chronic topical application of DMBA directly to the rabbit sclera is capable of inducing hyper­plastic melanocytic lesions of the choroid that can be doc­umented by fundus photography.9 In the current report, the characteristics of a wide spectrum of clinical and his­tologic primary uveal melanocytic lesions induced by DMBA in rabbit eyes are described. Data from this ex­perimental model are related to what is known about the evolution of primary human choroidal melanoma.

Materials and Methods

For all experiments, 10 ~-tl of a 1% solution of DMBA (Aldrich Chemical, Milwaukee, WI) in acetone was used as an initiator of the neoplastic process. Although expe­rience with the effects of DMBA on cutaneous melano­cytes of animals indicates that it is capable of both initi­ating and promoting these cells to melanoma (a complete carcinogen), 10- 12 the formation of experimental cutaneous melanoma is facilitated by the promotion of DMBA-ini­tiated melanocytes by croton oil. 14·15 Whereas phorbol 12-myristate 13-acetate (PMA) also is a potent promotor and has been reported to stimulate human 16 and mouse17

epidermal melanocytes, we found this agent to induce an unacceptably severe allergic response when applied to rabbit eyes in any concentration, a feature noted by others who developed an animal model of orbital myositis. 18

Therefore, we used croton oil as a promoting agent. In a series of preliminary investigations, we determined that the maximum tolerable dose of croton oil in acetone for chronic application to the rabbit eye was between 0.25% and 0.5% and also showed that chronic application of croton oil to the rabbit eye without DMBA did not induce either clinically detectable lesions or high-grade cytologic atypia.

Dutch (pigmented) rabbits weighing between 1 and 3 kg (approximately 6 weeks of age) were examined ex­ternally and by fundus photography to exclude pre-exist­ing lesions. Eyes demonstrating uneven fundus pigmen­tation were excluded from study. Before any manipula­tion, each rabbit received an intramuscular injection of a mixture containing ketamine ( 40 mg/kg), xylazine ( 10

978

mg/kg), and acepromazine (1 mg/kg), 0.5 to 0.75 ml, fol­lowed by topical proparacaine HCl 0.5% (Allergan Phar­maceuticals, Irvine, CA). Only one eye of each animal was manipulated.

In the current investigation, 93 rabbits were separated into 7 treatment protocols summarized below and in Ta­ble 1. Briefly, after performing a peritomy extending from the lateral aspect of the superior rectus muscle to just below the insertion of the lateral rectus muscle, the lateral rectus was disinserted to permit free rotation of the globe inferonasally and to expose the equator. Loose episcleral tissue was removed without thinning the sclera. This pre­paratory procedure was performed once, fashioning a painting site approximately 1 cm2 in diameter that cor­responded to a zone in the fundus near the edge of the medullary ray temporally. Subsequently, all topical agents were applied to the eye in a manner identical to our orig­inal protocol with weekly debridement of episcleral gran­ulation tissue before the application of agents.9

Group 1 eyes received only preparatory surgery and were observed throughout the study period to determine the possible effect of this preparatory procedure on cho­roidal melanocytes.

Group 2 eyes received preparatory surgery and weekly applications of the topical anesthetic proparacaine (pro­paracaine has not been reported to have any carcinogenic potential and was used in all eyes receiving acetone, DMBA, or croton oil).

Group 3 eyes received weekly applications of topical acetone (acetone is the vehicle for both DMBA and croton oil and is neither an initiator nor a promotor) after pre­paratory surgery.

Group 4 eyes received 4 weekly treatments of 1% DMBA in acetone after preparatory surgery.

Group 5 eyes received 4 weekly treatments of 1% DMBA in acetone after preparatory surgery, followed by weekly applications of acetone.

Group 6 eyes received 4 weekly treatments of 1% DMBA in acetone after preparatory surgery, followed by weekly applications of the promotor, croton oil, 0.25% in acetone.

Group 7 eyes received 4 weekly treatments of 1% DMBA in acetone after preparatory surgery, followed by weekly applications of the promotor, croton oil, 0.5% in acetone.

After the last treatment, all eyes were observed for 11 to 34 weeks, the post-treatment observation period (Table 1 ). Fundus photographs were taken monthly during the treatment and observations phases of these studies. At the conclusion of the observation period, all rabbits were killed, subjected to a complete necropsy, and the eyes were removed for histologic study. All protocols for this study were approved by the University of Iowa Animal Welfare Committee.

