Telocytes are the physiological counterpart of inflammatory fibroid polyps and PDGFRA‐mutant GISTs

Abstract PDGFRA mutations in the gastrointestinal (GI) tract can cause GI stromal tumour (GIST) and inflammatory fibroid polyp (IFP). Hitherto no cell type has been identified as a physiological counterpart of the latter, while interstitial Cajal cells (ICC) are considered the precursor of the former. However, ICC hyperplasia (ICCH), which strongly supports the ICC role in GIST pathogenesis, has been identified in germline KIT‐mutant settings but not in PDGFRA‐mutant ones, challenging the precursor role of ICC for PDGFRA‐driven GISTs. Telocytes are a recently described interstitial cell type, CD34+/PDGFRA+. Formerly considered fibroblasts, they are found in many organs, including the GI tract where they are thought to be involved in neurotransmission. Alongside IFPs and gastric GISTs, GI wall “fibrosis” has been reported in germline PDGFRA‐mutants. Taking the opportunity offered by its presence in a germline PDGFRA‐mutant individual, we demonstrate that this lesion is sustained by hyperplastic telocytes, constituting the PDGFRA‐mutant counterpart of germline KIT mutation‐associated ICCH. Moreover, our findings support a pathogenetic relationship between telocyte hyperplasia and both IFPs and PDGFRA‐mutant GISTs. We propose the term “telocytoma” for defining IFP, as it conveys both the pathogenetic (neoplastic) and histotypic (“telocytary”) essence of this tumour, unlike IFP, which rather evokes an inflammatory‐hyperplastic lesion.


| INTRODUCTION
Inflammatory fibroid polyps (IFPs) are benign, polypoid lesions usually centred in the submucosa of the gastrointestinal (GI) tract, with a predilection for gastric antrum. They consist of a proliferation of spindle, stellate or epithelioid mesenchymal cells harboured in a vascularized stroma, infiltrated by inflammatory cells (often with numerous eosinophils). Immunohistochemically, the lesional mesenchymal cells express CD34 and platelet-derived growth factor receptor alpha (PDGFRA). Most IFPs bear PDGFRA mutations. 1,2 A subset of GI stromal tumours (GISTs), accounting for IFPs. 3 This GIST subgroup, which manifests a strong predilection for the stomach, is frequently at least partly epithelioid and, unlike other GISTs, is often negative or only faintly/patchy positive for CD117.
No physiological counterpart has hitherto been described for IFPs, although it has been suggested that they may develop from PDGFRA-positive mesenchymal cells found along the villus membrane in the small intestine. 2,4 Conversely, GISTs as a whole are commonly considered to be related to interstitial cells of Cajal (ICCs). 5 Beyond the immunophenotype shared between most GISTs and ICCs (CD117+, DOG1+), the presence of ICC hyperplasia (ICCH) both in human germline KIT-mutant kindreds and in mice engineered to express KIT mutations found in human GISTs strongly supports a relationship between ICCs and GISTs. 6,7 However, to date, no ICCH has been detected either in human germline PDGFRA-mutants or in mice generated for the conditional expression of mutant PDGFRA, arguing against a relationship between ICCs and PDGFRA-driven GISTs. Conversely, GI stromal cell hyperplasia has been described in these conditions, especially at submucosal level. [7][8][9][10] Telocytes (TCs) are a type of interstitial cell, formerly named interstitial Cajal-like cells, described in the connective tissue of several organs, including the GI tract. [11][12][13][14][15][16][17][18][19][20][21][22][23] For a long time, they went unrecognized and were labelled simplistically as fibroblasts. In the GI tract, TCs are CD34+, PDGFRA+ and CD117-, and are thought to play a role in neurotransmission, possibly by spreading the low waves generated by ICCs. 24 A relationship between TCs and GISTs (without genotypic distinction) or IFPs has been sometimes suggested, based on shared immunohistochemical markers and/or, in the former instance, to explain the occurrence of extragastrointestinal cases. 8,25 We previously published a kindred bearing a germline mutation (CC?TT at 1957-58) that led to a Leu for Pro substitution at 653 (P653L) in PDGFRA gene. 9 The proband featured GI tumours encompassing GISTs, IFPs (including tumours previously described as "fibrous tumours" 26 ) and a lipoma, and an intramural diffuse stromal cell proliferation, heavily affecting the submucosa, in the stomach and proximal duodenum. We carried out a morphological and immunophenotypical investigation of the histotype of these stromal cells in order to establish their possible morphogenetic and/or pathogenetic relationship with synchronous IFPs and GISTs.

