Rosen's Breast Pathology, 4e

367

Ductal Carcinoma In Situ

FIG. 11.48.  In situ carcinoma with ductal and lobular features. A: Carcinoma in a duct ( right ) and distended lobular glands ( left ). B: Small and large cell populations are present. Microlumina have been formed in an area composed of small cells in a duct. The lesion was immunoreactive for E-cadherin.

quantitative descriptors of necrosis are present, focally pres- ent, or absent. In the Van Nuys classification system, only the “zonal” type of necrosis classifies DCIS as exhibiting necrosis; however, occasional desquamated or individually necrotic cells do not qualify as DCIS in this scheme. Sneige et al. 129 studied interobserver reproducibility among six pathologists who assessed nuclear grade, accord- ing to Lagios’ criteria (Table 11.3), in 125 examples of DCIS. Complete agreement on nuclear grade was reported in 43 cases (35%), and five of six pathologists agreed in 45 (36%). The generalized kappa for distinctions between grades 1 and 2 and between 2 and 3 were 0.29 and 0.48, respectively (stan- dard error = 0.02). These levels of agreement were regarded as fair and moderate by the authors. Pair-wise correlations between individual pathologists and the consensus grade in- cluded a range of kappa values from 0.44 to 0.76, with 5 of 6 having values greater than 0.60, representing “substantial” agreement. Some authors have suggested that apocrine andmicropap- illary DCIS be listed as separate categories and not included in a three-tiered grading scheme. 157 This proposal appears to derive from perceived difficulty in assigning these lesions to one of the three conventional grades because of inconsis- tent expression of individual criteria for grading. However, in practice, DCIS with apocrine or micropapillary features express the same range of histologic variation as nonapo- crine and nonmicropapillary DCIS, and separate character- ization is not recommended. In regard to apocrine lesions, nuclear grade is based on comparison with nuclei in usual benign apocrine metaplasia. The distribution of architectural patterns and necrosis is not different in apocrine and non- apocrine DCIS. Micropapillary DCIS is subject to variations in nuclear grade and to necrosis (illustrated in this chapter), which do not differ from other architectural types of DCIS.

Agreement onDCIS grading has been shown to be achieved more commonly with the Van Nuys classification system, compared with those of Lagios and Holland. 158 Pinder et al. 159 have identified a subdivision of DCIS with “very poor prog- nosis.” This “very high-grade” type of DCIS has high cytonu- clear grade, solid (more than 50%) architecture, and extensive comedo-type necrosis (more than 50%). This “novel” classifi- cation for DCIS offered better prognostic discrimination for ipsilateral recurrence than “cytonuclear grade” classification alone when applied to cases assembled in a clinical trial. 159 Preliminary studies show promise in assessing the mo- lecular “grade” of DCIS through the use of array-based com- parative genomic hybridization (CGH) and other related techniques. Such molecular profiling has the potential to improve the clinical evaluation of DCIS and is becoming in- creasingly possible. 160 Angiogenesis Studies of the microvascular pattern of capillaries in breast tissue from patients with DCIS have demonstrated increased periductal vascularity associated with some but not all DCIS. The most reliable information has been obtained from his- tologic sections immunostained with vascular markers such as factor VIII, CD31, or CD34. Increased microvessel size has been observed at sites of DCIS compared with normal breast tissue. 161 Simple hyperplasia has been shown to have a 22-fold greater degree of microvascular density than nor- mal ducts. 162 Neovascularization has been described around DCIS associated with invasive lesions. 163 Periductal neo- vascularity found around 21 (38%) of 55 examples of pure DCIS studied by Guidi et al. 164 was not related to histologic subtype, the presence of necrosis, proliferative index (PI),