Ja. Brainard et Jr. Goldblum, STROMAL TUMORS OF THE JEJUNUM AND ILEUM - A CLINICOPATHOLOGICAL STUDYOF 39 CASES, The American journal of surgical pathology, 21(4), 1997, pp. 407-416
Recently, cell size, cell density, and growth pattern were found to be
reliable histologic parameters in separating benign from malignant du
odenal stromal tumors. However, there are few data on the histologic f
eatures and important prognostic parameters of stromal tumors from oth
er parts of the small bowel. Thus, we studied the clinical and patholo
gic features of 39 stromal tumors of the jejunum and ileum to determin
e which parameters would be most useful in distinguishing a benign fro
m a malignant tumor. In all cases, the following histologic parameters
were recorded: (a) predominant growth pattern (organoid, fascicular,
solid, or mixed), (b) cellularity (low or high), (c) predominant cell
type (spindled, epithelioid, or mixed), (d) nuclear pleomorphism (mini
mal, moderate, or severe), (e) the presence or absence of tumor cell n
ecrosis, (f) the presence or absence of mucosal infiltration, (g) the
presence or absence of skeinoid fibers, and (h) the number of mitotic
figures per 50 high-power microscopic fields (HPF). Clinical follow-up
was obtained in all cases, and the patients were considered to have s
uffered an adverse outcome if they developed metastatic disease or die
d as a complication of their tumor. In the absence of these features,
patients were not considered to have suffered an adverse outcome. Twen
ty-five patients suffered an adverse outcome. Twenty-one patients died
of disease from 1 month to 9 years (median:;2 years). One patient die
d at 4 days because of postoperative complications. Three patients wer
e alive with metastatic disease at 6 months, 6 years, and 7 years. Twe
nty-four of these 25 patients developed metastatic disease, most commo
nly to the liver. Fourteen patients did not suffer an adverse out come
. Eleven patients were alive without disease from 2 to 11 years (media
n: 3 years), and three patients died of unrelated causes at 1, 1, and
3 years. Although there was some overlap in features between clinicall
y benign and malignant tumors, features that were significantly associ
ated with an adverse outcome included tumor size >5 cm, mitotic counts
>5 mitotic figures per 50 HPF, high cellularity, the absence of a pre
dominant organoid growth pattern, the absence of skeinoid fibers, the
presence of severe nuclear pleomorphism, and the presence of mucosal i
nfiltration and tumor cell necrosis (p < 0.05 using the chi-square and
Fisher's exact tests). Features that were significantly associated wi
th decreased survival included tumor size >5 cm, mitotic counts >5 mit
otic figures per 50 HPF, high cellularity, the absence of skeinoid fib
ers, and the presence of tumor cell necrosis (p < 0.05 using the Mante
l-Haenszel log-rank test). Given the fact that there is some overlap i
n these features between clinically benign and malignant tumors, a mul
tiparametric analysis using the above features is the most effective w
ay of predicting clinical behavior.