Article Text
Abstract
Aims To fully elucidate the clinicopathological features of breast carcinoma in sclerosing adenosis (SA-BC).
Methods Clinical and histological characteristics of 206 SA-BCs from 180 patients were retrospectively evaluated. Immunohistochemical phenotype was examined. The clinicopathological relevance of the topographical pattern of SA-BCs was analysed.
Results Overall, up to 46 patients (25.6%) had contralateral cancer, either SA associated or not. Of 99 cases who underwent core needle biopsy (CNB), 36 were underestimated as adenosis or atypical ductal hyperplasia at CNB, 5 invasive cases were misinterpreted as in situ carcinomas, whereas 4 ductal carcinoma in situ (DCIS) cases were overdiagnosed as invasive carcinoma. Microscopically, 163 tumours were in situ, including 136 DCIS, 19 lobular carcinomas in situ (LCIS) and 8 mixed DCIS/LCIS; of these carcinomas in situ (CIS), 37 had microinvasion. The DCIS group exhibited low, intermediate and high grades in 53.7%, 34.6% and 11.8% of cases, respectively, mostly with solid (43.4%) or cribriform (41.9%) pattern. Forty out of 43 invasive cases were invasive ductal carcinoma (IDC), mostly DCIS predominant. Immunophenotypically, luminal A phenotype was identified in 55.1%, 63.2% and 45.0% of DCIS, LCIS and IDC cases, respectively. Topographical type A group (carcinoma being entirely confined to SA, n=176) was characterised by smaller size, less invasiveness, lower grade and more frequency of luminal A immunophenotype compared with type B group (≥ 50% but not all of the carcinomatous lesion being located in SA, n=30) (all P<0.05).
Conclusions CIS, especially non-high-grade DCIS, represents the most common variant of SA-BC, and luminal A is the most predominant immunophenotype. CNB assessment might be challenging in some SA-BCs. The topographical pattern has great clinicopathological relevance. Careful evaluation of the contralateral breast and long-term follow-up for patients with SA-BC is necessary given its high prevalence of bilaterality.
- breast pathology
- histopathology
- immunohistochemistry
Statistics from Altmetric.com
Introduction
Sclerosing adenosis (SA) is a common type of breast adenosis originating in the terminal duct-lobular units, which can be observed in about 12.0%–27.8% of breast biopsies without cancer.1 2 Clinically, it can present as a frequent incidental finding under microscope, a lesion on mammography or a palpable mass.1 3 Histologically, SA is characterised by lobulocentric proliferation of glandular components accompanied with stromal sclerosis, and the peripheral myoepithelial cell layers are preserved, as well as the investing basement membrane.4 Occasionally, atypical hyperplasia and carcinoma can develop in SA, and ductal carcinoma in situ (DCIS) in SA (SA-DCIS) sometimes closely mimics infiltrating carcinoma both grossly and microscopically, leading to dilemmas in the diagnosis. Only a limited number of studies have evaluated breast cancer in SA (SA-BC), which suggested that SA-BC might have an increased risk of bilateral occurrence, less invasive behaviour and lower histological grade.5 6 However, the prior studies were mainly case reports or based on small sample sizes, and mostly restricted to DCIS. The clinical and pathological characteristics of SA-BC have not been well documented to date. The value of core needle biopsy (CNB), a reliable method in the preoperative diagnosis of breast lesions, in the assessment of SA-BC has never been discussed in the literature, either. On the other hand, data regarding the immunoprofile of SA-BC are limited as yet. Some studies indicated that negativity for oestrogen receptor (ER), progesterone receptor (PR) as well as human epidermal growth factor receptor 2 (HER2) was significantly more common in SA-DCIS than in non-SA-DCIS.5 6 Whereas others recently found that the proportion of HER2-positivity was almost the same in SA-BC and non-SA-BC.7 Hence, fulfilling a thorough illustration of SA-BC for the purpose of better understanding its features has substantive clinical value.
