The "neurothekeoma": immunohistochemical analysis distinguishes the true nerve sheath myxoma from its mimics.

In contrast with the myxoid variant of neurothekeoma (nerve sheath myxoma), evidence of neurosustentacular (NS) differentiation in the so-called cellular and mixed (intermediate) variants of neurothekeoma remains controversial. In this study, we selected 22 tumors coded as neurothekeoma or nerve sheath myxoma from the Soft Tissue Registry of the AFIP. Each tumor was histologically subtyped as either a myxoid/hypocellular neurothekeoma (MN) (N = 11) or as a "cellular" or "mixed" (intermediate) neurothekeoma variant (C&MV) (n = 11) and analyzed immunohistochemically. The MNs were composed of small, cytologically bland cells arranged in a loose cellular network or in files within highly myxomatous nodules delineated by dense collagen. The tumors showed clear-cut evidence of NS differentiation by exhibiting consistent immunoreactivity for S-100 protein (11 of 11 cases) and low-affinity nerve growth factor receptor, p75(NGFR), (NGFR) (10 of 10), and variable reactivity for glial fibrillary acidic protein (GFAP) (10 of 11) and CD57 (Leu-7) (5 of 9). They also showed pericellular collagen type IV (CIV) expression (9 of 9), scattered intralesional CD34-positive spindled cells (10 of 10), epithelial membrane antigen (EMA)-positive spindled cells located within the adjacent dense collagen (8 of 11), and immunoreactivity for alpha-smooth muscle actin (SMA) (3 of 10) and calponin (4 of 9). In 4 cases, scattered intralesional neuraxons were detected by the Bodian histochemical method or immunohistochemically with anti-neurofilament protein. The tumors had a male-to-female ratio of 6:5, a peak incidence in the 4th decade of life, and an anatomic distribution that included the upper and lower limbs and back. The C&MVs included 9 "mixed" and 2 "cellular" variants. C&MVs differed histologically from MNs by their higher cellularity and presence of larger spindled or epithelioid cells with vesicular nuclei. Immunohistochemically, the tumor cells expressed CIV (9 of 10), calponin (7 of 9), SMA (5 of 10), Leu-7 (1 of 7), S-100 protein (1 of 11), but not NGFR, GFAP, or CD34. EMA-positive spindled cells surrounded tumor fascicles in 1 case. Intralesional neuraxons were not identified. Clinically, these tumors differed from the MNs by exhibiting a male-to-female ratio of 4:7, a peak incidence in the 2nd decade, and an upper body distribution. Our results indicate that the MN shows NS differentiation and is the bona fide nerve sheath tumor, whereas the C&MVs fail to show convincing evidence of NS differentiation and probably warrant a separate classification.