Diagnostic Cytology
transhepatic cholangiography. Generally, tissue biopsies for his-
tologic examination are not performed due to a relatively high
rate of complications including direct leakage with peritonitis,
hemorrhage, and subsequent scar formation with stricture. A
major impediment to the use of direct cytologic examination of
bile and pancreatic juice has been a very low level of diagnos-
tic sensitivity.187,188 This is due, in part, to the fact that a signifi-
cant proportion of cholangiocarcinomas are well differentiated
adenocarcinomas and as a result it may be difficult to distin-
guish these low-grade neoplasms from benign reactive changes
due to inflammation. Additionally, many of these neoplasms
are associated with severe sclerosis and fibrosis of the sur-
rounding stroma, preventing exfoliation of the tumor cells and
thus leading to low cellularity. De Paralta-Venturina et al. have
emphasized a conservative approach in this setting to avoid a
potential false-positive diagnosis.189 The cytologic examination
of pancreatic juice has a somewhat higher diagnostic yield with
sensitivities ranging from 30 to 79%.190,191 Kondo et al. utilized
conventional pancreatic juice cytology and molecular PCR for
mutations in the oncogene which was present in two-thirds of
carcinomas that had a negative cytology.191
Of the specimen types the brushing specimens had the high-
est accuracy ranging from 36 to 89% with most studies showing
an average sensitivity of 50%. The specificity is similar to that
seen in other GI cytology ranging from 93 to 100%.31,189,192-204
In brushing smears, adenocarcinomas present as solitary,
intact malignant cells and three-dimensional aggregates31,194,205
(Fig. 14.32). Within the clusters, the abnormal nuclei are piled
up on one another and crowded. Individual cells have enlarged
nuclei with smooth to irregular contours, coarse hyperchromatic
chromatin, and often well-developed nucleoli. The N/C ratio is
variable but generally high. The major constituent of the cyto-
morphologic differential diagnosis is benign reparative atypia.
Several studies have analyzed the cytomorphologic features of
the biliary tract brushings with statistical analysis. Cohen et al.
analyzed a series of 90 biliary tract brushings obtained at ERCP
by regression analysis.31 Eighteen different cytomorphologic fea-
tures were included. The three that emerged as significant major
criteria were (1) chromatin clumping, (2) increased N/C ratios,
and (3) nuclear molding. The diagnostic sensitivity for all three
features was 43%, whereas the specificity was 100%. If two of
the three criteria were present, the sensitivity and specificity were
83 and 93%, respectively. Other helpful minor criteria that the
authors discussed included enlarged nuclei, variability in nuclear
size, irregular nuclear contours, nuclear groves, and a macro-
nucleoli. The authors did not, however, evaluate the presence
of individually dispersed abnormal epithelial cells as a marker of
reduced intercellular cohesion. Additionally, the specimens eval-
uated were cytospins of cellular material suspended from brush-
ings rather than direct smears. It is the authors' opinions that the
finding of individually dispersed abnormal epithelial cells is an
extremely important criteria for separating benign from malig-
nant proliferations.194,206 Vadmal et al. did not find isolated cells
as a helpful feature in distinguishing benign from malignant
lesions.207 Renshaw et al. examined sensitivity and specificity of
the criteria proposed by Nakajima et al. and Cohen et al.31,206,208
The three most important features for diagnosing malignancy
included loss of polarity, nuclear molding, and chromatin
crumping (sensitivity 36% and specificity 92%). Additionally
they assessed the observers' "gestalt" of benign verses malignant
without using any specific individual cytomorpholigic features.
This aspect proved to be the single best predictor of malignancy.
Overall, Renshaw et al. found excellent interobserver reproduc-
ibility as measured by kappa statistics as for overall assessment
of malignancy and for crumping of chromatin granules. In addi-
tion to the previous findings Koss emphasized other important
features from his experience: three-dimensional loose aggrega-
tion and single cells of carcinoma compared with flat plaques
and fewer single cells of benign reactive atypias.209 The major
cytologic features of biliary tract adenocarcinoma are summa-
rized in Table 14.8.
Reactive atypia usually resembles normal glandular epithe-
lium; however, varying degress of atypia are present but fall
short of malignancy. Features include cohesive monolayered
sheets with fairly uniform cells that maintain polarity207,210
(Fig. 14.31). The sheets typically contain sharply defined smooth
edges, and intact atypical and individually dispersed cells are
extremely rare. Generally, a honeycomb arrangement with mild
nuclear crowding and mildly enlarged round-to-oval nuclei with
Fig. 14.32 (A) Bile duct adenocarcinoma. The carcinoma cells clearly contrast with the adjacent flat cohesive sheet of benign reactive ductal cells. The
malignant cells are larger with pleomorphic nuclei with irregular nuclear membranes, high N/C ratios, crowding, and prominent nucleoli. The adjacent benign
group, in contrast, is cohesive with an even honeycomb distribution of cells with bland nuclear features (HP). (B) Adenocarcinoma. Small clusters of loosely
cohesive malignant cells with high N/C ratios, thick nuclear membranes, hyperchromatic chromatin, and prominent nucleoli (Papanicolaou x HP).
previous page 398 ComprehensiveCytopathology 1104p 2008 read online next page 400 ComprehensiveCytopathology 1104p 2008 read online Home Toggle text on/off