2. Phenotype of Interstitial cells of Cajal as cell counterparts to GIST cells
As in other groups of tumors, differentiation or phenotype is important in GISTs because it encapsulates criteria for practical histopathological diagnosis: it also provides an appreciation of what GIST tumor cells are biologically, and this can be important in understanding how they behave, especially in terms of how they might react to therapeutic agents.

One of the most important concepts of recent years has been that GISTs are showing the differentiation of the Interstitial cell of Cajal (ICC)(Kindblom et al 1998). ICCs form a network within the gut musculature with pacemaker activity, which controls the smooth-muscle cell contractions of peristalsis. In the light of this concept, it is appropriate to define the features of ICCs, because they form the basis for diagnosing GIST, and because there are misconceptions in the literature on the nature of ICCs, which in turn impact on our appreciation of the nature of GISTs.

Human ICCs often appear as solitary cells with very elongated profiles; they are found in specific locations in relation to smooth-muscle layers and more numerously around plexuses; they express the c-kit proto-oncogene as tyrosine kinase receptor, KIT (the latter being identifiable by CD117 immunostaining); and are vimentin-positive and desmin-negative (Faussone-Pellegrini et al 1989; Rumessen and Thuneberg 1991; Rumessen et al 1993a; Rumessen et al 1993b; Rumessen 1994; Faussone-Pellegrini and Thuneberg 1999). There appear to be CD34⁺ and CD34^- subpopulations (Vanderwinden et al 1999; Robinson et al 2000).

With regard to ultrastructure, in spite of subtle site-specific differences, ICCs have several smooth-muscle cell features – in particular, caveolae, lamina, and myofilaments – which form the basis of regarding ICCs as cells showing a primitive level of smooth-muscle differentiation, and in fact these cells have been referred to as 'myoid' (Rumessen and Thuneberg 1991). There are, however, additional features which are found in ICCs which are less typical of smooth-muscle cells, although found in them (Eyden et al 1992; Faussone-Pellegrini and Cortesini 1985): abundant intermediate filaments; dense bodies amid these intermediate filaments; and smooth endoplasmic reticulum. The dense bodies amid abundant intermediate filaments (Rumessen and Thuneberg 1991) have also been seen in an unambiguous leiomyosarcoma (Eyden and Banerjee 2002) and in a transplanted leiomyosarcoma (Sawada et al 1992), and may represent a marker of a primitive level of smooth-muscle differentiation.

In summary, ICC ultrastructure suggests a primitive myoid cell characterized by a low expression of typical smooth-muscle features, as well as certain other features consistent with but perhaps not absolutely specific for or characteristic of smooth-muscle.

ICCs are not neuronal. Ramon y Cajal may have regarded the cells named after him as neuronal but the detailed ultrastructural studies by the Faussone-Pellegrini and Rumessen/Thuneberg groups suggest a myoid cell, an unusual myoid cell maybe, but not a neuronal cell, and in particular a cell without neuroendocrine granules. This needs to be emphasised because it conflicts with certain published views on the ICC – for example, " … ultrastructural features described as being diagnostic of GANT [see below] are, in fact, characteristics of gastrointestinal pacemaker cells " (Kindblom et al 1998); and that "[Interstitial cells of Cajal] … have immunophenotypic and ultrastructural features of both smooth muscle and neuronal differentiation …" (Fletcher et al 2002).

3. Ultrastructure and immunophenotype of GIST
Is it possible to identify GI tumors which correlate phenotypically with ICCs and which can, therefore, be designated as GIST? On the basis of our own recent study of 82 mesenchymal GI tumors (Eyden et al 2002), GISTs can indeed be identified by ICC features, although the picture is complicated by phenotypic overlap with gastrointestinal autonomic nerve tumor (GANT) and, as already suggested, true smooth-muscle tumor. 36 tumors were identified as having varying expression of processes containing abundant intermediate filaments, so-called solitary focal densities amid these intermediate filaments, and smooth endoplasmic reticulum, as well as a varying but usually low expression of more typical smooth-muscle features such as myofilaments, attachment plaques, caveolae and lamina. Most were also KIT-positive, a lesser percentage were CD34 positive, but a significant proportion showed smooth-muscle expression as indicated by a-smooth-muscle actin, pan-actin (HHF35) and, less frequently, desmin.

4. Gastrointestinal autonomic nerve tumor (GANT) and its relationship to GIST
The foregoing ultrastructural definition of GIST also assumes the absence of neuroendocrine granules, cell structures which characterize GANTs. GANTs, however, also have partial phenotypic overlap with GIST: most are KIT positive and some are CD34 positive; there is some smooth-muscle differentiation as determined immunophenotypically and ultrastructurally; and they have processes containing intermediate filaments (less commonly microtubules), smooth endoplasmic reticulum, while some have solitary focal densities amid intermediate filaments.

Despite a distinctive ultrastructure, GANT has been either equated with GIST (Schmid and Wegmann 2000) or regarded as a variant of GIST (Lee et al 2001). This is largely because of the KIT-positivity of GANTs, although it has to be said that the GANTs from these studies have been diagnosed ultrastructurally on the basis of granules which have been pointed out as looking more like lysosomes than true neuroendocrine granules (Eyden 2001, 2002). That having been said, many of the ultrastructurally confirmed GANTs containing bona fide neuroendocrine granules in our own study (Eyden et al 2002) are indeed KIT-positive. So, we regard GIST and GANT as having partial immunophenotypic and ultrastructural overlap, but we also maintain that these are different from each other. It seems inconceivable that a cell with a primitive level of smooth-muscle differentiation should not be biologically different from a neuronal cell: differences in behaviour could be identifiable with appropriate further investigation.