The inclusion criteria for SA-BC varied from study to study.5–8 Ogura et al classified SA-DCIS into ‘inner type’ (≥50% cancer in SA) and ‘outer type’ (<50% cancer in SA).6 Given the quite different clinicopathological features of these two types, the ‘inner type’ was considered to be cancer genuinely arising in SA in its true sense. We thus adopted the inclusion criteria of ≥ 50% of cancer being located in the SA lesion in the current study. Several studies also have analysed the topographical patterns of SA-BC, whereas the methods for topographical classification differed.6–8 Moritani et al divided 24 cases of carcinoma in situ (CIS) involving SA into topographical type A (CIS entirely surrounded by SA) and type B (CIS involved SA at least focally but was not confined to the SA area), and revealed that most type A CIS cases were non-high-grade and ER+/PR+/HER2− immunophenotype predominated, whereas most type B cases were high-grade and nearly half were ER−/PR−.8 However, the clinicopathological relevance of the topographical pattern of SA-BC cases, including those in situ and invasive ones, has not yet been comprehensively evaluated in larger series.
Herein, we presented so far the largest series of SA-BC and attempted to fully elucidate its clinical, histological characteristics and immunoprofile, and also evaluated the correlation between the topographical pattern and clinicopathological features.
Materials and methods
Case selection
Altogether 206 lesions of SA-BC in 180 patients, diagnosed between March 2006 and May 2015, were retrieved from the archival files of the Department of Pathology, Fudan University Shanghai Cancer Center (Shanghai, China). H&E-stained sections for each case were independently reviewed by two experienced senior pathologists. Clinical information was obtained from the medical records.
We further divided our cases into topographical type A (carcinoma being entirely confined to SA) and type B (≥ 50% but not all of the carcinomatous lesion being located in SA), according to the topographical relationship between cancer and SA (figure 1).
Immunohistochemistry and fluorescence in situ hybridisation
Immunohistochemical (IHC) staining was performed using the Ventana Bench Mark ultra autostainer (Ventana Medical System Inc, Roche Tuson, Arizona, USA) and the Ventana ultra view universal DAB detection kit. The primary antibodies included ER, PR, HER2, androgen receptor (AR), E-cadherin, p120, Ki-67, AE1/AE3, calponin and p63. All of these antibodies were from Roche Ventana. Appropriate positive and negative controls were carried out simultaneously for all stains. For cases of HER2 IHC 2+, Abbott-Vysis HER2 FISH assay was employed to further confirm the status of HER2 gene amplification.
The status of ER, PR and HER2 were evaluated using the scoring criteria of the American Society of Clinical Oncology/College of American Pathologists guideline.9 10 The Ki-67 labelling index was visually calculated for the percentage of nuclei-positive tumour cells and estimated to the closest decile; the hot spots were included in cases with heterogeneous staining. The cut-off point for Ki-67 was set to 14%, according to the recommendation.11 Subsequently, the IHC surrogate subtype (luminal A, luminal B, HER2-overexpression and triple-negative) of each case was determined based on the procedures proposed by 2013 St Gallen International Expert Consensus.12 E-cadherin and p120 were employed when a lobular carcinoma was suspected. E-cadherin is membranous reactivity in ductal carcinoma but negative in lobular carcinoma. P120 is membranous stained in tumour cells of ductal carcinoma whereas cytoplasmic stained in those of lobular carcinoma. Both AE1/AE3/p63 and AE1/AE3/calponin double immunostainings were performed to evaluate the presence of myoepithelial cells in each case.
Statistical analysis
The statistical analyses were performed using the SPSS software package (SPSS V.19.0, Chicago, Illinois, USA). Categorical variables were compared by the χ2 test, and measurement data were analysed using the Pearson correlation analysis. The P values of <0.05 were considered statistically significant.
Results
Clinical characteristics
Altogether, 206 tumours of SA-BC were identified in 180 female patients, with a mean age of 49.9 years at diagnosis (range, 30–79 years). Eighty-six tumours were located in the left and 68 in the right breast; of these, 20 patients had contralateral non-SA-BC, and each 10 were synchronous and metachronous (table 1). The remaining 26 patients presented with bilateral SA-BC, either simultaneous (n=22) or asynchronous (n=4).
Clinically, 73 (35.4%) cases were symptomatic, with palpable breast mass as the most frequent complaint (n=60). Other symptoms included breast tenderness (n=7), nipple discharge (n=4), nipple retraction (n=1) and breast discomfort (n=1). The other 133 (64.6%) asymptomatic cases were accidentally found by auxiliary examination. Regarding the physical examination, palpable masses, as the most common manifestation, were found in 113 (54.9%) cases. The localised thickening of breast was identified in 32 (15.5%), and there were no findings in the other 61 (29.6%) cases. The clinical features are summarised in table 2.
