Nucleolar size in choroidal and ciliary body melanomas and corresponding hepatic metastases
Abstract.
Purpose: This study aimed to investigate the relationship between hepatic metastasis and the mean diameter of the 10 largest nucleoli (MLN) in uveal melanoma.
Methods: A cross-sectional histopathological analysis of 37 metastases (13 surgical or needle biopsies, 24 autopsies) and corresponding primary choroidal and ciliary body melanomas was conducted, using statistical tests appropriate for paired data. The largest nucleoli were measured from digital photographs of silver-stained sections along a 5-mm-wide linear field. Confounders considered were presence of epithelioid cells and microvascular density (MVD), counted as the number of discrete elements labelled by monoclonal antibody QBEND/10 to the CD34 epitope.
Results: Hepatic metastases had more frequent epithelioid cells (p = 0.0047) and a higher MVD (median difference, 7.5 counts/0.313 mm2 more; p = 0.044) than their corresponding primary tumours. Hepatic metastases, especially in autopsy specimens rather than surgical biopsies, tended to have a smaller MLN (median 3.6 μm) than the corresponding primary tumour (median difference, 0.55 μm; p = 0.066). The MLN in hepatic metastases was not associated with presence of epithelioid cells and MVD. Overall survival after diagnosis of metastasis was comparable whether hepatic metastases had a large or small MLN (p = 0.95), whereas a high MVD tended to be associated with shorter survival (p = 0.096) among the 13 patients with known survival.
Conclusions: The results suggest that MLN is not a useful marker for assessing prognosis after diagnosis of hepatic metastasis from uveal melanoma.
Introduction
Uveal melanoma is a slow-growing tumour, which often appears to seed distant metastases before the primary tumour is treated (Collaborative Ocular Melanoma Study Group 1998; Eskelin et al. 2000; Diener-West et al. 2001). Local recurrence is infrequent and regional lymph node metastasis is exceptional. Metastasis occurs mainly to the liver via the blood stream (Eskelin et al. 2000). Most patients die of hepatic rather than extrahepatic metastases. The haematological route is predominant because the eye does not posses a lymphatic drainage system and thus microvascular factors are considered important in uveal melanoma (Folberg et al. 1992; McLean 1993).
The overall 10-year survival in uveal melanoma is estimated to be 68% (Kujala et al. 2003). Death can be delayed for several decades after the primary tumour has been definitively cured; in such cases the metastatic cells reside in apparent dormancy and the events that lead to delayed progressive clinical metastasis are unknown (Kujala et al. 2003).
We have recently shown that presence of epithelioid cells and microvascular density (MVD) most closely paralleled the progression of uveal melanoma from primary tumour to metastasis (Toivonen et al. 2004). These two characteristics may be interrelated. Further, high MVD in metastasis may help to predict survival if the metastasis is biopsied (Toivonen et al. 2004).
A large mean diameter of the 10 largest nucleoli (MLN) has consistently been associated with a high chance of death from uveal melanoma (Huntington et al. 1989; Gamel et al. 1992; Sørensen et al. 1993a, 1993b; McLean et al. 1997; Seregard et al. 1998; Moshari & McLean 2001). We have proved earlier that MLN is an independent predictor of survival in primary choroidal and ciliary body melanomas (Al Jamal et al. 2003). We compared MLN in primary uveal melanoma and corresponding hepatic metastasis to determine whether it is associated with tumour progression and whether measuring MLN from hepatic metastasis might aid in predicting survival after dissemination of uveal melanoma.
Materials and Methods
Inclusion criteria and data collection
The study followed the tenets of the Declaration of Helsinki and was approved by our institutional review board. All patients in the district covered by Helsinki University Central Hospital, Helsinki, in whom a choroidal and ciliary body melanoma had been enucleated between 1962 and 1981 and which had later metastasized were eligible. Enucleation was the standard treatment for all patients except those with the smallest melanomas during this period, making the series essentially population-based and unselected.
Inclusion criteria required that ≥ 50% of the primary tumour remained in the tissue block and that the remaining part was not entirely on the vitreal side of Bruch’s membrane (Mäkitie et al. 1999a), and that one or more core needle biopsy, biopsy or autopsy specimens with a surface area of ≥ 0.35 mm2 were available from hepatic metastases. This is roughly the minimum area needed to measure MVD (Mäkitie et al. 1999b).
