Infiltrating macrophages in extratumoural tissues after brachytherapy of uveal melanoma

Authors


Päivi Toivonen, MD
Department of Ophthalmology
Helsinki University Central Hospital
Haartmaninkatu 4 C
PL 220
FI-00029 HUS
Helsinki
Finland
Tel: + 358 9 4717 3131
Fax: + 358 9 4717 5100
Email: paivi.toivonen@hus.fi

Abstract.

Purpose:  To compare distribution of macrophages in extratumoural ocular tissues in enucleated eyes with irradiated and nonirradiated uveal melanomas to find out how irradiation affects distribution of macrophages so as to gain insight into their potential routes of migration and changes in local inflammatory responses.

Methods:  Thirty-four matched pairs of primarily enucleated nonirradiated and secondarily enucleated irradiated eyes with choroidal and ciliary body melanoma were stained with mAb PG-M1, and the extratumoural immunopositive elements were counted under the microscope. Main outcome variables were the number of macrophages in the sclera underlying the tumour, in the choroid adjacent to the tumour, and in the ciliary body. The number of macrophage aggregates in the anterior ipsi- and contralateral episclera adjacent to the limbus was also counted.

Results:  Macrophages were more numerous within the sclera under the tumour in irradiated eyes when compared to primarily enucleated eyes (median 1514 versus 619/mm², p = 0.0001), and more aggregates of episcleral macrophages adjacent to the limbus were found after irradiation (ipsilateral side, median 132 versus 0, p = 0.0034; contralateral side, median 79 versus 0, p = 0.014). In primarily enucleated eyes, increasing numbers of tumour-infiltrating macrophages were associated with presence of higher numbers of macrophages in the ciliary body (p = 0.003) and the adjacent choroid (p = 0.044), whereas in the irradiated eyes, increasing numbers of tumour-infiltrating macrophages (p = 0.010) and increasing extent of necrosis (p < 0.001) were associated with higher numbers of intrascleral macrophages underlying the tumour.

Conclusions:  Resident macrophages are present in extratumoural tissues in eyes with uveal melanoma. Brachytherapy may alter their route of migration and increase the number of macrophages in the sclera and episclera. Histopathologically detectable episcleral aggregates of macrophages adjacent to the limbus are detected predominantly after irradiation, a population of which is clinically visible as episcleral deposits after irradiation.

Introduction

Both primarily enucleated and irradiated uveal melanomas contain tumour-infiltrating macrophages (Messmer et al. 1992; Schilling et al. 1997; Mäkitie et al. 2001; Clarijs et al. 2003; Toivonen et al. 2003). Larger numbers of them in the former are associated with shorter survival (Mäkitie et al. 2001) and monosomy 3 (Maat et al. 2008), which in itself is one of the strongest prognostic markers of metastatic disease known at the moment (Prescher et al. 1996; Sisley et al. 1997; White et al. 1998; Aalto et al. 2001). On the other hand, macrophages may be a future tool in cancer immunotherapy because they have a potential role in cancer vaccines as manipulated antigen-presenting cells targeted against tumour cells (Schmollinger & Dranoff 2004).

Even in normal human uvea, resident macrophages are the third most numerous cell type (Hogan et al. 1971). Macrophages play a role in physiological processes, e.g. by phagocytosing the regressing vessels of the tunica vasculosa lentis (Djano et al. 1999; McMenamin et al. 2002). Thus, macrophages are not always a sign of a pathological process.

In Europe and North America, the first line of treatment for uveal melanoma is brachytherapy (Gunduz et al. 2000; Lommatzsch et al. 2000; Shields et al. 2000; Diener-West et al. 2001). After brachytherapy, clinically visible pigmented episcleral deposits appear and persist for several years in the majority of eyes both over and distant from the tumour base (Toivonen & Kivelä 2006). These deposits, which range from black to brownish spots to small patches up to 3 mm in diameter, are likely to be macrophage related. Their number is associated with plaque size, ruthenium as compared to iodine isotope and extent of tumour regression. Also, radiation atrophy of retinal pigment epithelium and choroid may contribute to the formation of the deposits.