Tissue for light microscopy was fixed in 10% buffered formalin, embedded in paraffin, and multiple levels were sectioned through the block to ensure that the painting site was examined (the painting site was identifiable his­tologically by the presence of episcleral scar tissue).9 Ad­jacent sections were stained with hematoxylin-eosin with

Pe' er et al · Animal Model of Primary Uveal Melanocytic Lesions

Table 1. Summary of Current Treatment Protocols

Group

Group 1 (control) Group 2 (proparacaine)

Group 3 (acetone)

Group 4 (DMBA 4 wks)

Group 5 (DMBA and chronic acetone)

Group 6 (DMBA and chronic 0.25% croton oil)

Group 7 (DMBA and chronic 0.50% croton oil)

Number of Rabbits

11 17

17

6

21

9

12

Protocol

Preparatory surgery only

Preparatory surgery followed by 15 to 16 weekly applications of proparacaine and scleral debridement

Preparatory surgery followed by 12 to 16 weekly applications of proparacaine, scleral debridement, and 10 ~1 of acetone

Preparatory surgery followed by 4 weekly applications of 10 ~1 of a 1% solution of DMBA in acetone

Preparatory surgery followed by 4 weekly applications of 10 ~1 of a 1% solution of DMBA in acetone and 12 weekly applications of acetone

Preparatory surgery followed by 4 weekly applications of 10 ~1 of a 1% solution of DMBA in acetone and 12 additional weekly treatments of 0.25% croton oil in acetone

Preparatory surgery followed by 4 weekly applications of 10 ~1 of a 1% solution of DMBA in acetone and 12 additional weekly treatments of 0.50% croton oil in acetone

Weeks of Post-treatment Observation

26-34 11-19

11-19

22

11

11

11

and without permanganate bleaching of melanin to assist in the assessment of cytologic atypia. A selected focus of choroidal melanocytic hyperplasia identified by light mi­croscopy was excised from the paraffin block, and the tissue was reprocessed for transmission electron micros­copy according to previously described methods. 19

ules, a "chromatophore" (Fig 1).9 Histologically, mela­nocytic hyperplasia was defined by the occurrence of choroidal thickening by melanocytes in the painting area compared with nontreated zones of the same eye and nor­mal eyes; a negative result was defined by a density of melanocytes in the painting site that was identical to that seen in the unpainted portion of the treated eye and in normal eyes; and a equivocal histologic finding was de­fined as the presence of an increase in choroidal mela­nocytes within the painting site but without sharp contrast to the rest of the eye or with normal eyes. Foci of hyper­plasia were designated as either diffuse (Fig 2), nodular (Fig 3) or mixed. 9 For all eyes interpreted histologically as equivocal or positive (melanocytic hyperplasia), step sections were taken to exclude the possibility of proximity to emissary vessels or nerves, regions normally thickened by melanocytes in the rabbit eye.

Clinical and histologic effects of treatments were eval­uated as previously described.9 Briefly, a positive clinical result was defined by the appearance of new fundus pig­mentation compared with baseline photographs in areas other than those adjacent to the vortex veins or ciliary nerves, zones that are normally hyperpigmented in pig­mented rabbits; a negative clinical result was defined as no clear difference between baseline and post-treatment photographs; and a fundus was classified as equivocal by the appearance of subtle, but not dramatic changes in pigmentation.

Histologically, tissue sections were evaluated for evi­dence of melanocytic hyperplasia and cytologic atypia. Normally, three types of pigmented cells can be identified in or adjacent to the rabbit choroid: retinal pigment ep­ithelium; choroidal melanocytes; and a syncytial cell con­taining melanin and complex melanin-lipofuscin gran-

Cytologically, normal choroidal melanocytes lack nu­cleoli, chromatin dispersion and a nuclear fold (Fig 1 ). Atypical cells feature a nucleus that is more basophilic than seen in normal cells and exhibit evidence of chro­matin dispersion and one or more nucleoli with or without nuclear folds. Evaluation of cytologic features was limited

979

Ophthalmology Volume 99, Number 6, june 1992

Figure 1. Normal untreated rabbit choroid. Three types of pigmented cells are identified: retinal pigment epithelium (arrow), choroidal mela­nocytes (M), and "chromatophores" (p) (left: hematoxylin-eosin; original magnification, X87).

to choroidal melanocytes, because both retinal pigment epithelium and "chromatophores" may demonstrate nu­cleoli normally.

