Eye color change in one patient during the 6-month treatment period with latanoprost. A and B, Right and left eyes before treatment. C and D, Right and left eyes of the same patient after 6 months of treatment with latanoprost (C) and 6 months after trabeculectomy(D).
Light micrograph of a semithin section through an iridectomy specimen (fellow eye specimen; toluidine blue, original magnification ×150). Bar indicates 100 µm.
Electron micrograph showing a clump cell in the iris stroma (fellow eye specimen; uranyl acetate and lead citrate, original magnification ×16 000). Bar indicates 1 µm.
Electron micrograph showing free melanin granules (arrowheads) in the collagen of the iris stroma. The granules are close to a melanocyte (M). The finding was rare in fellow eye specimens and in test eye specimens that were treated with latanoprost (fellow eye specimen; uranyl acetate and lead citrate, original magnification ×17 000). Bar indicates 1 µm.
Electron micrographs of the anterior border layer from the fellow B2 (A) and test B18 (B) iridectomy specimens from the patient who underwent eye color change during the study (uranyl acetate and lead citrate, original magnification ×8000). Bar indicates 1 µm.
Electron micrographs of stromal melanocytes from the fellow B2 (A) and test B18 (B) iridectomy specimens from the patient who underwent eye color change during the study (A, original magnification×9000; B, uranyl acetate and lead citrate, original magnification ×13 000). Bar indicates 1 µm.
Electron micrographs of melanin granules. A, Stromal melanocyte shows melanin granules in the early stages of clumping (arrowheads) (test eye specimen; original magnification ×14 000). B, Vacuolation (arrowheads) of melanin granules is common in iris epithelium(fellow eye specimen; original magnification ×6000). C, Higher-power view of a stromal melanocyte shows premelanosomes (arrowheads) and granules with incomplete melanin filling (asterisk) (fellow eye specimen; original magnification ×34 000). D, Premelanosomes (arrowheads) can also be seen in the stromal melanocyte, as can a giant melanin granule (g) (fellow eye specimen; original magnification ×16 000). (A-D, uranyl acetate and lead citrate.) Bars indicate 1µm.
Electron micrograph of 2 melanocytes that are part of the anterior border of the iris. Their nuclei (N) show atypical features such as nuclear invagination and aggregation of the dark-staining heterochromatin at the nuclear margin (fellow eye iridectomy specimen; uranyl acetate and lead citrate, original magnification ×13 000). Bar indicates 1 µm.
Pfeiffer N, Grierson I, Goldsmith H, Appleton P, Hochgesand D, Winkgen A. Fine Structural Evaluation of the Iris After Unilateral Treatment With Latanoprost in Patients Undergoing Bilateral Trabeculectomy (The Mainz II Study). Arch Ophthalmol. 2003;121(1):23-31. doi:10.1001/archopht.121.1.23
To investigate by masked electron microscopy whether 6 months of topical latanoprost caused pathological changes in the peripheral iris of patients with glaucoma.
Seventeen patients with bilateral primary open-angle glaucoma requiring trabeculectomy were recruited for this study. The iridectomy taken during surgery on the first eye served as a control. The second test eye was treated topically with latanoprost for 6 months before its trabeculectomy. Fourteen patients completed the treatment arm of the study, and 1 of these underwent marked color change. As a result, 31 iridectomy specimens were fixed, coded, and evaluated.
The specimens were evaluated for evidence of stromal inflammation, vascular alterations, and stromal and posterior epithelial degeneration. None of these features was evident in any of the 31 iridectomy specimens. There was evidence of the incidence of free melanin granules in the stroma, melanin turnover, abundance of stromal clump cells, atypical cellular features in melanocytes, and prominence of the anterior border. After code breaking, it was evident that none of these features distinguished the test from the fellow irises in our group. The patient with color change in the test iris did not stand out from the others in this analysis. Qualitative reexamination of further sections after unmasking gave the impression of increased melanin granule numbers in the melanocytes of the anterior border region.
The ultrastructure of the iridectomies from the latanoprost-treated eyes and the fellow eyes conformed to published standards for normal iris. There was no evidence of early ultrastructural changes, which might have been the harbingers of latanoprost-induced iris abnormality.
LATANOPROST, THE intraocular pressure–lowering prostaglandin analogue, has few, if any, systemic side effects1,2 but has ocular side effects that include darkening of eye color.3- 7 The iris color change occurs in hazel eyes, particularly those with a green-brown or yellow-brown coloration, and is rare in those with a homogeneous color.7 As yet, little is known about the morphologic and ultrastructural effects of latanoprost on the iris after either short- or long-term topical use. In addition, the mechanism by which the eye color change takes place is not entirely clear.
