A Study of Iridectomy Histopathologic Features of Latanoprost- andNon–Latanoprost-Treated Patients

Objectives  To examine the histopathologic features of iridectomy specimens frompatients undergoing glaucoma surgery and to compare histologic abnormalitiesin a group of patients with a history of latanoprost therapy with those ina group of patients who had no history of prostaglandin therapy (controls).

Methods  Iridectomy specimens and patient history forms were submitted to thecentral Latanoprost Pathology Center. These were independently examined by3 ophthalmic pathologists in a masked fashion. Specimens were evaluated formalignant, premalignant, and other changes including differences in levelsof pigmentation, degrees of cellularity, inflammation, vascular abnormalities,and changes in the iris pigment epithelium.

Results  Specimens were received from 449 patients with a history of latanoprosttreatment and 142 patients who had no history of treatment with latanoprostor other prostaglandin analogues. No evidence of malignant or premalignantchanges was found. In latanoprost-treated irides, the prevalence of iris freckleswas higher (P = .001) than in control irides,as was the combined number of stromal fibroblasts and melanocytes (P<.001). In a subgroup of specimens received through June 2002,there was no significant difference in mean melanocyte counts (P=.35) obtained by immunohistochemical staining techniques betweenthe latanoprost-treated and control groups.

Conclusions  These findings support previous studies indicating that latanoprost-inducedeye color changes are due to an increased amount of melanin within the irisstromal melanocytes. The increased numbers of freckles may be a focal manifestationof this effect.

New glaucoma drugs including latanoprost, isopropyl unoprostone, travoprost,and bimatoprost are associated with increased pigmentation of the iris insome patients.¹ Our knowledge of this phenomenonis based primarily on clinical observations, with relatively limited histopathologicinvestigation.^2,3 This study wasdesigned to evaluate, in a systematic manner, key histopathologic findingsin iris specimens from patients undergoing glaucoma surgery and to comparethe results in groups defined by a history of latanoprost therapy or no historyof therapy with latanoprost or any prostaglandin analogues. Assessments ofmalignant or premalignant changes including the presence of melanoma, increasedcell number, and mitotic figures were recorded. Other characteristics evaluatedincluded the degree of pigmentation of the anterior border layer and stroma,inflammation, abnormalities of the stromal blood vessels, and pigment epithelialchanges. This is the first large histopathologic study to address the effectsof latanoprost on the human iris, including the safety aspects of latanoprost-inducediris pigmentation.

Iris specimens, obtained at the time of iridectomy during glaucoma surgery,were received in 10% neutral-buffered formalin at the Latanoprost PathologyCenter (LPC) and labeled by the LPC coordinator with a unique study accessionnumber. The formalin-fixed tissue specimens were processed overnight on anautomated tissue processor, embedded in paraffin, serially sectioned at 5 μm,and mounted onto treated slides. The first 3 slides were stained with hematoxylin-eosin,the next 3 slides were bleached with potassium permanganate prior to stainingwith hematoxylin-eosin, and the following 7 slides were left unstained. Allslides were labeled with the unique study accession number assigned to thespecimen. If any tissue remained after sectioning, the paraffin block wasarchived.

Sets of microscopic slides (1 hematoxylin-eosin slide, 1 bleached hematoxylin-eosinslide, and 1 unstained slide) were sent to 3 reviewing pathologists (D.M.A.,H.E.G., and W.R.G.) who examined the set independently and with no knowledgeof therapy history. They completed the accompanying pathology grading form,including an assessment of the quality of the specimens, their location, andtheir orientation. These forms were returned to the LPC coordinator, who completeda composite grading form using specific rules for combining data from the3 grading forms.

An iris freckle (ephelis) was identified by the pathologists as a focalarea of increased pigmentation in the anterior border layer melanocytes withoutan increased number of melanocytes.⁴ This isconsistent with the general definition of an iris freckle in the field ofeye pathology. Using all specimens received by June 2002 and the techniquespreviously described,⁵ 1 of the unstained slideswas stained with S100a and fluorescein isothiocyanate. This stained slidewas examined by a single reviewer (in the laboratory of D.M.A.) who was maskedto treatment history and identified melanocytes and other cells (fibroblastsand clump cells) by their staining characteristics, counted them, and enteredthe counts into a spreadsheet. Specimens were adjudicated using a protocolformulated by the 3 pathologists prior to the beginning of the study.

The patient history form accompanying the specimen included iris color(blue, hazel, or brown); type of glaucoma; ocular history; type of surgery;treatment history, including the start of medical treatment; medication history,including specific medications used; duration of latanoprost therapy (<3months, 3 months to 3 years, or >3 years); other glaucoma medications used;and presence or absence of increased pigmentation after glaucoma medication.The latanoprost group was defined by a history of latanoprost therapy, andthe control group was defined by no history of therapy with latanoprost orany other prostaglandin analogue.