5. Conclusions
The Interstitial cell of Cajal has the immunophenotype and ultrastructure of a relatively primitive smooth-muscle cell but with certain unusual additional features: however, these cells are not neuronal and they lack neuroendocrine granules. Mesenchymal tumors of the GI can be identified with the features of ICCs and can therefore be designated as GISTs. Neuroendocrine granules allow the distinction from GANT, which on this basis is a biologically different tumor. However, GANT and GIST exhibit partial immunophenotypic and ultrastructural co-expression, suggesting a complex genetic and post-genomic control over cellular features.

References

1. Kindblom LG, Remotti HE, Aldenborg F, Meis-Kindblom JM. Gastrointestinal pacemaker cell tumor (GIPACT). Gastrointestinal stromal tumors show phenotypic characteristics of the interstitial cells of Cajal. Am J Pathol 152: 1259-1269, 1998
2. Faussone-Pellegrini MS, Pantalone D, Cortesini C. An ultrastructural study of the interstitial cells of Cajal of the human stomach. J submicrosc Cytol Pathol 21: 439-460, 1989
3. Rumessen JJ, Thuneberg L. Interstitial cells of Cajal in human small intestine. Ultrastructural identification and organization between the main smooth muscle layers. Gastroenterology 100: 1417-1431, 1991
4. Rumessen JJ, Peters S, Thuneberg L. Light- and electron microscopical studies of interstitial cells of Cajal and muscle cells at the submucosal border of human colon. Lab Invest 68: 481-495, 1993a
5. Rumessen JJ, Mikkelsen HB, Qvortrup K, Thuneberg L. Ultrastructure of interstitial cells of Cajal in circular muscle of human small intestine. Gastroenterology 104: 343-350, 1993b
6. Rumessen JJ. Identification of interstitial cells of Cajal. Significance for studies of human small intestine and colon. Dan Med Bull 41: 275-293, 1994
7. Faussone-Pellegrini M-S and Thuneberg L. Guide to the identification of Interstitial cells of Cajal. Microsc Res Tech 47: 248-266, 1999
8. Vanderwinden J-M, Rumessen JJ, De Laet M-H, Vanderhaeghen J-J, Schiffmann SN. CD34⁺ cells in the human intestine are fibroblasts adjacent to, but distinct from, Interstitial cells of Cajal. Lab Invest 79: 59-65, 1999
9. Robinson TL, Sircar K, Hewlett BR, Chorneyko K, Riddell RH, Huizinga JD. Gastrointestinal stromal tumors may originate from a subset of CD34-positive Interstitial cells of Cajal. Am J Pathol 156: 1157-1162, 2000
10. Eyden BP, Hale RJ, Richmond I, Buckley CH. Cytoskeletal filaments in the smooth muscle cells of uterine leiomyomata and myometrium: an ultrastructural and immunohistochemical analysis. Virchows Archiv 420: 51-58, 1992
11. Faussone-Pellegrini MS and Cortesini C. The muscle coat of the lower esophageal sphincter in patients with achalasia and hypertensive sphincter. An electron microscopic study. J submicrosc Cytol Pathol 17: 673-685, 1985
12. Eyden B and Banerjee SS. Leiomyosarcoma showing partial synthetic-phase differentiation. Ultrastruct Pathol 26: 335-340, 2002
13. Sawada M, Terada N, Yamatoto R, Nishizawa Y, Wada A, Mori Y, Sakamoto H, Tanizawa O.
14. Establishment and characterization of human uterine leiomyosarcoma heterotransplanted into nude mice. Int J Cancer 52: 124-129, 1992
15. Fletcher CDM, Berman JJ, Corless C, Gorstein F, Lasota J, Miettinen M, O'Leary TJ, Remotti H, Rubin BP, Shmookler B, Sobin LH, Weiss SW. Diagnosis of gastrointestinal stromal tumors: a consensus approach. Int J Surg Pathol 10: 81-89, 2002
16. Eyden B, Chorneyko KA, Shanks JH, Menasce LP, Banerjee SS. Contribution of electron microscopy to understanding cellular differentiation in mesenchymal tumors of the gastrointestinal tract: a study of 82 tumors. Ultrastruct Pathol 26: 269-285, 2002.
17. Schmid S, Wegmann W. Gastrointestinal pacemaker cell tumor: clinicopathological, immunohistochemical, and ultrastructural study with special reference to c-kit receptor antibody. Virchows Arch 436: 234-242, 2000
18. Lee JR, Joshi V, Griffin JW, Lasota J, Miettinen M. Gastrointestinal autonomic nerve tumor. Immunohistochemical and molecular identify with gastrointestinal stromal tumor. Am J Surg Pathol 25: 979-987, 2001
19. Eyden B. Gastrointestinal pacemaker cell tumour. Virchows Arch 438: 632-633, 2001
20. Eyden B. Gastrointestinal autonomic nerve tumor. Am J Surg Pathol 26: 396-397, 2002

Acknowledgements.
I would like to thank my histopathologist colleagues, Dr Kathy Chorneyko (Hamilton, Canada), Dr Lia P Menasce, Dr Jonathan H Shanks and Dr S Sankar Banerjee, for providing the histopathological background to this presentation.