The following cytologic grading system for the assess­ment of cytologic atypia was applied to each eye: grade 0 (Fig 4) was defined as 0 to 2 atypical cells per painting site; grade I (Fig 5) was defined as 3 to 5 atypical cells per painting site without the formation of a cluster (at least 4 contiguous atypical cells); grade 2 (Fig 6) was de­fined as more than 5 atypical cells or one cluster per painting site; and grade 3 (Fig 7) was defined by the pres­ence of at least two clusters. Cytologic atypia was graded independently of the histologic assessment of hyperplasia.

In our previous investigation into the effect of chronic exposure of DMBA on rabbit choroidal melanocytes, we analyzed fundus photographs for the emergence of pig­mented lesions and studied tissue sections to determine

Figure 2. Diffuse melanocytic hyperplasia in a rabbit treated with DMBA and 0.25% croton oil. The retina is artifactitiously detached. The choroid is diffusely thickened by melanocytes (hematoxylin-eosin; original mag­nification, X87).

980

Figure 3. Nodular melanocytic hyperplasia in a rabbit treated with DMBA and 0.50% croton oil. Multiple micronodules of melanocytes are identified in the choroid (hematoxylin-eosin; original magnification, X87).

the presence or absence of hyperplasia, but we did not evaluate these tissue sections for evidence of cytologic atypia.9 In that study, 22 rabbits were divided into 2 groups as summarized in Table 2: 18 rabbit eyes received 5 to 14 weekly topical applications of 1% DMBA in acetone to the sclera ( 10 JLl per application) and 4 rabbit eyes formed a control group, receiving 10 JLl of HPLC-grade acetone (the protocol for this control group is identical to the current treatment group 3, Table 1). Tissue sections from that study were therefore bleached and graded for cytologic atypia.

Fundus photographs from the current study were in­terpreted by two investigators (JP and RPV) who had no role in the treatment ofthese animals and both of whom were masked to the treatment protocols. The analysis of histologic sections obtained from the current study for hyperplasia and cytologic atypia in sections of eyes from

Figure 4. Grade 0 atypia in an untreated rabbit. Notice the lack of nuclear enlargement and the absence of chromatin dispersion and a nucleolus, all features of atypical cells (hematoxylin-eosin; original magnification, X177).

Pe' er et al · Animal Model of Primary Uveal Melanocytic Lesions

Figure 5. Grade 1 atypia in an eye receiving DMBA and 0.25% croton oil. Notice the randomly dispersed atypical cells designated by the arrows (hematoxylin-eosin with permanganate bleach; original magnification, X177).

both the previous9 and current studies was performed by an investigator (JP), who did not treat any of the animals and was masked to the results of fundus photographs and the treatment protocols.

Statistical analysis compared each treatment group with the control group (group 1 received only preparatory sur­gery) and with every other treatment group using the Fisher's exact test for an association between treatment protocol and the development offundus lesions, histologic evidence of hyperplasia and cytologic atypia. For these statistical comparisons, the two doses of croton oil were consolidated into one group because we detected no sig­nificant difference between the effects of the two doses on the emergence of clinically detectable fundus lesions and histologic evidence of hyperplasia and atypia. Fundus photographs graded as "equivocal" were excluded from statistical analysis. In the statistical analysis of cytologic atypia, grades 0 and 1 were consolidated into one group,

Figure 6. Grade 2 atypia in an eye receiving DMBA and 0.50% croton oil. The atypical cells are more numerous than in grade 1 (hematoxylin­eosin with permanganate bleach; original magnification, X 177).

Figure 7. Grade 3 atypia in an eye receiving DMBA and 0.50% croton oU. The atypical cells are confluent (hematoxylin-eosin with permanganate bleach; original magnification, X 177).

termed "low grade," and grades 2 and 3 were consolidated into another group, termed "high grade."

