Described by Stewart and Foote in 1945,1 columnar cell lesions have recently received attention because of their frequent detection by mammography and occasional association with atypical epithelial proliferations.2
Despite documented histologic and molecular associations with atypical ductal hyperplasia (ADH), low-grade ductal carcinoma in situ (DCIS), lobular neoplasia, and tubular carcinoma,3 the long-term significance and management of columnar cell lesions remains controversial.4 Recently, Abdel-Fatah et al proposed a “low-grade mammary neoplasia family” hypothesis that includes columnar cell lesions as a precursor lesion.5 This hypothesis is based on molecular linkage analyses available in the literature, as well as a comprehensive morphologic study comparing the association of different putative precursor lesions and the development of some special types of low-grade invasive mammary carcinoma. These analyses were based on columnar cell lesions coexisting with more advanced lesions in the same specimen. To investigate this proposed precursor role of columnar cell lesions, we hypothesized that if patients having biopsies containing columnar cell atypia are followed long enough, formal patterns of atypical hyperplasia (AH) as well as tubular and low-grade carcinomas will develop, and these should occur in the same breast that originally contained the columnar cell lesion.
To date, our long-term follow-up studies, as well as those of others, have shown columnar cell lesions to be an indicator of minimally increased relative risk for later cancer development in either breast.6-8 We have more importantly documented that columnar cell lesions are often associated with a more meaningful risk indicator lesion, atypical ductal hyperplasia.6
The purpose of this study is to test the abovementioned precursor hypothesis by examining the clinicopathologic characteristics of breast carcinomas arising after a diagnosis of columnar cell lesion. If columnar cell lesions are precursor lesions, the expectation is that subsequent carcinomas will be ipsilateral, and of low grade.
MATERIALS AND METHODS
Details on the 77 women from the Nashville Breast Cohort included in this study have been previously published.6, 9, 10 Entry biopsies were performed between 1965 and 1982. The authors reviewed the available histologic slides from 54 women (70%), and for the remaining 23 women, information was obtained from pathology reports (n = 13), subject interviews or medical records (n = 8), and death certificates (n = 2). The following characteristics were recorded: laterality of entry biopsy and subsequent cancer, time interval between entry biopsy and subsequent cancer, columnar cell lesion subtype of entry biopsy, histologic type and grade of subsequent cancer, and additional clinical information as available. Where histologic slides were not available, only laterality information was included. Analysis of the characteristics of subsequent carcinomas was restricted to patients who had columnar cell lesions as the most concerning lesion in the entry biopsy. The columnar cell lesions chosen for evaluation correlate best with Wellings and Jensen's Atypical lobule-type A1 (ALA I).11 Specifically, they lacked more than mild usual-type epithelial hyperplasia and were devoid of patterns of the formally defined AH.12 Columnar cell lesions were classified according to the following scheme based on the definitions by Schnitt and Vincent-Salomon: columnar cell change represents columnar cell lesions lacking hyperplasia or atypia, columnar cell hyperplasia represents columnar cell lesions with columnar-type hyperplasia, and columnar cell atypia represents columnar cell lesions with mild cytologic atypia.13 Note that columnar cell atypia lacks architectural atypia and therefore does not meet criteria for AH. Subsequent carcinomas were graded using the Nottingham Grading System.14
The 95% confidence intervals for the proportion of invasive breast cancers in the contralateral breast were calculated using Wilson's method.15
All patients were women, and the median age at cancer diagnosis was 61 years. The median interval between entry biopsy and cancer diagnosis was 11 years (range, 1-25 years).
Entry Biopsy Characteristics and Subsequent Cancer Laterality
Thirty-five of 77 (45%) of the columnar cell lesions occurred in the right breast, 40 of 77 (52%) occurred in the left breast, and 2 of 77 (3%) were bilateral. Columnar cell lesions were classified as columnar cell change (24 of 77, 31%), columnar cell hyperplasia (39 of 77, 51%), and columnar cell atypia (14 of 77, 18%).
Seventy-two women (94%) subsequently developed invasive carcinoma, whereas 5 (6%) developed DCIS. The side of cancer relative to the side with columnar cell lesion could be established for 74 (96%) women; 33 (45%) developed an ipsilateral carcinoma, and 40 (54%) developed a contralateral carcinoma; a single woman (1%) developed bilateral disease.
There were 54 women who have complete follow-up cancer information (Table 1); those with columnar cell change numbered 14 (26%), with columnar cell hyperplasia 27 (50%), and with columnar cell atypia 13 (24%). Cancer developed in the ipsilateral breast relative to the breast with columnar cell lesion in 27 women (50%), and in the contralateral breast in 26 women (48%); 1 woman (2%) presented with bilateral invasive breast cancer.