CNB results and surgery procedures
Ninety-nine cases in our series received preoperative CNB. Twenty-two (22.2%) lesions were considered as invasive carcinoma, 3 of which were corrected to be DCIS with microinvasion and 1 was DCIS on excision. Twenty-six (26.3%) cases were diagnosed as DCIS; of these, 3 were upgraded to invasive ductal carcinoma (IDC) with DCIS predominance and 7 were DCIS accompanied with microinvasion based on the resected specimen. Eight (8.0%) cases were considered as DCIS with suspected infiltration, among which 3 were confirmed to be IDC on excision, 3 were DCIS accompanied with microinvasion and the remaining 2 cases were DCIS. Five (5.0%) were lobular carcinomas in situ (LCIS) based on the IHC results, whereas small foci of invasion were observed in surgical specimen in 2 of these cases. Two (2.0%) cases were deferred, with suspicion of invasive carcinoma, which were proved on surgical excision. Twenty-four (24.2%) cases were interpreted as atypical ductal hyperplasia (ADH). Neither atypia nor malignancy was observed in the other 12 (12.1%) cases, 7 of which were diagnosed as SA. Still, further surgery was proceeded for these cases because of the clinical/radiological–pathological discordance.
Modified radical mastectomy was performed in 38 breasts and simple mastectomy was in 140 breasts (121 with sentinel lymph node biopsy (SLNB)). Breast conservative surgery was performed in 20 (10 with SLNB) and lumpectomy was in 8 breasts. Nine out of the 169 patients (5.3%) had lymph node metastasis (table 2).
Gross findings
Grossly, 58 (28.2%) cases had firm mass with obscure boundaries, and 19 (9.2%) cases showed well-demarcated but non-encapsulated nodules. Ninety-eight (47.6%) cases revealed abnormal texture on cut surface, including firm texture with unclear margin (n=64), rubbery consistency (n=7), granulated (n=24) or scar-like appearance (n=3). Eight (3.9%) cases merely demonstrated darker colour, either dark red or grayish-brown, compared with the adjacent breast tissue. There were no grossly visible findings in the remaining 23 (11.2%) cases, for which the tumour size was evaluated under the microscope. The mean tumour size (the largest diameter of cancer) was 1.7 cm (range, 0.2–6.0 cm), and the majority (n=157, 76.2%) measured no more than 2.0 cm (table 2). The mean size of the in situ and invasive carcinomas was 1.6 and 2.1 cm, respectively.
Histological and IHC findings
Microscopically, the overall lobulocentric architecture of SA can be recognised at low-power magnification in each case, and the cancer area was associated with SA, which was the most remarkable and constant feature. CIS consisted of up to 79.1% (n=163) of the total cases, among which 37 cases were accompanied with microinvasion (figure 2), and the remaining 43 (20.9%) were invasive (table 2). In the following section, we evaluated the features of CIS (including those with microinvasion) and invasive cases separately.
Histological findings of cases with SA-CIS
DCIS cases occupied the overwhelming majority of CIS group (n=136, 83.4%); the remaining cases included 19 (11.7%) LCIS and 8 (4.9%) mixed DCIS/LCIS. Most of the DCIS cases demonstrated low (53.7%) or intermediate (34.6%) nuclear grade; only 11.8% were high grade. The tumour cells in DCIS cases typically grew in solid (43.4%) or cribriform (41.9%) pattern, whereas occasionally had a micropapillary structure predominance (2.2%). A mixed growth pattern, consisting of two or more different patterns in one case, occurred in the remaining 12.5% cases. Extensive apocrine metaplasia was found in 19 cases (figure 3). Prominent comedo necrosis was commonly observed in intermediate–high grade cases. Worthy to be noted, such necrosis was typically found in the cancer area beyond the SA contour in type B group (figure 1D). Microcalcifications were common in low–intermediate grade DCIS, mostly in the ductules of the adjacent SA area, and also scattered in some duct lumens of DCIS. The histological features of SA-DCIS are summarised in table 3.