After review by the LPC coordinator, the patient history form and thecomposite grading form, each identified only by the unique accession number,were sent to the Statistical Data Analysis Center (Department of Biostatisticsand Medical Informatics, University of Wisconsin, Madison) for data entry(Oracle database; Oracle, Redwood Shores, Calif) and statistical analysis.The numbers of melanocytes and other cells from the subset of specimens weretransferred separately from the laboratory for their inclusion in this article.This study received institutional review board approval from the Universityof Wisconsin Health Sciences Human Subjects Committee.

Comparisons of groups were based on the Wilcoxon rank sum test for continuousand ordinal variables and the Pearson χ² test for categoricalvariables. Analyses were performed using analysis of variance models for continuousand ordinal variables and logistic regression models for dichotomous variables,which included sex and iris color as covariates. Statistical significancewas set at P<.05. No formal adjustment for multiplecomparisons was used.

Between September 1998 and January 2003, 591 iris specimens were receivedat the LPC through the cooperation of 41 glaucoma surgeons and 22 centers,listed in Table 1. Specimens were receivedfrom 449 patients with a history of latanoprost treatment and 142 patientswho had no history of treatment with latanoprost or other prostaglandin analogues.The duration of latanoprost treatment was recorded for 346 (77%) of the 449patients in the latanoprost group; 57 used latanoprost for less than 3 months,247 for 3 months to 3 years, and 42 for more than 3 years. Table 2 summarizes, by treatment group, relevant information fromthe patient history form and the year of completion of the pathology gradingform. Race and ethnicity, iris color, type of glaucoma, and ocular historywere statistically similar for the latanoprost and control groups; however,there were more women in the control group compared with the latanoprost group(P = .01). With regard to the year of completionof the pathology grading form, there was a statistically significant difference(P<.001) in the distribution of specimens by treatmentgroup.

Evaluation of the composite pathology grading forms indicated that thelatanoprost and control groups were comparable in terms of quality, location,and orientation of the iridectomy specimens (data not shown). All specimenswere obtained from the peripheral iris.

No melanomas were reported on the pathology grading form in either thelatanoprost or control group (data not shown). No statistically significantdifferences in the presence of nevi were noted either in all irides or inblue, hazel, or brown irides (Figure 1).There was, however, a statistically significant difference in the presenceof freckles in the latanoprost group compared with the control group in allirides (P = .001) and in the hazel (P = .01) and brown (P<.001)irides (Figure 2). This difference remainedstatistically significant in multivariable logistic regression models, whichincluded sex and iris color (P = .009).

There was a statistically significant difference between treatment groupsin the number of combined stromal fibroblasts and melanocytes in all irides(P<.001) and in the brown irides (P<.001). In analyses by sex, there was no statistically significantdifference between groups in the number of stromal fibroblasts and melanocytesin the irides of all men (P = .24) or inthe irides of brown-eyed men (P = .32);there were statistically significant differences between groups in the iridesof all women (P<.001) and in the irides of brown-eyedwomen (P<.001). (Table 3).

No statistically significant differences in the degree of pigmentationin the melanocytes of the anterior border layer or stroma, in the mean thickness,or in many stromal features were noted between the latanoprost and controlgroups (Table 4). Free pigment granulesin the stroma (P = .05) were not statisticallysignificantly different; rubeosis, although rare, was statistically significantlydifferent (P = .002) (Table 4). The proportion of specimens determined to have a normaliris did not differ between the 2 groups (data not shown).

Prior to iridectomy, 21 patients were reported to have darkening ofthe iris. All of these patients had received latanoprost treatment. In 6 cases,there was photographic documentation of the darkening. Data from patientswith clinically observed darkening of the iris were included with data fromother latanoprost-treated patients without darkening. No histopathologic abnormalitieswere seen in these specimens. It is presumed that subclinical or unnoticedeye color changes occurred in additional patients.

Using the subset of specimens received by June 2002, no statisticallysignificant differences in melanocyte counts were noted. However, there wasa statistically significant difference in other cell counts (fibroblasts andclump cells) between the latanoprost and control groups for all irides (P = .005; n = 371), brown irides (P = .02; n = 182), and men with blueirides (P = .03; n = 56) (Table 5).

Differences in the iris color of normal eyes are the result of variableamounts of melanin granules within a constant number of melanocytes in thesuperficial stroma of the iris.^5-8 Irismelanocytes seem to reach their genetically determined amount of melanin inearly childhood, and their melanin content usually remains constant in adulthood.Iris color, however, can be affected by a variety of ocular disorders.⁹

Latanoprost, a phenyl-substituted analogue of prostaglandin F_2α, is effective for lowering intraocular pressure. In early trials inthe United States, United Kingdom, and Scandinavia, it was associated withincreased melanogenesis in the irides of patients treated in clinical trials.¹⁰ Typically this change was manifested by a concentricincrease in iris pigmentation appearing after about 6 months of treatment.¹⁰ The incidence of increased pigmentation is apparentlyrelated to the color of the iris and has been reported to range from 5% to70%.^11,12 More recently, otherprostaglandin analogues introduced for the treatment of glaucoma have beenassociated with increased pigmentation of the iris in some individuals.¹ Most of the data regarding this phenomenon involvelatanoprost, which was the first prostaglandin analogue available for thetreatment of glaucoma.