During the study period, 292 consecutive patients underwent the removal of an eye with a choroidal and ciliary body melanoma. Of these, 145 developed metastases which, in 92 cases, were cytologically or histologically confirmed. From these, 48 pairs of primary tumours and hepatic metastases that fulfilled the inclusion criteria were identified in our prior study (Toivonen et al. 2004). Of the 48 specimens, 37 were available for re-staining (Table 1).
| Characteristic | Primary tumour | Hepatic metastasis |
|---|---|---|
| (n = 37) | (n = 37) | |
| Median disease-free interval, years (range) | N/A | 5.0 (0.4–17.4) |
| Median overall survival, months (range) | N/A | 2.0 (0.8–31.4) |
| Median height, mm (range) | 7.0 (2.0–12.0) | N/A |
| Median LBD, mm (range) | 14 (6–21) | 17 (1.5–35)* |
| Cell type, n (%) | ||
| Spindle | 11 (31) | 2 (6) |
| Non-spindle | 25 (69) | 33 (94) |
| Median MVD, counts/0.313 mm2 (range) | 44.5 (12–104) | 56 (19–128) |
| Median MLN, μm (range) | 4.1 (3.3–5.2) | 3.6 (2.7–6.4) |
- * Diameter of specimen, if hepatic metastasis larger than the biopsy or autopsy specimen.
- N/A = not applicable; LBD = largest basal tumour diameter; MVD = microvascular density; MLN = mean diameter of the 10 largest nucleoli.
The metastases had been recognized by liver imaging or laparoscopy after they caused symptoms. The original sizes of the biopsied and autopsied metastases had not been recorded and, except in one case, the entire metastasis was not present in the specimen. The largest diameter in the biopsy was measured from the sections with a calliper. Of the 37 sections, one (3%) represented a core needle biopsy, 12 (32%) were surgical biopsies, and 24 (65%) were autopsy specimens. Median specimen sizes in these three groups were 1.5 mm, 9 mm (range 4–19 mm) and 20 mm (range 6.5–25 mm), respectively.
Silver staining
A comparative study of primary uveal melanoma found that measurements from silver-stained slides were easier to make and more strongly associated with prognosis than those from haematoxylin-eosin slides (Moshari & McLean 2001). The most frequent field selection for sampling has been a 5-mm-long linear strip from the centre of the melanoma. Linear sampling was recently reported to be comparable with scanning nucleoli from the entire tumour section in predicting outcome, although the mean value was 0.15 μm larger when MLN was sampled from the entire tumour (Moshari & McLean 2001). We chose silver staining and linear sampling for the purposes of the present study.
The sections were deparaffinized and bleached with 0.25% (w/v) potassium permanganate for 1 hour, then placed into 5% (w/v) oxalic acid solution for 5 mins. One-step silver staining was performed according to Moshari & McLean (2001) using two solutions. The first solution comprised 40 ml of 2.0% (w/v) gelatin (Bacto® Gelatin; Difco Laboratories, Detroit, MI, USA) and 0.88% (v/v) formic acid. The second solution consisted of 80 ml of 50% (w/v) silver nitrate in distilled water. The solutions were mixed in the dark and poured into a dish to cover the sections for 30 mins. The sections were then washed in distilled water, dehydrated and coverslips were mounted with Mountex® (Histolab Products AB, Gothenburg, Sweden).
Each slide was examined under a light microscope (Olympus BH-2; Olympus Corp., Tokyo, Japan) at 2× magnification to orient the central longest axis of the tumour for digital photography (Olympus DP-10 soft, Version 3.0; Soft Imaging System GmbH, Münster, Germany). A series of photographs at 40× optical magnification were then taken to image the nucleoli along the longest axis of the tumour. The central 5-mm distance was photographed, divided into 25 slightly overlapping images (resolution 1280 × 1024 pixels, image area 218 × 175 μm). If the largest basal tumour diameter was < 5 mm, the entire central longest axis of the tumour was photographed.
From each of the 25 photographs, the largest nucleoli were measured using the image analysis software (Olympus DP-10 soft, Version 3.0). A strip of 41 μm in height was scanned from the top of each photograph (total area 0.205 mm2) (Al Jamal et al. 2003). More than one strip was scanned if the largest axis photographed was < 5 μm. Between one and five nucleoli per image were measured (range 13–80 per section). The 10 largest nucleoli from each tumour were retained for statistical analysis.