| Tissue specimens
The gastric tissues investigated were from the only member of a previously published germline PDGFRA-mutant kindred who featured a remarkably marked stromal proliferation of the gastroduodenal wall together with IFPs and GISTs. 9 The procedures followed were in accordance with the ethical standards of the local institutional committee on human experimentation and with the Helsinki declaration of 1975, as revised in 1983. Informed written consent for this study was obtained from the patient.

| Histological characterization of GI wall diffuse stromal proliferation
A proliferation of stromal tissue involved the stomach wall (Figure 1A) and proximal duodenum. It started as two bands lining the inner and outer aspects of the submucosa. These showed a tendency to merge and ultimately completely obliterate this structure ( Figure 1C). Higher magnification revealed spindle-to-polygonal shaped cells set in a variably collagenized background, often displaying thin cytoplasmic processes, at times extremely long, at times apparently interconnecting ( Figure 1D). When involving muscularis propria, they dissociated smooth muscle fascicles according to an interstitial pattern of growth ( Figure 1F). These cells were diffusely and intensely positive for CD34 ( Figure 1B and I) and PDGFRA (Figure 1E and G) but turned out to be CD 117 negative ( Figure 1H).
Besides the interstitial PDGFRA-positive cells, CD34 IHC stained the endothelial lining of vessels, including the numerous capillaries present ( Figure 1I). Conversely, CD117 revealed another population of cells with elongated cytoplasmic processes, not highlighted by PDGFRA IHC, either at the interface between fibrous tissue and smooth muscle fascicles or inside the latter, in close contact with smooth muscle cells; the only cell type stained in the fibrous intermuscular interstitium by CD117 IHC were mast cells ( Figure 1H).

| Relationship between GI wall diffuse stromal proliferation and IFP
Several typical IFPs were detected, histologically superimposable to their sporadic counterpart, being composed of an admixture of fibroblast-like CD34+ mesenchymal cells and inflammatory cells, often with numerous eosinophils, in a myxoid collagenous background. Interestingly, IFPs stemmed from the aforementioned diffuse stromal proliferation (Figure 2A

| Relationship between GI wall diffuse stromal proliferation and GIST
Two gastric GISTs were present, whose histological features were indistinguishable from those of their sporadic PDGFRA-mutant counterpart, featuring epithelioid cytology and PDGFRA and CD117 (patchy-weak) positivity, as shown in Figure 3. Notably, near the gastric muscularis propria permeated by the previously described fibrous proliferation ( Figure 3B) the GIST was characterized by a cytology that was relatively spindled ( Figure 3E A CD34+ stromal proliferation altered the structure of gastric wall, forming two thick bands along the inner and outer aspects of submucosa, sometimes forming ill-defined nodules, and infiltrating the deeper wall layers with an interstitial pattern (A, H&E; B, CD34). At times, the stromal proliferation obliterated submucosa (C). At higher magnification, the stromal proliferation consisted of spindle-to-polygonal PDGFRA+ TCs set in a variably collagenized background, often displaying slender, sometimes extremely long, cytoplasmic processes (telopodes), at times interconnecting with each other (arrows highlight some TCs whose telopodes are clearly evident due to the favourable intersection by the histological cutting surface) (D, H&E; E, PDGFRA). The stromal proliferation with TCs (arrows, as in D and E) infiltrated muscularis propria with an interstitial pattern, separating smooth muscle fascicles, where no PDGFRA+ TCs are observed (F, H&E; G, PDGFRA). CD117 demonstrated ICC-IM, characteristically in close relation with smooth muscle fibres (arrowheads); mast cells (circled) were the only CD117+ cells populating the interstitial fibrosis (H). CD34 stained not only interstitial TCs (arrows), but also the endothelial lining of blood vessels: A small venule is evident at the centre of the picture; arrowheads indicate capillaries, identifiable because of their lumen (I).