It has been suggested that eye color darkening could be brought about by a fixed number of stromal melanocytes increasing their cytoplasmic melanin content3,8; this is in accordance with the known mechanism that distinguishes normal light-colored from dark-colored eyes.9,10 Perhaps there are alternative processes in operation, such as the migration of melanocytes within the stroma; proliferation of melanocytes has even been considered as a potential but worrying mechanism to account for the darkening process.3,8 However, a recent study from our group11 suggests that this latter possibility is unlikely.
Much remains unknown about the effects of topical prostaglandins on iris structure, in particular whether they produce fine structural alteration to the tissue. The present study was conducted to investigate, in a masked fashion, whether there were any cellular or extracellular differences between the iris of an eye that was treated for 6 months with latanoprost as compared with the iris of the fellow eye of the same patient that had not been exposed to latanoprost. Peripheral iridectomy specimens were obtained from patients scheduled for sequential bilateral trabeculectomy. We studied all the iridectomy specimens by light and transmission electron microscopy.
The present investigation was a combined study between 2 centers, with the clinical component being conducted in the Department of Ophthalmology at the University of Mainz, Mainz, Germany, and the morphologic and fine structural evaluation being performed in the Unit of Ophthalmology, University of Liverpool, Liverpool, England.
We recruited 17 patients aged 50 years or older with primary open-angle glaucoma or pseudoexfoliation glaucoma who were to undergo sequential bilateral trabeculectomies because of inadequate intraocular pressure (IOP) control. Exclusion criteria included presence of any features of angle-closure glaucoma, laser trabeculoplasty within 3 months of recruitment to the study, evidence of ocular inflammation or infection, and the wearing of contact lenses. Patients were informed fully about the purpose of the study, and each gave written informed consent (approvals for each aspect of the investigation were provided by the Ethical Review Committee of the Land Rheinland-Pfalz, Mainz). Thereafter they underwent a standard trabeculectomy with the removal of a 1 × 2-mm square tissue block centered on the Schlemm canal as described elsewhere.12 A basal iridectomy was taken at the 12-o'clock position.
The eye that was selected for operation first we designated the fellow eye; it was the eye with the higher IOP, the greater field loss, or both. The second eye (study eye) was scheduled for a trabeculectomy procedure 6 months later and, while waiting for surgery, was given 1 drop (50 µg/mL) of latanoprost nightly for the 6-month period. Six months is the usual waiting period at the University of Mainz between operations for bilateral trabeculectomy. All eyes received latanoprost for the first time. All eyes had been given maximal tolerated therapy before receiving latanoprost. No washout preceded the study period, and other antiglaucomatous therapy was permitted when it was required to reach an appropriate level of IOP control for a given patient.
The patient visits for screening were arranged at baseline, week 2, month 3, and month 4.5, and on the day of surgery at month 6. Six weeks after surgery, a follow-up visit was performed. At each visit a history was taken and, in addition, visual acuity testing, Goldmann applanation tonometry, slitlamp examination, and ophthalmoscopy were undertaken. To evaluate iris color change, standardized photographs were taken en face of the whole eye at baseline and 3, 4.5, and 6 months, and these were coded and then assessed by experienced masked observers. The patients' eye color was classified as brown, blue, green, gray, or hazel. The first 4 were of uniform color, while the hazel eyes were brown mixed with 1 of the other 3 colors (Table 1).
The treatment was open to the patients and the treating ophthalmologists in Mainz but masked to the analyzing morphologists in Liverpool. Of the 17 patients, all had surgery on the fellow eye, but only 14 underwent surgery on the study eye. Two patients who withdrew had finished the 6-month period of latanoprost treatment but, because of the level of IOP control obtained, retired from the program without surgery to the study eye. The third patient died 6 weeks after surgery to the fellow eye of problems unrelated to the study. In all, 31 peripheral iridectomy and trabeculectomy specimens were available for morphologic examination.
The specimens were coded and fixed in 2.5% glutaraldehyde buffered with 0.1M Sorensens phosphate buffer; the iridectomies were first placed, anterior border downward, on small strips of filter paper to minimize wrinkling and rolling up of the delicate tissue. The coded fixative bottles were transported to the morphology center, where the histologists remained unaware of the patient or treatment identity of the 31 iridectomies and trabeculectomy blocks up to the completion of the study.
