Clinicopathological characteristics and prognosis of breast carcinoma with apocrine differentiation: a propensity score-matched analysis of the SEER database

Introduction

Breast cancer remains the most prevalent malignancy among women worldwide, with an estimated 310,000 new cases diagnosed in the United States in 2024 (1). Among its diverse histopathological subtypes, carcinoma with apocrine differentiation (APO)—formerly referred to as apocrine carcinoma—represents a rare histological form of invasive breast cancer, accounting for approximately 0.3% to 4% of all cases (2). APO is characterized by glandular-like architecture and distinct cytological features. Clinically, it is defined by abundant granular eosinophilic cytoplasm, sharply demarcated cell borders, large nuclei with prominent nucleoli, and immunohistochemical positivity for markers such as the androgen receptor (AR) and gross cystic disease fluid protein-15 (GCDFP15) (3).

Despite its distinctive pathological features, the biological behavior and prognostic significance of APO remain incompletely understood, largely due to its rarity and the paucity of large-scale prospective studies. Most existing data have been derived from small case series, and previous analyses based on the Surveillance, Epidemiology, and End Results (SEER) database have yielded conflicting conclusions regarding survival outcomes compared with invasive ductal carcinoma of no special type (IDC-NST). Some studies have reported worse survival in APO patients (4), whereas others have demonstrated better outcomes (5). These discrepancies may reflect limitations in data completeness and the influence of multiple confounding factors inherent in large-scale epidemiologic research.

To address these gaps, we conducted a retrospective cohort study utilizing SEER data to compare survival outcomes between patients with APO and those with IDC-NST. We employed propensity score matching (PSM) to minimize selection bias by balancing key covariates between the two groups, thereby isolating the independent prognostic impact of APO. Additionally, we evaluated clinicopathological variables to identify factors associated with survival in this underexplored population. Our findings provide novel insights into the natural history of APO and may inform the development of evidence-based, subtype-specific management strategies for this rare but biologically distinct form of breast cancer. We present this article in accordance with the STROBE reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0622/rc).

Methods

Data source and study population

Patient data were extracted from the SEER database using SEER*Stat software (version 8.4.4), covering 17 cancer registries from 2000 to 2021 (submission date: November 2023). Maintained by the National Cancer Institute, the SEER program represents approximately 26.5% of the U.S. population and provides comprehensive demographic, clinicopathological, and treatment-related information on primary malignancies. Eligible cases were identified using the International Classification of Diseases for Oncology, 3^rd edition (ICD-O-3) codes: APO (ICD-O-3 8401/3) and IDC-NST (ICD-O-3 8500/3). Inclusion criteria required histologically confirmed diagnoses. Exclusion criteria were: incomplete data for estrogen receptor (ER), progesterone receptor (PR), or human epidermal growth factor receptor 2 (HER2) status; presence of distant metastases; diagnosis of in situ carcinoma; missing tumor-node-metastasis (TNM) staging information; diagnosis based solely on autopsy or death certificate; or incomplete follow-up. After applying these criteria, a total of 492,642 cases were included in the final analysis, comprising 1,066 APO cases and 491,576 IDC-NST cases. In this study, patients were classified into three subtypes based on receptor expression status: luminal type, HER2-positive, and triple-negative breast cancer (TNBC). This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

PSM

To address the imbalance in sample size and reduce selection bias, PSM was conducted using the “MatchIt” package in R. Cases of APO and IDC-NST were matched at a 1:5 ratio using the nearest-neighbor method with a caliper width of 0.01. Covariates included in the matching process were age, race, tumor grade, T stage, N stage, ER/PR/HER2 status, and treatment modalities (surgery, chemotherapy, radiotherapy). This approach ensured balanced baseline characteristics between the two groups for subsequent comparative analyses. A 1:5 matching ratio was selected to preserve statistical power by retaining more controls, as the IDC-NST group was considerably larger than the APO group. Sensitivity analyses using 1:1 and 1:4 matching ratios were also performed.

Statistical analysis

Continuous variables (e.g., age) were compared using Student’s t-test, and categorical variables were analyzed using Pearson’s Chi-squared test or Fisher’s exact test, as appropriate. Survival outcomes were evaluated using Kaplan-Meier (KM) survival curves and compared with log-rank tests. Overall survival (OS) was defined as the time from diagnosis to death from any cause or last follow-up, while breast cancer-specific survival (BCSS) was defined as the time from diagnosis to death specifically attributable to breast cancer, with censoring for non-cancer-related deaths. Cox proportional hazards regression models were employed to identify prognostic factors through univariate and multivariate analyses. A two-sided P value of <0.05 was considered statistically significant, and hazard ratios (HRs) were reported with corresponding 95% confidence intervals (CIs). All statistical analyses were performed using R software (version 4.4.2), with relevant packages including “stats”, “survival”, “survminer”, “MatchIt”, “ggplot2”, and “cobalt”.