Brachytherapy probably modulates factors such as microvascular density and the number and distribution of tumour-infiltrating macrophages in uveal melanoma (Toivonen et al. 2003). How irradiation affects the migration of macrophages within the normal tissues of eyes with uveal melanoma has not been studied. In general, when new antigen is introduced to the anterior chamber, macrophages migrate from the eye to the spleen and other lymphoid organs and act as antigen-presenting cells, contributing to anterior chamber-associated immune deviation (ACAID) (Wilbanks & Streilein 1991; Streilein et al. 1992). Alternatively, it has been postulated that part of the antigen flux from the anterior chamber may occur in a soluble, non-cell-associated form (Camelo et al. 2006). The routes of migration of macrophages that are associated with progression and regression of uveal melanoma are presently unclear.

We undertook a case–control analysis to compare the number of macrophages in normal intraocular tissues in irradiated and nonirradiated eyes with uveal melanoma. Our goal was to investigate how irradiation affects local inflammatory reactions as measured by the number of macrophages and to chart and compare the distribution of macrophages in these eyes to gain at least an indirect insight into the routes, which migrating macrophages may be taking.

Patients and Methods

Aims of the study

The primary aim was to compare the number of macrophages infiltrating extratumoural tissues in eyes with irradiated and nonirradiated uveal melanoma. A secondary aim was to identify aggregates of macrophages large enough to be visible as pigmented episcleral deposits after brachytherapy in these two group of eyes.

The study followed the tenets of the Declaration of Helsinki and was approved by the institutional review board.

Inclusion criteria and study population

All eyes with a choroidal and ciliary body melanoma removed after having been treated with brachytherapy with cobalt, ruthenium and iodine plaques between 1981 and 2002 at the Department of Ophthalmology, Helsinki University Central Hospital were eligible if tumour tissue remained in the block and if a matched pair with a nonirradiated, primarily enucleated melanoma was found (Toivonen et al. 2003). This centre managed an average of 95% of uveal melanomas treated with brachytherapy in Finland.

Files were searched from March 1981, when brachytherapy of uveal melanoma was for the first time used in this centre, to August 2002. A total of 56 consecutively enucleated irradiated eyes were identified. In two blocks, no residual tumour remained, leaving 54 of the 56 tumours for matching.

Matched pairs for the irradiated tumours were drawn from a consecutive series of 292 uveal melanomas that were enucleated between 1962 and 1981 as described previously (Toivonen et al. 2003). During this period, brachytherapy was not available making the series essentially population based and unselected.

Matching was based on four variables associated quantitatively or qualitatively with the presence of tumour-infiltrating macrophages (Mäkitie et al. 1999a,b, 2001). Efficient matching variables should be strongly associated with outcome variables but unassociated with each other (Schlesselman 1982). Because largest basal diameter (LBD) and tumour height are related, LBD was not matched, expecting that matching for height would sufficiently reduce bias because of LBD (Schlesselman 1982; Toivonen et al. 2003):

  • 1Tumour location: (i) ciliary body involved versus (ii) uninvolved.
  • 2Height of the tumour at primary treatment: (i) <8 mm versus (ii) ≥8 mm.
  • 3Cell type of the tumour at enucleation: (i) spindle versus (ii) nonspindle [mixed or epithelioid] versus (iii) necrotic.
  • 4Pigmentation at enucleation: (i) amelanotic to weak versus (ii) moderate versus (iii) strong, determined by sorting unstained sections on white tissue paper under incandescent light (Mäkitie et al. 2001).

From the 292 primarily enucleated eyes, a matched pair was found for 34 of the 54 irradiated eyes (inclusion ratio, 63%) (Toivonen et al. 2003). Of the 34 tumours, 4 (12%) were treated with a cobalt-60 plaque, 21 (62%) with a ruthenium-106 plaque, and 9 (26%) with an iodine-125 plaque; seven of these irradiated tumours (20%) underwent retreatment later.

As previously reported, the gender distribution was equal in the matched pairs, and the age at enucleation tended to be lower in the primary enucleation group. Matching balanced tumour height and, indirectly, LBD (Toivonen et al. 2003). In eyes managed with brachytherapy, significantly more necrosis, a lower microvascular density (MVD), and less frequent loops and networks, were found (Toivonen et al. 2003). MVD was lower in irradiated tumours than in the matched nonirradiated tumour in 21 (66%) of the 32 matched pairs in which MVD could be evaluated.