Results

Development of Clinically Detectable Fundus Lesions

F1at, pigmented lesions first appeared in 12 eyes between 5 and 15 weeks after the initial exposure to DMBA. Table 3 summarizes the relationship between treatment proto­cols and the development of pigmented fundus lesions. Only eyes that were exposed to DMBA developed clini­cally detectable pigmented lesions. When data for the two concentrations of croton oil are combined, statistical analysis using the Fisher's exact test indicates that the protocol of DMBA followed by croton oil has a signifi­cantly greater association with pigmented fundus lesions than does either DMBA alone or DMBA with acetone (P = 0.014).

Eight of the 12 pigmented lesions that were detected by fundus photography began to disappear clinically after treatment with acetone or croton oil was terminated (i.e., during the observation period). Figures 8 to 10 illustrate the time-course for the emergence and subsequent clinical regression of a flat pigmented lesion in an eye that received 4 weekly treatments with 1% DMBA in acetone followed by weekly applications of 0.5% croton oil in acetone (group 7, Table 1). After 4 weekly treatments with DMBA and 7 weekly applications of croton oil, no pigmented lesion is seen (Fig 8), but after 4 additional treatments with croton oil, a flat pigmented lesion is identified just behind and below the medullary ray (Fig 9). This eye received 1 more application of croton oil. Three weeks after this last application of croton oil, during the obser­vation period, the pigmentation is no longer visible (Fig 1 0). The histology of eyes that demonstrated clinical regression is discussed below.

981

Ophthalmology Volume 99, Number 6, June 1992

Table 2. Previous Treatment Protocols9

Group Number of

Rabbits Protocol Weeks of Post-treatment

Observation

Acetone 4

Chronic DMBA 18

Preparatory surgery followed by 8 to 10 weekly applications of proparacaine, scleral debridement and 10 JLl of acetone

Preparatory surgery followed by 5 to 14 weekly applications of 10 JLl of a 1% solution of DMBA in acetone

19

19

Histologic Changes

The relationship between the various treatment protocols and the development of histologically defined uveal me­lanocytic hyperplasia is summarized in Table 4. There was no significant association between the type of topical agent used and the presence of any particular form of hyperplasia (nodular, diffuse, or mixed). However, each of the treatments involving a topical agent resulted in a significantly greater proportion of hyperplastic lesions when compared with eyes that received only preparatory surgery. As suspected in our previous studies,9 the chronic application of proparacaine and acetone accompanied by weekly episcleral debridement did produce uveal mela­nocytic hyperplasia, but none of these hyperplastic lesions was detected by fundus photography. No mitotic figures were identified. Transmission electron microscopy per­formed on a selected lesion demonstrated melanocytes and not retinal pigment epithelial cells, "chromato­phores," or rilelanophages (Fig 11 ). There was no evidence of retinal pigment epithelial pathology.

Cytologic Atypia The relationship between treatment protocol and cytologic atypia is summarized in Table 5. Despite the development of hyperplastic lesions in groups 2 and 3 (Table 4, eyes not exposed to DMBA), most of these eyes were assigned to atypia grades 0 or 1 (low grade); none of these eyes demonstrated grade 3 atypia. These lesions may be rep-

resent a form of relatively "benign" hyperplasia. Although eyes exposed to DMBA alone also developed low grade atypia, the combination ofDMBA (initiator) followed by croton oil (promotor) has a very strong association with high grade (grades 2 and 3) cytologic atypia when com­pared with all other protocols (P < 0.0001 compared with protocols without DMBA exposure and P = 0.019 com­pared with DMBA alone or DMBA with acetone). There­fore, exposure to both an initiator and promotor has a tendency to produce cytologically "atypical" melanocytic hyperplasia.

Interestingly, there is no association between the grade of cytologic atypia and any particular pattern of hyper­plasia (nodular, diffuse, or mixed). There is, however, a strong association between cytologic atypia and the emer­gence of a clinically detectable fundus lesion (Fisher's exact test, P = 0.0002): although lesions that are cytologically atypical are not necessarily pigmented clinically, clinically pigmented lesions usually demonstrate histologic evidence of both hyperplasia and cytologic atypia.

Significant cytologic atypia may be detected histolog­ically in lesions that have regressed clinically. For example, Figure 7, illustrating grade 3 atypia, was taken from the same eye that demonstrated total clinical regression as illustrated in Figures 8 to 10. Five of eight eyes that dem­onstrated clinical regression oflesions after all treatments had ceased showed persistent histologic lesions. The his­tologic findings of eyes that demonstrated clinical regres­sion are summarized in Table 6.

