To the Editors: Several variants of atypical fibroxanthoma (AFX) are split based on cytomorphology, for example, clear cell, granular cell, and spindle cell nonpleomorphic types.1 Nine cases of an unusual variant of AFX with keloidal tumoral sclerosis were presented by Kim and McNiff,2 following in from an anecdotal report3 and series of “keloidal” AFX. In several cases, “the keloidal collagen also formed ring-shaped structures surrounding CD31-positive vascular structures.” This report conveys our research on the perivascular sclerosis in a subset of 10 AFX examples with collagenous arcs and rings (Table 1).TABLE 1: 10 Prospectively Selected Cases of Sclerosing AFXThis curious arc and ring pattern of hyalinized collagen in some AFXs is morphologically similar to the always superficially located perivascular linear and tubular sclerosis in a minority of dermatofibromas. It is cautioned that the keloidal collagen associated with atypical fibrohistiocytic cells in AFX may erroneously lead to the diagnosis of “keloidal dermatofibroma.4” Interesting in this regard, it has been said that the spectrum of fibrous histiocytoma (a diagnostic term utilized as a synonym for dermatofibroma) can be viewed somewhat more broadly, with dermatofibroma (DF) at one pole and with its “malignant analogue” AFX at the other.5 In both entities, AFX and DF, a sparse central component of CD31-positive cells in the sclerotic arcs and rings deemed endothelial cells led to our initial hypothesis that the “keloidal collagen” is actually vascular basement membrane material of effete capillaries. Laminins, nidogens, and perlecans are arranged into highly organized supramolecular architectures by vascular basement membranes.6 We proposed the adherence of these elements likely explained the homogeneous (hyalinized) appearance of this material by obscuring its fibrillar texture (Figs. 1, 2). However, reactions for collagen type IV (specific for vascular basement membrane material)7 were negative for the morphologically similar curvilinear and circular material in both our AFX and DF examples. Also unreactive were the coarse hyalinized collagen bands of a common keloid, which like dermal collagen, are predominantly collagen type I.8FIGURE 1: Atypical fibroxanthoma. A, Hyalin rings surround small vessels in lower right of this AFX overlain by impetiginized parakeratosis (×400). B, Pleomorphic histology includes the usual tumor giant cells (×200). C, Dual immunohistochemical reaction: central endothelial cells label red for CD31 and rest upon an actin-positive (brown) pericyte layer (×700).FIGURE 2: Dermatofibroma. A, Sclerotic arcs and rings surround central cells in effete vessels; hemosiderin is the brown pigment (×700). B, A dual immunohistochemical reaction labels central endothelial cells red for CD31 although actin-positive pericytes are decorated brown as they layer against the external sclerotic ring (×700). C, A collagen IV reaction labels a thin endothelial basement membrane brown that lies between the endothelial cells and pericytes; the thick sclerotic collagenous rind around the vessel is nonreactive, that is, not thickened endothelial basement membrane (×1000).Pericytes are actin-positive cells situated in the basement membranes outside the endothelium of capillaries and venules. During the onset of angiogenesis the intramural pericytes leave the microvascular wall and enter the perivascular space, where they become collagen producing stromal cells. Sundberg et al9 identified 4 stages of pericyte migration/progression from the endothelium to a collagen-synthesizing fibroblast in the interstitium. Pericytes have been identified in the wound-healing process via their expression of platelet-derived growth factor, a moiety which enhances neoformation of collagen and is associated with a collagen-rich dense tumor stroma, and have been shown to differentiate into collagen-producing cells during liver fibrosis.10 Double immunohistochemical reactions in our AFXs and DFs alike display concentric pattern: luminal CD31-positive endothelial cells within a ring of actin-positive pericytes that face the external rim of sclerotic material (Figs. 1C, 2B). Lesional cells of both DF and AFX show no consistent propensity to border the sclerotic material, discounting any role they might have in producing it. This leaves the pericyte stimulated to differentiate into fibroblasts when in a proliferative “fibrohistiocytic” environment, as the sole architect because it is consistently found within the rings as expected. Sclerotic hyalin rings similar to the material under discussion have been illustrated in hemangiopericytomas.10 The consistent localization of the pericyte to within the hyalin rings of the 10 sclerosing AFXs in this series supports the hypothesis that this material is synthesized by pericytes, a proposition that should be tested by more sophisticated studies. Larisa M. Lehmer, MA Bruce D. Ragsdale, MD Western Dermatopathology with Central Coast Pathology and Santa Barbara Pathology Laboratory, San Luis Obispo, CA 93401