Download Progress in Retinal and Eye Research

Progress in Retinal and Eye Research xxx (2011) 1e11
Contents lists available at ScienceDirect
Progress in Retinal and Eye Research
journal homepage: www.elsevier.com/locate/prer
Estimating prognosis for survival after treatment of choroidal melanoma
Bertil Damato a, *, Antonio Eleuteri b, Azzam F.G. Taktak b, Sarah E. Coupland c
a
Ocular Oncology Service, Royal Liverpool University Hospital, Prescot St, Liverpool L7 8XP, UK
Department of Medical Physics and Clinical Engineering, Royal Liverpool University Hospital, Prescot St, Liverpool L7 8XP,UK
c
Department of Pathology, Duncan Building, Royal Liverpool University Hospital, Daulby St, Liverpool L7 8XP, UK
b
a b s t r a c t
Keywords:
Melanoma
Uveal melanoma
Mortality
Chromosome aberrations
Prognostication
Histology
Mathematical-modeling
Choroidal melanoma is fatal in about 50% of patients. This is because of metastatic disease, which usually
involves the liver. KaplaneMeier survival curves based only on tumor size and extent do not give a true
indication of prognosis. This is because the survival prognosis of choroidal melanoma correlates not only
with clinical stage but also with histologic grade, genetic type, and competing causes of death. We have
developed an online tool that predicts survival using all these data also taking normal life-expectancy
into account. The estimated prognosis is accurate enough to be relevant to individual patients. Such
personalized prognostication improves the well-being of patients having an excellent survival probability, not least because it spares them from unnecessary screening tests. Such screening can be targeted
at high-risk patients, so that metastases are detected sooner, thereby enhancing any opportunities for
treatment. Concerns about psychological harm have proved exaggerated. At least in Britain, patients
want to know their prognosis, even if this is poor. The ability to select patients with a high risk of
metastasis improves prospects for randomised studies evaluating systemic adjuvant therapy aimed at
preventing or delaying metastatic disease. Furthermore, categorization of tissue samples according to
survival prognosis enables laboratory studies to be undertaken without waiting many years for survival
to be measured. As a result of advances in histologic and genetic studies, biopsy techniques and statistics,
prognostication has become established as a routine procedure in our clinical practice, thereby
enhancing the care of patients with uveal melanoma.
Ó 2011 Published by Elsevier Ltd.
Contents
1.
2.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Survival predictors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.
Clinical risk factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.1.
Age and sex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.2.
Tumor dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.3.
Ciliary body involvement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.4.
Extraocular spread . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.5.
TNM staging system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.
Histological risk factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.1.
Melanoma cytomorphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.2.
Extravascular matrix patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.3.
Microvascular density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.4.
Non-neoplastic cellular infiltrates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.5.
Cell proliferation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.
Genetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.1.
Chromosome 3 loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.2.
Chromsome 8 gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.3.
Chromosome 6p gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
00
* Corresponding author. Tel.: þ44 (0) 151 7063973; fax: þ44 (0) 151 7065436.
E-mail address: [email protected] (B. Damato).
1350-9462/$ e see front matter Ó 2011 Published by Elsevier Ltd.
doi:10.1016/j.preteyeres.2011.05.003
Please cite this article in press as: Damato, B., et al., Estimating prognosis for survival after treatment of choroidal melanoma, Progress in Retinal
and Eye Research (2011), doi:10.1016/j.preteyeres.2011.05.003
2
B. Damato et al. / Progress in Retinal and Eye Research xxx (2011) 1e11
3.
4.
5.
6.
7.
2.3.4.
Chromosome 1p loss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3.5.
Gene expression profiling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4.
Other predictors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Statistical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.1.
KaplaneMeier analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2.
Liverpool Uveal Melanoma Prognosticator Online (LUMPO) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Harvesting tumor tissue for prognostication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Scope of prognostication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.1.
Clinical care . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.2.
Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Putting theory into practice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1. Introduction
2. Survival predictors
Approximately 90% of all intraocular melanomas involve the
choroid, the remainder being confined to iris and ciliary body. Most
patients present with visual symptoms, such as blurred vision
(Damato, 2001). If neglected, these uveal tumors can cause visual
loss, a painful and unsightly eye, and death from intracranial spread
or metastatic disease. Such metastatic disease almost always
involves the liver and is only rarely detectable at the time of ocular
treatment.
It is widely stated that patients with uveal melanoma have a 5050 chance of dying of their disease. Although this summary statistic
holds true when applied to all patients, it is mostly false when
applied to individuals. This is because in most patients the prognosis is actually very much better or considerably worse than this
estimate. An important aim of this article is, therefore, to explain
how the accuracy of prognostication can be enhanced to the level at
which it is relevant to individual patients.
Most cancer prognostication is based on the TNM (tumor, node,
metastasis) staging system of the American Joint Committee on
Cancer (AJCC) and the Union for International Cancer Control
(UICC). The TNM staging of uveal melanomas has recently been
improved, and we have contributed significantly to this project
(Finger, 2009). The TNM staging system currently uses only clinical
data. Can it be enhanced by taking account of histologic and genetic
predictors? If so, how? In this review we discuss a possible way
forward.
Survival after treatment of cancer tends to be displayed using
KaplaneMeier analysis, but is this methodology satisfactory? We
have developed an online tool that provides a better impression of
disease-free survival after treatment of choroidal melanoma and in
this article we present our approach.
Treatment of metastatic disease is only rarely effective, and
doubts have been expressed regarding the value of screening and
treatment (Augsburger et al., 2009; Kim et al., 2010). However,
some recent results are encouraging (Sato, 2010). Is prognostication
harmful or does it actually enhance quality of life? The answers to
these questions are further aims of this article.
Our ultimate objective in this review is to describe how we
currently translate basic science into our clinical practice. We hope
to show how statistics can be applied to clinical, pathologic, and
genetic data so as to enhance patient care and quality of life, possibly
even maximising any opportunities for prolonging life itself.
This article is not meant to be encyclopaedic and only information relevant to prognostication is provided. It is confined to
choroidal tumors (i.e., tumors involving choroid) so as to allow indepth analysis and to prevent discussion from being diluted by
excessive tumor diversity.
2.1. Clinical risk factors
00
00
00
00
00
00
00
00
00
00
00
00
00
00
2.1.1. Age and sex
It has been suggested that older age and male gender correlate
with reduced survival (Seddon et al., 1983; Virgili et al., 2008).
However, it is uncertain as to what extent these correlations are the
result of bias arising from competing risks (i.e., death from other
causes) (Kujala et al., 2003).
2.1.2. Tumor dimensions
Numerous studies have shown a strong correlation between
largest basal tumor diameter and reduced survival probability
(Damato and Coupland, 2009). One explanation for this association
is that large tumors have been present for a longer time and have,
therefore, had more opportunity to develop metastatic capability
and disseminate. This hypothesis assumes that metastatic spread
tends to start after the tumor has grown large. The same correlation
is partly explained by the likelihood that a large uveal melanoma
has been metastasising for a longer period so that any metastases
have also been growing for a longer time (i.e., ‘lead-time bias’).