Brachytherapy

Ruthenium plaques were bought from BEBIG Isotopen- und Medizintechnik GmbH (Berlin, Germany). Iodine applicators were crafted to conform with the ruthenium plaques. (Heikkonen et al. 1992) Cobalt treatment was performed by Stallard’s applicators. Five 0.5- mm-thick unrimmed, noncollimating ruthenium and iodine plaques were used: CCA (diameter 15 mm, circular), CCB (20 mm, circular), CCC (25 mm, circular), COB (20 mm, notch for the optic nerve) and CIB (20 mm, notch for the limbus). Iodine seeds were attached with silicone rubber, which increased plaque thickness to 1.0–1.5 mm. The diameters of the cobalt plaques, used for four patients, were 15 mm (CKA-3) and 20 mm (CKA-4). The radiation dose to the tumour apex and base were calculated from the dose rate and treatment time. If the patient had had multiple treatments, the cumulative dose was used.

The median dose to tumour apex was 86 Gy, and the median time from brachytherapy to enucleation was 1.5 years (range, 0.13–8.5). Of the irradiated eyes, 18 (53%) were enucleated after partial regression (because of tumour regrowth or nonresponse) and 16 (47%) after maintained regression (because of complications).

Clinical data

The height and LBD of the tumour and the presence or absence of tumour growth through Bruch’s membrane were based on A- and B-scan ultrasonography (US), clinical examination or both. The last US measurement preceding enucleation and, for tumours managed by primary enucleation, the macroscopic measurement made by the ophthalmic pathologist was taken to be tumour height at enucleation.

Immunoperoxidase staining

The paraffin blocks were cut at 5 μm. Immunostaining was performed using the avidin–biotinylated peroxidase complex method (Vectastain ABC Elite Kit, Mouse IgG; Vector Laboratories, Burlingame, CA, USA) as described previously in detail (Fuchs et al. 1992).

The mouse monoclonal antibody (mAb) PG-M1 (IgG3; lot 2562; Dakopatts; diluted 1:50) to the CD68 epitope was used to label macrophages. CD68 is an intracytoplasmic 110-kDa glycoprotein of lysosomal granules, which is expressed by macrophages in most human tissues. PG-M1 immunostains tumour-infiltrating macrophages in uveal melanoma more consistently than other tested anti-CD68 antibodies (Mäkitie et al. 2001).

Pretreatment with 0.4% (wt/vol) pepsin (2500 FIP-U, E. Merck, Darmstadt, Germany) in 0.01 m hydrochloric acid for 15 min at 37°C enhanced immunostaining and reduced background with the antibody used. To differentiate immunoreaction from melanocytes of the choroid and ciliary body and to evaluate it in pigmented tumours, the peroxidase reaction was developed with 3,3′-diaminobenzidine tetrahydrochloride, and melanin was then bleached overnight with 3.0% (vol/vol) hydrogen peroxide and 1.0% (wt/vol) disodium hydrogen phosphate, as described previously (Kivelä 1995).

Histopathologic data

The matched set of 34 irradiated and primarily enucleated eyes was analysed under a light microscope (Olympus BH-2; Olympus, Tokyo, Japan) to count CD68-positive macrophages within the sclera, choroid, ciliary body and episclera.

Intrascleral macrophages under the tumour

The area with visually densest immunopositive elements within the sclera underneath the tumour base was identified under 10 × magnification and photographed for counting under 40 × magnification (area, 218 × 174 μm).

Macrophages in the choroid adjacent to the tumour

The outermost etched rectangle of the photography eyepiece (Olympus WK 10 × /20) was aligned with the edge of the tumour on both sides under 10 × magnification. The centre crosshair identified the area of the choroid to be photographed for counting under 40 × magnification (distance from tumour edge, 0.65 mm). The thickness of the choroid was also measured if the choroid did not fill the entire image height. This count was excluded if this area coincided with the optic disc or ciliary body.

Macrophages in the ciliary body

The outermost etched rectangle was aligned with the chamber angle under 10 × magnification, and the crosshair identified the area of the ciliary body to be photographed for counting under 40 × magnification (distance from chamber angle, 0.65 mm). Immunopositive elements were counted primarily from the ciliary body ipsilateral to the tumour. In case of equal distance to tumour margins, both sides were photographed and the mean count was used for analysis. When the tumour infiltrated the ciliary body, the contralateral ciliary body was evaluated.

For each of the three areas photographed, all CD68-immunopositive elements at least 3 μm in size and clearly separated from one another were counted from the digital photographs using image analysis software (Olympus DP-10 Soft, version. 3.0; Soft Imaging System GmbH, Münster, Germany).