The masked peripheral iridectomy specimens, while still on their support filter paper, and the trabeculectomies were processed for light and electron microscopy in a standardized manner. The tissues were removed from primary fixative, washed in phosphate buffer, and then secondarily fixed in 1% osmium tetroxide in 0.1M Sorensens buffer for 1 hour. After an additional buffer wash, the specimens were dehydrated in graded alcohols, cleared in propylene dioxide, embedded in an epoxy resin mixture (Araldite/Epon; Agar Scientific Ltd, Essex, England), and cured overnight in an oven. Semithin sections (0.5-1.5µm) and ultrathin sections (50 nm) were cut on an ultramicrotome (Ultracut E; Reichert Ophthalmic Instruments, a division of Leica Microsystems Inc, Buffalo, NY). Semithin sections, stained with toluidine blue, were viewed with a Polyvar light microscope (Reichert Ophthalmic Instruments) with the use of conventional or differential interference contrast optics. Ribbons of ultrathin sections where mounted on single-slot copper grids supported by a Formvar film (Agar Scientific Ltd). The ultrathin sections were stained with uranyl acetate and lead citrate before being examined in a transmission electron microscope (Phillips CM10).
At least 3 sequences of semithin and ultrathin sections were examined for each iridectomy specimen. A photographic portfolio was made of each specimen that included conventional light microscopy mapping of the specimen, differential interference contrast microscopic documentation where appropriate, construction of a montage using overlapping electron micrographs, and higher-power electron micrographs concentrating on cellular detail (particularly in the stromal melanocytes).
Several features were subjected to particular scrutiny. The posterior epithelial cells and the iris stroma were examined for evidence of degeneration. In addition, the stromal blood vessels were studied for evidence of structural change; these and the surrounding tissue were examined for the presence of inflammatory cells. The incidence of stromal clump cells was noted, as was the presence of extracellular melanin in the tissue. The thickness of the anterior border layer was assessed and stromal and anterior border melanocytes were examined for ultrastructural evidence of melanin turnover (presence of melanosomes types 1, 2, and 3; evidence of melanin granule degeneration) and cellular atypia.11,13
The previously mentioned features were given a subjective score by one of us (I.G.) after full examination of each coded specimen; other investigators provided interobserver checks of the scoring. The scores were as follows:+++, marked; ++, obvious; +, present; +/−, marginal; and −, absent. These were broad categories that were allotted to each feature in each specimen on the basis of assessment of ultrathin sections during the screening sessions at the electron microscope and then subsequent detailed examination of the electron micrographs generated during the investigations. Throughout the analysis, the 31 specimens remained fully masked to the sectioners and the observers. Assignment of a given score was based on abundance, prominence, or both where appropriate (eg, the score for free melanin granules was based on abundance, thickness of the anterior border was scored mostly on prominence, and clump cells had both features taken into consideration).
At the baseline evaluation, the mean ± SD IOP with maximal tolerated therapy was 25.0 ± 5.0 mmHg in the fellow eyes and 23.5 ± 4.6mmHg in the study eyes. The IOP in the fellow eyes decreased after surgery so that, at the 6-month postoperative follow-up, IOP was 16.1 ± 5.5mmHg. The mean IOP of the study eyes was 18.1 ± 3.7 mmHg by the second week and 19.8 ± 3.6 mmHg at the last (6-month) assessment before surgery.
Of the 17 patients, 5 were graded as having brown irises, 6 as hazel, and 6 as blue, green, or grey (Table 1). In all cases the irides of both eyes were graded as having the same color. The individual grading is given in Table 2. Of the 17 patients who started the study, 2 developed an eye color change by the end of the 6-month period of treatment with latanoprost. Both were initially graded as having a brown color. In one patient the color change was sufficiently striking that it was evident to our clinical assessors at the 6-month assessment and also from the photographic evaluation conducted later (Figure 1). The treated eye was originally brown close to the pupil, with more green than brown toward the periphery, and it became uniformly and intensely brown after exposure to latanoprost. The other patient had a more subtle color change that became apparent only at the photographic evaluation conducted by masked observers. This patient also had similar green-brown iris coloration that became more uniformly brown, but not with the same intensity seen in the previous patient. This second patient was 1 of the 3 who did not undergo trabeculectomy on the study eye because effective IOP control was established with latanoprost. He declined surgery within the study period.
The iris specimens kept their shape and orientation on the filter paper used as a support before fixation and up to the latter stages of processing(Figure 2). Unfortunately, 3 specimens exhibited features of excessive dehydration, which may have been produced by the combined action of the filter paper and the drying caused by theater lights. General examination of the specimens showed no distinctive pathological feature in any of the 31 specimens. Some specimens exhibited disruption of the posterior cuboidal epithelium that was thought to have arisen either from surgery or from handling before fixation. In our experience, trauma of this kind is often seen in peripheral iridectomy specimens when subjected to morphologic study.