Another possibility is that metastatic spread commences early,
when the tumor size is still small, so that large tumor size is mostly
an indicator of increased tumor malignancy and greater growth
rate. Many patients with uveal melanoma experience long delays
before treatment (Ah-Fat and Damato, 1998). In such cases, the
more malignant melanomas are likely to grow more than those that
are have less metastatic potential. Our view is that all these
explanations are equally plausible and that different mechanisms
apply to different tumors according to whether they start to metastasise before or after they grow large.
Greater tumor thickness is associated with higher metastatic
mortality and some consider this to be an important survival
predictor (Shields et al., 2009). Tumor thickness is one of the factors
used to estimate survival with the TNM uveal melanoma staging
system (see 2.1.5) (Finger, 2009). In our multivariate analyses,
thickness does not appear to be a significant independent risk
factor when diameter and other predictors are taken into account
(Damato and Coupland, 2009). Thickness is influenced not only by
neoplastic growth but also by tumor prolapse into sub-retinal
space, and interstitial edema.
2.1.3. Ciliary body involvement
Several studies show that ciliary body involvement is associated
with a poorer prognosis for survival (Seddon et al., 1983). The
explanation for this correlation is uncertain. Large uveal melanomas are more likely to involve ciliary body (Damato and
Please cite this article in press as: Damato, B., et al., Estimating prognosis for survival after treatment of choroidal melanoma, Progress in Retinal
and Eye Research (2011), doi:10.1016/j.preteyeres.2011.05.003
B. Damato et al. / Progress in Retinal and Eye Research xxx (2011) 1e11
Coupland, 2009). Furthermore, they more likely to show adverse
histological and genetic risk factors (Damato et al., 2007).
2.1.4. Extraocular spread
Extraocular spread is associated with increased mortality
(Coupland et al., 2008). In our analysis of 847 eyes enucleated for
uveal melanoma and assessed at our hospital, we found extraocular
spread to correlate significantly with other adverse, clinical,
histological and genetic risk factors for metastasis (Coupland et al.,
2008). Whether extraocular extension contributes to metastatic
risk or whether it is only an indicator of increased tumor malignancy is uncertain.
2.1.5. TNM staging system
The 7th TNM staging system categorises tumors according to
their size, categorized by tumor basal diameter and height, also
taking account of ciliary body involvement and extraocular spread
(i.e., nil, <5.1 mm and >5.0 mm). Risk of metastatic death is estimated according to: ocular tumor stage (with categories having the
same prognosis grouped into the same stage); regional lymph node
involvement; and presence of known metastases (Finger, 2009).
2.2. Histological risk factors
2.2.1. Melanoma cytomorphology
The presence of epithelioid melanoma cells is associated with
a worse prognosis (Fig. 1a and b) (McLean et al., 1978). Differentiation between spindle and epithelioid cells is subjective and,
furthermore, there is no consensus at present as to how many cells
need to show an epithelioid cytomorphology for the tumor to be
categorised as ‘mixed’ or ‘epithelioid’. Nucleolar size, specifically
the mean of the ten largest nucleoli (MLN) also correlates with
3
increased mortality and is widely used as a prognostic indicator (Al
Jamal et al., 2003). Nucleolar assessment is facilitated by methods
such as silver-staining.
2.2.2. Extravascular matrix patterns
In 1993, Folberg et al. described nine patterns of what they then
considered to be blood vessels, but which were subsequently found
to lack endothelial cells (i.e., ‘vasculogenic mimicry’) (Lin et al.,
2005; Folberg et al., 1993). Metastatic death correlates most
strongly with the presence of at least three, back-to-back, closed
loops (i.e., ‘networks’). These patterns are best appreciated by light
microscopy on sections stained with the periodic acid Schiff (PAS)
reagent, without hematoxylin counter-staining (Fig. 1c). We and
others have confirmed that PAS-positive networks correlate with
other risk factors for metastasis and with increased mortality
(Coupland et al., 2008; Damato et al., 2007).
2.2.3. Microvascular density
The microvascular density (MVD) is measured by counting the
number of blood vessels per unit area in the most vascularised parts
of the tumor (‘hot spots’), using anti-endothelial antibodies to
identify these vessels (e.g., anti-CD34 antibodies). MVD is higher in
tumors with epithelioid cells, high-risk extravascular matrix
patterns, and increased mortality (Mäkitie et al., 1999).
2.2.4. Non-neoplastic cellular infiltrates
Macrophages are commonly present in uveal melanomas and
their identification is facilitated by immunohistochemistry, using
the anti-CD68 antibody, after melanin bleaching. The number of
macrophages is higher in tumors with epithelioid cells, increased
microvascular density and large size as well as in tumors that have
proved fatal (Mäkitie et al., 2001b). This is possibly because
Fig. 1. Histological features of uveal melanoma: epithelioid cells (A) (H&E stain, 40 magnification); spindle-cells (B) (H&E stain, 40 magnification); connective tissue closed loops
(C) (PAS without hematoxylin counterstain, 10 magnification) and mitoses (D) demonstrated with Ser-10 antibody (APAAP stain, 20 magnification).
Please cite this article in press as: Damato, B., et al., Estimating prognosis for survival after treatment of choroidal melanoma, Progress in Retinal
and Eye Research (2011), doi:10.1016/j.preteyeres.2011.05.003
4
B. Damato et al. / Progress in Retinal and Eye Research xxx (2011) 1e11
macrophages are involved in angiogenesis and because they may
suppress anti-tumor immune responses (Jager et al., 2011). Tumorinfiltrating lymphocytes (TILs), particularly T-lymphocytes, are also
associated with increased metastatic mortality (Whelchel et al.,
1993). Increased HLA classes I and II expression, as determined by
immunohistochemistry, is associated with poor survival and it has
been suggested that such expression suppresses immune responses
to the tumor (Ericsson et al., 2001; Maat et al., 2009).
2.2.5. Cell proliferation
There are several indicators of cell proliferation. The mitotic
count per specified number of high-power fields is widely accepted
and correlates strongly with risk of metastatic death (Coupland
et al., 2008; Damato et al., 2007). In sections stained with hematoxylin and eosin, however, it can be difficult to recognise mitotic
figures or to differentiate them from occasional apoptotic bodies.
Dividing cells can be identified with antibodies such as PHH3 (also
known as Ser-10) (Fig. 1d) (Angi et al., 2009). However, staining
with this antibody fades over time and is used only in a few centers.
Other markers such as Ki-67 and PC-10 are more widely used and
have been shown to correlate with other predictors of metastasis
and with increased mortality (Al-Jamal and Kivelä, 2006; Seregard
et al., 1998). There is no uniform way of counting positive cells and
for preparing specimens so that results from different centers vary
considerably.
