Thill et al. demonstrated a variety of putative stem cell markers in uveal melanoma and these included CD133, Pax6, Musashi, nestin, Sox2 and ABCB5; this was done by examining eight uveal melanoma cell lines and eight paraffin embedded primary uveal melanomas, using fluorescence activated cell sorting, immunohistochemistry and reverse transcriptase PCR [42]. A limitation of this study, however, as recognised by the authors themselves, was the inability of these markers to unequivocally identify the cancer stem cell population. Indeed, the expression levels of cancer stem cell markers, such as CD133, CD44 and ABCB transporters, has been highly variable between studies for a particular tumour type and often differs depending on the experimental techniques employed for identification [60, 61].

To overcome these limitations and to characterise downstream molecules, it is necessary to develop standardised functional assays that identify tumour cells with stem cell like properties.

Few investigators have searched for cancer stem cells in uveal melanoma despite the availability of cell lines derived from both primary and metastatic tumours.

Using in vitro clonal analyses, serial passaging, re-plating assays, immunophenotyping and reverse transcriptase-PCR, we recently provided functional evidence that uveal melanoma cell lines contain a subpopulation of self-renewing cancer cells as well as cells that can proliferate and differentiate [41]. In this study, uveal melanoma cell lines showed distinct clonal morphologies in adherent culture akin to holoclones, meroclones and paraclones. These cells formed melanomaspheres (MS) when grown at clonal density in non-adherent culture, which could be serially propagated for several generations. MS demonstrated antigenic heterogeneity expressing markers associated with both a primitive migratory neural crest phenotype (Sox10, Pax3, Notch1 and slug) and a more differentiated phenotype (MelanA and HMB45). Moreover, the uveal melanoma cells surviving cisplatin treatment produced significantly more holoclones than untreated cells, suggesting enrichment for this cancer stem cell-like subpopulation. These data are consistent with studies using cancer cell lines from a range of tumour types, including prostate, breast and head and neck, which have all demonstrated retention of a subpopulation with stem cell-like characteristics [62–65]. Such studies have shown that in vitro clonogenicity correlates well with in vivo tumour initiating abilities [62–64].

Another functional property of the normal stem cell population is their ability to efflux cytotoxic chemicals, thereby protecting them from damage and death. To achieve this, they express high levels of the ABC group of drug transporters [66, 67]. These same transporters have also been shown to afford protection to cancer stem cells contributing to multidrug resistance [68]. The multidrug resistance proteins, P-glycoprotein (also known as ABCB1), multidrug resistance-associated protein (MRP1) and lung resistance protein (LRP) have been identified in primary uveal melanoma specimens and cell lines [69, 70]. More recently, ABCB1 was shown to be expressed in primary uveal melanomas, but not in normal uveal melanocytes [71]. In the same study, ABCB1− and ABCB1+ cell populations were isolated from the uveal melanoma cell line OCM1A and examined for their tumourigenic and metastatic propensity. The ABCB1+ cells exhibited enhanced clonogenicity and anchorage-independent growth. Furthermore, although both ABCB1+ and ABCB1− cells formed tumours in mice, the ABCB1+ cells did so more efficiently and produced significantly larger tumours than the ABCB1− cells. Similarly, when injected into the tail vein of NOD SCID gamma mice, ABCB1+ cells formed metastatic growths in the liver that tended to be larger and more frequent in number than when the ABCB1− cells were injected. Based on our current knowledge that efflux pumps afford protection to cancer stem cells, shielding them from the adverse effects of therapeutic insult, there is much scope for investigating these proteins and the mechanisms by which they may contribute to therapeutic resistance in uveal melanoma.

Metastasis is a multi-step process involving tumour–stromal interactions that enable tumour angiogenesis, tumour cell migration and invasion into the secondary site. Each type of cancer tends to have its own target organs for metastases. For example, uveal melanomas metastasise preferentially to the liver [72, 73], suggesting that the microenvironment in the secondary organ must be very important, as originally hypothesised by Paget’s “seed and soil” theory (reviewed in [74]).

It is well known that following ablation or surgical removal of the primary uveal melanoma, some patients develop metastatic disease after many years of clinical remission [75]. One explanation for this phenomenon of “tumour dormancy” is that the dormant metastatic cells are cancer stem cells in a quiescent state.

Hepatic metastases from uveal melanoma are usually very numerous and widely dispersed throughout the liver. It is not known whether such miliary metastases represents an enhanced cancer stem cell pool in the original primary tumour or de-differentiation of uveal melanoma cells by the local liver microenvironment to a cancer stem cell phenotype.

The molecular characterisation of uveal melanoma stem cells is still at a very early stage. The origin of uveal melanoma stem cells has yet to be determined. Whether uveal melanoma stem cells are derived from melanocyte stem cells, melanocyte progenitors, or more mature melanocytes that have de-differentiated remains unknown and requires further investigation.

Many significant experimental challenges remain. There is a need for standardised and sensitive self-renewal assays, as well as models that mimic uveal melanoma cells and their local microenvironment.

If cancer stem cells are indeed the driving force behind uveal melanoma development, progression and metastasis, this would have profound therapeutic implications. We could at last achieve the breakthroughs that have eluded us for so many years.