In none of the 31 specimens examined was there any evidence of degeneration of the posterior pigment epithelium, stromal degeneration, vascular changes, or inflammation (Table 3). On the other hand, the incidence of clump cells (Figure 3) was marked in 10 of the specimens but virtually absent from others. Extracellular melanin granules were not identified in 6 specimens and not abundant in any. Where "free" melanin granules were identified, they were usually found on the anterior border layer or between the collagen fibrils of the stroma close to the anterior border layer (Figure 4). The anterior border layer in approximately half of our specimens was not pronounced, consisting of little more than 2 or 3 cell layers, whereas in the other half it was more evident, reaching to about 5 cells thick. The incidence of melanin granules in the anterior border melanocytes was variable(Figure 5 and Table 3) from specimen to specimen and within each specimen.
The melanocytes of the stroma (Figure 6) and of the anterior border layer (Figure 5) were examined for the presence of premelanosomes, immature melanosomes, and incomplete melanization of melanin granules (Figure 7A, C, and D). Evidence of granule degradation, such as giant melanosome formation (Figure 7D) and melanin clumping (Figure 7A), was also noted. Collectively, these features were taken as being fine structural evidence of melanin turnover. Only 4 of the 30 specimens were graded as having obvious melanin turnover, whereas in 6 we failed to observe any ultrastructural features of turnover in the iris melanocytes (Table 3). The posterior pigment epithelial cells were not included in this part of the analysis, but occasional premelanosomes and immature granules could be seen in their cytoplasm, while vacuolated granules and giant melanin granules were commonplace (Figure 7B).
Characteristic features of nuclear atypia include margination of nuclear heterochromatin, prominent nucleoli, heavy nuclear indentation, and nuclear vacuolation. Such features are usually recorded by conventional light microscopy; however, they can also be seen by electron microscopy. If anything, they are more obviously visible by electron microscopy (Figure 8) than by light microscopy. In about half of the specimens atypical nuclei were present, but never in large numbers, while in the remainder, atypia was either not observed or marginal (Table 3).
On the basis of the scores given to each coded specimen and the general light and electron microscopic appearance, a survey was carried out. The morphologic analysis team predicted whether the irises had been treated with latanoprost, and predictions were subdivided into suspect and highly suspect. After code breaking, it was evident that correct prediction was poor, with only 5 (36%) of 14 eyes properly allocated to the latanoprost treatment group (study eyes) and 7 (44%) of 16 control specimens correctly placed in the nonlatanoprost group (fellow eyes).
Comparison of grading scores from the appropriate study and fellow eye pairs also provided no support for the premise that latanoprost treatment was predictable on morphologic grounds. Three patients did not undergo surgery on their study eye, so there were 14 complete pairs (Table 2). The 5 features for which there was morphologic variation were listed, and if all the features were more marked in the test than the fellow iris, then there would be 5 pluses for that patient. Two patients had 3 pluses, whereas 8 patients had only 1 plus or none. Patient 2, who underwent eye color change in the study eye, had 2 pluses for clump cell incidence and atypia, but the other 3 features had an equal score (Table 2). The study iris (B18) of this patient did have obvious clump cells in the stroma; atypia was present, but our morphologic assessment of melanin turnover was less than that of some other specimens in the series(Table 3).
After code breaking, iridectomy specimens B2 and B18 from patient 2 were reexamined under the electron microscope because this patient underwent marked color change in the treated eye (see Figure 1) (iridectomy specimen B18). Careful comparison was made between melanocytes in the anterior border layer (Figure 5) and in the stroma (Figure 6). It did seem that in the anterior border region the melanin granules were apparently more abundant and more variable in size in the test specimen(Figure 5B) than in the control specimen. No such difference was noted when stromal melanocytes were compared(Figure 6).
A previous study of latanoprost effects on the iris by our group13 failed to show any evidence of tissue alteration; in particular, there was no evidence of cellular proliferation in the adult iris, latanoprost treated or otherwise. The present study has several advantages over the previous investigation. The latanoprost treatment period was twice as long (6 months), the control specimens came from the fellow eyes of each of the test patients, and every specimen was subjected to fine structural analysis rather than just wax histologic examination. Thus, there were key strengths in the study design that allowed for a more thorough and effective structural examination of latanoprost-treated tissue than has been possible so far. The emphasis in this work was to attempt to identify early ultrastructural changes that might have been the harbingers of latanoprost-induced iris disease.