2.3. Genetic
2.3.1. Chromosome 3 loss
In 1996, Prescher et al. correlated chromosome 3 loss (i.e.,
‘monosomy 3’) with metastatic death (Prescher et al., 1996). Fatality
occurred exclusively in patients with monosomy 3 and almost all
patients with this abnormality died by the end of the study. Others
have confirmed these findings using a variety of methods (Damato
et al., 2007; Damato et al., 2010; Shields et al., 2011; Trolet et al.,
2009; Young et al., 2007). Fatality can occur also with partial chromosome deletions, which can be missed with methods such as FISH
(Damato et al., 2007) (Fig. 2H). Higher-resolution techniques are
therefore replacing FISH. Our current approach is to perform
multiplex ligation-dependent probe amplification (MLPA) in the
first instance (Fig. 3) (Damato et al., 2010). In selected cases, we
perform additional: (a) a CGH for tumors with apparent disomy 3
with high-grade histological malignancy; and (b) microsatellite
analysis (MSA) when the DNA concentration of the sample is too
small for MLPA or when MLPA results are inconclusive. We hope to
be able to compare our techniques with gene expression profiling
(GEP) (see 2.3.5).
Apart from inadequate tumor sample cellularity, there are other
causes of failure, such as isodisomy 3, in which both copies of the
same chromosome are from the same parent. Furthermore, uveal
melanomas can show considerable genetic intra-tumoral heterogeneity (Dopierala et al., 2010). This has given rise to concerns about
sampling error. Further studies are required to determine whether
prognostication relying on tumor biopsies is as reliable as that based
on large samples obtained from local resection and enucleation
specimens. Recent studies have proposed a critical area of chromosome 3 loss (e.g. BAP1): these results require independent validation but suggest that that sensitivity for prognostication in uveal
melanoma is likely to improve in future (Harbour et al., 2010).
2.3.2. Chromsome 8 gain
This abnormality can arise as a result of abnormal splitting of the
chromosome during cell division. Instead of the two chromatids
moving to different cells, both short arms of chromosome 8 move to
one cell with both long arms of the same chromosome going to the
other, producing ‘isochromosome 8q’. Another mechanism is
trisomy 8 formation. Gains of chromosome 8q correlate with
increased mortality (Fig. 2G). White et al. have suggested that
chromosomes 3 loss and 8q gain must be present together for
metastatic disease to occur (White et al.,1998). Using MLPA, we have
found that metastatic death can rarely occur with monosomy 3
alone (Damato et al., 2010) (Fig. 3); however, further studies using
more sensitive tests are needed to determine whether in these fatal
cases chromosome 8q gains are indeed absent or merely undetected.
2.3.3. Chromosome 6p gain
Chromosome 6p gain tends to arise as a result of isochromosome 6p. It is associated with a relatively good prognosis (Parrella
et al., 1999). This is because monosomy 3 is less common in
tumors with chromosome 6p gain (Damato et al., 2010). Furthermore, when both these abnormalities occur together, survival is
better than with monosomy 3 alone (Damato et al., 2010).
2.3.4. Chromosome 1p loss
Chromosome 1p loss correlates with chromosome 3 loss and
increased mortality (Damato et al., 2010; Kilic et al., 2005).
However, it loses significance in our multivariate analyses.
2.3.5. Gene expression profiling
Uveal melanomas cluster into two molecular classes according
to their gene expression profile, with metastatic disease occurring
almost exclusively with class 2 tumors, which have a high mortality
(Onken et al., 2004). A 15-gene assay has been developed, which
shows high sensitivity and specificity, even with very small samples
and formalin-fixed specimens (Onken et al., 2010).
2.4. Other predictors
Increased mortality is associated with many other factors,
whose discussion is beyond the scope of this article. These include:
insulin-like growth factor-1 receptor; inducible nitric oxide synthase; increased ezrin expression; and low Heat-Shock-Protein-27
(HSP-27) expression (Ericsson et al., 2001; All-Ericsson et al.,
2002; Johansson et al., 2009; Makitie et al., 2001a; Jmor et al.,
2010). Interestingly, G protein alpha subunit q (GNAQ), which
occurs in about 50% of uveal melanomas, does not correlate with
survival (Bauer et al., 2009).
3. Statistical methods
3.1. KaplaneMeier analysis
KaplaneMeier analysis is the standard method of reporting
cancer survival and plots the proportions of patients surviving at
different time intervals. Each death is represented by a step in the
curve with the size of the step depending on the number of survivors
at the start of the interim period. When the curves represent
metastatic mortality, non-metastatic deaths are treated like other
causes of incomplete follow-up and are censored, with censoring
indicated by ticks in the curve. Because of censoring of nonmelanoma deaths, it is not possible to deduce from KaplaneMeier
curves a patient’s chances of dying from melanoma or the chances of
surviving a specified time. This is because such censoring exaggerates the apparent metastatic mortality to give a false impression of
the survival probability (Kujala et al., 2003). For example, older
patients appear to have an increased risk of metastatic death (Fig. 4).
When statistical methods take account of competing risks, however,
older people are actually found to be less likely to die of metastasis
because more of them have died of other causes before the metastatic disease has had time to develop (Fig. 4). Similar difficulties
Please cite this article in press as: Damato, B., et al., Estimating prognosis for survival after treatment of choroidal melanoma, Progress in Retinal
and Eye Research (2011), doi:10.1016/j.preteyeres.2011.05.003
B. Damato et al. / Progress in Retinal and Eye Research xxx (2011) 1e11
5
Fig. 2. KaplaneMeier survival curves showing metastatic mortality to correlate strongly with: (A) Ciliary body involvement (logrank, p ¼ 0.001); (B) Basal tumor diameter (logrank,
p < 0.001); (C) Extraocular spread (logrank, p ¼ 0.002); (D) Epithelioid cytomorphology (logrank, p < 0.001); (E) Closed loops (logrank, p < 0.001); (F) Mitotic rate (logrank,
p < 0.001); (G) Chromosome 8 gain (logrank, p < 0.001); and (H) Chromosome 3 loss (logrank, p < 0.001). Metastatic death occurred in 12 patients without apparent chromosome 3
loss, encouraging the authors to use MLPA instead of FISH. With permission from Damato et al. (2007).
occur when comparing males with females, because competing
risks are greater in males. KaplaneMeier analysis has other limitations, such as not allowing multivariate analysis and not coping with
continuous variables such as basal tumor diameter. For these
reasons, if used for purposes other than comparing randomized
groups of patients, KaplaneMeier survival curves are all-too easily
mis-interpreted by those who are not aware of all the causes of bias.
They do not allow accurate prognostication. A variety of methods
Please cite this article in press as: Damato, B., et al., Estimating prognosis for survival after treatment of choroidal melanoma, Progress in Retinal
and Eye Research (2011), doi:10.1016/j.preteyeres.2011.05.003
6
B. Damato et al. / Progress in Retinal and Eye Research xxx (2011) 1e11
uveal melanomas, this shortcoming is significant because many
patients die at home, far from the oncology center that treated
them. Hepatic metastases from colon cancer can mistakenly be
attributed to uveal melanoma, for example, whereas fatal cardiac
arrhythmia from metastases may be mis-diagnosed as ischemic in
origin (Makitie and Kivela, 2001). Fortunately, because uveal
melanomas are so rare, it is possible to estimate the metastatic
mortality without depending on the certified cause of death
(Hakulinen and Dyba, 2007). This is done by comparing the allcause mortality of the patient population with that of the general
population matched for age and sex. We are fortunate in obtaining
reliable mortality data from the National Health Service (NHS)
Cancer Registry on all patients resident in mainland Britain, who
are flagged by their NHS number.