Table 3. Development of Clinically Detectable Pigmented Lesions

982

Positive

No Protocol Regression Regression Negative

Group 1 (control) 0 0 11 Group 2 (proparacaine) 0 0 16 Group 3 (acetone*) 0 0 18 Group 4 (DMBA 4 wks) 0 1 5 Group 5 (DMBA with chronic acetone) 1 1 16 Group 6 (DMBA with croton oil 0.25%) 3 2 3 Group 7 (DMBA with croton oil 0.50%) 0 4 5 Chronic DMBA t 4 0 10

• Includes 4 eyes treated identically with HPLC-grade acetone in the previously published study.9

t Data from previously published study,9 provided for comparison.

Equivocal

0 1 3 0 3 1 3 4

Pe'er et al · Animal Model of Primary Uveal Melanocytic Lesions

No evidence of metastasis was detected during a thor­ough necropsy of every animal in this study.

Discussion

Most cancers develop through evolutionary stages, ac­quiring new characteristics at each step ("tumor progres-

Top left, Figure 8. Fundus photograph after 4 weekly treatments of DMBA and 7 weekly applications of croton oil, 0.5%. o pigmented fundus lesions are seen.

Top right, Figure 9. Fundus photograph of sam.e animal as illustrated in Figure 8 after an additional 4 treatments with croton oil, 0.5%. A flat pigmented lesion is seen within and just below the center of the medullary ray.

Bottom, Figure 10. Fundus photograph of same animal as illustrated in Figures 8 and 9, taken 3 weeks after the last application of croton oil. The pigmented lesion in Figure 9 is no longer visible. The histology of this eye demonstrated grade 3 atypia as shown in Figure 7.

sion").20 Recently, Clark21 separated both human and ex­perimentally induced cancers into the following stages of tumor development: class lA (e.g. , including typical cu­taneous nevi) lesions usually differentiate or disappear; class IB lesions (e.g., abnormal growth in a cutaneous melanocytic nevus) demonstrate a failure to differentiate accompanied by disorderly growth; classIC lesions (e.g., cutaneous dysplastic nevus) include randomly dispersed

Table 4. Development of Histologically Defined Uveal Melanocytic Hyperplasia

Results

Protocol Positive Negative

Group 1 (control) 0 10

Group 2 (proparacaine and chronic debridement) 7 6

Group 3 (proparacaine and acetone*) 6 10

Group 4 (DMBA [4 treatments]) 2 4 Group 5 (DMBA with chronic acetone) 9 6 Group 6 (DMBA with croton oil 0.25%) 5 2

Group 7 (DMBA with croton oil 0.50%) 9 3

Chronic DMBAt 9 6

• Includes 4 eyes treated identically with HPLC-grade acetone in the previously published study9

t Data from previously published study,9 provided for comparison.

Equivocal

1

4 5 0 6 2 0 3

983

Ophthalmology Volume 99, Number 6, June 1992

Figure 11. Electron micrograph of the choroid from a zone of nodular hyperplasia. The pigmented cells are melanocytes, not melanophages (original magnification, X9025).

atypical cells constituting a formal cancer precursor; class II lesions include in situ and microinvasive proliferations; class III lesions are associated with some potential to es­tablish metastases; and class IV lesions are the metastases themselves.

Conjunctival melanoma seems to conform to this scheme. Primary acquired melanosis progresses through at least two phases, one lacking atypia and one demon­strating atypical melanocytic hyperplasia. 3 Each phase of primary acquired melanosis may co-exist with a nevus.4

By contrast, ophthalmic pathologists tend to separate uveal melanocytic lesions into two groups, nevi and mel­anomas, 22 and do not discuss the possibility of a spectrum of changes from benign hyperplastic lesions through atypical hyperplastic lesions.

The data from our series of experiments suggest, as Gass7 suspected from his clinical observations, that uveal melanocytic proliferations may pass through at least three phases of tumor progression: from hyperplastic lesions that are relatively benign cytologically (similar to lesions developing in eyes not exposed to DMBA), through hy-

perplastic lesions demonstrating randomly distributed cells with cytologic atypia (our cytologic grade 1, Fig 5), and through a stage of confluent atypical cells forming a cluster (our cytologic grades 2 and 3 atypia, Fig 7).