All of the 19 cases with LCIS invariably displayed solid pattern. Most (78.9%) of these cases were of classic type (figure 4), and the other 4 (21.1%) were pleomorphic (table 3). In the remaining 8 mixed DCIS/LCIS cases, two CIS components intermingled with each other in some cases whereas occupied discrete areas in others. Seven of the DCIS parts were low grade and one was intermediate. As for the LCIS parts, 6 fell into the classic and 2 into the pleomorphic subtype (table 4).
Immunophenotypes of cases with SA-CIS
Only in situ components were evaluated in 15 cases of CIS with microinvasion because the microinvasive foci were unavailable during IHC detection, including 13 DCIS and 2 mixed CIS cases. All of the remaining CIS cases with microinvasion shared the same immunophenotype between in situ and microinvasive components. Luminal A phenotype was identified in 75 (55.1%), and luminal B in 26 (19.1%) DCIS cases. Triple-negative phenotype accounted for 17.6% (n=24), approximately two-thirds of which exhibited distinct apocrine features. The remaining 11 (8.1%) DCIS cases were categorised as HER2-overexpression subtype (table 3). Of the 19 LCIS cases, most (63.2%) were of luminal A phenotype; luminal B, triple-negative and HER2-overexpression subtype constituted 26.3%, 5.3% and 5.3%, respectively (table 3). All except 2 of the mixed CIS cases demonstrated the same immunoprofile in DCIS and LCIS components (table 4).
Histological findings and immunophenotype of invasive carcinomas in SA
Most invasive cases were IDC (n=40, 93.0%); only 2 (4.7%) were ILC and 1 (2.3%) was mixed IDC/ILC. All except 5 were accompanied with CIS within SA. CIS represented the predominant component of malignancy in 32 cases; the invasive area was <0.5 cm in 11 of these cases. The majority of the IDCs were of grade 1 (45.0%) or grade 2 (47.5%) (table 5). Invasive tumour cells were diffusely distributed or formed nests, cords and trabeculae. Immunohistochemically, 45.0% of the IDCs were considered as luminal A, 32.5%, 17.5% and 5.0% cases were of luminal B, triple-negative and HER2-overexpression phenotype, respectively. Both of the ILCs were LCIS predominant, with invasive lesions measuring <0.5 cm, and both showed luminal A phenotype. The case with mixed IDC/ILC also had limited invasive elements; the ductal and lobular carcinoma was of luminal A and triple-negative phenotype, respectively.
The comparison between the two topographical groups
Overall, there were 176 cases of topographical type A and 30 of type B. Invasive cases in type A (18.2%) group were much less than those in type B (36.7%, P=0.021), and CIS cases of type A had a dramatically lower frequency of microinvasion (17.4%) than those of type B (63.2%) (P<0.001, table 2). Both DCIS and IDC cases of type A tended to display more commonly low nuclear grade compared with those of type B ones (both P<0.001). The most common growth pattern in type A DCIS group was solid (50.0%), followed by cribriform (43.2%), whereas mixed pattern appeared as the most dominant in type B group (66.7%), indicating a significant difference between the two types (P<0.001). The detail is illustrated in table 3.
Immunohistochemically, luminal phenotype (80.5%), especially luminal A (61.0%), was the most predominant in type A DCIS cases, whereas both triple-negative (38.9%) and HER2-overexpression (27.8%) occupied significantly higher proportions in type B DCIS group (P<0.001, table 3). Luminal A (62.1%) also was the most prevalent in type A IDC group, followed by luminal B (27.6%); in comparison, luminal B (45.5%) appeared to be the most frequent in type B IDC and then triple-negative (36.4%) (P<0.001, table 5).
Discussion
As noted in prior studies, SA is a very common finding in the breast of female adults,2 whereas SA-BC is less common and its features are still less well defined. We thus conducted this comprehensive study to fully elucidate the clinicopathological characteristics of this unique disease. To the best of our knowledge, the current dataset of SA-BC might represent so far the largest one.