3.2. Liverpool Uveal Melanoma Prognosticator Online (LUMPO)
Fig. 3. KaplaneMeier survival curves showing metastatic mortality according to
chromosomes 3 loss and 8q gain as determined by MLPA. All 133 patients without
chromosome 3 loss survived so that the curves for patients without any chromosome
abnormality and those with only chromosome 8q gain overlie each other. Three of 56
patients with chromosome 3 loss died despite having no apparent chromosome 8q
gain. It is not yet known whether subtle chromosome 8q gain was missed or whether
metastatic death can occur without this abnormality. With permission from Damato
et al. (2010).
have been developed to take competing risks into account, as in the
study by Kujala et al., cited earlier.
A problem with mortality statistics is that they depend greatly
on the accuracy with which the cause of death is diagnosed. With
We have developed an online tool that generates an all-cause
mortality curve according to age, sex, TNM size category (based
on basal tumor diameter and tumor height), ciliary body involvement, melanoma cytomorphology, closed loops, mitotic count,
chromosome 3 loss, and presence of extraocular spread (www.
ocularmelanomaonline.com). It also categorizes tumors according
to their TNM stage.
Previously, we relied on neural networks, which were trained
using data from patients treated by local resection or enucleation
(Damato et al., 2008). However, these were found not to cope well
with imputation procedures dealing with missing data, such as
closed loops and mitotic count, which could not be assessed in tiny
Fig. 4. Survival after treatment of choroidal melanoma. KaplaneMeier survival curves showing metastatic mortality from choroidal melanoma with basal tumor diameter of
11.5e14.4 mm, in 40e50-year-old men (A) and 70e80-year-old men (B). LUMPO plots of all-cause mortality from choroidal melanoma with basal diameter of 13 mm, in 45-year-old
men (C) and 75-year-old men (D). According to KaplaneMeier analysis, older age was associated with an increase in metastatic mortality from 13% to 33% whereas with LUMPO
older age was associated with a decrease in the metastatic mortality from 11% to 9%. With permission from Damato et al. (2008). See Figure 6 for explanation of survival curves.
Please cite this article in press as: Damato, B., et al., Estimating prognosis for survival after treatment of choroidal melanoma, Progress in Retinal
and Eye Research (2011), doi:10.1016/j.preteyeres.2011.05.003
B. Damato et al. / Progress in Retinal and Eye Research xxx (2011) 1e11
7
to more than 20 years. The performance of the model was assessed
by bootstrap re-sampling (i.e., randomly splitting the entire dataset
into training and test datasets 200 times) (Eleuteri et al., in press).
LUMPO produces estimates that are accurate enough to be relevant
to individual patients (Fig. 5) (Eleuteri et al., in press). This is
because the multivariate analysis includes not only clinical features
but also histologic grade of malignancy and genetic tumor type.
Fig. 6 shows the importance of including all these factors in the
prognostic model.
The program also generates a survival curve for the general
British population matched for age and sex (Fig. 6). This enables the
metastatic mortality to be estimated, as mentioned before (3.2).
The 95% confidence intervals are plotted to indicate the reliability of
each prediction. A pictogram is also drawn, to facilitate communication with patients.
Such ‘personalised prognostication’ has significantly changed
our clinical practice. Further studies will be needed to determine
whether LUMPO can be used with non-Caucasians and with
patients in other centers and countries, whose life-expectancy may
be quite different to that of British patients treated at our hospital.
4. Harvesting tumor tissue for prognostication
Fig. 5. Comparisons of observed mortality versus the mortality predicted with LUMPO
in (a) the clinical model (i.e., tumor diameter and thickness, ciliary body involvement
and extraocular spread) and the laboratory model (i.e., merging clinical model with
melanoma cytomorphology, closed loops, mitotic count per 40 high-power fields, and
chromosome 3 status). A perfect fit lies on the 45 line. The figure was obtained by
bootstrap re-sampling, so for every point on the graph the model was trained on
a subset of the data, and tested on a disjoint dataset. The confidence intervals are wider
in the laboratory model because of smaller numbers. From Eleuteri et al. (in press),
with permission.
biopsy samples (Eleuteri et al., in press). Our latest tool therefore
uses the Accelerated Failure Time (AFT) model, which specifies that
predictors act additively, each predictor accelerating the progress of
the disease and shortening the survival time (Harrell, 2001). We
were unable to use the Cox proportional hazards model because
predictors such as basal tumor diameter did not satisfy the required
proportionality of hazards assumption (i.e., different basal tumor
diameters produced survival curves of different shape). LUMPO also
interpolates data, thereby adjusting for erroneous results to some
extent. For example, although metastatic death from disomy-3
uveal melanomas is very rare, our tool would nevertheless indicate a poor survival probability if the tumor is not found to have
monosomy 3 but if it shows epithelioid cytomorphology, closed
loops, ciliary body involvement, and large size. Chromosome 8q
gain and chromosome 6p gain are not included because they do not
improve the model, possibly because of insufficient data; however,
this may change in future, once we have more patients with longer
follow-up.
Our current online survival predictor was trained with our own
dataset of 3653 uveal melanoma patients, with follow-up ranging
Between 1999 and 2007, we performed genetic typing and
histologic grading only on tumors treated by enucleation or local
resection; prognostication on irradiated tumors was based only on
tumor dimensions and extent. In 2007, we started performing
prognostic biopsy on such cases. This was because our experience had
convinced us of the clinical benefit of accurate prognostication and
the need for multivariate analysis using histologic and genetic data.
For patients treated with brachytherapy, we perform fineneedle aspiration biopsy using a 25-G beveled needle, which is
passed obliquely through the sclera overlying the tumor immediately before inserting the radioactive plaque. In patients having
a post-equatorial tumor and those treated with proton beam
radiotherapy, we perform the biopsy within days of completing the
radiotherapy, using the sutureless, trans-conjunctival, 25-G
vitrectomy system. We are still auditing these procedures, to
assess the adequacy of the tumor samples and determine the
incidence and outcome of complications such as episcleral tumor
seeding, vitreous hemorrhage, rhegmatogenous retinal detachment, and endophthalmitis. Interim results suggest that our
success rates have not been as high as those reported by some other
authors (Shields et al., 2011). Biopsy forceps have been developed
to enhance the yield (Akgul et al., 2011).