Darkening of iris coloration is a common side effect of topical latanoprost treatment associated mostly with hazel rather than homogeneously colored irises.1,3,7,8 The basis of the color change is suspected to be prostaglandin-provoked melanogenesis3,8 rather than any melanocyte proliferation.8,13 To date, however, the morphologic and biological investigations of prostaglandin-provoked melanogenesis are few, although it has been shown in the monkey iris, on the basis of the incorporation of a radiolabeled false melanin precursor, that latanoprost treatment associated with iris color change produces an increased uptake of the precursor in stromal melanocytes.14 Latanoprost does not seem to cause cultured melanocytes to proliferate10 but does increase tyrosinase transcription in these cells.15,16 With some reservations, 17 on the whole the in vitro findings are in line with the observation from monkey eyes with latanoprost-induced color change, that there is an increase in the melanin granule content of the iris melanocytes but no increase in cell number.3
Excessive melanogenesis itself could be a potential pathological problem. Overproduction of melanin granules might lead to their release into the stroma to provoke an inflammatory reaction18 or cytotoxicity.19 Indeed, if free melanin granules were to escape into the aqueous humor, then pigmentary glaucoma may ensue.20 The findings of the present study indicate that, in the short-term period of 6 months, none of the previously mentioned events are probable. To that end, the latanoprost-treated iridectomy specimens were free of any signs of inflammation or degeneration and, as such, the findings were consistent with those of a previous case report11 and wax histologic study13 by our group. Indeed, the ultrastructure conformed to published standards for normal iris tissue.21
Extracellular melanin granules were found sporadically in the iris stroma, but they were not abundant and their presence bore no relationship to latanoprost treatment. Usually the granules were large and of the size normally associated with the posterior epithelium.21,22 We suspect they had been released and dispersed by the trauma of handling during processing or at the time of surgery. The scavaging phagocytes of the iris stroma are clump cells, and they engulf debris, including free melanin granules.21,23 It seemed reasonable that greater mobilization of these cells would be a consequence of disordered melanin production; however, we found no association between clump cell incidence and latanoprost treatment.
Whether the incidence of melanin granules in the melanocytes was increased after latanoprost therapy, particularly in the patient who underwent eye color change, remains to be determined by quantitative analysis. However, there was no superficial morphologic evidence of this on the basis of masked conventional light and electron microscopic examination, so the difference was not gross. The higher incidence of melanin granules that distinguishes the iris melanocytes of brown from blue irises needs carefully computerized evaluation to be shown clearly, 9 so our findings are hardly surprising. Recently, latanoprost was reported to increase tyrosinase messenger RNA expression in about half of 13 brown irises and in only 1 of 7 hazel irises in organ cultures.16 In our study, the 2 eyes that underwent an iris color change were also brown.
The attempt to assess melanin turnover on the basis of ultrastructural features is crude but not novel. The morphologic features considered important by Schraermeyer24 for the identification of melanogenesis in ocular cells were the presence of (1) premelanosomes, (2) early-stage melanosomes, (3) melanin granules only partly melanized, and (4) mature melanin granules with features of degeneration. Evidence of melanogenesis in our iris stromal melanocytes was present, but it was variable between specimens. Since the work of Hu and Mah, 25 it has been thought likely that iris melanocytes synthesize melanin throughout life, so that some evidence of melanin turnover, in both the fellow and test iridectomies, was expected.
The variable fine structural evidence of melanin turnover, in this study, bore no clear relationship to latanoprost treatment, and, for that matter, the one iris that underwent color change during the study did not stand out from the rest in this context. However, it would need to have been a pronounced difference to have shown up by our masked scoring system. Our qualitative observation, after unmasking of the specimens, that there may be more melanin granules in the anterior border melanocytes of the color-changed iris compared with its control is in itself weak. The other subjective observation that we could not find a clear-cut difference in the abundance of melanin granules between the stromal melanocytes of the same 2 iridectomy specimens is also weak. What is evident is that differences, if they exist, are not profound and will require entensive morphometrics to highlight them. Currently we are subjecting the iris with color change and its partner from the fellow eye to intense quantitative evaluation by computerized image analysis similar to that used by others to compare melanocyte pigmentation in brown and blue eyes.9,10
Corresponding author and reprints: Norbert Pfeiffer, MD, Department of Ophthalmology, Johannes Gutenberg-Universität, Langenbeckstraße 1, 55131 Mainz, Germany (e-mail: firstname.lastname@example.org).
Submitted for publication July 10, 2001; final revision received August 7, 2002; accepted September 3, 2002.
This study was supported by Pharmacia and Upjohn, Stockholm, Sweden.
We thank Kathryn P. B. Cracknell, BSc (Hons), for her technical contribution to this project and Mary D. Kelley for her secretarial assistance.