Episcleral macrophages adjacent to the limbus

The outermost etched rectangle was aligned with the chamber angle ipsilateral and contralateral to the tumour under 2 × magnification. The episclera and the conjunctiva, when present, and the outermost sclera coinciding with the crosshair were photographed under 40 × magnification (distance from chamber angle, 3.2 mm).

All CD68-immunopositive elements and clusters at least 8 μm in size were counted as aggregates of cells potentially visible clinically as deposits. The size limit was based on knowledge that erythrocytes, the diameter of which is 8 μm, can be routinely visualized with biomicroscopy. The ipsilateral and contralateral counts were analysed separately.

Tumour-infiltrating macrophages

The number of tumour-infiltrating macrophages was estimated semiquantitatively according to Mäkitie et al. (2001) by comparing the density of CD68-positive cells in nonnecrotic areas of the tumour to published standard photographs (few versus moderate versus high numbers of cells). Confluent immunopositive cells in necrotic areas did not influence the grading.

Tumour necrosis

The area of necrosis as percentage of the entire tumour was recorded from haematoxylin–eosin stained sections.

Statistical analysis

All analyses were performed with the Stata statistical software package (release 9.0; Stata Co., College Station, TX, USA), StatXact-3 (Cytel Software, Cambridge, MA, USA) and GraphPad Prism (release 4.01; GraphPad Software, San Diego, CA, USA).

The Wilcoxon signed-rank test was used to compare distributions of continuous variables, and the Stuart–Maxwell test and its trend version to compare unordered and ordered contingency tables, respectively.

When analysing the interrelationships of tumour characteristics and number of macrophages, Spearman rank correlation coefficient, nonparametric test for trend and Kruskal–Wallis test were used to compare continuous variables, and Kruskal-Wallis test was used to compare singly ordered contingency tables. The number of macrophages was given per 1 μm². For tabulation, tumour necrosis was divided in tertiles (<5%, 5–29%, 30–99%), and p-values for both categorized and continuous data analysis are given, when appropriate. p < 0.05 was considered statistically significant, and all tests were two-tailed.

Results

Interrelationship between macrophages in extratumoural tissues and tumour characteristics

The CD68-immunopositive elements in the sclera, choroid and ciliary body (Fig. 1) could be counted in 71%, 59% and 44% of the matched pairs, respectively. Because 13 (38%) of the 34 matched tumours involved the ciliary body, only the contralateral ciliary body was available for counting in these eyes.

Figure 1.

 Haematoxylin–eosin staining of an irradiated, secondarily enucleated eye with a uveal melanoma. Boxes represent the photographed areas of extratumoural tissues from which the CD68-immunopositive macrophages were counted: 1) the sclera underneath the tumour base, 2) the normal choroid adjacent to the tumour, 3) the ciliary body, and 4) the ipsi- and contralateral episclera near the limbus.

Nonirradiated eyes

No significant associations were found with tumour characteristics and the number of macrophages in extratumoural tissues (Table 1).

Table 1.   Interrelationship between the number of extratumoural macrophages and tumour characteristics.
CharacteristicNumber of extratumoural macrophages (per 1 mm²)Number of macrophage-related aggregates (per 1 mm²)
Within sclera beneath the tumour median (range)Within choroid adjacent to the tumour median (range)Within ciliary body closest to the tumour median (range)Within limbal episclera ipsilateral to the tumour median (range)Within limbal episclera contralateral to the tumour median (range)
  1. The number of macrophages is scaled to 1 mm² for clarity. LBD, largest basal tumour diameter.

  2. * Kruskal–Wallis test (categorical)/Spearman rank correlation (continuous).

  3. Nonparametric test for trend (categorical)/Spearman rank correlation (continuous).

  4. Nonparametric test for trend.

  5. § Kruskal–Wallis test.