5. Scope of prognostication
5.1. Clinical care
We have found that almost all of our patients want to know their
prognosis for survival, whether this is good or bad and even when
they are told that prognostication is most unlikely to improve their
chances of prolonging life (Cook et al., 2008). In-depth psychological studies have shown that patients given a poor outlook only
rarely regret their decision to have prognostication (Cook et al.,
2008). Although bad news is indeed upsetting, patients develop
compensatory mechanisms and feel a sense of empowerment over
their future planning, which they value (Cook et al., 2010). Interestingly, patients find a poor prognosis less upsetting than an
uncertain prognosis (Cook et al., 2010). The patients who are most
distressed are those who cannot be given an accurate prognosis
because genetic testing has failed. Some might question whether
patients are able to provide informed consent for prognostication,
especially when they also need to decide on their ocular treatment
Please cite this article in press as: Damato, B., et al., Estimating prognosis for survival after treatment of choroidal melanoma, Progress in Retinal
and Eye Research (2011), doi:10.1016/j.preteyeres.2011.05.003
8
B. Damato et al. / Progress in Retinal and Eye Research xxx (2011) 1e11
Fig. 6. Liverpool Uveal Melanoma Prognosticator Online, showing all-cause mortality for a 60-year-old female with a 16 mm cilio-choroidal melanoma and no extraocular spread.
(A) Survival curves based solely on clinical features (i.e., tumor dimensions, ciliary body involvement and extraocular spread), with line P corresponding to patient survival and G
indicating survival of general British population matched for age and gender; (B) Pictogram showing same results; (C) Survival curve for same patient if laboratory studies show no
epithelioid cells, no closed loops, low mitotic count (i.e., <2/40 high-power fields) and no monosomy 3; and (D) survival curve for same patient if laboratory studies show
epithelioid cells, closed loops, mitotic rate exceeding 7/40 high-power fields, and monosomy 3. The 8-year metastatic mortality can be deduced by subtracting the patient mortality
from the mortality of the matched general population and is estimated to be: 40% if only clinical features are known; 6% if the tumor is found to be of low-grade, disomy-3 type; and
70% if the tumor is of high-grade malignancy with monosomy 3 (Damato et al., unpublished data).
at a very stressful time. However, this is not a significant problem,
as our own studies have shown (Cook et al., 2010).
Patients receiving a good prognosis tend to be reassured by the
news, but some remain sceptical so that special measures need to
be taken to convince them of their good fortune (e.g., informing
them of evidence base). One patient became extremely upset on
being informed that his diffuse melanoma was of spindle-cell,
disomy-3 type and therefore likely to be non-lethal: he wrongly
concluded that his eye had been enucleated unnecessarily. He had
been sent the good news by mail but had ignored our advice to
discuss this letter with our specialist nurse by phone, instead
making a formal complaint directly to the chief executive of our
hospital. We were, however, able to reassure him of the necessity of
the operation, despite the tumor’s relatively low-grade malignancy.
This mishap taught us that the response to prognostication can be
unpredictable so that it is important to proactively encourage
questions and discussion, if possible in person.
Patients’ attitudes to prognostication, as well as their reaction to
their survival predictions, are likely to vary according to their age,
gender, social class and nationality, not to mention the way in
which they are counseled by their medical practitioner. Further
studies are therefore needed to determine whether our experience
is replicated in other centers and countries.
Thanks to our prognostication methods, we have been able to
screen high-risk patients for metastatic disease and have identified
resectable hepatic metastases in a small number of patients, some
of whom have enjoyed prolonged survival (Fig. 7).
Screening for metastases from uveal melanoma is expensive and
causes significant stress to patients, especially if they have to travel
long distances or if there is a false-positive result, which is not
uncommon. Further studies are needed to determine whether
prognostication and screening actually prolong survival.
5.2. Research
There is scope for randomised trials evaluating different forms
of systemic adjuvant therapy, in the hope that metastatic disease
and death might be delayed or even prevented altogether, as has
been achieved with orbital rhabdomyosarcoma, breast cancer and
other malignancies. Unfortunately, previous studies with uveal
melanoma have failed to show statistically-significant benefit
(Desjardins et al., 1998). A recent study investigating intrahepatic
fotemustine showed a non-significant trend toward improved
survival (Voelter et al., 2008). However, the high-risk patients were
selected only on the basis of large tumor size, and from our own
data we estimate that about 25% of these are likely to have had
a non-lethal melanoma. Presuming that the observed difference
was real and not due to chance, if prognostication had also been
based on histologic and genetic data, albeit with slower accrual,
then patients with non-lethal melanoma would not have been
Please cite this article in press as: Damato, B., et al., Estimating prognosis for survival after treatment of choroidal melanoma, Progress in Retinal
and Eye Research (2011), doi:10.1016/j.preteyeres.2011.05.003
B. Damato et al. / Progress in Retinal and Eye Research xxx (2011) 1e11
9
Fig. 7. A 48-year-old man was referred in 2007 with an amelanotic choroidal melanoma in the right eye (A). The tumor measured 8.7 mm 7.9 mm in basal diameter with
a thickness of 2.2 mm (B). The patient had a biopsy and a trial of photodynamic therapy was commenced, the patient accepting that this was not standard treatment. Because of the
small tumor size we would not normally have recommended life-long screening for metastasis, estimating the 10-year metastatic mortality to be approximately 10%. Histology
revealed epithelioid cells and MLPA showed chromosome 3 loss and gains in chromosome 8q (C). The photodynamic therapy was abandoned and the patient was treated with
proton beam radiotherapy, which is more reliable. Multivariate analysis indicated a 19% risk of metastatic death in 8 years and the patient underwent 6-monthly MRI. In 2009, two
lesions were detected in the right lobe of the liver and these were resected without complication. A year later, metastases developed in the left lobe of the liver as well as in bone
and lungs. The patient was treated with ipilimumab and was still alive in mid-2011, when this article was written. The patient may not have survived as long if biopsy had not been
performed and if his metastases had not been detected at a later stage.
enrolled in this study and a statistically-significant result might
have been obtained, changing clinical practice.
There is also a need for randomised trials evaluating the impact
of ocular treatment on metastatic disease. It is not known whether
ocular treatment of uveal melanoma ever influences survival and if
so in whom (Damato, 2010). Enucleation has at various times been
considered to prolong life, shorten life, and to be inconsequential.
Similarly, it is not known whether the risk of metastatic death is
iatrogenically increased by surgical manipulation, local tumor
recurrence and treatment delays. Randomised trials have either
been impossible or under-powered (e.g., Collaborative Ocular
Melanoma Study (COMS)) (Damato, 2007). This is because
approximately 1000 patients would need to be studied (Tishkovskaya, S et al., unpublished data). Because many patients are likely
to be lost to accrual or follow-up, at least twice this number would
need to be enrolled and this would require international, multicenter studies, which would be difficult and expensive to organize.
