338
Chapter 11
random sections disclose foci of DCIS in sections of one or
more segmental ducts with no apparent lobular connection
even when the lesion is traced with serial sections. The rela-
tive frequency of origin from the TDLU or from larger duct
structures, and the clinical significance of this distinction
remain to be determined.
The microscopic classification of DCIS became the sub-
ject of heightened interest after the widespread introduction
of breast conservation therapy. Interest in factors associated
with the success or failure of this therapy directed attention
not only to variants described on the basis of growth pattern,
but also to finer cytologic details.
The spectrum of histologic patterns of DCIS in men does
not differ appreciably from the appearance of the disease in
women. There is a higher proportion of papillary DCIS in
men, and comedo DCIS is less frequent than in women.
In standard histologic sections, DCIS is confined within
the lumens of ducts and lobules involved in the process.
When studied by immunohistochemistry (IHC) for laminin
or type IV collagen, basement membranes in DCIS appear
intact or focally discontinuous.
99–101
The presence or absence
of mitotic figures is not a definitive feature in the diagnosis
of DCIS, because mitoses may also be found rarely in normal
and hyperplastic epithelium. However, the finding of one or
more mitoses per 10 high-power fields (HPFs) suggests DCIS.
Myoepithelium in DCIS
Myoepithelial cells are often retained but attenuated inDCIS,
and they are occasionally hyperplastic at the periphery of the
duct (Fig. 11.4). Myoepithelial cells do not generally accom-
pany the neoplastic epithelial proliferation within the duct
lumen in DCIS except for certain types of carcinoma arising
in a papilloma or in solid papillary DCIS.
Experimental evidence suggests that myoepithelial cells
may have a paracrine tumor suppressor effect on DCIS, act-
ing to inhibit invasion.
102
Tumor suppression capabilities of
myoepithelial cells include inhibition of invasion
103
and of
angiogenesis.
104
In vitro
, myoepithelial cells have been shown
to have the capacity to inhibit breast carcinoma cell growth
and to induce apoptosis.
105
These tumor-inhibiting properties
have been attributed in part to the expression of maspin, a pro-
tease inhibitor, by myoepithelial cells.
106
Other tumor suppres-
sor genes expressed by myoepithelial cells include cytokeratin
5 (CK5), smooth muscle actin (SMA), and caveolin-1.
107
The functional activity of myoepithelial cells in
in situ
carcinoma is significantly changed compared with normal
epithelium. These alterations include overexpression of che-
mokines CXCL12 and CXCL14, which bind to and enhance
the invasiveness of carcinoma cells.
108
In particular, CXCL12
and its receptor CXCR4 appear to promote breast carcinoma
cell growth and metastases.
109,110
Cell Types in DCIS
The range of subtle differences in cell type found inDCIS usu-
ally engenders little comment, but certain distinct variants
have been identified and described by specific names.
Signet
comedo type) tends to be a well-defined, tan tumor with white
to pale yellow flecks composed of necrotic DCIS (comedos)
that extrude from the cut surface when the lesion is com-
pressed. Abundant calcification in the lesion can impart a
gritty sensation upon cutting. Although these findings are
suggestive of this type of carcinoma, a similar gross appear-
ance is found in some instances of duct ectasia and mastitis.
Most classifications of DCIS have been based on histo-
pathologic features of the lesions, but some investigators
have drawn attention to the distinction between grossly
apparent and microscopic lesions. Gump et al.
96
studied
70 consecutive patients treated in one institution for lesions
classified as DCIS on an initial biopsy. Fifty-four (77%) had
carcinomas classified as “gross” because the patient pre-
sented with a palpable tumor, nipple discharge, or Paget dis-
ease. The majority of these patients (48 of 54 or 89%) had a
mass. Microscopic DCIS in 16 (23%) was nonpalpable, and
it was detected by mammographic calcifications or it was
an incidental finding. Invasive carcinoma was found in six
(12%) surgical specimens subsequent to an initial biopsy
that revealed gross DCIS, but not in the patients who had
microscopic DCIS. Axillary lymph node (ALN) metastases
were found in only one patient with a gross lesion, and not
in any patient with microscopic DCIS.
A slightly more complex classification based on anatomic
distribution was proposed by Andersen et al.
97
They identi-
fied three types of growth patterns, which occurred individ-
ually or in combinations. “Microfocal” lesions involved “one
or a few lobules and/or ducts” measuring up to 5 mm. “Dif-
fuse” DCIS involved a region of 5 to 10 mm or an entire seg-
ment of the breast, and the “tumor-forming” type consisted
of closely connected glandular structures that occupied an
area of 60 to 70 mm, resulting in a palpable mass. Microfo-
cal and diffuse types of DCIS were typically not palpable. A
population-based review of cases revealed that 18 of 35 pa-
tients (51%) with DCIS had microfocal lesions, 13 (37%) had
the diffuse type, and 4 (11%) had tumor-forming DCIS.
21
No
ALN metastases were found in any of the patients who had
an axillary dissection.
Microscopic Pathology
General Histologic Features
The microanatomic site of origin of many DCIS appears to
be in the
terminal duct lobular unit
(TDLU). The most con-
vincing evidence for this conclusion comes from the sub-
gross microdissection studies of Wellings et al.
98
Expanded
TDLUs sometimes resemble primary or secondary segmen-
tal ducts, but their lobular origin is suggested by an excessive
number of duct structures within a low-power microscopic
field and by the accompanying stroma. Recently character-
ized columnar cell lesions lend support to this conclusion.
Exceptions can be found to the concept of the TDLU ori-
gin of DCIS. For example, it does not readily describe DCIS
limited to major central lactiferous ducts, sometimes asso-
ciated with Paget disease or nipple discharge. Occasionally,