Results

Patient characteristics of APO and IDC-NST

A total of 492,642 patients meeting the inclusion and exclusion criteria were analyzed, including 1,066 cases of APO and 491,576 cases of IDC-NST. The median follow-up duration was 67 [interquartile range (IQR), 43–97] months for APO patients and 52 (IQR, 24–90) months for IDC-NST patients. A summary of demographic and clinicopathological characteristics is presented in Table 1.

Compared with IDC-NST, APO patients were significantly older at diagnosis (62.9 vs. 59.3 years, P<0.001) and had a lower proportion of White individuals (74.9% vs. 77.9%). APO tumors were more likely to be of higher histological grade, larger in size, and associated with more advanced lymph node involvement. APO cases were predominantly ER-negative (72.5% vs. 17.9%, P<0.001) and PR-negative (78.2% vs. 28.0%, P<0.001), with a higher rate of HER2 positivity (26.3% vs. 15.9%, P<0.001). Chemotherapy was more frequently administered in the APO group (60.5% vs. 40.6%, P<0.001). No significant differences were observed in gender distribution, radiotherapy use, or surgical intervention rates between the two groups (P>0.05).

Survival analysis and prognostic factors

Survival outcomes were assessed using KM analysis and log-rank tests to compare OS and BCSS between APO and IDC-NST patients. For BCSS analysis, 1,161 cases (0.2% of the total cohort) with unknown causes of death were excluded to ensure analytical validity.

Log-rank testing revealed that APO was associated with significantly poorer OS compared to IDC-NST (P<0.001, Figure 1A). However, no significant difference was found in BCSS between the two groups (P=0.24, Figure 1B).

Figure 1 KM curves of all patients with APO vs. IDC-NST before PSM. (A) APO patients were associated with significantly poorer OS compared to IDC-NST (P<0.001). (B) However, no significant difference was found in BCSS between the two groups (P=0.24). APO, apocrine differentiation; BCSS, breast cancer-specific survival; IDC-NST, invasive ductal carcinoma of no special type; KM, Kaplan-Meier; OS, overall survival; PSM, propensity score matching.

Univariate and multivariate Cox proportional hazards analyses identified age at diagnosis, sex, race, tumor grade, T stage, N stage, ER/PR/HER2 status, and receipt of surgery, chemotherapy, or radiotherapy as independent prognostic factors for both OS and BCSS (P<0.05) (Table 2). Poorer outcomes were associated with Black race, male sex, higher tumor grade, and advanced T/N stage, while ER/PR/HER2 positivity and receipt of standard therapies were associated with improved survival. Although histological subtype (APO vs. IDC-NST) was not an independent predictor of OS (HR =0.912, 95% CI: 0.80–1.03, P=0.17), APO was found to be an independent favorable prognostic factor for BCSS (HR =0.72, 95% CI: 0.58–0.88, P=0.001), indicating a potential subtype-specific biological advantage.

PSM

To mitigate potential confounding biases, PSM was conducted, resulting in a balanced cohort of 6,375 patients—1,065 with APO and 5,310 with IDC-NST. One patient in the APO group was excluded after matching because no suitable control with a propensity score within the predefined caliper (0.01) could be identified. The median follow-up time was 67 months for both groups (IQR, 43–97 months for APO and 37–100 months for IDC-NST). Post-matching evaluation confirmed balanced distributions across all covariates, including age, sex, race, tumor grade, T/N stage, ER/PR/HER2 status, and treatment modalities (surgery, chemotherapy, radiotherapy), with standardized mean differences <0.1 (Table 1). This confirmed the comparability of the matched cohorts and enabled valid assessment of outcomes based on histological subtype.

Survival analysis after PSM

Post-PSM KM analysis showed no significant difference in OS between APO and IDC-NST (P=0.21, Figure 2A). For BCSS, 22 cases (0.3%) were excluded due to unknown causes of death. APO patients demonstrated significantly better BCSS compared to IDC-NST (P=0.08, Figure 2B). After matching using 1:1 and 1:4 matching ratios, consistent survival results were obtained. KM analysis demonstrated that APO was not associated with OS but was associated with better BCSS (Figures S1,S2), confirming the robustness of our findings.