6. Putting theory into practice
When counseling new patients with uveal melanoma, we offer
prognostication, discussing the risks and benefits, taking account of
each individual’s general health and personality. We provide them
with an audio-recording of the consultation on CD-ROM or cassette
to help them remember what they were told. Consent is requested
for the use of their samples and data for research. Fresh tumor
samples are obtained from excision or biopsy specimens for histologic and genetic studies. These investigations are now performed in
our hospital, by experienced staff working in an accredited
pathology laboratory, and, in addition to free text, synoptic reports
are issued to avoid ambiguity. Quality control laboratory studies are
performed in collaboration with other centers. Data are prospectively computerised into our ocular oncology database/patient
management system by a full-time data manager and these data are
used routinely for clinical care, audit and research. The clinical,
histologic and genetic data are entered into an online form (i.e.,
LUMPO) (3.2) to obtain a personalised survival curve. A printout of
this survival curve is sent to the patient’s referring ophthalmologist,
family doctor and to our medical oncologist, together with
a covering letter, explaining the significance of all results and the
proposed management plan. The charts are reviewed so that this
letter takes into account of the patient’s age and general health. In
our covering letter and in any counseling, we note the width of the
confidence intervals relating to the survival curves and explain that
Please cite this article in press as: Damato, B., et al., Estimating prognosis for survival after treatment of choroidal melanoma, Progress in Retinal
and Eye Research (2011), doi:10.1016/j.preteyeres.2011.05.003
10
B. Damato et al. / Progress in Retinal and Eye Research xxx (2011) 1e11
estimates of life-expectancy may turn out to be wrong or misleading.
For example, of the patients who are reassured by a 90% chance of
surviving ten years, there will be 10% who develop metastatic
disease sooner than this. Even though all patients are informed from
the outset that prognostication involves approximation, there is
scope for formal psychological studies assessing how such patients
react when metastatic disease develops unexpectedly. It has been
suggested that molecular testing of uveal melanomas must
demonstrate high reproducibility, reliability and validity before it is
introduced into clinical care (Harbour, 2009). Although such an ideal
is desirable, if clinical deployment of genetic testing were to be
delayed until high levels of reliability are proven, then it would be
necessary to rely on crude indicators such as tumor size and patients
would be deprived of meaningful prognostication. We believe that
a more realistic approach is to offer the best methods available,
following the principles of good laboratory and medical practice
whilst addressing technical limitations and clinical uncertainties
appropriately.
If the patient has an excellent survival probability, a letter is sent
to the patient giving the good news, explaining that there is little
scope for systemic screening. This letter is followed by a telephone
call from specialist oncology nurse to discuss the result. An individualised screening policy is formulated on a case-by-case basis,
taking the patient’s wishes and fears into account.
If the patient has a high risk of metastatic disease, an appointment is made at our clinic to discuss the result and its implications,
to undergo liver imaging and liver function tests, to receive counseling from our health psychologist, and to make arrangements for
further management. Depending on the level of risk and the
patient’s ability to commute, further care is continued at our clinic
or at our oncologist’s hospital or at the patient’s local hospital, in
which case a detailed information pack is sent to the local hospital.
Patients developing metastatic disease are entered into various
trials, being led by the Liverpool Clinical Trials Unit and supported
by the National Cancer Research Institute Melanoma Group. It is
hoped that before long it will be possible to enroll high-risk
patients into trials of systemic adjuvant therapy.
Quality of life and other patient-related outcomes (e.g., difficulty
reading) are measured by sending patients a questionnaire after six
months and then on every anniversary of their treatment.
7. Future directions
It is likely that over the next few years there will be further
advances in biopsy techniques as well as histologic and genetic
methods. There is scope for greater standardisation with regards to
measuring tumor dimensions, performing laboratory investigations, categorising melanoma cytomorphology, defining genetic
tumor type and diagnosing cause of death.
We anticipate that we will continue training and evaluating our
prognostic model (i.e., LUMPO) and that as the dataset expands we
should be able to incorporate more predictors (e.g., chromosome 6p
gain). There is scope for investigating whether the multivariate
model would be enhanced by host factors, such as lymphocytic and
macrophage infiltrations and macrophage density, which we have
not so far recorded. With patients having rare combinations of
predictors we expect that in time greater numbers should make it
possible to improve the precision of survival estimates, so that
confidence intervals become narrower. We also hope to lengthen
the survival curves well beyond ten years as follow-up times
increase. At present, we use general population data to estimate the
likelihood of metastatic death, but we hope to be able to make
personalised adjustments if an individual patient’s life-expectancy
is known to be diminished by known lifestyle factors or disease.
Progress in prognostication would be accelerated by multicenter
collaboration, which should become more feasible as biopsy,
histologic grading and genetic typing become more widespread.
There is scope for multicenter studies to determine whether
methods used in one unit are reliable elsewhere and to compare
rival techniques and strategies. We are in the process of developing
an online tool to enable workers from other centers to determine
whether LUMPO is relevant to their data.
If prolongation of life is achieved in more patients as a result of
hepatic or systemic treatment, then statistical methods will need to
be developed to deal with any bias caused by such therapy.
Until now, we have relied solely on qualitative methods to
measure psychological responses to prognostication. There is scope
for quantitative studies statistically correlating quality of life with
the survival prognosis and for several years we have been collecting
data prospectively for this purpose.
The TNM staging manual does not yet advise on merging clinical
and laboratory tumor data. There are only recommendations for
collecting data and suggestions for histologic grading. However,
moves are afoot to improve the TNM system so that it uses all
available data.
In conclusion, the accuracy of prognostication is greatly
enhanced by multivariate analysis of clinical, histologic and genetic
data, also taking age and sex into account. KaplaneMeier analysis is
inadequate because it does not cope with continuous variables and
competing risks. Cox proportional hazards model is not possible
because the hazards are not proportional. We have developed and
validated online prognostication tools that are accurate enough to
be relevant to individual patients.
Patients with a “low-risk” uveal melanoma are more likely to
enjoy a good quality of life if they can confidently and reliably be
informed of their excellent prognosis, especially if they are spared
unnecessary screening for metastatic disease. Special precautions
are needed to avoid misunderstandings from arising.
Patients with a high risk of metastatic disease tend to develop
coping mechanisms so that their quality of life is far better than one
might expect. At least in Britain, they do not regret being informed
of their prognosis because they want to be empowered to sort out
their personal affairs and to do all they can to improve their chances
of survival.
Improved identification of “high-risk” patients has led to targeted screening for metastatic disease and better opportunities for
clinical trials evaluating different treatments for metastatic disease.
More accurate prognostication also enhances prospects for randomised trials evaluating systemic adjuvant therapy as well as basic
science research.
These benefits have come about because of simple yet robust
protocols and infrastructures that have enabled us to prospectively
collect and computerise clinical and laboratory data and outcomes
uninterruptedly for over one-quarter of a century.
The prognostic methods we have developed for ocular oncology
should be widely applicable, not only in ophthalmology and
oncology but also in other fields of medicine.
Acknowledgements
We are grateful to the National Specialist Commissioning Group
(NSCG) of the National Health Service of England for sponsoring our
service, including genetic studies, pathology, database, and statistical support.
References
Ah-Fat, F.G., Damato, B.E., 1998. Delays in the diagnosis of uveal melanoma and
effect on treatment. Eye 12, 781e782.
Please cite this article in press as: Damato, B., et al., Estimating prognosis for survival after treatment of choroidal melanoma, Progress in Retinal
and Eye Research (2011), doi:10.1016/j.preteyeres.2011.05.003
B. Damato et al. / Progress in Retinal and Eye Research xxx (2011) 1e11
Akgul, H., Otterbach, F., Bornfeld, N., Jurklies, B., 2011. Intraocular biopsy using
special forceps: a new instrument and refined surgical technique. Br. J. Ophthalmol. 95, 79e82.
