The frequency of diagnosis of lobular neoplasia in our study was 2.4% in a consecutive series of routinely removed breast specimens in a general hospital. The rate of LCIS ranges from 0.5% to 3.6% of breast specimens [2
]. Because there are no obvious clinical or radiological features, the true incidence of LN in the general population is unknown [22
The diagnosis of LN is typically related to an incidental finding on breast biopsies that are performed for other indications. With the increasing use of mammography, lobular neoplasia has been observed in association with microcalcifications in up to 40% of cases that are diagnosed by core needle biopsy [24
]. Microcalcifications rarely form within LNs and they usually correlate with other benign or malignant breast lesions--the diagnosis of LN is most often incidental [23
Columnar cell lesions (CCLs) comprise a spectrum of morphological alterations of the duct epithelial lining, acquiring a columnar cell appearance and involving variably dilated acini of the terminal duct lobular unit (TDLU) [25
]. There has been recent, increasing interest in these lesions, because they are detected in up to 42% of the breast biopsies that are performed due to the presence of microcalcifications by mammography [26
For instance, many terms have been used to describe CCLs, from "blunt duct adenosis" to "clinging carcinoma" [25
]. Nevertheless, Schnitt and Vincent-Salomon's nomenclature and diagnostic criteria of CCL have been the most widely used [18
], whereas in the most recent WHO guidelines, CCL was included under the term "flat epithelial atypia" (FEA) [6
]. After the release of the WHO classification, Schnitt began referring to CCC and CCH with atypia as "flat atypia" [28
CCLs have been linked to lobular neoplasia, low-grade DCIS, and invasive carcinoma. Further, similar genetic abnormalities have been found in CCC and CCH with atypia or FEA and the associated low-grade DCIS and invasive carcinoma. These findings have led to the reasonable conclusion that CCC and CCH with atypia are the earliest morphologically identifiable precursor lesions of low-grade DCIS and invasive carcinoma [8
Yet, there are no prospective randomized trials, and few epidemiological studies with patients with only CCC and CCH with atypia have evaluated the prognosis of these lesions. Several studies, comprising a limited number of cases, demonstrated little or no risk for progression to invasive carcinoma [29
]. Thus, there remains no consensus on the ideal treatment for these atypical lesions.
In our series, LN and CCL coexisted in 78.5% of cases, most often as mild forms of the spectrum of CCL--eg, CCC without atypia (67.7%). Our data are consistent with a recent study that examined 68 core needle biopsy specimens with a diagnosis of LN due to the excision of microcalcifications. The authors demonstrated an association between LN and CCL in 54% of cases, none of which presented with CCC or CCH with atypia after wide excision biopsy [15
]. However, after analyzing 111 breast biopsies with LN but no other in situ
or invasive carcinomas, Leibl et al.
noted that LN was associated with FEA--ie, CCC and CCH with atypia in 86.5% of cases [33
]. Our studies and other reports have observed a frequent association of CCL with LN, but they differ regarding the presence or absence of atypia.
There are many terms for LCC. Moreover, the WHO morphological definition of FEA is imprecise and does not describe the cytological and architectural features that are necessary for its diagnosis. In our study, all cases were reviewed by 2 observers, including a well-trained breast pathologist (HG). We used well-defined diagnostic criteria per Schnitt and Vincent-Salomon and noted fewer cases of CCC and CCH with atypia than what has been reported [10
]. We believe that in many series and cases in our Breast Consulting Laboratory, FEA is being overdiagnosed, which could lead to the implementation of more aggressive treatments [34
The frequency of invasive carcinomas that were associated with LN in our series was 45.6%, and we observed a similar frequency of ILC (47.2%) and invasive ductal carcinoma (IDC). However, when LN subtypes were analyzed separately, we observed a 4-fold higher frequency of IDC that was associated with ALH versus ILC and a greater link between ILC (33.3%) and LCIS compared with IDC. We also noted a 12-fold increase in the correlation between ILC and LCIS (33.3%) compared with ALH (2.8%).
Our data are consistent with a series of 775 cases of LN [14
]. Bratthauer and Tavassoli stratified the LNs as "lobular intraepithelial neoplasias" (LINs) and evaluated the frequency of association between LIN subtypes (1, 2, and 3) and invasive carcinoma. The percentage of LIN 1 (equivalent to ALH) that was associated with invasive carcinoma was 14%, and 89% of these tumors were IDCs. In the patients with LIN 3 (equivalent to LCIS), the frequency of association with IDC and ILC was 23% and 86%, respectively. The authors concluded that the advance from LIN 1 to LIN 3 was linked to a 64% increase in the frequency of invasive carcinoma and a greater than 700% rise in the likelihood of ILC [14
Our results corroborate other studies and suggest that lobular neoplasia is not only a risk indicator but also a nonobligate precursor of invasive breast carcinoma [23
]. Invasive carcinomas that develop after a diagnosis of ALH are 3 times more likely to arise in the ipsilateral rather than contralateral breast [35
Lobular neoplasia and ILC are detected together frequently in the same specimen and location of the tumor--in up to 90% of cases of ILC [10
]. These lesions have similar immunohistochemical profiles, including the loss of expression of E-cadherin and β-catenin and the cytoplasmic localization of p120-catenin [36
]. Invasive and in situ
lobular carcinomas confer similar genetic gains and losses, often bearing the same mutations in the gene that encodes E-cadherin (CDH1
However, it is unknown why LCIS carries a greater risk of progression to invasive disease and is associated more frequently with invasive lobular carcinoma compared with ALH. Mastracci et al.
demonstrated that somatic alterations in CDH1
are a hallmark of LCIS but not ALH [39
]. This disparity suggests that mutations that inactivate CDH1
can distinguish LNs that are able to progress to invasive disease, explaining our morphological data [39