Sx. Jiang et al., LARGE-CELL NEUROENDOCRINE CARCINOMA OF THE LUNG - A HISTOLOGIC AND IMMUNOHISTOCHEMICAL STUDY OF 22 CASES, The American journal of surgical pathology, 22(5), 1998, pp. 526-537
Large cell neuroendocrine carcinoma (LCNEC) of the lung is defined as
a poorly differentiated and high-grade neuroendocrine turner that is m
orphologically and biologically between atypical carcinoid and small c
ell lung carcinoma (SCLC). During a survey concerning bcl-2 protein ex
pression in the subtypes of lung cancer, we noticed that two previousl
y diagnosed non-SCLCs met the criteria for LCNEC. Because LCNEC is a n
ewly recognized clinicopathologic entity and because all reported case
s have been retrieved from the so-called ''neuroendocrine tumor file,'
' we suspected that LCNEC had been underdiagnosed. In the present stud
y, we histologically reviewed 766 surgically resected lung cancers and
were able to diagnose 22 (2.87%) LCNECs with the neuroendocrine featu
res subsequently confirmed by immunostaining for multiple neuroendocri
ne markers. Each case stained positively For at least three general ne
uroendocrine markers, and 12 (54.5%) also were positive for neuroendoc
rine hormones. Histologically, most LCNECs showed a marked decrease in
or a loss of organoid architecture and could be mistaken for poorly d
ifferentiated adenocarcinoma or squamous cell carcinoma. Because our L
CNECs are the first to be identified by retrospective review of routin
ely diagnosed lung cancers, and 18 had been classified as non-SCLC, th
ey may represent cases relatively difficult to diagnose. The present s
tudy shows that the most difficult diagnostic factor of LCNEC is the r
ecognition of its light microscopic neuroendocrine features, and LCNEC
must be distinguished not only from atypical carcinoid or SCLC, but a
lso from common non-SCLC. Histologically, when an organoid architectur
e is subtle or absent, the rosettelike structure becomes the best mark
er for the recognition of neuroendocrine differentiation. Clinically,
the prognosis for our LCNECs was significantly worse than that for sta
ge-comparable non-SCLCs (p = 0.046).