Assessment of microvascular density
Microvessels were identified with the monoclonal antibody (mAb) QBEND/10 to the CD34 epitope of endothelial cells (lot 121202, diluted 1 : 25; Novocastra Laboratories, Newcastle-upon-Tyne, UK) (Ramani et al. 1990). They were counted at 400× magnification from the most highly vascularized area (‘hot spot’, identified under 100× magnification), using an eyepiece with an etched graticule corresponding to 0.313 mm2 (Olympus WK 10×/20L-H) (Mäkitie et al. 1999b). Any immunolabelled element, clearly separate from those adjacent and totally inside the graticule or touching its top or left border, was counted as a microvessel. Microvessels were counted three times by one grader and the highest count was registered.
Statistical analysis
All analyses were performed using stata (Release 11.0; Stata Corp., College Station, TX, USA) and StatXact-3 (Cytel Co., Cambridge, MA, USA) statistical software packages. A p-value < 0.05 was considered statistically significant. All tests were two-tailed.
The median and range are given as descriptive statistics. Kruskal–Wallis test was used to compare continuous variables between categories. Spearman’s rank correlation was used to analyse interrelationships between two continuous variables. Wilcoxon signed rank test was used to compare distributions of paired continuous data and Stuart–Maxwell test to compare unordered paired contingency tables (Fleiss 1981).
Overall survival was calculated as the time from the date of diagnosis of metastases to death. Survival was analysed with the Kaplan–Meier product-limit method and log-rank test. Results for MLN and MVD were divided into two categories according to their medians.
Results
The 37 hepatic metastases had more frequent epithelioid cells (p = 0.0047, Stuart–Maxwell test) (Fig. 1A, Table 1) than the primary uveal melanomas from which they had spawned.
Scatterplots of (A) presence of epithelioid cells, (B) type of metastatic specimen, (C, D) mean diameter of the 10 largest nucleoli (MLN) and (E, F) microvascular density (MVD) in hepatic metastasis compared with the corresponding primary tumour (A, C, E) and the diameter of the metastatic specimen (D, F). Jitter was applied to display individual observations for categorical variables (A, B). When the open circles cluster above and below the diagonal, higher and lower categories, respectively, predominate in the metastatic tumours (A, C, E). Lines represent linear regression with 95% confidence intervals (D, F). p-values are according to (A) Stuart–Maxwell, (B) Kruskal–Wallis, (C, E) Wilcoxon rank sum, and (D, F) Spearman’s rank correlation tests.
Mean diameter of the 10 largest nucleoli
Nucleoli could be reliably identified using the silver staining method in all 37 metastatic specimens and in 26 corresponding primary tumour specimens. One of the primary tumours was necrotic, in four the nucleoli were indistinct, and in six the specimen was too small to be analysed.
The median MLN in the primary tumours was 4.1 μm (range 3.3–5.2 μm, mean 4.2 μm, standard deviation [SD] 0.52) (Table 1). Metastatic specimens fell into two subgroups. The first subgroup included 12 surgical biopsies and the second 24 autopsy specimens. The median MLN was 3.9 μm (range 3.2–6.4 μm, mean 4.3 μm, SD 1.14) in the former group and 3.5 μm (range 2.7–5.7 μm, mean 3.7, μm SD 0.72) in the latter (Fig. 1B). The autopsy specimens thus tended to have a smaller MLN than surgical specimens (p = 0.065, Kruskal–Wallis test). The MLN in the single core needle biopsy was 3.8 μm.
Of the patients for whom autopsy specimens were available, three had received either chemotherapy or irradiation for metastases and 12 had not received any treatment. Median MLN in both groups was 3.6 μm (means 3.6 μm and 3.9 μm, respectively).
Because the trend towards a smaller median MLN in autopsy specimens might indicate post-mortem artefact, we also compared MLN measured from normal hepatocytes in preserved areas of the liver, present in two surgical and four autopsy specimens. The median MLN was 2.2 μm (range 2.0–2.4 μm) in the former group and 2.7 μm (range 2.1–3.0 μm; p = 0.16) in the latter.
In all types of specimen combined, hepatic metastases tended to have a smaller MLN (median, 3.6 μm, range 2.7–6.3 μm, mean 3.9 μm, SD 0.91) than the 26 matched primary tumours (mean difference between pairs, 0.55 μm; p = 0.066, Wilcoxon signed rank test) (Fig. 1C).
When analysed by subgroup, five of eight primary tumours had a numerically larger MLN than the matched surgically biopsied metastasis, but the two types of tumour did not differ overall from one another (mean difference between pairs, 0.07 μm; p = 0.89, Wilcoxon signed rank test). Likewise, 12 of 17 primary tumours had a numerically larger MLN than the matched autopsied hepatic metastasis and, in this subgroup, the primary tumours also had a larger MLN than the metastases did as a group (mean difference between pairs, 0.45 μm; p = 0.044).