Al Jamal, R.T., Mäkitie, T., Kivelä, T., 2003. Nucleolar diameter and microvascular
factors as independent predictors of mortality from malignant melanoma of the
choroid and ciliary body. Invest. Ophthalmol. Vis. Sci. 44, 2381e2389.
Al-Jamal, R.T., Kivelä, T., 2006. KI-67 immunopositivity in choroidal and ciliary body
melanoma with respect to nucleolar diameter and other prognostic factors.
Curr. Eye Res. 31, 57e67.
All-Ericsson, C., Girnita, L., Seregard, S., Bartolazzi, A., Jager, M.J., Larsson, O., 2002.
Insulin-like growth factor-1 receptor in uveal melanoma: a predictor for
metastatic disease and a potential therapeutic target. Invest. Ophthalmol. Vis.
Sci. 43, 1e8.
Angi, M., Damato, B., Kalirai, H., Dodson, A., Taktak, A., Coupland, S.E., 2009.
Immunohistochemical assessment of mitotic count in uveal melanoma. Acta
Ophthalmol..
Augsburger, J.J., Correa, Z.M., Shaikh, A.H., 2009. Effectiveness of treatments for
metastatic uveal melanoma. Am. J. Ophthalmol. 148, 119e127.
Bauer, J., Kilic, E., Vaarwater, J., Bastian, B.C., Garbe, C., de Klein, A., 2009. Oncogenic
GNAQ mutations are not correlated with disease-free survival in uveal melanoma. Br. J. Cancer 101, 813e815.
Cook, S.A., Damato, B., Marshall, E., Salmon, P., 2010. Reconciling the principle of
patient autonomy with the practice of informed consent: decision-making
about prognostication in uveal melanoma. Health Expect., 10e7625.
Cook, S.A., Damato, B., Marshall, E., Salmon, P., 2008. Psychological aspects of
cytogenetic testing of uveal melanoma: preliminary findings and directions for
future research. Eye. doi:10.1038/eye.2008.54.
Coupland, S.E., Campbell, I., Damato, B., 2008. Routes of extraocular extension of
uveal melanoma: risk factors and influence on survival probability. Ophthalmology 115, 1778e1785.
Damato, B., 2010. Does ocular treatment of uveal melanoma influence survival? Br.
J. Cancer 103, 285e290.
Damato, B., 2001. Detection of uveal melanoma by optometrists in the United
Kingdom. Ophthalmic Physiol. Opt. 21, 268e271.
Damato, B., 2007. Legacy of the collaborative ocular melanoma study. Arch. Ophthalmol. 125, 966e968.
Damato, B., Coupland, S.E., 2009. A reappraisal of the significance of largest basal
diameter of posterior uveal melanoma. Eye (Lond.) 23, 2152e2160.
Damato, B., Duke, C., Coupland, S.E., Hiscott, P., Smith, P.A., Campbell, I., Douglas, A.,
Howard, P., 2007. Cytogenetics of uveal melanoma: a 7-year clinical experience.
Ophthalmology 114, 1925e1931.
Damato, B., Eleuteri, A., Fisher, A.C., Coupland, S.E., Taktak, A.F., 2008. Artificial
neural networks estimating survival probability after treatment of choroidal
melanoma. Ophthalmology 115, 1598e1607.
Damato, B.E., Dopierala, J.A., Coupland, S.E., 2010. Genotypic profiling of 452
choroidal melanomas with multiplex ligation-dependent probe amplification.
Clin. Cancer Res. 16, 6083e6092.
Desjardins, L., Dorval, T., Lévy, C., Cojean, I., Schlienger, P., Salmon, R.J., Validire, P.,
Asselain, B., 1998. Randomised study on adjuvant therapy by DTIC in choroidal
melanoma. Ophtalmologie 12, 168e173.
Dopierala, J., Damato, B.E., Lake, S.L., Taktak, A.F., Coupland, S.E., 2010. Genetic
heterogeneity in uveal melanoma assessed by multiplex ligation-dependent
probe amplification. Invest. Ophthalmol. Vis. Sci.
Eleuteri, A., Damato, B., Coupland, S.E., Taktak, A.F.G., in press. Enhancing survival
prognostication in patients with choroidal melanoma by integrating pathologic,
clinical and genetic predictors of metastasis. J. Biomed. Eng. Tech.
Ericsson, C., Seregard, S., Bartolazzi, A., Levitskaya, E., Ferrone, S., Kiessling, R.,
Larsson, O., 2001. Association of HLA class I and class II antigen expression and
mortality in uveal melanoma. Invest. Ophthalmol. Vis. Sci. 42, 2153e2156.
Finger, P.T., 2009. The 7th edition AJCC staging system for eye cancer: an international language for ophthalmic oncology. Arch. Pathol. Lab. Med. 133,
1197e1198.
Folberg, R., Rummelt, V., Parys-Van Ginderdeuren, R., Hwang, T., Woolson, R.F.,
Pe’er, J., Gruman, L.M., 1993. The prognostic value of tumor blood vessel
morphology in primary uveal melanoma. Ophthalmology 100, 1389e1398.
Hakulinen, T.R., Dyba, T.A., 2007. Recent developments in relative survival analysis.
In: Taktak, A.F.G., Fisher, A.C. (Eds.), Outcome Prediction in Cancer. Elsevier,
Amsterdam, pp. 43e64.
Harbour, J.W., 2009. Molecular prognostic testing and individualized patient care in
uveal melanoma. Am. J. Ophthalmol. 148, 823e829.
Harbour, J.W., Onken, M.D., Roberson, E.D., Duan, S., Cao, L., Worley, L.A.,
Council, M.L., Matatall, K.A., Helms, C., Bowcock, A.M., 2010. Frequent mutation
of BAP1 in metastasizing uveal melanomas. Science 330, 1410e1413.
Harrell, F.E.J., 2001. Parametric survival models. In: Regression Modeling Strategies.
Springer, pp. 413e441.
Jager, M.J., Ly, L.V., El Filali, M., Madigan, M.C., 2011. Macrophages in uveal melanoma and in experimental ocular tumor models: friends or foes? Prog. Retin.
Eye Res. 30, 129e146.
Jmor, F., Kalirai, H., Taktak, A., Damato, B., Coupland, S.E., 2010. HSP-27 protein
expression in uveal melanoma: correlation with predicted survival. Acta Ophthalmol., 10e3768.
11
Johansson, C.C., Mougiakakos, D., Trocme, E., All-Ericsson, C., Economou, M.A.,
Larsson, O., Seregard, S., Kiessling, R., 2009. Expression and prognostic significance of iNOS in uveal melanoma. Int. J. Cancer 126, 2682e2689.
Kilic, E., Naus, N.C., van, Gils W., Klaver, C.C., van Til, M.E., Verbiest, M.M., Stijnen, T.,
Mooy, C.M., Paridaens, D., Beverloo, H.B., Luyten, G.P., de Klein, A., 2005.
