Tubular Adenomas of the Breast Are Cytologically Distinct from Fibroadenomas

Tubular adenomas were regarded as one type of fibroepithelial lesions of the breast [1], until the 5th edition of the World Health Organization classification of tumors of the breast [2], where it is currently reclassified as an epithelial tumor. Tubular adenoma and the most common fibroepithelial lesion, namely fibroadenoma, are similar in clinical behavior [3, 4]. The epithelial component of both tubular adenomas and fibroadenomas are generally bland and cytologically benign [5]. However, histologically, tubular adenomas lack the characteristic stromal expansion seen in fibroadenomas [6, 7]. Recent evidence demonstrated different molecular profiles underpinning tubular adenomas and fibroadenomas [8]. As now tubular adenoma is increasingly regarded a distinct entity from fibroadenomas and fibroepithelial lesions, it is worthwhile to characterize the cytological features of tubular adenoma of the breast. In this study, the largest cohort, to date, of histologically confirmed aspirates of tubular adenomas were reviewed and compared with aspirates of fibroadenomas. Cytological features of tubular adenomas that are diagnostically indicative, and those that are useful in differentiation from fibroadenomas, are identified. By a comprehensive comparison and description, tubular adenomas of the breast can be accurately defined on fine-needle aspiration cytology (FNAC).

Computerized searches of the pathology archives in the involved institutions from the year 2000 to 2021 were performed. All aspirates with histological follow-up confirming a diagnosis of tubular adenoma of the breast were retrieved. Consecutive aspirates with histological and/or radiological confirmation of a diagnosis of fibroadenoma, from the year 2002 to 2009, from one of the involved institutions (Prince of Wales Hospital) were also retrieved. Demographical data, including patient age and laterality of lesion, were obtained from the cytology reports. All cytologic slides were reviewed by three of the authors (J.J.X.L., J.K.M.N., and G.M.T.) with a list of parameters (Table 1) evaluated. Specimens with insufficient cellularity (absence of any epithelial element) and presence of artifacts (e.g., excessive drying or poor staining quality) precluding assessment were excluded. Discrepancies were resolved by reviewing the slides together on a multiheaded microscope until a consensus was reached. The study was approved by the institutions’ Ethical Committees. Parameters evaluated include those pertaining to the following:

• Epithelium: cellularity, cohesiveness, fragment size, tubule (percentage of epithelial fragments in tubular formation) and tubular fragments (tissue fragments containing multiple tubular structures with or without stroma)

• Stroma: cellularity, percentage of naked stromal fragments, and fragment size

• Background features: presence of calcifications, histiocytes, multinucleated giant cells, apocrine metaplastic cells, and debris

• Myoepithelium: presence of background bipolar nuclei and myoepithelial cells within epithelial structures

• Epithelial cytomorphologic features: nuclear features, chromatic pattern, nuclear/cytoplasmic ratio, and mitosis

Materials were generated by both imaging-guided and free-hand FNAC with 22-, 23-, or 25-gauge needles attached to 10-mL syringes. For most cases, two to four direct smears accompanied by two to four liquid-based slides were prepared. Some cases only had direct smears liquid-based preparations. At least one Papanicolaou or hematoxylin and eosin-stained slide (smear or liquid-based preparation) was present for each case. Air-dried preparations and cell block preparations were reviewed if present but not included for assessment for cytomorphologic parameters.

Statistical analysis was performed using SPSS statistical software (Windows version 23; IBM Corporation, Armonk, NY, USA). The χ² test was used to compare differences between tubular adenoma and fibroadenomas as categorical variables. The Mann-Whitney-Wilcoxon test was used in comparison of continuous variables, including demographical (age) and cytological (epithelium and stroma) parameters. Multivariate binary logistic regression with backward Wald method was performed to identify parameters differentiating tubular adenomas from fibroadenomas. A p value of <0.05 was considered as significant.

A total of 43 aspirates of tubular adenoma and 94 aspirates of fibroadenoma were included, after excluding 37 specimens (16 tubular adenomas and 21 fibroadenomas). Patients with tubular adenomas in this cohort were on average younger (32.02 vs. 36.28 years, p = 0.016). No difference was observed in the laterality of lesion (Table 2).