Nonirradiated eyesn = 30n = 27n = 24n = 22n = 17
All tumours619 (132–1711)814 (237–3952)737 (132–2422)0 (0–316)0 (0–105)
Tumour height at enucleation*p = 0.50/0.96p = 0.39/0.22p = 0.48/0.61p = 0.53/0.10p = 0.54/0.32
 < 8 mm632 (132–1711)737 (237–2229)711 (132–1790)0 (0–316)0 (0–105)
 ≥ 8 mm527 (158–1343)814 (441–2026)632 (316–1553)26 (0–105)0 (0–0)
Tumour LBD at enucleationp = 0.59/0.60p = 0.39/0.30p = 0.31/0.60p = 0.13/0.22p = 0.65/0.96
 ≤ 10 mm632 (132–1711)751 (434–1001)645 (132–922)0 (0–0)0 (0–79)
 >10–15 mm658 (263–1343)913 (237–3952)790 (342–2422)26 (0–105)0 (0–105)
 >15 mm606 (158–1053)1020 (553–1130)632 (316–1790)0 (0–316)0 (0–53)
The area of tumour necrosisp = 0.55/0.54p = 0.50/0.42p = 0.27/0.70p = 0.35/0.29p = 0.37/0.20
 <5%632 (158–1711)802 (237–3952)711 (316–2422)0 (0–26)0 (0–79)
 5–29%632 (132–1343)1023 (441–1163)1224 (132–1369)0 (0–105)0 (0–105)
 30–99%606 (369–632)533 (251–814)895 (632–1790)53 (0–316)26 (0–53)
Tumour pigmentationp = 0.33p = 0.64p = 0.64p = 0.75p = 0.30
 Amelanotic to weak500 (132–632)553 (251–2026)843 (132–1553)0 (0–26)0 (0–0)
 Moderate816 (158–1343)1065 (441–2229)764 (316–1290)13 (0–53)0 (0–105)
 Strong606 (369–1711)907 (237–3952)711 (395–2422)0 (0–316)0 (0–79)
Tumour-infiltrating macrophagesp = 0.46p = 0.044p = 0.003p = 0.91p = 0.90
 Few395 (395–395)448 (448–448)
 Moderate632 (132–1711)737 (237–2229)632 (132–1290)0 (0–105)0 (0–105)
 High619 (158–1395)1027 (434–3952)1290 (579–2422)0 (0–316)0 (0–79)
Irradiated Eyesn = 28n = 26n = 22n = 25n = 22
All tumours1514 (500–3844)1051 (87–2273)1014 (26–1869)132 (0–606)79 (0–237)
Tumour height at primary treatment*p = 0.40/0.60p = 0.45/0.24p = 0.27/0.70p = 0.48/0.69p = 0.31/0.39
 <8 mm1790 (685–3844)993 (316–2273)1080 (26–1869)158 (0–606)118 (0–237)
 ≥ 8 mm1343 (500–3133)1053 (87–2260)1184 (764–1316)53 (0–184)39 (0–132)
Tumour LBD at primary treatmentp = 0.32/0.26p = 0.91/0.98p = 0.66/0.69p = 0.64/0.78p = 0.090/0.11
 ≤ 10 mm1948 (553–3844)1644 (316–2131)1369 (369–1869)132 (0–263)0 (0–105)
 >10–15 mm1343 (685–3133)1065 (87–2167)849 (26–1211)26 (0–211)92 (0–184)
 >15 mm1224 (500–2212)985 (474–2273)1106 (395–1316)158 (26–606)105 (0–237)
Tumour height at enucleation*p = 0.18/0.10p = 0.19/0.35p = 0.14/0.92p = 0.67/0.92p = 0.22/0.85
 <8 mm1948 (685–3844)934 (316–2273)987 (26–1869)158 (0–606)105 (0–237)
 ≥ 8 mm1343 (500–2370)1420 (87–2260)1066 (764–1843)53 (0–184)26 (0–105)
Tumour LBD at enucleationp = 0.23/0.19p = 0.49/0.66p = 0.96/0.96p = 0.76/0.68p = 0.26/0.25
 ≤ 10 mm1264 (553–3844)1221(474–2131)948 (395–1869)105 (0–263)118 (0–237)
 >10–15 mm1922 (500–2975)895 (87–2273)1106 (369–1843)132 (0–606)53 (0–237)
 >15 mm1237 (974–2370)1051 (448–1947)948 (26–1316)53 (0–184)26 (26–105)
Tumour reductionp = 0.60/0.59p = 0.53/0.88p = 0.68/0.87p = 0.32/0.60p = 0.98/0.72
 <−1 mm1343 (843–2370)1049 (87–2167)948 (658–1843)53 (0–184)53 (0–237)
 ≥ −1–2 mm1646 (553–2975)869 (316–2131)882 (369–1185)184 (0–606)79 (0–237)
 >2 mm1804 (500–3844)1059 (448–2273)1211 (26–1869)158 (0–263)66 (0–132)
The area of tumour necrosisp < 0.001/0.003p = 0.64/0.60p = 0.41/0.58p = 0.91/0.64p = 0.11/0.046
 < 5%1053 (553–2212)921 (474–2131)948 (395–1264)158 (0–606)132 (0–237)
 5–29%1343 (500–1948)1532 (87–2273)1132 (26–1843)39 (0–184)79 (0–105)
 30–99%2370 (1185–3844)1053 (316–1947)1159 (369–1869)158 (26–263)26 (0–132)
Plaque diameterp = 0.55p = 0.39p = 0.68p = 0.43p = 0.79
 152291 (685–3844)1303 (599–1947)1080 (895–1869)158 (0–606)26 (0–237)
 201343 (553–2975)1143 (87–2273)948 (369–1553)132 (0–211)105 (0–132)
 251685 (500–3133)887 (448–2260)1040 (26–1843)92 (0–184)53 (0–237)
Plaque isotope§p = 0.76p = 0.48p = 0.82p = 0.66p = 0.39
 Ruthenium1343 (500–3844)1221 (87–2260)948 (369–1869)145 (0–263)39 (0–237)
 Cobalt/Iodine1685 (711–3133)921 (448–2273)1132 (26–1316)53 (0–606)105 (0–237)
Tumour pigmentationp = 0.11p = 0.098p = 0.57p = 0.11p = 0.80
 Amelanotic to weak1185 (685–2212)895 (599–1053)882 (395–1843)145 (0–184)26 (0–184)
 Moderate1343 (553–2212)1049 (87–2131)1014 (711–1264)171 (53–606)53 (0–237)
 Strong1922 (500–3844)1532 (316–2273)1185 (26–1869)26 (0–263)105 (0–237)
Tumour-infiltrating macrophagesp = 0.010p = 0.44p = 0.98p = 0.22p = 0.096
 Few1343 (1343–1343)87 (87–87)764 (764–764)105 (105)53 (53)
 Moderate1159 (500–3133)1053 (474–2260)1172 (395–1843)53 (0–211)105 (0–237)
 High2185 (1237–3844)1351 (316–2273)948 (26–1869)171 (0–606)0 (0–237)