In our series, up to 25.6% of the overall patients with SA-BC presented bilateral breast cancer, either synchronously or metachronously, consistent with the previously reported proportion of 21%–38%.5–8 These results indicated a higher risk for bilateral cancer in patients with SA-BC, compared with the reported incidence of 1.6%–7% for bilateral occurrence in all patients with breast cancer.7 13 Hence, careful monitoring of the contralateral breast for patients with SA-BC is recommended.8
Clinically, as many as 64.6% of the lesions in our series lacked evident clinical manifestations and no palpable mass was detected in nearly half cases during physical examination. These results were in line with the previous indication that SA-BC was more likely to be asymptomatic compared with non-SA-BC.5 7 Similarly, there were no obvious masses in up to 62.8% of our cases macroscopically. The peculiar inherent nature of this disease made the grossing even more difficult. Given these challenges, the importance of enhancing the awareness and familiarity of this distinct tumour is worthy to be emphasised.
Preoperative CNB has become a more convenient and efficient method in the assessment of breast lesions, and its value in permitting one-stage surgical therapy for patients with breast cancer has been stressed.14 Nevertheless, CNB assessment might be challenging in some SA-BCs. According to the current study, about 36.4% cases of SA-BC were underestimated as benign adenosis or ADH at CNB. One possible explanation for these false-negative results might be that cancer within the extensive SA tends to be limited or occurs preferentially at the periphery of SA, peculiarly in those early stage tumours, therefore there is a risk of sampling inaccuracy/inadequacy for CNB.8 Moreover, the overall architecture of SA-BC is not consistently recognisable in those small CNB samples. Consequently, wide SA and SA accompanied with ADH in a CNB specimen should be evaluated with particular attention and further resection is recommended.14 Although there is excellent correlation between the findings of CNB and surgical specimen, previous experience has indicated that a number of diagnoses of DCIS by CNB are upgraded to invasive carcinoma on excision.4 In our series, 5 such CIS cases at CNB were upgraded to invasive carcinoma after resection, also reflecting the limitation of CNB for accurate interpretation in some SA-BCs. On the other hand, 4 DCIS cases (3 with microinvasion) in our series were overdiagnosed as invasive carcinoma at CNB, representing another tricky situation, which will be discussed in more detail hereinafter.
Although LCIS was previously announced to be the most prevalent non-invasive breast carcinomas presenting in SA,15 SA-LCIS is uncommon in the recently published studies,6–8 16 and accounted for merely 9.2% of SA-BCs in our study. Instead, DCIS, especially non-high-grade ones, constituted the majority of our series. The predominant histological patterns of SA-DCIS were solid and cribriform, consistent with the non-comedo type reported by Ogura et al.6 We encountered 8 cases of mixed DCIS/LCIS in SA, which had seldom been described before.17 These cases more frequently exhibited a combination of low-grade DCIS and classic LCIS, demonstrating similar histological features to their pure SA-DCIS or SA-LCIS counterparts.
To date, we are not aware of any literature establishing the features of invasive breast cancers in SA. Our results indicated that IDC was the most prevailing histological type of invasive SA-BC, mostly well-to-moderately differentiated. Notably, most invasive SA-BCs in our series were CIS predominant, similar to the findings by Ogura et al.6
There was no full-scale analysis on the IHC profile of SA-BC as yet. Gene expression profiling studies of DCIS have demonstrated that DCIS cases can also be classified into the ‘intrinsic’ molecular subtypes as invasive breast carcinomas.18–21 For the first time, we systematically analysed the surrogate intrinsic immunoprofiles of both in situ and invasive SA-BCs based on the St Gallen criteria. Luminal phenotypes, especially luminal A, strikingly exceeded others in SA-DCIS. Luminal A immunophenotype also appeared to be the most predominant in LCIS and invasive cases. These findings were alike to some prior reports focusing on in situ cases, which announced that SA-CIS was characterised by high positive rate for ER/PR and low HER2 expression.8
In some SA-CIS cases, extensive proliferated glands with angulated branches or closely packed epithelial nests were entrapped in dense fibrous stroma, and surrounding myoepithelial cells were attenuated and inconspicuous. This pattern might exactly resemble invasive carcinoma, which made differential diagnosis intractable. As mentioned above, three lesions of DCIS in our series, with/without microinvasion, were overinterpreted as IDC at CNB assessment. The relatively preserved architectural framework of SA at low-power magnification points to the possibility of SA-CIS. The stroma in SA-CIS is mainly composed of proliferated collagen and fibrosis, compared with more cellular desmoplastic stroma in invasive carcinoma. Identification of myoepithelial cells surrounding the atypical glands by careful inspection and auxiliary IHC staining can rule out the possibility of invasive carcinoma.8 Another situation that may cause confusion with SA-BC is SA with florid hyperplasia. The latter might impart a very ‘busy’ and cellular morphological appearance, with irregular and distorted branched structures, stromal fibrosis as well as occasionally scant myoepithelial cells, reminiscent of SA-DCIS or even IDC. In this circumstance, estimating the atypia of proliferated epithelial cells is useful, and immunostaining for CK5/6 and CK14 can be employed to facilitate the assessment when necessary.22 The presence of myoepithelial cells can also be highlighted by immunostaining.
