Statistics from Altmetric.com
Editor,—We read with great interest the article by Foss and associates published in the March issue of theBJO.1
In discussing different techniques for the detection of vascular structures in uveal melanomas, it is obviously necessary to avoid a misinterpretation of the pertinent criteria:
(1) Foss and associates interpreted the areas 1 and 2 shown in Figure 1A of their article as “area of silence”, and “area of normal vasculature”. We would not have classified the region marked as 1 in Figure 1A as a “silent area”, because many small vessels with a “white” lumen and a very thin PAS positive basement membrane can be identified, even at the magnification and the weak PAS stain given in their article. In addition, the two enlarged vessels marked as 2 would not have been classified as “normal” according to our original description.2-4 Therefore, we were not surprised that the authors could demonstrate a positive staining for vascular endothelium using factor VIII in Figure 1B. In our view, Foss and associates just misinterpreted findings with their PAS technique, which led to a misleading conclusion.
(2) Furthermore, in our experience we find it relatively easy to discriminate between basement membrane material and connective tissue as shown in Figure 2. In cases of doubt, we always stained serial sections for connective tissue using the Gomori’s trichrome stain. Blood vessels within a tumour contain endothelial cells, a basement membrane, and a lumen sometimes with detectable erythrocytes. This lumen can be identified in most of the tumour blood vessels using the PAS staining technique. In cases of doubt, the PAS stain could be compared with the adjacent haematoxylin and eosin stain of serial sections in order to identify a lumen and/or erythrocytes within a vessel lumen. The ultrastructure of the tumour blood vessels in melanocytic naevi and malignant melanomas of the uvea has been demonstrated in detail, previously.4 The blood vessels in malignant melanomas usually had a thick, multilaminated, and fragmented basement membrane with excessive amounts of connective tissue. These findings were the morphological basis for the detection of tumour blood vessels using the modified PAS stain without haematoxylin counterstaining.2-4
(3) Finally, Foss and associates assume in their article that the detection of vascular networks in malignant melanomas using the fluorescein conjugated lectin, Ulex europaeus 1, was mainly due to autofluorescence. As we used appropriate control sections, dyes, and filter blocks we can rule out the possibility of autofluorescence in our studies.2 5 For example, the highly vascularised choriocapillaris in tissue sections of surgically enucleated eyes served as an internal positive or negative control, and did not show any autofluorescence, but did stain positive with F-UEAI agglutinin. Moreover, Foss and associates also claimed that our fluorescent patterns “could partially resist bleaching”. We only examined patterns demonstrated by fluorescein conjugated markers with laser scanning confocal microscopy, and we never bleached tissue thus examined because we demonstrated that the laser scanning confocal microscope can image through melanin.2 5
We feel that the routine detection of tumour blood vessels in malignant uveal melanomas can be performed easily using the simple PAS stain without haematoxylin counterstaining.2-7 After learning how to interpret the results,3 4 this staining technique is as sensitive as other cost intensive and time consuming immunohistochemical staining techniques of blood vessels in uveal melanocytic lesions. In our view, the detection of the microcirculation architecture is the most important prognostic factor for the survival of patients with uveal melanoma.
Editor,—We are grateful for the opportunity to respond to the letter from Rummelt and Naumann. It must be unusual for differences in tinctorial staining to be discussed in the pages of an ophthalmology journal! They reiterate a number of points made by Folberg to which we would respond as follows.
We do not believe that a PAS stain can be considered as specific or as sensitive a method for the detection of blood vessels as immunohistochemistry for endothelial cells. However, we would emphasise the fact that we did find the patterns to be statistically significant, as have other groups. We stand by our histological interpretation of the “silent” areas and would point out that transverse sections of blood vessels can be identified within the silent areas shown in Folberg’s original paper, Figure 2. Of the defined Folberg patterns, we feel that this term best describes the area shown in our Figure 1.
We would argue that blood vessels with collagen around them are fibrovascular structures and that new canalised or non-canalised capillary buds are unlikely to show such structure. The Folberg patterns contain mature vessels but they are not the sole source of perfusion for which the tumour depends on for oxygenation—the “silent” areas in some tumours in the PAS stained sections are too large to support an actively growing tumour and we would expect much more necrosis in such tumours if no blood vessels were present. The F8 stain shows that there are capillaries.
The autofluorescence we found was primarily associated with collagen, of which there is little around vessels in the choriocapillaris. Autofluorescence is a well known phenomenon which can easily be used to show fibrovascular trabeculae in unstained melanoma and other tissues. We note that the legend of Figure 3 of Folberget al’s original description describes autofluorescence in their preparations. It is sometimes difficult to convince editors, but unstained controls are essential in fluorescence studies and should be illustrated. We note that the controls referred to by Rummelt and Naumann in their letter applied to their later work which used cyanine and not fluorescein which was used in the original paper.