Irradiated eyes

The larger the area of tumour necrosis was, the higher was the number of macrophages within the sclera under the tumour base (p < 0.001, nonparametric test for trend and p = 0.003, Spearman; Table 1). A trend towards a smaller number of contralateral episcleral deposits (p = 0.11 and p = 0.046) and towards a higher number of macrophages within the sclera and choroid with increasing tumour pigmentation (p = 0.11 and p = 0.098, respectively, nonparametric test for trend; Table 1) was observed. Tumour height and LBD at brachytherapy were unassociated with the number of macrophages in all extratumoural tissues (Table 1), except for a trend towards more contralateral episcleral deposits with larger LBD (p = 0.090, nonparametric test for trend; p = 0.11, Spearman; Table 1).

Interrelationship between macrophages in extratumoural tissues and tumour-infiltrating macrophages

The CD68 epitope could be assessed in all but one of the matched pairs.

Nonirradiated eyes

The number of CD68-immunopositive macrophages in the choroid and ciliary body increased with increasing numbers of tumour-infiltrating macrophages (p = 0.044 and p = 0.003, respectively; Table 1).

Irradiated eyes

Increasing numbers of tumour-infiltrating macrophages were associated with increasing numbers of CD68-immunopositive macrophages within the sclera under the tumour base (p = 0.010 nonparametric test for trend; Table 1).

Pairwise comparison of macrophages in extratumoural tissues

Irradiated eyes contained significantly more CD68-immunopositive macrophages in the sclera beneath the tumour than nonirradiated eyes (Wilcoxon signed-rank test, p = 0.0001; Figs 2A,B and 3A). The number of macrophages in the choroid and ciliary body adjacent to the tumour did not differ statistically between the matched pairs (p = 0.41 and p = 0.17, respectively; Figs 2C–F and 3B,C).

Figure 2.

