On 2014 Mar 29, Tero Kivelä commented:

Thank you very much for useful clarifications!

On 2014 Mar 24, Robert M Verdijk commented:

The current TNM classification was published when this study was ongoing. The aim of this study was to evaluate the histopathologic confirmed presence of extraocular extension and therefore we only included enucleated eyes in our study. Determination of the size of the extraocular extension after eye-conserving therapy is not completely independent from interpretation, and every other type of examination than histopathologic examination will not be as specific to determine microscopic extraocular extension. In our multivariate analysis a larger episcleral diameter of extraocular extension (HR=1.078) and presence of chromosome 8q gain (HR= 2.874) were correlated with worse survival. As chromosome 8q gain is associated with a larger tumor size (van den Bosch et al., 2013) we added tumor prominence and tumor size in our model in order to diminish the effect of these parameters. To answer the questions from Kivelä we also analyzed the tumors enucleated before 1999 and after 1999 separately, and found no differences in survival. The parameter that was strongly associated with survival before and after 1999 was gain of chromosome 8q, consistent with what we already reported, nevertheless, indeed we agree with Kivelä as discussed in our paper that one should be careful in extrapolation of the data.

In addition we did not evaluate macrophage infiltration separately in this study, and no immunohistochemistry was used. Extracellular matrix patterns were evaluated within this study, but no immunohistochemistry for microvessel density measurement was performed. We did not choose to evaluate these parameters. For extracellular matrix patterns closed loop networks were evaluated and defined as at least three back to back loops. Furthermore these were evaluated using non-counterstained PAS stain. A dark green filter was not used since in our experience closed loop networks can be identified without the filter. In all our studies closed loop networks correlated with prognosis.

We reported in our paper that two patients treated with fractionated stereotactic radiotherapy had extraocular extension. For both patients the B-scan did not show evident extraocular extension at time of diagnosis. After enucleation one patient had an extraocular extension of 0.10 mm, and for the other patient the extension was 3.0 mm. The indication for enucleation in these patients was secondary glaucoma and intraocular tumor progression, respectively.

As requested by Kivelä we constructed Kaplan-Meier curves according to the separate groups for extraocular extension, as subdivided in the 7th edition of the TNM classification for uveal melanoma. The survival curves and show a decreased survival in larger episcleral diameter of extraocular extension. These supplementary Figures 1 and 2 could not be introduced in this reply, but can be requested by email and will be available through my Researchgate account.

References: van den Bosch T, van Beek JG, Vaarwater J, Verdijk RM, Naus NC, Paridaens D, de Klein A, Kiliç E. Higher percentage of FISH-determined monosomy 3 and 8q amplification in uveal melanoma cells relate to poor patient prognosis. Invest Ophthalmol Vis Sci. 2012 May 14;53(6):2668-74.

On 2014 Mar 11, Tero Kivelä commented:

This is an important contribution, because it is the first study that lends independent support to the new Tumor, Node, Metastasis (TNM) classification subcategories for extraocular extension of uveal melanoma that were introduced in the 7th edition of this classification (Kujala E, 2013). The new size categories already had been independently confirmed (Shields CL, 2013).

The subdivisions for extraocular extension in TNM were based on a smaller data set than the size categories, and the independent significance of extraocular extension moreover had been challenged (Coupland SE, 2008) because another dataset suggested that extraocular extension reflected tumor malignancy and would not be significant after adjusting for tumor size, presence of epithelioid cells, closed extravascular matrix loops, high mitotic rate, and monosomy 3. The TNM bulding data set did not contain histopathologic and genetic variables, and could not address this criticism, whereas the present study adjusted for all these factors and still found extraocular extension, especially its size, to be independently associated with metastatic death. This is encouraging but not yet final proof.

All studies have some limitations and the following will affect interpretation of the present one:

1: The material was enriched in large tumors: it was unselected between 1987 and 1999, but between 1999 and 2011 only large melanomas (diameter >16 mm and thickness >12 mm) were enucleated and thus were available for analysis. The reported statistics thus do not directly apply to consecutive, unselected uveal melanomas (because statistics can only be extrapolated to a population that is similar to the sample).