The diameter of the metastatic specimen and MLN were significantly correlated (p = 0.029, Spearman’s rank correlation) (Fig. 1D). The MLN in hepatic metastasis was not associated with presence of epithelioid cells (p = 0.39, Kruskal–Wallis test) or MVD (p = 0.54, Spearman’s rank correlation).
Microvascular density
Microvascular factors could be reliably determined from all of the 37 metastatic and 36 of the 37 primary tumour specimens (Table 1).
Median MVD in the 26 matched primary melanomas was 44.5 counts/0.313 mm2 and was significantly lower than the 56 counts/0.313 mm2 in the specimens from hepatic metastases (median difference, 7.5 counts/0.313mm2 more, range 38 counts/0.313 mm2 less to 70 counts/0.313 mm2 more; p = 0.044, Wilcoxon signed rank test) (Fig. 1E). The MVD in metastasis was not associated with the diameter of the specimen (p = 0.62, Spearman’s rank correlation) (Fig. 1F).
Survival after metastasis
Overall survival rates among the 13 patients in whom metastasis was biopsied before death were comparable between those with hepatic metastases with an MLN smaller than the median value and those with a larger MLN (1.1 months versus 1.8 months; p = 0.95, log-rank test) (Fig. 2A). Overall survival tended to be longer if hepatic metastases had an MVD smaller rather than larger than the median value (6.5 months versus 1.5 months; p = 0.096, log-rank test) (Fig. 2B).
Kaplan–Meier plots of overall survival after diagnosis of metastases from uveal melanoma according to (A) mean diameter of the 10 largest nucleoli and (B) microvascular density (MVD) in hepatic metastases. Mortality of patients whose metastases had an MVD higher than the median tended to be higher than that of patients whose metastases had a lower MVD. Numbers (a, b) below the graphs show patients at risk. p-values are derived from log-rank test.
Discussion
Analysis of enucleated primary choroidal and ciliary body melanomas and corresponding matched hepatic metastases revealed that the metastases tended to have a smaller MLN than the primary tumours. However, MVD was on average higher in the metastases compared with the matched primary tumours, despite the small sample size. This difference was especially clear in matched pairs that included an autopsied metastasis.
The MLN measured from autopsy specimens was on average significantly smaller than MLN measured from surgical biopsies. This, and the difference between primary and metastatic lesions, was unlikely to reflect simple post-mortem change, given that no such difference was found in MLN between normal hepatocytes in preserved liver tissue, present in some of the same specimens. In fact, MLN in some metastases was larger than in the matched primary tumours, indicating heterogeneity among hepatic metastases. It is possible that the variation may reflect differences in underlying tumour biology, such as the dormancy of some metastases. Although all patients had progressive metastases, which led to death, it is still feasible that many of the sampled metastasis might have represented a minority that were dormant, and thus had small nucleoli.
Microvascular density is a histopathological indicator associated with tumour microvessels and serves as an independent indicator of shorter survival after enucleation in uveal melanoma (Foss et al. 1996; Mäkitie et al. 1999b; Chen et al. 2002). In primary uveal melanoma, MVD may reflect tumour angiogenesis, aggressiveness or both. High MVD in primary uveal melanoma is associated with the presence of epithelioid cells (Mäkitie et al. 1999b), which also tended to be more frequent in the hepatic metastases studied than the matched primary tumours.
Analogous with the finding that high MVD in primary uveal melanoma is associated with shorter time to metastasis than low MVD (Foss et al. 1996; Mäkitie et al. 1999b; Chen et al. 2002), high MVD in hepatic metastasis tends to be associated with shorter survival after diagnosis of metastasis (Toivonen et al. 2004). We have previously confirmed that MLN is a predictor of survival after enucleation in primary uveal melanoma (Al Jamal et al. 2003). However, survival rates after the diagnosis of metastatic disease were comparable whether metastases had a high or a low MLN, and MLN in metastatic specimens was thus not a useful prognostic indicator.
Acknowledgements
This study was supported by grants from the Helsinki University Central Hospital Research Fund (TYH2008203, TYH5210), Sigrid Jusélius Foundation, Eye Foundation, Eye and Tissue Bank Foundation, Finnish Medical Foundation, Ahokas Foundation, Instrumentarium Research Foundation, Paulo Foundation, Biomedicum Foundation, and the Evald and Hilda Nissi Foundation, Finland.