 Extratumoural CD68-immunopositive macrophages in irradiated (A, C, E, G, H) and matched primarily enucleated (B, D, F, H, J) eyes with uveal melanoma. Representative images of immunopositive macrophages in the visually densest area of labelled cells within the sclera underneath the tumour base (A–B), in the normal choroid adjacent to the tumour (C–D), in the ciliary body (E–F), in the ipsilateral episclera (G–H) and in the contralateral episclera near the limbus (I–J). Scl, scleral side; conj, conjunctival side.

Figure 3.

 Scatterplot of CD68-immunopositive macrophages in the sclera (A), the normal choroid (B), the ciliary body (C), the ipsilateral episclera (D) and the contralateral episclera (E) in irradiated, secondarily enucleated eyes with uveal melanoma as compared to the matched, primarily enucleated eyes. When the open squares cluster above the diagonal, the irradiated eyes had higher numbers of macrophages. Scatterplot of ipsilateral as compared to contralateral episcleral CD68-immunopositive aggregates of macrophages in eyes with irradiated uveal melanoma (F). Wilcoxon signed-rank test, two-tailed.

It was possible to assess ipsilateral CD68-immunopositive aggregates of macrophages in the episclera near the limbus in 25 (74%) and 22 (65%) irradiated and nonirradiated eyes, and contralateral aggregates in 22 (65%) and 17 (50%) of the 34 eyes, respectively. Ipsilateral aggregates were found in 21 (84%) and 9 (41%) of the irradiated and nonirradiated eyes, and contralateral aggregates in 15 (68%) and 5 (29%) eyes, respectively. Considering the matched pairs, the aggregates could be counted in 44% ipsilaterally and in 32% contralaterally to the tumour. Significantly more aggregates were present in the irradiated eyes both ipsilaterally (p = 0.0034, Figs 2G–H and 3D) and contralaterally (p = 0.014, Figs 2I, J and 3E) when compared to nonirradiated ones. Both irradiated and nonirradiated eyes contained statistically as many ipsi- as contralateral deposits (p = 0.17, Fig. 3F, and p = 0.59, respectively). Numerically, more ipsi- than contralateral deposits were found in the irradiated eyes (median 132 versus 79).

Pairwise comparison of tumour-infiltrating macrophages

The number of CD68-positive cells was higher approximately as often in the irradiated member as in the primarily enucleated member of the pair (p = 0.67, Fig. 4).

Figure 4.

 Scatterplot of tumour-infiltrating CD68-immunopositive macrophages in the matched, primarily enucleated uveal melanomas as compared to irradiated, secondarily enucleated melanomas. Jitter has been applied to show individual observations for categorical variables. When the open squares cluster above the diagonal, higher categories predominate in the more regressed state. Mod, moderate numbers of macrophages.

Discussion

The present results suggest that irradiation increases both the number of CD68-immunopositive macrophages in the sclera beneath the tumour and the number of ipsi- and contralateral episcleral aggregates of macrophages near the limbus. In eyes with irradiated tumours, increasing proportion of necrosis and a higher number of tumour-infiltrating macrophages were associated with increasing numbers of intrascleral macrophages beneath the tumour. In eyes with nonirradiated tumours, the higher the number of tumour-infiltrating macrophages was, the more CD68-immunopositive cells were found in the adjacent choroid and the ciliary body. Thus, the distribution of macrophages in eyes with uveal melanoma seems to change in several ways after brachytherapy.

The strength of our series is that it is essentially unselected and representative of all uveal melanomas secondarily enucleated after brachytherapy. The main limitations are its cross-sectional design and the limited number of specimens. The largest group of irradiated tumours, those which were successfully managed without complications or recurrence, was unavailable for analysis. Moreover, it was possible to assess the number of macrophages in the ciliary body in less than one half of the matched pairs because of technical reasons.

The higher number of episcleral macrophages near the limbus in irradiated eyes when compared to primarily enucleated ones is in agreement with our previous clinical observations regarding the appearance and distribution of clinically visible deposits related to macrophages in irradiated eyes (Toivonen & Kivelä 2006). Our histopathological study also confirms that the number of limbal aggregates is unassociated with tumour height, diameter and the grade of tumour pigmentation, (Toivonen & Kivelä 2006) and this was also true of eyes with nonirradiated tumours. Clinically, the deposits were more numerous ipsilateral to the tumour and least in abundant in the opposite sector (Toivonen & Kivelä 2006). This study was unable to confirm this difference in distribution statistically even though numerically there were more aggregates in the ipsilateral limbus.