What are the clinical implications of these experimental findings? The cytologically benign hyperplastic lesions in­duced by chronic irritation to the eye (our treatment groups 2 and 3) were entirely subclinical (i.e., not detected by fundus photography). These subclinical lesions (Fig 2) match Clark's description of class lA lesions and also match the histologic description rendered by Naumann et ae3 of human uveal melanocytic nevi. The observation that our experimentally induced benign melanocytic le­sions were not detected clinically is not surprising because the study by Naumann et ae3 that described the histology of human uveal melanocytic nevi was performed on 100 eyes, only 15 of which were removed for a pigmented lesion. In fact, Naumann et ae3 concluded that" ... we believe that it must be extremely difficult or impossible to detect such a lesion by ophthalmoscopy. This fact probably accounts for the varying frequency of such nevi indicated by the clinical literature."

The human counterparts of the second stage of devel­opment, randomly dispersed atypical cells in an otherwise benign lesion (Fig 5), are less clearly defined. However, Gass 7 reported 1 patient with a choroidal nevus who was followed by serial photography for 10 years; the eye was obtained at autopsy. In the legend that accompanies the photomicrograph (Gass' figure 7), Gass7 describes a "pre­dominantly small spindle cell tumor. Some of the larger cells contained nucleoli." The lesion that Gass7 illustrated corresponds to the lesions that we classified as grade 1 atypia (containing scattered significant numbers of atyp­ical cells) and the lesion that Clark designated as class IC.21

The experimentally induced lesions classified by us as grade 3 atypia (Fig 7) are, by definition, composed of con­fluent patches of atypical cells and may represent the bio­logical equivalent of an early cancer (melanoma) lacking the capacity for metastasis (i.e., Clark's class II lesion).21

Therefore, many of the precursor lesions induced in this experimental model do have human counterparts,

Table 5. Effect of Protocol on Cytologic Atypia

984

Low Grade High Grade

Protocol Grade 0 Grade 1 Grade 2

Group 1 (control) 9 2 0 Group 2 (proparacaine) 6 8 3 Group 3 (acetone*) 7 10 4 Group 4 (DMBA [4 treatments]) 3 1 2 Group 5 (DMBA and acetone) 1 10 7 Group 6 (DMBA and croton oil 0.25%) 1 1 4 Group 7 (DMBA and croton oil 0.50%) 0 3 6 Chronic DMBA t 5 5 7

• Includes 4 eyes treated identically with HPLC-grade acetone in the previously published study.9

t Data from previously published study,9 provided for comparison.

Grade 3

0 0 0 0 3 3 3 1

Pe'er et al · Animal Model of Primary Uveal Melanocytic Lesions

Table 6. Histology of the 8 Eyes That Demontrated Clinical Regression after Cessation of All Treatments

Number Protocol Histologic Findings

1 Group 4 (DMBA 4 weeks) Diffuse hyperplasia, grade 2 atypia 2 Group 5 (DMBA and acetone) Mixed nodular and diffuse hyperplasia,

grade 1 atypia 3 Group 6 (DMBA and 0.25% croton oil) Diffuse hyperplasia, grade 3 atypia

4 Group 6 (DMBA and 0.25% croton oil) Equivocal 5 Group 7 (DMBA and 0.50% croton oil) Nodular hyperplasia and grade 3

atypia (Figs 7, 8, 9, and 10) 6 Group 7 (DMBA and 0.50% croton oil) Mixed nodular and diffuse hyperplasia,

grade 2 atypia

7 Group 7 (DMBA and 0.50% croton oil) Negative 8 Group 7 (DMBA and 0.50% croton oil) Negative

and, like most human cancers, the development of human uveal melanoma may involve multiple evolutionary stages in addition to the extreme poles of totally benign nevus on the one hand, and melanoma capable of producing metastasis on the other.

We did not induce any lesions that showed unrestricted growth, either forming a tumefaction (Clark's class III) or spawning a metastasis (Clark's class IV).21 However, we did not extend our experiment for a relatively long period of time. Experimental models of cutaneous melanoma10

-12 and conjunctival melanocytic lesions13

required nearly a year of repetitive exposure to initiators and promoters before tumefactions evolved. Our data in­dicate that the protocol most likely to produce an exper­imental model of primary uveal melanoma capable of metastasis requires initiation with DMBA and promotion with croton oil for longer periods of time than used in the current study.