Table 1. Clinicopathologic Characteristics of Cancer Cases Following CCL Subtypes
|CCL||27a||26a||16 [30%]||23 [42%]||15 [28%]||1.56 (0.1-4.5)||19/54|
Histopathologic Characteristics of Subsequent Carcinomas
Characteristics of the carcinomas are summarized in Table 1. The carcinomas had the following combined histologic grades: 16 (30%) were grade 1, 23 (42%) were grade 2, and 15 (28%) were grade 3. Further analysis based on the different columnar cell lesion subtypes on the initial biopsy is also presented in Table 1. There was no association between the columnar cell lesion subtypes and the grade or type of subsequent carcinomas. The carcinoma subtypes were distributed as follows: 39 invasive mammary carcinomas of no special type, 8 no special type carcinomas with lobular features, 2 no special type carcinomas with tubular features, 3 pure tubular carcinomas, 1 tubulolobular variant carcinoma, and 1 pure mucinous carcinoma. Of the 3 cases (6%) of tubular carcinoma, 2 followed columnar cell hyperplasia, and 1 followed columnar cell atypia.
Grade 3 invasive carcinoma followed columnar cell atypia in 46% of cases, columnar cell change in 29%, and columnar cell hyperplasia in 19%. The most common carcinoma grade after columnar cell hyperplasia and columnar cell change was grade 2 (44% and 50%, respectively). There was no evidence that the subsequent carcinomas were either more or less likely to occur in the ipsilateral breast than the contralateral breast (P = .48).
Note that columnar cell lesions were detected in only 4 (2 columnar cell change and 2 columnar cell hyperplasia) of the 54 subsequent excisions (7%). The carcinomas in those cases were 2 of intermediate grade and 2 of high grade, respectively.
Despite the attention that columnar cell atypia has attracted recently, 2 important questions remain: 1) should isolated columnar cell atypia be excised after identification on a core needle biopsy, and 2) does columnar cell atypia represent a precursor lesion to low-grade in situ and invasive carcinoma. The first question, which pertains to the diagnostic and therapeutic relevance of columnar cell atypia, is still being debated in the literature16-19; the second question addresses the long-term significance and possible precursor nature of this lesion. In this study, we have addressed the second question with an analysis of the follow-up data from the Nashville Breast Cohort.
To date, the available evidence linking columnar cell atypia to formal patterns of AH and carcinoma is based on co-occurrence and cumulative molecular alterations rather than documented clinicopathologic evolution and progression. Concomitant presence of columnar cell atypia and other, more advanced lesions such as ADH, low-grade DCIS, atypical lobular hyperplasia (ALH), and tubular carcinoma has been solidly documented.3 The association between columnar cell atypia and tubular cancer was first reported in 1997,20 whereas the association between columnar cell atypia, ALH, and tubular cancer was reported in 1999 by Rosen,21 and later substantiated by Abdel-Fatah and several others.3, 22, 23 This association between tubular carcinoma and a background of columnar cell atypia has been documented in as many as 90% of tubular carcinomas. It is often the case that ADH or DCIS appears to arise in a background of columnar cell atypia, in the same lobular unit or cluster of lobular units, rather than in 2 separate, unrelated foci.24, 25 This seems to at least suggest a link between columnar cell atypia and the family of low-grade breast cancers, either as a concomitant lesion, or as a nonobligate early precursor.3, 23
From a morphological standpoint, it is true that the constituting cells in columnar cell atypia and architecturally atypical lesions are identical and seamlessly merge into each other. Also, as far as tubular carcinoma is concerned, it is notable that cellular morphology is similar, and apical snouts are equally characteristic.
Immunophenotypical similarities are also notable. Columnar cell atypia and associated more advanced lesions are uniformly estrogen and progesterone receptor-positive and do not express HER-2/neu. Cyclin D1 and p53 expression also tends to be identical within the atypical columnar cell lesion and the contiguous invasive carcinoma.26, 27 Moreover and as expected, the available molecular evidence supports the morphologic and immunophenotypic similarities. This can be summarized by the consistent loss of 16q in columnar cell atypia as in this entire family of low-grade breast lesions, and by the similar loss of heterozygosity and nonrandom X chromosome inactivation profiles between columnar cell atypia and adjacent tubular carcinomas.28 An accumulation of genetic abnormalities has also been demonstrated in lesions of increasing severity (columnar cell atypia to invasive carcinoma), and elegantly tabulated by Abdel-Fatah and colleagues.3
As compelling as these findings may be, they do not a priori establish a causal relationship between columnar cell atypia and AH or carcinoma. As documented by Page et al and others in their long-term follow-up evaluations of the natural history of DCIS treated by biopsy only, reaffirming low-grade DCIS as a nonobligate precursor to invasive cancer, we would expect columnar cell atypia to evolve, as a precursor, into the lesion it presumably precedes.29-33 In other words, the most persuasive argument for progression remains documentation of actual progression rather than mere co-occurrence. So far, the available studies have failed to substantiate this hypothesis. They have even failed to establish columnar cell atypia as a marker of significant risk for subsequent cancer on long-term follow-up.