2: Inflammation was roughly determined by presence of obvious clusters of lymphoid inflammatory cells, whereas macrophage infiltration also is associated with survival and with monosomy 3 in uveal melanomas (e.g. Mäkitie T, 2001, Maat W, 2008, Bronkhorst IH, 2011).

3: Microvascular density was not analyzed (e.g. Mäkitie T, 1999, Chen X, 2002); it is associated with prognosis of uveal melanoma independent of cell type, extravascular matrix patterns and inflammation (macrophages) and might have entered the model instead of another variable.

4: Kaplan-Meier graph by the diameter of the extraocular extension, which would have allowed direct comparison with the TNM data, is absent.

5: The paper does not explicitly mention:

5.1: Whether secondarily enucleated eyes had an extraocular extension already at the time of radiotherapy or developed it later.

5.2: What extracellular matrix patterns refer to in this study. They were coded as being absent or present. However, all uveal melanomas have at least one of the nine described patterns (even absence of vessels is a pattern, i.e. silent; Folberg R, 1993).

5.3: Were extracellular matrix patterns identified from non-counterstained sections under a dark green filter as originally described (Folberg R, 1993) or from counterstained sections in which they may be less obvious (McLean IW, 1997).

Conflict of interest: I was one author of the uveal melanoma chapter of the 7th edition of the TNM classicifation, American Joint Committee on Cancer.

On 2014 Mar 11, Tero Kivelä commented:

This is an important contribution, because it is the first study that lends independent support to the new Tumor, Node, Metastasis (TNM) classification subcategories for extraocular extension of uveal melanoma that were introduced in the 7th edition of this classification (Kujala E, 2013). The new size categories already had been independently confirmed (Shields CL, 2013).

The subdivisions for extraocular extension in TNM were based on a smaller data set than the size categories, and the independent significance of extraocular extension moreover had been challenged (Coupland SE, 2008) because another dataset suggested that extraocular extension reflected tumor malignancy and would not be significant after adjusting for tumor size, presence of epithelioid cells, closed extravascular matrix loops, high mitotic rate, and monosomy 3. The TNM bulding data set did not contain histopathologic and genetic variables, and could not address this criticism, whereas the present study adjusted for all these factors and still found extraocular extension, especially its size, to be independently associated with metastatic death. This is encouraging but not yet final proof.

All studies have some limitations and the following will affect interpretation of the present one:

1: The material was enriched in large tumors: it was unselected between 1987 and 1999, but between 1999 and 2011 only large melanomas (diameter >16 mm and thickness >12 mm) were enucleated and thus were available for analysis. The reported statistics thus do not directly apply to consecutive, unselected uveal melanomas (because statistics can only be extrapolated to a population that is similar to the sample).

2: Inflammation was roughly determined by presence of obvious clusters of lymphoid inflammatory cells, whereas macrophage infiltration also is associated with survival and with monosomy 3 in uveal melanomas (e.g. Mäkitie T, 2001, Maat W, 2008, Bronkhorst IH, 2011).

3: Microvascular density was not analyzed (e.g. Mäkitie T, 1999, Chen X, 2002); it is associated with prognosis of uveal melanoma independent of cell type, extravascular matrix patterns and inflammation (macrophages) and might have entered the model instead of another variable.

4: Kaplan-Meier graph by the diameter of the extraocular extension, which would have allowed direct comparison with the TNM data, is absent.

5: The paper does not explicitly mention:

5.1: Whether secondarily enucleated eyes had an extraocular extension already at the time of radiotherapy or developed it later.

5.2: What extracellular matrix patterns refer to in this study. They were coded as being absent or present. However, all uveal melanomas have at least one of the nine described patterns (even absence of vessels is a pattern, i.e. silent; Folberg R, 1993).

5.3: Were extracellular matrix patterns identified from non-counterstained sections under a dark green filter as originally described (Folberg R, 1993) or from counterstained sections in which they may be less obvious (McLean IW, 1997).

Conflict of interest: I was one author of the uveal melanoma chapter of the 7th edition of the TNM classicifation, American Joint Committee on Cancer.