| DISCUSSION
We herein report a detailed morphological and immunophenotypical analysis of the diffuse stromal proliferation described in the GI wall in PDGFRA-mutant syndrome 7 and of its relationship with IFP and GIST, by exploiting the exceptional opportunity offered by its remarkable manifestation, together with the presence of these tumours, in an individual from a kindred previously published by the authors. 9 As expected, germline PDGFRA mutations are able to originate multiple IFPs and GISTs. 7 However, a third type of GI lesion, consisting of an intramural diffuse stromal proliferation variably referred to as "submucosal fibrous thickening," "intestinal fibrosis," "increased submucosal connective tissue," "thickened submucosa" or "expanded submucosa", has often been described in this genetic setting, both in humans and in mice. [8][9][10] It consists of a CD34+ stromal proliferation, that variously involves the entire thickness of the GI tract, often heavily affecting the submucosa ( Figure 1A-C). In a previous paper describing PDGFRA-mutant syndrome, we found this stromal proliferation remarkably evident in the stomach and duodenum of an individual, demonstrating that, in the submucosa, it often consisted of two thick bands running along its inner and outer limits; furthermore, its constitutive cells were positive not only for CD34, but also for PDGFRA, whereas both CD117 and DOG1 were negative. 9 Here, besides confirming these characteristics, we show that these stromal cells often display long, slender cytoplasmic processes that sometimes appear to connect with each other (Figure 1D-G) 27 and, before completely obliterating it ( Figure 1C), they form two thick bands along its inner and outer aspects ( Figure 1A-B), that is exactly where the TCs physiologically form a monolayer 24 ; (iv) when in the muscularis propria, they are preferentially located within the interstitial "fibrosis," whereas the The panoramic view shows a GIST originating from muscularis propria, whose smooth muscle bundles appear variously dissociated by the presence of interstitial TC hyperplasia, bulging on the external side of the gastric wall (A). As in Figure 2, higher magnification panels refer to areas encircled in grey in panel A, so as to underline the topographic location of the illustrated microscopic fields to better appreciate the transition between TC hyperplasia and GIST; corresponding areas are connected by arrows. Thin fibrous bands, populated by PDGFRA+ TCs, separate smooth muscle bundles near the periphery of GIST (B, H&E; C, PDGFRA; D, CD117, highlighting ICC-IM and mast cells). GIST (right) at the interface with muscularis propria (left) showed a relatively spindled cell cytology (E: H&E), and displayed an intense PDGFRA positivity (F), but only a weak and focal CD117 staining (G), thereby immunophenotypically resembling the neighbouring hyperplastic TCs rather than ICC-IM. At a relative distance from the muscularis propria the GIST assumed a wholly epithelioid cytology (H, H&E), maintaining the intense PDGFRA immunoreactivity (I) and sometimes acquiring a comparatively strong positivity for CD117 (J). (Scale bars: A: 2 mm; B-J: 65 lm) typical of germline KIT mutations. Of note, this finding confutes the histotype concealed by two of the terms previously used for defining the referred stromal proliferation, that is "submucosal fibrous thickening" and "intestinal fibrosis": In fact, TCs are decreased in conditions determining a genuine fibrosis of GI wall such as Crohn's disease and ulcerative colitis. 28,31 We went on to ascertain whether it was possible to detect morphological and/or immunophenotypical clues potentially linking TC hyperplasia with IFP or GIST in PDGFRA-mutant syndrome.
With regard to IFPs, we have already described examples stemming from the "submucosal thickening" found in an individual affected by PDGFRA-mutant syndrome. 9 Herein, we show also that there appears to be a gradual cytologic progression from TCs constituting the aforementioned thickening to stromal cells forming IFP, with a gradual loss of both slender cytoplasmic processes and relative cell elongation, accompanied by the acquisition of a relatively plumper morphology in the presence of a persisting, although overall less intense, positivity for PDGFRA ( Figure 2C-G). We therefore con-  (Figures 1 and 3). Under these circumstances, it is remarkable that an analysis of the transition between the gastric muscular wall and GIST in this latter condition demonstrates Differences in TC features, possibly concealing different TC subpopulations and varying with both GI segment and GI wall layer, 24 together with possible site-specific differences in gene expression, 32 could explain how TCs as a whole can originate diverse tumour types, and why, unlike IFPs, PDGFRA-mutant GISTs show a strong (albeit not entirely exclusive) gastric predilection. 33,34 Our results are based on the study of the only individual featuring a prominent diffuse stromal proliferation of GI wall from a PDGFRAmutant kindred we previously investigated. 9 In absolute terms, this constitutes a limitation to the significance of our conclusions. However, it must be considered that germline PDGFRA mutations are an exceedingly rare event in humans, with only five to six examples (considering kindreds or single individuals) reported in the literature 7,8 ; on top of that, the presence of GI diffuse stromal proliferation together with IFPs and/or GISTs, that is the ideal setting for investigating the relationship between these latter tumours and TCs, has been reported in a fraction only of mutants (in particular, with regard to the kindred we previously published, only in the proband, that is member II-2-). 8,9 The study of sporadic IFPs and PDGFRA-mutant GISTs is probably of limited utility for defining a possible link between these tumours and TCs as, under these circumstances, only the fully developed neoplastic morphological and immunophenotypical characters of the former tumours are observable, features which could be deemed not to be specific enough in terms of cell lineage in the absence of any type of transition with respect to physiological TCs In conclusion, our results support TCs as the physiological counterpart of both IFPs and PDGFRA-mutant GISTs, possibly pathogenetically related to both of these tumour types. Furthermore, on the basis of our results we herein propose "telocytoma" as a more