Both the rarity and under-recognisation of SA-BC limit the clinicians’ ability to standardise the procedures for clinical management. Whereas partial mastectomy for patients with SA-DCIS was preferred by Yoshida et al, who argued that total mastectomy was unnecessary.5 Simple mastectomy with SLNB was the most prevalent strategy adopted in our series. Further optimal treatment strategies are required, especially for those cases in which SA is extremely extensive or multifocal whereas malignancy is limited. And the prognosis of patients with SA-BC is warranted to be further investigated after a longtime follow-up.
We further evaluated the topographical pattern of SA-BC and its clinicopathological relevance. Given the remarkable differences in tumour size, histological grade and growth pattern, biological behaviours as well as immunoprofile revealed in the current study, we supposed that the formation of the two types might be different. For topographical type A SA-BCs where all the neoplastic proliferation is situated within SA, one could safely conclude that the tumour arises within the adenosis. As for topographical type B ones, it would be harder to distinguish whether carcinoma arises from SA and then grows beyond it, or originally intrudes from the area outside SA. Moritani et al considered that type B might comprise heterogeneous groups of lesions, that is, some might arise within SA while others would originate external to it.8 How one decides which one is the case might resort to complex molecular methods, not just histological cutoffs and clinicopathological context. Whether the two topographical types employ diverse pathways of pathogenesis depends on further exploration with larger cohorts and more equal distribution between type A and B.
In conclusion, non-high-grade DCIS was the major histological subtype and luminal A was the most prevalent immunophenotype in both in situ and invasive SA-BCs. Better understanding the features of SA-BC is of paramount importance for accurate diagnosis and appropriate treatment. CNB assessment might be challenging in some SA-BCs due to either inaccurate/inadequate sampling or confusing morphological appearance. Given its high risk of bilateral occurrence, careful evaluation of the contralateral breast and long-term follow-up is of great value. The topographical pattern of SA-BC had great clinical, histological and immunoprofile relevance. Further studies are encouraged to elucidate the pathogenesis of this unique disease and to determine whether the two topographical types are subjected to diverse biological mechanisms.
Take home messages
Carcinomas in situ (CIS), especially non-high-grade ductal carcinoma in situ (DCIS), represented the most common variant of breast cancer in sclerosing adenosis (SA-BC), and luminal A was the most predominant immunophenotype in both in situ and invasive SA-BCs.
Core needle biopsy assessment might be challenging in some SA-BCs due to either inaccurate/inadequate sampling or confusing morphological appearance.
The topographical pattern of SA-BC had great clinicopathological relevance. Topographical type A SA-BCs were characterised by smaller size, less invasiveness, lower grade and more frequency of luminal A phenotype compared with topographical type B ones.
Careful evaluation of the contralateral breast and long-term follow-up for patients with SA-BC is necessary given its high prevalence of bilaterality.
Abstract translation
References
Footnotes
BHY and SXT are co-first authors and both contributed equally to the paper.
BHY and SXT contributed equally.
Handling editor Dhirendra Govender.
Contributors WTY: conceived the study and revised the manuscript. BHY: designed the study and wrote the manuscript. SXT: collected and analysed the data and reviewed the manuscript. XLX, YFC, RB, RHS, XYT and HFL: performed the morphological study and IHC evaluation. XYZ: performed data interpretation and reviewed the manuscript.
Funding This study was funded by Science and Technology Commission of Shanghai Municipality (15495810300).
Competing interests None declared.
Ethics approval the Institutional Review Board of Fudan University Shanghai Cancer Center, Shanghai, China.
Provenance and peer review Not commissioned; externally peer reviewed.