Figure 2 KM curves of all patients with APO vs. IDC-NST after PSM. (A) No significant difference in OS between APO and IDC-NST (P=0.21). (B) However, APO patients demonstrated significantly better BCSS compared to IDC-NST (P=0.008). APO, apocrine differentiation; BCSS, breast cancer-specific survival; IDC-NST, invasive ductal carcinoma of no special type; KM, Kaplan-Meier; OS, overall survival; PSM, propensity score matching.

Multivariate Cox regression in the matched cohort identified histological subtype, age, T stage, ER/HER2 status, and receipt of surgery, chemotherapy, or radiotherapy as independent predictors of OS and BCSS (Table 3). APO histology remained a favorable prognostic factor for both OS (HR =0.861, 95% CI: 0.746–0.993, P=0.04) and BCSS (HR =0.672, 95% CI: 0.538–0.839, P<0.001). Other favorable predictors included early T stage, ER positivity (OS: HR =0.733, 95% CI: 0.608–0.885, P=0.001; BCSS: HR =0.641, 95% CI: 0.484–0.850, P=0.002), HER2 positivity (OS: HR =0.817, 95% CI: 0.721–0.925, P=0.001; BCSS: HR =0.637, 95% CI: 0.531–0.765, P<0.001), and receipt of standard treatments. In contrast, age ≥60 years was independently associated with worse outcomes for both OS (HR =2.428, 95% CI: 2.106–2.799, P<0.001) and BCSS (HR = 1.694, 95% CI: 1.417–2.025, P<0.001).

Subgroup analysis after PSM

Among the matched cohort, 3,306 patients (51.9%) had TNBC, including 561 APO and 2,745 IDC-NST cases. Baseline characteristics were well balanced (Table S1). KM analysis showed significantly improved BCSS for APO patients compared to IDC-NST (P<0.05), though OS did not differ significantly (Figure 3). Multivariate analysis confirmed APO as an independent protective factor for BCSS (HR =0.72, 95% CI: 0.61–0.85, P<0.001), but not for OS (HR =0.87, 95% CI: 0.72–1.05, P=0.15) (Tables 4,5).

Figure 3 KM curves of TNBC patients with APO vs. IDC-NST after PSM. (A) No significant difference in OS between APO and IDC-NST (P=0.29). (B) However, APO patients demonstrated significantly better BCSS compared to IDC-NST (P<0.001). APO, apocrine differentiation; BCSS, breast cancer-specific survival; IDC-NST, invasive ductal carcinoma of no special type; KM, Kaplan-Meier; OS, overall survival; PSM, propensity score matching; TNBC, triple-negative breast cancer.

For the HER2-positive subgroup (n=1,197; 185 APO vs. 1,009 IDC-NST) and luminal subtypes (n=1,361; 319 APO vs. 1,556 IDC-NST), no significant differences in OS or BCSS were observed (P>0.05) (Figures 4,5), and APO histology did not emerge as an independent prognostic factor (HER2 positive: OS, HR =0.651, 95% CI: 0.364–1.166, P=0.15; BCSS, HR =0.419, 95% CI: 0.159–1.105, P=0.08; luminal, OS, HR =1.035, 95% CI: 0.790–1.355, P=0.80; BCSS, HR =1.164, 95% CI: 0.779–1.739, P=0.46) (Tables 6-9).

Figure 4 KM curves of HER2 positive patients with APO vs. IDC-NST after PSM. No significant difference in (A) OS and (B) BCSS between APO and IDC-NST. APO, apocrine differentiation; BCSS, breast cancer-specific survival; HER2, human epidermal growth factor receptor 2; IDC-NST, invasive ductal carcinoma of no special type; KM, Kaplan-Meier; OS, overall survival; PSM, propensity score matching.

Figure 5 KM curves of luminal patients with APO vs. IDC-NST after PSM. No significant difference in (A) OS and (B) BCSS between APO and IDC-NST. APO, apocrine differentiation; BCSS, breast cancer-specific survival; IDC-NST, invasive ductal carcinoma of no special type; KM, Kaplan-Meier; OS, overall survival; PSM, propensity score matching.

