Myxoid Cellular Neurothekeoma

Abstract

To the Editor: We are slightly puzzled by the letter concerning a case of S-100 negative myxoid neurothekeoma (nerve sheath myxoma, NSM) (1). It appears that an analogous observation was reported previously (2), but the histologic features described in this paper were definitely reminiscent of a tumor presently known as cellular neurothekeoma (CNT) (3). Although these two entities share a few similarities that might suggest a histogenetic relationship, we believe that the bulk of recent evidence argues in favor of a fundamental difference. Nerve sheath myxoma has a peak incidence in the 4th decade and manifests as a dermal tumor with a multilobular or plexiform growth pattern delineated by dense collagenous tracks (4,5). It is composed of slender spindle-shaped and stellate cells loosely interconnecting within an abundant myxoid stroma that contains high amounts of sulphated glucosaminoglycans. Ultrastructurally, the tumor cells show distinct signs of nerve sheath or schwannian differentiation, which is confirmed by consistent expression of S-100 protein and nerve growth factor receptor (NGFR) (6–8) in the absence of positive staining for EMA and markers histiocytic differentiation. Cellular neurothekeoma, a tumor occurring mainly in children and young adults with a slight female predominance, displays a similar growth pattern at low magnification (3). However, it is characterized by higher cell density, larger and occasionally epithelioid cells that contain plump vesicular nuclei, and may occasionally exhibit some degree of atypia (9), and absent or sparse mucinous matrix. By immunohistochemistry, these tumors typically express neither S-100 protein nor NGFR (8,10) while showing consistent reactivity with NK1C3 (CD57) and the pan-monocyte marker Ki-M1p (8,11). Our observations entirely confirm these immunohistochemical data. In a series of 15 cases including nine classic NSM and six typical CNT, we found that the expression of S-100/NGFR and NK1C3/Ki-M1p were mutually exclusive. Other antigens have been described in either entity or both, but they appear to be of little diagnostic utility because of their variable expression. We did not assess PGP9.5, an allegedly neural or neuroendocrine antigen claimed to be a distinctive marker for CNT (12), because we observed that this glycoprotein is characterized by a very low specificity even without antigen retrieval (unpublished, August 1999). To augment the diagnostic confusion, mixed or intermediate forms of CNT exist, which contain focal myxoid areas reminiscent of NSM (13). However, in four such cases that could be retrieved from our archive, we found that the immunophenotype of the myxoid parts perfectly matched that of the cellular parts, being positive for NKIC3 and Ki-M1p and negative for S-100 and NGFR. More remarkably, we were recently confronted with a case that appeared to be a typical NSM by conventional microscopy (Fig. 1). To our great surprise, this tumor expressed neither S-100 nor NGFR, but reacted strongly with NKIC3 and Ki-M1p. Consequently, this case was signed out as “myxoid cellular neurothekeoma.”FIG. 1.: Myxoid cellular neurothekeoma: a lobulated dermal tumor composed of slender fusiform cells within an abundant myxoid matrix. The tumor expressed neither S-100 nor NGFR while being strongly positive for Ki-M1p and NKIC3The persistence of the “cellular” immunophenotype in certain neurothekeomas with myxoid features outlines a broad morphologic spectrum in this entity. Thus, since NKIC3 and Ki-M1p apparently were not assessed by the authors (1,2), the two cases described as S-100 negative myxoid neurothekeoma may actually represent examples of myxoid CNT. Conversely, we never observed any cellular variants of NSM as yet. Nevertheless, reports of an S-100 expression in rare cases of CNT (8,12) suggest that these tumors might in fact represent NSM with a higher cell density and sparse myxoid matrix. As illustrated e.g., by the variety of small round blue cell tumors, morphology alone does not necessarily allow a definite classification. We thus find it difficult to believe that NSM and CNT represent extremes of a common spectrum, since the divergent immunoprofiles in NSM and CNT suggest at least a disparate differentiation. Indeed, S-100 and NGFR are established markers of neural/schwannian differentiation, and Ki-M1p is a reliable indicator of monocytic lineage (14). Thus, regardless of the histologic presentation, NMS is likely to be schwannian, and CNT fibrohistiocytic in nature. We would not venture as far as Zelger et al., who proposed that CNT be a variant of dermatofibroma (11), though we can conceive a relationship with plexiform histiocytic tumor. In any case, we consider that the diagnostic differential between myxoid and cellular neurothekeoma should rely on thorough immunophenotyping rather than on morphologic aspects. Pierre Rudolph, M.D., Ph.D. Christoph Schubert, M.D., Ph.D.