Comparing the features of epithelial and stromal components in the specimens, aspirates of tubular adenomas showed lower cohesiveness (p = 0.010) with a trend of more small epithelial fragments (p = 0.061) and less large epithelial fragments (p = 0.033) (Fig. 1a), indicating an overall smaller size of epithelial fragment in tubular adenomas. Tubules were more abundant in tubular adenomas (p = 0.009) (Fig. 1b) and tubular fragments were seen exclusively in tubular adenomas (p < 0.001) (Fig. 1c, d). Stromal fragments were more commonly seen in tubular adenomas (p = 0.011), but those seen in fibroadenomas were more likely to be naked or of large size (p = 0.006) (Fig. 2a, b) (Table 3).

As for background features, there were no differences in all parameters assessed including presence of calcifications, histiocytes, multinucleate giant cells, apocrine metaplastic cells, and debris (p > 0.05) (Fig. 3a–c). Background bipolar nuclei and myoepithelial cells within epithelial structures were identified in all cases (Table 4). Epithelial cytomorphologic features between tubular adenomas and fibroadenomas were similar in terms of nuclear size, nuclear size variation, nuclear membrane irregularity, nucleolar features, and chromatic character. The average nuclear/cytoplasmic ratio of epithelial cells in tubular adenomas were slightly lower than that of fibroadenomas (difference = 0.033, p = 0.006) (Table 5).

Multivariate analysis showed independent statistical associations in four parameters. A predominant (≥50%) composition of large epithelial fragments (p = 0.003, OR = 0.083, 95% CI: 0.016–0.434) (Fig. 4a) and naked stromal fragments (p < 0.001, OR = 0.154, 95% CI: 0.056–0.421) (Fig. 4b) were predictive of fibroadenoma, whereas presence of stromal fragments (p = 0.042, OR = 2.437, 95% CI: 1.018–5.830) (Fig. 4c, d) and a composition of tubules of ≥5% (p < 0.001, OR = 11.710, 95% CI: 3.681–37.254) (Fig. 4e, f) were predictive of tubular adenomas (Table 6).

Although tubular adenomas and fibroadenomas of the breast share considerable overlaps in terms of clinical behavior and morphologic appearance [5, 6], the evolving understanding on the molecular features of fibroadenomas indicate that these two entities may not be biologically similar. MED12, and less frequently RARA, and TERT promoter mutations are identified in fibroepithelial lesions including fibroadenomas and phyllodes tumors [7, 9]. In comparison, tubular adenomas were shown to harbor MET and FGFR3 mutations [8]. These updates are reflected on the addition of tubular adenoma to the category of epithelial lesions in the 5^th edition of the World Health Organization classification of tumors of the breast [2].

Despite the advances in molecular profiling of tubular adenomas (and fibroadenomas), tubular adenomas remain a poorly characterized entity on FNAC. Studies describing the cytological features were largely limited to case reports or case series [10‒14], with the largest series consisting of 26 cases reported by Sengupta and colleagues [13]. Kumar et al. [14] compared six aspirates of tubular adenomas to ten of fibroadenomas and suggested tubular structures (tubules) and three-dimensional cohesive structures to be useful in diagnosing tubular adenomas. However, the small case number precluded statistical analysis. In this study, the largest cohort to date of aspirates of histologically confirmed tubular adenomas (n = 43) were reviewed and compared to a comparable cohort of fibroadenomas (n = 94).

At low-power examination, tubular adenomas are usually of moderate cellularity with highly cohesive epithelial structures (Table 3). Compared to fibroadenomas, the epithelial fragments in tubular adenomas are of smaller size (p = 0.033). Aspirates of tubular adenoma are more likely to contain stromal fragments (p = 0.011) and when stromal fragments are present, more likely to be associated with epithelial fragments (p = 0.042). These differences correspond to the characteristic cytological description of fibroadenomas, namely, the presence of large “staghorn/antler horn” or “honeycomb” epithelial sheets and naked stromal fragments [15]. As for the increased amount of stromal fragments in tubular adenomas, it may be explained by preferential sampling of thin and interspersing stromal tissue amount tubules [6] (Fig. 5a–d), which are delicate and easily fragmented by narrow-bore needles used in FNAC. Tubules are common in tubular adenomas (n = 26/43, 60.47%) and increased compared to fibroadenomas (p = 0.009). A highly indicative feature for tubular adenoma was observed. “Tubular fragment” that contains multiple tubular structures were only seen in aspirates of tubular adenomas (p < 0.001).

