Classification of peripheral primitive neuroectodermal tumors (pPNETs)
is fraught with controversy. Some authors report a more aggressive cl
inical outcome in patients with peripheral neuroepithelioma (PN) when
compared to Ewing's sarcoma (ES). However, tumor classification varies
between different studies, making interpretation of these data diffic
ult. It is unknown whether pPNETs with immunohistochemical evidence of
neural differentiation, but without evidence of neural differentiatio
n by light microscopy, are more clinically aggressive than those witho
ut immunohistochemical evidence of neural differentiation. We immunohi
stochemically evaluated 38 cases of pPNET (35 osseous and three extrao
sseous) without histologic evidence of neural differentiation, i.e., r
osettes, using antibodies to four ''neural'' markers, including S-100
protein, Leu-7, synaptophysin, and neuron-specific enolase. All cases
were stained for CD99 as well as for cytokeratin (AE1/AE3), as some au
thors have purported a more aggressive clinical course for cytokeratin
-positive pPNETs. Thirty-three of 38 (87%) cases were CD99-positive. O
nly three (8%) cases showed focal AE1/AE3 positivity, two of which co-
expressed at least three neural markers. Thirty-three (87%) cases expr
essed at least one neural marker, nine (24%) cases expressed only one
neural marker, nine (24%) cases expressed two neural markers, and 15 (
39%) cases expressed at least three neural markers. There was no signi
ficant difference in outcome between those that expressed one or more
neural markers when compared with those that did not. Similarly, there
was no significant difference in survival in patients whose tumors we
re positive for two or more neural markers or three or more neural mar
kers versus those with fewer neural markers. Patients with pelvic tumo
rs had significantly worse survival when compared to patients with tum
ors in axial-nonpelvic and extremity locations (p < 0.01). None of the
above statistical analyses changed if patients with metastatic diseas
e at presentation were excluded from analysis. Within a given site, ex
pression of neural markers did not correlate with clinical outcome, ex
cept for those patients with pelvic tumors (p = 0.019). In conclusion,
immunohistochemical expression of neural markers in pPNETs that lack
evidence of neural differentiation by light microscopy is not predicti
ve of more clinically aggressive tumors. Thus, routine immunohistochem
ical evaluation for neural differentiation of pPNETs that lack evidenc
e of neural differentiation by light microscopy does not appear to be
warranted. In addition, although cytokeratin immunoreactivity is rarel
y seen in this group of tumors, there are not enough data to support o
r refute its prognostic significance.