In vitro studies in which human iris tissue or melanocyte cultures wereincubated with latanoprost indicate that the iris darkening associated withlatanoprost treatment is caused by the induction of tyrosinase expression.^13-16 Inan additional in vitro study, Dutkiewicz et al¹⁷ demonstratedthat latanoprost induced tyrosinase activity but did not increase the mitoticindex in a panel of human uveal and cutaneous melanoma cell lines. That studyalso suggested that the adverse effect of latanoprost on in vivo iris pigmentationdid not result from increased cell division but from elevated tyrosinase activity.

The cynomolgus monkey has been described as an effective animal modelfor studying the increased pigmentation of the iris with latanoprost treatment.^10,18,19 Darkening of theiris was noted after about 2 months of treatment, with latanoprost inducingincreased pigmentation in sympathectomized eyes. Nevi and freckles on theiris conjunctiva or eyelids were not affected. The authors concluded thatlatanoprost treatment increased the normal low melanin synthesis in the iridialmelanocytes of the cynomolgus monkey. In latanoprost-treated eyes, the amountof eumelanin increased from 3- to 7-fold, whereas the variation of pheomelanindid not exceed 25%.²⁰

Zhan et al²¹ used Dutch-belted rabbitsfollowing unilateral superior cervical ganglionectomy as an experimental modelfor studying prostaglandin-induced iris color darkening. These investigatorsdemonstrated that sympathetic innervation is required for age-related physiologicdarkening of iris color and that prostoglandins may compensate for sympatheticdenervation to produce the darkening.

Individual case reports and small series of iris specimens studied histopathologicallyfollowing darkening from latanoprost treatment have been reported.^2,3,22-24 Theselight and electron microscopic studies indicate that latanoprost-induced eyecolor change is due to an increased amount of melanin within the iris stromalmelanocytes rather than any increase in melanocyte number.

In our study, no evidence of malignant or premalignant changes was observedin either group based on the absence of melanomas, atypical nevi, cells withan atypical appearance, or increased numbers of mitotic figures. There wasan increased prevalence of freckles in the latanoprost-treated group. We suggestthat the increased number of freckles is a manifestation of focal increasedtyrosinase expression and that the same pathogenesis exists for the more diffusedarkening. An extreme of this change may be the latanoprost-associated, diffuse,uniform dark velvet-brown appearance simulating a diffuse iris melanoma, asdescribed by Tsai et al.² We do not believethat the increase in iris freckles has malignant potential or can lead toany adverse clinical effect on the eye.

There was a statistically significant increase in the number of combinedstromal melanocytes and fibroblasts, most strongly evident in brown-eyed women.It should be noted that with the hematoxylin-eosin staining used for thisstudy, stromal melanocytes and fibroblasts cannot be reliably differentiated.Because of this, immunohistochemical staining was performed for the subsetof specimens received by June 2002, and no statistically significant differencein melanocyte count was seen in all irides or any subgroup. The combined stromaland fibroblast increase noted in the hematoxylin-eosin–stained slidesappears to be due to an excess of fibroblasts and/or clump cells but not melanocytes.

Prostaglandin F_2α as well as prostaglandins E₁ and E₂ has been shown to act as an extracellular factorto regulate cell proliferation.²⁵ ProstaglandinF_2α stimulated DNA synthesis and cell proliferation in quiescentSwiss mouse 3T3 cell cultures^25,26 andin 1 of 2 clones of 3T3 cells from the National Institutes of Health (Bethesda,Md).²⁶ In addition, prostaglandin F_2α can modulate the growth and differentiation of corneal epithelial cellscultured in vitro. Consequently, it is not surprising that an increase inthe number of iris fibroblasts and/or clump cells occurs in some latanoprost-treatedeyes.²⁷

In summary, our results are consistent with previous findings that latanoprost-inducedeye color change is due to an increased amount of melanin within the irisstromal melanocytes and does not involve an increase in melanocyte number.Further examination of several histopathologic characteristics, as noted previously,shows no adverse histopathologic effects in latanoprost-treated irides ascompared with controls.

Correspondence: Daniel M. Albert, MD, MS,Department of Ophthalmology and Visual Sciences, F4/344 Clinical Science Center,600 Highland Ave, Madison WI 53792-3284 (dalbert@wisc.edu).

Submitted for Publication: July 17, 2003; finalrevision received January 26, 2004; accepted April 21, 2004.

Financial Disclosure: None.

Funding/Support: This study was supported bya grant from Pharmacia Corporation, a Pfizer, Inc, company (New York, NY).

Acknowledgment: We thank Kirsten L. Hope, BA,for her diligent effort as an editor.