One reason for this may be that we were able to count clinically all deposits in every clock hour, whereas histopathologically, the aggregates were counted only in one plane representing two opposite clock hours. Clinically, the deposits are unevenly distributed, and any section studied represents a random sample of them. On the other hand, the average number of macrophage-related aggregates detected histopathologically was much higher than the average number of clinically observed deposits. An obvious reason is that not all immunopositive deposits were pigmented. It is also possible that the smallest deposits, the size of which was around 8 μm, could remain clinically undetected even when pigmented.

We earlier postulated that the more necrotic the irradiated tumour would be, the more episcleral deposits would be present (Toivonen & Kivelä 2006). The number of histopathologically identifiable aggregates was, however, not associated with the degree of histopathologically observable necrosis within the tumour, whether the tumour had been irradiated or not. In the brachytherapy group, the degree of tumour reduction also was unassociated with the degree of necrosis. We have previously shown that irradiated tumours contain more necrosis than nonirradiated ones, but the number of tumour-infiltrating macrophages in nonnecrotic areas of the tumour does not differ between these groups (Toivonen et al. 2003).

A positive association with increasing extent of necrosis was the increasing number of intrascleral macrophages beneath the tumour in irradiated eyes. Perhaps plaque radiotherapy stimulates macrophages to migrate via this route by occluding vessels within the tumour and potentially by modifying the integrity of the sclera. Choroidal vascular occlusion and later scarring around the regressing tumour might also contribute to preferential migration through the sclera. Although these scars are more pronounced after brachytherapy with ruthenium, we did not, however, observe a statistical difference when compared to other isotopes.

An association was also found between the number of tumour-infiltrating macrophages and intrascleral macrophages beneath the tumour in irradiated eyes. Contrary to irradiated eyes, primarily enucleated ones revealed an association between the number of tumour-infiltrating macrophages and macrophages in the adjacent choroid and ciliary body. Our results suggest that there may be differing migration pathways for macrophages in treated and nontreated eyes with uveal melanoma. For example, it could be hypothesized that in untreated eyes, macrophages preferentially migrate to and, perhaps, from the uveal melanoma through the uveal tract or represent spillover to the adjacent choroid. After irradiation, the situation may reverse: macrophages may migrate from the tumour and, perhaps, in to it preferentially through the sclera or they are recruited from the adjacent choroid. Nonirradiated tumours have a higher microvascular density than irradiated matched pairs, and extravascular matrix loops and networks tend to be less frequent after brachytherapy (Toivonen et al. 2003). Thickening of vascular basement membranes and sclerosis and thrombosis of larger tumour vessels are likewise seen after irradiation (Liszauer et al. 1990; Messmer et al. 1992; Schilling et al. 1997) and potentially change the routes of migrating macrophages. These differences in microvascular attributes may contribute to the different distribution of macrophages within the matched pairs. As mentioned, macrophages reside in the normal human eyes, but we did not have access to this kind of control material (Hogan et al. 1971; Schroedl et al. 2008).

To the best of our knowledge, this study is the first to study macrophages in normal tissues of eyes with irradiated uveal melanomas. Several previous studies about tumour-infiltrating macrophages exist (Fuchs et al. 1988; de Waard-Siebinga et al. 1996; Mäkitie et al. 1998, 2001; The Collaborative Ocular Melanoma Study Group 1998; Polak et al. 2007; Maat et al. 2008; Schroedl et al. 2008). The eyes apparently were primarily enucleated, although some studies might not have excluded other types of treatment before enucleation. Direct comparisons between these studies are not possible.

To test the hypothesis of preferential migration through the choroid before irradiation and through the sclera after irradiation and to understand the influence of these routes on tumour progression and regression, one would need to be able to label and track activated macrophages in vivo. In animal models, such studies are becoming possible (Jung et al. 2000), and hopefully similar methods suitable for clinical research will be developed as well. Detailed knowledge of macrophage behaviour might help to develop biologic tools against progressive uveal melanoma in the future.

Acknowledgements

Supported by grants from the Helsinki University Central Hospital Research Funds (TYH5210, TYH2008203), The Finnish Cultural Foundation, The Eye Foundation, Finska Läkaresällskapet, The Sigrid Juselius Foundation, The Paulo Foundation, The Evald and Hilda Nissi Foundation, The Biomedicum Helsinki Foundation, Research and Science Foundation of Farmos, and The Eye and Tissue Bank Foundation, Finland.

Ancillary