Calcifications can be seen radiologically in tubular adenomas and fibroadenomas [16, 17] and were both occasionally present in aspirates of either entity. Histiocytes, multinucleated giant cells, apocrine metaplastic cells, and debris are cytological features associated with fibrocystic changes of the breast [18]. These features were variably observed in aspirates of tubular adenomas and fibroadenomas without statistical difference (p > 0.05).

There was an apparent difference in nuclear/cytoplasmic ratio of 0.033 (p = 0.006) in the epithelial cells between tubular adenoma and fibroadenoma. The difference was practically unobservable in routine light microscopy assessment and quantification of nuclear/cytoplasmic ratio below a factor of 0.1 is impractical without the use of digital image analysis [19]. No difference was observed in all other epithelial cytomorphological features. Cytomorphologically malignant features (marked nuclear membrane irregularity, presence of large distinct nucleoli, and a coarse chromatin pattern) were not identified in any aspirates of this cohort. Mitosis was only seen in one case of fibroadenoma. The absence of alarming cytomorphological features may be useful in ruling out high-grade carcinomas [20]. However, an abundance of epithelial fragments in tubules may give rise to the suspicion of malignancy [21], in particular tubular carcinoma, of which a case of overdiagnosis on FNAC has been reported [10].

On multivariate analysis, four independent parameters were identified. Presence of tubules and stromal fragments favored tubular adenoma while a predominance of large epithelial fragments and naked stromal fragments favored fibroadenoma (Table 6). These findings not only echo the histological differences among tubular adenomas and fibroadenomas [3] but also confirms that tubular adenomas display cytologically distinguishable features from fibroadenomas and that the diagnosis of tubular adenoma can be made on FNAC.

Review of this large cohort of aspirates showed that tubular adenomas generally display moderate epithelial cellularity with high cohesiveness, with stromal fragments that contain epithelium. Tubules are common in aspirates of tubular adenoma (n = 26/43, 60.47%) and the presence of “tubular fragments” (tissue fragments containing multiple tubular structures with or without stroma) is a unique and pathognomonic feature of tubular adenomas (p < 0.001). Calcification and fibrocystic changes were variably seen in tubular adenomas, but myoepithelial cells are always present. Cytomorphologically malignant features and mitoses were absent in all aspirates of tubular adenoma. Comparison with fibroadenomas demonstrated presence of tubules and stromal fragments as independent factors associated with tubular adenomas, whereas a predominance of large epithelial fragments and naked stromal fragments were associated independently with fibroadenomas. Tubular adenoma of the breast is not only a histologically and molecularly separate entity from fibroepithelial lesions but can also be diagnosed on FNAC based on its distinctive cytological features.

The study was approved by the Joint Chinese University of Hong Kong – New Territories East Cluster Clinical Research Ethics Committee (approval number: 2020.320) and the New Territories West Cluster Research Ethics Committee (approval number: NTWC/REC/18080). The study was granted the exemption of requiring written informed consent by the Joint Chinese University of Hong Kong – New Territories East Cluster Clinical Research Ethics Committee and the New Territories West Cluster Research Ethics Committee.

The authors declare that there is no conflict of interest regarding the publication of this paper.

The authors have no funding to declare.

Joshua J.X. Li: conceptualization, data curation, investigation, methodology, visualization, and writing – original draft; Joanna K.M. Ng: conceptualization, investigation, and methodology; Billy S.W. Lai and Ka-Ho Shea: investigation and resources; Conrad H.C. Lee: validation and visualization; Julia Y. Tsang: validation and formal analysis; and Gary M. Tse: conceptualization, investigation, methodology, supervision, and writing – review and editing.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.