Two features of our experimental model have not been described in human uveal melanoma. First, we demon­strated clinical regression of eight experimentally induced lesions after promotion was withdrawn. This is a well­described clinical phenomenon in the evolution of human conjunctival melanoma (e.g., the "waxing and waning" of acquired melanosis, described clinically by Reese24

) as well as in the animal model of primary acquired melanosis induced by DMBA, 13 but the histology of the regressed lesion in the conjunctiva has never been reported. It is therefore important to note that a uveal melanocytic lesion was demonstrated histologically in five of the experimen­tally induced lesions that clinically regressed (Table 6). Second, a micronodular form of uveal melanocytic hy­perplasia (Fig 3) also has not been described in humans, but this response is seen frequently in other animal models involving chemical carcinogens. In fact, a multinodular response may be followed by selective overgrowth of one or more of these nodules.25

How relevant is a rabbit model of primary uveal me­lanocytic proliferations to the study of human uveal mel­anoma? Spontaneous uveal melanoma has developed in rabbit eyes demonstrating the full cytologic spectrum of

spindle and epithelioid cells, 26 but reports of primary uveal melanomas in this species are so rare27 that one would not expect such a tumor to develop spontaneously in a large colony of rabbits. The rabbit eye was selected for study because of its relatively large size, permitting serial funduscopy and fundus photography, the relatively low cost of procuring and maintaining rabbits compared with other animals with comparably sized eyes, the relatively long lifespan of the rabbit, and the well-characterized anatomy and physiology of the rabbit eye. 28

How relevant is an animal model of melanoma based on chemical carcinogenesis? Although clusters of individ­uals exposed to chemicals have developed uveal mela­noma,29 most epidemiologic investigation has been fo­cused on the hypothesis that early childhood exposure to ultraviolet radiation may be associated with the subse­quent development of this cancer.30 Interestingly, DMBA­induced cutaneous melanocytic lesions in animal models have been promoted to melanoma after exposure to ul­traviolet light.31 Furthermore, because the application of ultraviolet light to mouse skin pretreated with DMBA and croton oil enhanced the development of cutaneous melanomas in mice,32 our model of primary choroidal melanocytic lesions may be used eventually to experi­mentally investigate the role of ultraviolet radiation in the pathogenesis of human uveal melanoma.

References

I. Clark WH Jr, From L, Bernardino EA, Mihm MC. The histogenesis and biologic behavior of primary human ma­lignant melanomas of the skin. Cancer Res 1969;29:705-27.

2. Clark WH Jr, Elder DE, Guerry D IV, et al. A study of tumor progression: the precursor lesions of superficial spreading and nodular melanoma. Hum Pathol 1984; 15: 1147-65.

3. Folberg R, McLean IW, Zimmerman LE. Primary acquired melanosis of the conjunctiva. Hum Pathol 1985;16:129-35.

985

Ophthalmology Volume 99, Number 6, June 1992

4. Folberg R, McLean IW, Zimmerman LE. Malignant mel­anoma ofthe conjunctiva. Hum Patholl985;16:136-43.

5. Folberg R, McLean IW. Primary acquired melanosis and melanoma of the conjunctiva: terminology, classification, and biologic behavior. Hum Pathol1986;17:652-4.

6. Jakobiec FA, Folberg R, Iwamoto T. Clinicopathologic characteristics of premalignant and malignant melanocytic lesions of the conjunctiva. Ophthalmology 1989;96:147-66.

7. Gass JDM. Problems in the differential diagnosis of cho­roidal nevi and malignant melanomas. The XXXII Edward Jackson Memorial Lecture. Am J Ophthalmol1977;83:299-323.

8. Albert DM, Shadduck JA, Liu HS, et al. Animal models for the study of uveal melanoma. Int Ophthalmol Clin 1980;20(2): 143-60.

9. Folberg R, Baron J, Reeves RD, et al. Primary melanocytic lesions of the rabbit choroid following topical application of 7, 12-dimethylbenz[ a)-anthracene: preliminary observa­tions. J Toxicol Cutaneous Ocul Toxicol 1990;9:313-34.