Discussion

Breast carcinoma with APO is histologically characterized by tumor cells exhibiting abundant eosinophilic cytoplasm, prominent nucleoli, and ≥90% apocrine morphology (6). Although APO accounts for fewer than 4% of all breast malignancies (2,7,8), its rarity poses challenges for conducting large-scale clinical studies. The SEER database, which encompasses approximately 30% of the U.S. population and includes standardized clinicopathological and survival data, offers a robust resource for investigating rare tumor types. Leveraging this comprehensive dataset, our study elucidated the distinct molecular features (e.g., HER2 positivity, ER/PR negativity) and prognostic differences between APO and IDC-NST, thereby overcoming the limitations of prior single-institution analyses.

Our findings demonstrated that APO exhibits more aggressive clinicopathological characteristics than IDC-NST. Specifically, APO tumors were associated with larger size, higher histological grade, more extensive nodal involvement, and a higher prevalence of HER2 positivity, while ER and PR expression rates were markedly lower. These features contributed to a greater proportion of APO tumors being classified as TNBC or HER2-enriched, which may explain the higher rates of chemotherapy utilization, despite comparable surgical treatment rates between the two histological subtypes. These observations are consistent with previous reports describing the hormonal receptor negativity and HER2 overexpression that typify APO (7,9,10). Although APO has generally been associated with older age at diagnosis in previous studies (4,5,10,11), Saridakis et al. reported the opposite trend, observing a younger median age in APO patients than in those with non-apocrine breast carcinomas (62 vs. 65 years) (10). Notably, their study also reported findings consistent with ours—larger tumors, higher grade, and greater nodal involvement in APO (10).

Paradoxically, triple-negative apocrine carcinoma (TNAC) demonstrates a more indolent clinical course relative to conventional TNBC. TNAC is often associated with lower tumor grade, smaller tumor size, and less nodal involvement (5,12-14). This seemingly contradictory behavior may reflect underlying molecular heterogeneity, particularly the frequent overexpression of the AR in APO, which is hypothesized to inhibit epithelial-mesenchymal transition and metastatic progression.

Our KM survival analysis revealed that patients with APO had significantly lower 5-year OS than those with IDC-NST, although no significant difference in 5-year BCSS was observed. Importantly, multivariable Cox regression analysis identified APO as an independent favorable prognostic factor for BCSS but not for OS. To further control for confounding effects due to imbalanced baseline characteristics, we applied PSM at a 1:5 ratio (15), which yielded well-matched groups. Post-matching analysis demonstrated that, in the entire cohort, the KM curves showed comparable OS between the APO and IDC-NST groups, and significantly better BCSS in the APO group. Notably, although the unadjusted analysis did not reach statistical significance for OS, multivariate Cox regression subsequently identified APO as an independent favorable prognostic factor for OS, as well as for BCSS. This apparent discrepancy likely reflects a suppression effect: the protective effect of APO was partially confounded by other prognostic variables (such as age, tumor size, nodal status, and grade) that were not perfectly balanced even after PSM. After adjusting for these covariates, the independent favorable impact of APO became apparent. Other favorable predictors included younger age (<60 years), lower N stage, ER/HER2 positivity, and receipt of multimodal therapy (surgery, chemotherapy, radiotherapy).

Previous studies have reported inconsistent findings regarding the prognostic significance of APO. Zhang et al. observed inferior OS and BCSS for APO relative to IDC (4), whereas Tanaka et al. reported no difference (9). In contrast, most recent studies suggest a survival advantage in APO (5,8,16). These discrepancies may be attributed to variations in diagnostic definitions (histological vs. molecular), unmeasured confounding variables, and the limited statistical power of small sample sizes.

To explore the impact of molecular subtype on prognosis, we conducted stratified analyses. In TNBC, APO was associated with significantly improved BCSS compared to IDC-NST, although OS did not differ. Multivariate analysis confirmed APO as an independent favorable prognostic factor for BCSS in TNBC (HR =0.42, 95% CI: 0.27–0.64), in line with previous SEER-based studies demonstrating superior survival outcomes for TNAC (5). Despite an inferior response to neoadjuvant chemotherapy, TNAC has been shown to have better long-term survival than non-apocrine TNBC, potentially due to its lower proliferative index and distinct molecular signature, as suggested by Hu et al. (16).

Conversely, in HER2 positive and luminal subtypes, we observed no significant differences in OS or BCSS between APO and IDC-NST, even in multivariate models. This contradicts earlier studies that linked luminal APO with larger tumor size, higher grade, and worse prognosis (10,15). The discrepancy may reflect residual confounding in earlier unmatched cohorts or variability in subtype classification criteria.