The first study of these lesions was by Eusebi and colleagues, who retrospectively identified 21 patients with so-called ”clinging carcinoma” of the flat, monomorphic type (corresponding to what we would now diagnose as columnar cell atypia) from 4000 benign breast biopsies performed between 1965 and 1971. After an average follow-up period of 19.2 years, only 1 of the patients had recurrence of a histologically identical lesion without evidence of DCIS or invasive tumor.34 Also, Bijker and colleagues, within the European Organization for Research and Treatment of Cancer Trial, reported no local recurrence in 59 patients with so-called clinging carcinoma, after an average follow-up of 5.4 years.35 In a more recent study, de Mascarel and colleagues examined co-occurrence and subsequent incidence of carcinoma associated with various types of atypia, and although columnar cell atypia, which they referred to as flat epithelial atypia, had a 17% risk of concomitant invasive cancer, none of the 84 patients with flat epithelial atypia as an isolated diagnosis experienced malignant recurrence after 10 years of follow-up.36 Most recently, Martel and colleagues recognized 63 cases of columnar cell atypia (flat epithelial atypia/flat duct intraepithelial neoplasia, DIN1) of 1751 core biopsies, with invasive carcinoma present in 7 ipsilateral follow-up excisions performed between 2 and 9 years after the initial procedure. Their conclusion amounted to a mild increase in absolute cancer risk of 11% after a diagnosis of flat epithelial atypia on needle core biopsy.37 In our study of 2008, a nested case-control analysis examining the different types of columnar cell lesions with or without hyperplasia and atypia showed no statistically significant difference between columnar cell atypia and the different columnar cell lesions lacking atypia with respect to subsequent breast cancer risk. This would suggest that the overall relative risk associated with columnar cell atypia is equal to that of columnar cell lesions in general, that is, mild and on the order of 1.5.6, 7
Our current results echo the previous studies and fail to support a precursor hypothesis for columnar cell lesions. This is manifested through 2 main outcomes, 1 being an equal distribution between ipsilateral and contralateral subsequent carcinomas, and relatively random distribution of grades in those carcinomas. The results noted for columnar cell lesions in general are, within the confines of small numbers and limited statistical power, identical to those noted for columnar cell atypia, and more supportive of the marker rather than precursor status of either lesion. This was similarly demonstrated in carcinomas after ADH with nonpreferential tumor type, grade, or laterality (manuscript in preparation). Moreover, although not statistically significant, the number of high-grade tumors was greatest with lesions diagnosed as columnar cell atypia, whereas the expected natural history—were the precursor theory upheld—should be of evolution into tubular, or at least low-grade infiltrating mammary carcinoma.
This apparent contradiction invites several possible explanations. First, the nature of the sample in the vast majority of the cases under study is excisional, performed because of a clinically palpated mass, in the premammographic, prestereotactic biopsy era. It is therefore reasonable to assume that the excision, at least in a fraction of cases, may have entirely removed the lesion, hence eliminating its potential for progression to a more atypical epithelial proliferation, but obviously not eliminating its role as a marker of future breast cancer risk.
Alternatively, a closer look at the different genetic alterations affecting columnar cell atypia and more advanced lesions (ADH, DCIS, and tubular carcinoma) reveals that random overlap is conspicuous, and straightforward stepwise progression is far from definitive, unlike what is seen between lobular carcinoma in situ and invasive lobular carcinoma.38 It appears that greater genetic similarity lies between columnar cell change and tubular carcinoma than between columnar cell atypia and tubular carcinoma.39, 40 This brings forth a more robust and fact-compatible hypothesis that would explain the majority of instances of columnar cell atypia without eliminating the possibility of progression, only rendering it a statistically much less likely event once an isolated columnar cell atypia lesion is formed. In other words, we postulate that several genetic alterations usually take place relatively simultaneously and randomly, and tend to include the alterations found in those lesions fulfilling criteria for ADH and low-grade DCIS. Those genetic abnormalities would allow for both a basement membrane-limited proliferation (flat epithelial atypia/columnar cell atypia) and intraluminal proliferative abilities (architectural atypia in ADH and DCIS). The 2 phenomena would therefore be more likely to occur simultaneously than evolve from 1 to another. This explanation would equally apply to tubular carcinoma.
A potentially effective method to test this hypothesis would be to study the genetic alterations in the different parts of clustered lesions (a relatively common occurrence) that exhibit ADH/DCIS with a background of flat epithelial atypia with or without adjacent invasive tubular carcinoma. Simpson and colleagues attempted this approach in 1 of their studied cases, although the morphologic associations, in our opinion, were not entirely compelling, and the single examination of a case is unlikely to be sufficient for definitive inferences and generalizations.39
In conclusion, our results reinforce the idea that columnar cell atypia, although occurring frequently in conjunction with more atypical lesions and special type invasive carcinoma, does not seem to progress to these co-occurring atypical lesions. Further follow-up studies focusing on needle core biopsy specimens and additional careful and detailed molecular analyses as suggested above are likely to provide us with more definitive answers regarding this challenging group of lesions. In the meantime, the available information advises us, when encountering columnar cell atypia, to look for more advanced lesions, and to ascribe to isolated columnar cell atypia a mild increase in future cancer risk, similar to that associated with nonatypical columnar cell lesions of the breast.