Concurrent loss of chromosome arm 1p and chromosome 3 predicts
a decreased disease-free survival in uveal melanoma patients. Invest. Ophthalmol. Vis. Sci. 46, 2253e2257.
Kim, I.K., Lane, A.M., Gragoudas, E.S., 2010. Survival in patients with presymptomatic diagnosis of metastatic uveal melanoma. Arch. Ophthalmol. 128, 871e875.
Kujala, E., Mäkitie, T., Kivelä, T., 2003. Very long-term prognosis of patients with
malignant uveal melanoma. Invest. Ophthalmol. Vis. Sci. 44, 4651e4659.
Lin, A.Y., Maniotis, A.J., Valyi-Nagy, K., Majumdar, D., Setty, S., Kadkol, S., Leach, L.,
Pe’er, J., Folberg, R., 2005. Distinguishing fibrovascular septa from vasculogenic
mimicry patterns. Arch. Pathol. Lab. Med. 129, 884e892.
Maat, W., van der Slik, A.R., Verhoeven, D.H., Alizadeh, B.Z., Ly, L.V., Verduijn, W.,
Luyten, G.P., Mulder, A., van Hall, T., Koning, F., Jager, M.J., van Bergen, J., 2009.
Evidence for natural killer cell-mediated protection from metastasis formation
in uveal melanoma patients. Invest. Ophthalmol. Vis. Sci. 50, 2888e2895.
Mäkitie, T., Carpen, O., Vaheri, A., Kivelä, T., 2001a. Ezrin as a prognostic indicator
and its relationship to tumor characteristics in uveal malignant melanoma.
Invest. Ophthalmol. Vis. Sci. 42, 2442e2449.
Mäkitie, T., Kivelä, T., 2001. Cardiac metastasis from uveal melanoma. Arch. Ophthalmol. 119, 139e140.
Mäkitie, T., Summanen, P., Tarkkanen, A., Kivelä, T., 2001b. Tumor-infiltrating
macrophages (CD68(þ) cells) and prognosis in malignant uveal melanoma.
Invest. Ophthalmol. Vis. Sci. 42, 1414e1421.
Mäkitie, T., Summanen, P., Tarkkanen, A., Kivelä, T., 1999. Microvascular loops and
networks as prognostic indicators in choroidal and ciliary body melanomas.
J. Natl. Cancer Inst. 91, 359e367.
McLean, I.W., Zimmerman, L.E., Evans, R.M., 1978. Reappraisal of Callender’s spindle
a type of malignant melanoma of choroid and ciliary body. Am. J. Ophthalmol.
86, 557e564.
Onken, M.D., Worley, L.A., Ehlers, J.P., Harbour, J.W., 2004. Gene expression profiling
in uveal melanoma reveals two molecular classes and predicts metastatic
death. Cancer Res. 64, 7205e7209.
Onken, M.D., Worley, L.A., Tuscan, M.D., Harbour, J.W., 2010. An accurate, clinically
feasible multi-gene expression assay for predicting metastasis in uveal melanoma. J. Mol. Diagn. 12, 461e468.
Parrella, P., Sidransky, D., Merbs, S.L., 1999. Allelotype of posterior uveal melanoma:
implications for a bifurcated tumor progression pathway. Cancer Res. 59,
3032e3037.
Prescher, G., Bornfeld, N., Hirche, H., Horsthemke, B., Jöckel, K.H., Becher, R., 1996.
Prognostic implications of monosomy 3 in uveal melanoma. Lancet 347,
1222e1225.
Sato, T., 2010. Locoregional management of hepatic metastasis from primary uveal
melanoma. Semin. Oncol. 37, 127e138.
Seddon, J.M., Albert, D.M., Lavin, P.T., Robinson, N., 1983. A prognostic factor study of
disease-free interval and survival following enucleation for uveal melanoma.
Arch. Ophthalmol. 101, 1894e1899.
Seregard, S., Spangberg, B., Juul, C., Oskarsson, M., 1998. Prognostic accuracy of the
mean of the largest nucleoli, vascular patterns, and PC-10 in posterior uveal
melanoma. Ophthalmology 105, 485e491.
Shields, C.L., Furuta, M., Thangappan, A., Nagori, S., Mashayekhi, A., Lally, D.R., Kelly, C.C.,
Rudich, D.S., Nagori, A.V., Wakade, O.A., Mehta, S., Forte, L., Long, A., Dellacava, E.F.,
Kaplan, B., Shields, J.A., 2009. Metastasis of uveal melanoma millimeter-bymillimeter in 8033 consecutive eyes. Arch. Ophthalmol. 127, 989e998.
Shields, C.L., Ganguly, A., Bianciotto, C.G., Turaka, K., Tavallali, A., Shields, J.A., 2011.
Prognosis of uveal melanoma in 500 cases using genetic testing of fine-needle
aspiration biopsy specimens. Ophthalmology 118, 396e401.
Trolet, J., Hupe, P., Huon, I., Lebigot, I., Decraene, C., Delattre, O., Sastre-Garau, X.,
Saule, S., Thiery, J.P., Plancher, C., Asselain, B., Desjardins, L., Mariani, P., PipernoNeumann, S., Barillot, E., Couturier, J., 2009. Genomic profiling and identification of high-risk uveal melanoma by array CGH analysis of primary tumors and
liver metastases. Invest. Ophthalmol. Vis. Sci. 50, 2572e2580.
Virgili, G., Gatta, G., Ciccolallo, L., Capocaccia, R., Biggeri, A., Crocetti, E., Lutz, J.M.,
Paci, E., 2008. Survival in patients with uveal melanoma in Europe. Arch.
Ophthalmol. 126, 1413e1418.
Voelter, V., Schalenbourg, A., Pampallona, S., Peters, S., Halkic, N., Denys, A.,
Goitein, G., Zografos, L., Leyvraz, S., 2008. Adjuvant intra-arterial hepatic fotemustine for high-risk uveal melanoma patients. Melanoma Res. 18, 220e224.
Whelchel, J.C., Farah, S.E., McLean, I.W., Burnier, M.N., 1993. Immunohistochemistry
of infiltrating lymphocytes in uveal malignant melanoma. Invest. Ophthalmol.
Vis. Sci. 34, 2603e2606.
White, V.A., Chambers, J.D., Courtright, P.D., Chang, W.Y., Horsman, D.E., 1998.
Correlation of cytogenetic abnormalities with the outcome of patients with
uveal melanoma. Cancer 83, 354e359.
Young, T.A., Burgess, B.L., Rao, N.P., Gorin, M.B., Straatsma, B.R., 2007. High-density
genome array is superior to fluorescence in-situ hybridization analysis of
monosomy 3 in choroidal melanoma fine needle aspiration biopsy. Mol. Vis. 13,
2328e2333.
Please cite this article in press as: Damato, B., et al., Estimating prognosis for survival after treatment of choroidal melanoma, Progress in Retinal
and Eye Research (2011), doi:10.1016/j.preteyeres.2011.05.003