10. Pawlowski A, Haberman HF, Menon lA. Junctional and compound pigmented nevi induced by 9,10-dimethyl-1,2-benzanthracene in skin of albino guinea pigs. Cancer Res 1976;36:2813-31.

11. Clark WH JR, Min BH, Kligman LHL. The developmental biology of induced malignant melanoma in guinea pigs and a comparison with other neoplastic systems. Cancer Res 1976;36:4079-91.

12. Pawlowski A, Haberman HF, Menon lA. Skin melanoma induced by 7, 12-dimethylbenzanthracene in albino guinea pigs and its similarities to skin melanoma of humans. Cancer Res 1980;40:3652-60.

13. Folberg R, Baron J, Reeves RD, et al. Animal model of conjunctival primary acquired melanosis. Ophthalmology 1989;96: 1006-13.

14. Berkelhammer J, Oxenhandler RW, Hook RR Jr, Hennessy JM. Development of a new melanoma model in C57BL/6 mice. Cancer Res 1982;42:3157-63.

15. Takizawa H, Sato S, Kitajima H, et al. Mouse skin mela­noma induced in two stage chemical carcinogenesis with 7, 12-dimethylbenz[a]anthracene and croton oil. Carcino­genesis 1985;6:921-3.

16. Eisinger M, Marko 0, Weinstein lB. Stimulation of growth of human melanocytes by tumor promoters. Carcinogenesis 1983;4:779-81.

986

17. Kanno J, Matsubara 0, Kasuga T. Histogenesis of the in­tradermal melanocytic tumor in BDF1 mice induced by topical application of 9, 1 0-dimethyl-1 ,2-benzanthracene (DMBA) and 12-o-tetradecanoylphorbol-13-acetate (TPA). Acta Pathol Jpn 1986;36:1-14.

18. Fries PD, Fohrman DS, Char DH. Phorbol ester-induced orbital myositis. Arch Ophthalmol 1987;105:1273-6.

19. Zimmerman LE, Font RL, Ts'o MOM, Fine BS. Application of electron microscopy to histopathologic diagnosis. Trans Am Acad Ophthalmol Otolaryngol1972;76:101-7.

20. Foulds L. Neoplastic Development. Vol. 1. London: Aca­demic Press; 1969;41-96.

21. Clark WH. Tumor progression and the nature of cancer. Br J Cancer 1991 ;64:631-44.

22. Zimmerman LE. Malignant melanoma of the uveal tract. In: Spencer WH, ed. Ophthalmic Pathology: An Atlas and Textbook, 3rd ed. Vol. 3. Philadelphia: WB Saunders, 1986;2072-139.

23. Naumann G, YanoffM, Zimmerman LE. Histogenesis of malignant melanomas of the uvea. I. Histopathologic char­acteristics of nevi of the choroid and ciliary body. Arch Ophthalmol 1966;76:784-96.

24. Reese AB. Tumors of the Eye, 3rd ed. Hagerstown, MD: Harper & Row; 1976;242-62.

25. Farber E, Sarma DSR. Hepatocarcinogenesis: a dynamic cellular perspective. Lab Invest 1987;56:4-22.

26. Brown WH, Pearce L. Melanoma (sarcoma) of the eye in a syphilitic rabbit. J Exp Med 1926;43:807-13.

27. Beniashvili DSh. Spontannaia melanoblastoma krolika. Vopr Onkol 1972;18:84-5.

28. Prince JH, ed. The Rabbit Eye in Research. Springfield, IL; Charles C. Thomas, 1964.

29. Albert DM, Robinson NL, Fulton AB, et al. Epidemiological investigation of increased incidence of choroidal melanoma in a single population of chemical workers. Int Ophthalmol Clin 1980;20(2):71-92.

30. Tucker MA, Shields JA, Hartge P, et al. Sunlight exposure as risk factor for intraocular malignant melanoma. N Eng! J Med 1985;313:789-92.

31. Epstein JH, Epstein WL, Nakai T. Production of melanomas from DMBA-induced "blue nevi" in hairless mice with ul­traviolet light. J Nat! Cancer Ins 1967;38: 19-30.

32. Donawho CK, Kripke ML. Photoimmunology of experi­mental melanoma. Cancer Metastasis Rev 1991; 10: 177-88.