Logistic regression analysis of small round cell neoplasms: A cytologic study

The diagnosis of small round cell neoplasms (SRCN) in children is difficult by both histologic and cytologic methods. These neoplasms are unified morp hologically by the scanty cytoplasm surrounding relatively round nuclei con taining a primitive chromatin pattern. Further categorization is achieved h istologically by the recognition of architectural differentiation and cytop lasmic features. Numerous series and case reports documenting the cytologic features of SRCN have been published in the English-language literature, b ur relatively few studies have statistically analyzed the diagnostic utilit y of these features. Our logistic regression analysis of 59 cases of SRCN i ndicates that the presence of an extremely scanty cytoplasm Savors the diag nosis of Ewing's sarcoma (P = 0.004), as does the positive expression of cy toplasmic vacuoles (P = 0.02). Strap or tadpole cells closely correlate wit h the diagnosis of rhabdomyosarcoma (P < 0.001). The positive expression of rosettes strongly supports the presence of a neuroblastoma (P < 0.001). Th e diagnosis of Wilms' tumor is confirmed by the expression of tubules (P < 0.001). The presence of lymphoglandular bodies strongly favors the diagnosi s of lymphoma (P < 0.001). While some cytologic features ave highly correla ted with specific SRCN, these features are not invariably present, and defi nitive diagnosis may require immunohistochemical or ultrastructural analysi s performed on cell block material. Diagn. Cytopathol. 1999;20:271-277. (C) 1999 Wiley-Liss, Inc.