Article Text

Discussion of paper by Foss et al
  1. University of Iowa, 100 Medical Research Center, Room 233, Iowa City, IA 52242–1182, USA

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Uveal melanoma is one of the few forms of cancer treated before a pathologist can examine tissue to assess the risk of metastasis by assigning a histological grade. Study of the microcirculation architecture in these tumours was initially aimed at evaluating the prognostic significance of microcirculatory patterns in tissue sections of eyes removed for choroidal or ciliary body melanoma and also to determine if these patterns could be imaged by ultrasonographic or angiographic techniques.

    Patterns formed by the microcirculation architecture in uveal melanomas were found to be strongly associated with death from metastatic melanoma by our research group1 2-7 and by others.8 9 Ophthalmic pathologists have shown interest in using a simple histological technique developed in our laboratory to detect these microcirculatory patterns and compare features of prognostic significance in uveal melanomas.10 By using a periodic acid Schiff (PAS) stain without haematoxylin counterstaining for nuclei, the amount of visual ‘noise’ is reduced, improving the sensitivity of detecting small microvessels. After a green filter is placed into the light path of the microscope, visualisation of the microcirculation is enhanced. The PAS stain resists bleaching techniques to remove melanin in highly pigmented tumours.1The PAS stain does not identify endothelium directly, but it does identify extracellular matrix components that are vessel associated.11

    In the paper above, Foss and associates raise some important issues. They contend that the prognostically significant patterns we have described are not based upon the microcirculation and therefore this invalidates attempts to image them angiographically.12

    Firstly, contrary to the belief of Foss et al, we reiterate that PAS staining is not specific for blood vessels1 2 but that the patterns we have found follow precisely those obtained by staining with fluorescein conjugatedUlex,1 4 by transmission electron microscopy,1 3 with three dimensional reconstructions using laser scanning confocal microscopy,4 7 by comparison with other specific markers for vascular endothelium such as CD31,7 13 and by tracing these patterns in serial sections to the vortex veins and choriocapillaris.7

    Foss and associates, in fact, concur with our view that the patterns we describe represent in part perivascular connective tissue. Indeed, oneexpects to find elements other than endothelium in a microcirculatory bed and we take this point further in that we have described how the contribution of extracellular matrix components, especially type VI collagen, adds to the formation of prognostically significant microvascular patterns in choroidal and ciliary body melanomas.11 Similar patterns have been observed in other tumours. For instance, photomicrographs of back to back vascular loops in two dimensional sections of tumours have been published (for example, in neural and neuroendocrine tumours14). Moreover, the presence of a ‘chicken wire’ microcirculation (that is, networks) is a hallmark of myxoid liposarcoma (in two dimensional histological sections) and is demonstrated routinely by the PAS stain.15

    With regard to the issue of autofluorescence as a confounding factor in assessing these patterns, Figure 3 in the article by Foss et al speaks for itself by beautifully illustrating the microvascular composition of these patterns (note the loops surrounding islands of tumour all connect to structures with well defined lumina).

    Caution is appropriate in using microvascular density in uveal melanoma as a dominant prognostic factor. Figure 1B in the article by Foss et al shows a section of a uveal melanoma stained by an antibody to factor VIII in which, at the top of the figure, midway between the edges, there are four nearly vertically oriented linear regions of staining arranged perfectly in a head to tail configuration. This pattern could be interpreted as four separate vessels or one vessel (by connecting the segments). The same broken line appearance is present throughout the illustration (see the lower right portion of the figure). Antibodies to factor VIII are known to be insensitive markers for immature endothelium of the type seen in tumour angiogenesis16 17 and we have demonstrated that this antibody decorates the microcirculation incompletely. The prognostic significance of microvascular counts (Foss et al, p 240)18 based on this technique is therefore subject to question. Indeed, with more sensitive markers for endothelium, microvascular loops and networks have been demonstrated in two dimensional tissue sections of choroidal and ciliary body melanomas,7 13 making vessel counts even more difficult to interpret.19

    The main issue in question is: does the study of the microcirculation architecture of choroidal and ciliary body melanomas provide clinically useful information? Two approaches have been taken to image melanoma microcirculatory patterns in patients. With laser scanning confocal ophthalmoscopy and indocyanine green angiography, parallel vessels with cross linking20 and microcirculatory loops21 have been seen in posterior choroidal melanomas. With ultrasound tissue characterisation, it has been possible to separate patients with ciliary body and choroidal melanomas into prognostic risk groups13 22 based upon a histological microcirculatory pattern classification system.3

    We feel, therefore, that the questions raised by Foss and associates have already been resolved, mostly in the published literature, and also by the correct interpretation of the authors’ own data. Recently, Mehaffey and associates19 discovered that by using the most powerful Cox proportional hazards model available to them to assess prognosis in uveal melanoma, only roughly one third of what could be known about factors contributing to metastasis was included in their model. Perhaps now is the time to direct our energies towards these other factors.


    Supported by a grant from the National Institutes of Health (EY10457) and, in part, by an unrestricted grant from Research to Prevent Blindness, Inc, New York, USA.