We agree with Rummelt and Naumann that the detection of the microcirculation is of prime importance in determining prognosis in patients with choroidal melanoma. There is now a considerable body of published research indicating the importance of angiogenesis in uveal melanoma. In contrast with other tumours in which angiogenesis is thought to be important, there is no effective therapy for metastatic disease and death from uveal melanoma therefore reflects metastatic spread. There can be little doubt that the prognostic significance of tumour vascularity is due either to the ability of highly vascular tumours to metastasise more readily or the ability of metastatic cells, from such tumours, to be able to grow at sites of implantation.
It is quite clear that both the PAS patterns and the microvessel count have prognostic significance in uveal melanoma. It is not clear what the PAS patterns represent (we suggest hypotheses in our paper) but as the patterns do include mature blood vessels, it can be argued that they, in some way, represent the ability of the tumour to stimulate the formation of a mature blood supply. Microvessel density, by contrast, reflects the proliferation of capillary endothelium within the tumours and not all of the F8 stained structures may not be canalised. These facets of the biology of uveal melanoma are clearly related and important to tumour growth. However, at present we fail to see the point of routinely using either the Folberg or Foss methods for assessing angiogenesis in melanoma: there are no treatment decisions to be made! Nevertheless, we accept that an anti-angiogenesis strategy may be feasible and we hope that a measure of the angiogenic capacity of uveal melanoma will then prove useful in selecting patients for appropriate adjuvant therapy. Perhaps this is where we should now concentrate our efforts.
Editor,—The issues that separate us from Foss and Cree extend considerably beyond the subject of tinctorial staining of histological tissue sections. The overall goal of our research has been to design a non-invasive technique that would permit ophthalmologists to grade the biological behaviour of patients with ciliary body and choroidal melanomas at the time of diagnosis. As we have noted previously, uveal melanomas are among the few forms of cancer that are treated before a pathologist can examine tissue and assign a biological grade. Our research strategy was designed to identify a characteristic from tissue sections of eyes removed for ciliary body or choroidal melanomas that would be a reliable and strong marker for metastasis and that would be detectable by a non-invasive imaging technique. To date, we have focused our attention on the architecture of the microcirculation of uveal melanomas as such a marker.
Foss and coworkers have repeatedly challenged our assertion that patterns detected by our modification of the PAS stain actually detect blood vessels (an observation which, if correct, would invalidate the use of the microcirculation as a clinically relevant marker of malignant potential). At first, they declared that our microcirculation patterns were an artefact of autofluorescence. They repeat this assertion in their latest letter, but concede that their objection is restricted to a caption from one photomicrograph (they do not contest other work that supports our claim). Microvessels consist of more than tubules of endothelium. Even microvessels are associated with a extracellular matrix sheath. The modified PAS stain detects extracellular matrix elements associated with the microcirculation. Perhaps the most conclusive evidence to support our claim that the modified PAS stain accurately reflects the microcirculation in choroidal and ciliary body melanomas comes from a new report that shows an excellent correlation between angiograms of patients which choroidal melanomas taken with indocyanine green angiography with laser scanning confocal microscopy just before enucleation: careful mapping of histological sections from these tumours showed a correspondence between microcirculation patterns detected clinically and microcirculation patterns demonstrated with our modified PAS stain.
Angiogenesis is more than the production of new vessels: the microcirculation remodels and microcirculatory patterns in uveal melanoma therefore represent not only the new vessels but also the remodelling of this new vascular supply. Study of microcirculation architecture is therefore different from the study of vascular density. Every independent investigator who has studied microcirculation architecture in uveal melanomas has confirmed the association between the presence of microvascular loops or networks with metastasis (also M Jager, MD, personal communication, May 1997). However, at least two groups have now failed to confirm Foss’ study linking microvascular density with metastasis (also M Jager, MD, personal communication, May 1997).
Finally, we are dismayed by Foss and Cree’s failure to see the point of routinely assessing angiogenesis in uveal melanomas because “there are no treatment decisions to be made”. This would apply to all basic research concerning diseases that cannot be treated effectively today. Progress is being made in the laboratory, albeit slowly, in designing effective strategies to treat metastatic uveal melanoma,even though no effective treatment is available presently. Until such treatment does become available, it is reasonable to gather as much information as possible from prognostic features such as the tumour’s microcirculation architecture.