This study has several limitations. First, the lack of data on targeted therapies and neoadjuvant treatments prevented evaluation of their impact on survival outcomes, particularly for HER2-directed agents or PARP inhibitors. Second, although PSM minimized baseline confounding, unmeasured variables such as comorbidities, genomic alterations, or treatment compliance could still introduce residual bias. Third, we acknowledge that our subtype classification was based on immunohistochemistry (IHC) surrogates (ER/PR/HER2) rather than gene expression profiling (e.g., PAM50). Therefore, it may not fully capture the molecular heterogeneity of breast cancer, and future studies using true molecular subtyping are warranted.

Conclusions

In summary, APO exhibits more aggressive clinicopathological features than IDC-NST, yet is associated with better BCSS and comparable OS following adjustment for baseline characteristics. Multivariate analysis identified APO as an independent protective factor for both BCSS and OS. These prognostic advantages were most pronounced in TNBC subtype. Our findings highlight the biological distinctiveness of APO and underscore the importance of subtype-specific therapeutic strategies in its clinical management.

Acknowledgments

The authors thank all SEER program staff for their dedication in facilitating this retrospective study.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0622/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0622/prf

Funding: This work was supported by the Chutian Talent Project (No. CTYC002).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-0622/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Vranic S, Gatalica Z. An Update on the Molecular and Clinical Characteristics of Apocrine Carcinoma of the Breast. Clin Breast Cancer 2022;22:e576-85. [Crossref] [PubMed]
3. Nishida H, Kato A, Kaimori R, et al. Relationship between androgen receptor and androgen receptor-related protein expression in breast cancers focusing on morphologically identified carcinoma with apocrine differentiation. Sci Rep 2025;15:2892. [Crossref] [PubMed]
4. Zhang N, Zhang H, Chen T, et al. Dose invasive apocrine adenocarcinoma has worse prognosis than invasive ductal carcinoma of breast: evidence from SEER database. Oncotarget 2017;8:24579-92. [Crossref] [PubMed]
5. Wu W, Wu M, Peng G, et al. Prognosis in triple-negative apocrine carcinomas of the breast: A population-based study. Cancer Med 2019;8:7523-31. [Crossref] [PubMed]
6. Tan PH, Ellis I, Allison K, et al. The 2019 World Health Organization classification of tumours of the breast. Histopathology 2020;77:181-5. [Crossref] [PubMed]
7. Agarwal C, Pujani M, Sharma N, et al. Apocrine carcinoma of breast: A rare entity posing cytological challenge. Diagn Cytopathol 2017;45:1156-8. [Crossref] [PubMed]
8. Skenderi F, Alahmad MAM, Tahirovic E, et al. HER2-positive apocrine carcinoma of the breast: a population-based analysis of treatment and outcome. Breast Cancer Res Treat 2022;193:523-33. [Crossref] [PubMed]
9. Tanaka K, Imoto S, Wada N, et al. Invasive apocrine carcinoma of the breast: clinicopathologic features of 57 patients. Breast J 2008;14:164-8. [Crossref] [PubMed]
10. Saridakis A, Berger ER, Harigopal M, et al. Apocrine Breast Cancer: Unique Features of a Predominantly Triple-Negative Breast Cancer. Ann Surg Oncol 2021;28:5610-6. [Crossref] [PubMed]
11. Sasahara M, Matsui A, Ichimura Y, et al. Overexpression of androgen receptor and forkhead-box A1 protein in apocrine breast carcinoma. Anticancer Res 2014;34:1261-7.
12. Zhao S, Ma D, Xiao Y, et al. Clinicopathologic features and prognoses of different histologic types of triple-negative breast cancer: A large population-based analysis. Eur J Surg Oncol 2018;44:420-8. [Crossref] [PubMed]
13. Mills MN, Yang GQ, Oliver DE, et al. Histologic heterogeneity of triple negative breast cancer: A National Cancer Centre Database analysis. Eur J Cancer 2018;98:48-58. [Crossref] [PubMed]
14. Liao HY, Zhang WW, Sun JY, et al. The Clinicopathological Features and Survival Outcomes of Different Histological Subtypes in Triple-negative Breast Cancer. J Cancer 2018;9:296-303. [Crossref] [PubMed]
15. Han Y, Wang J, Xu B. Clinicopathological characteristics and prognosis of breast cancer with special histological types: A surveillance, epidemiology, and end results database analysis. Breast 2020;54:114-20. [Crossref] [PubMed]
16. Hu T, Liu Y, Wu J, et al. Triple-Negative Apocrine Breast Carcinoma Has Better Prognosis despite Poor Response to Neoadjuvant Chemotherapy. J Clin Med 2022;11:1607. [Crossref] [PubMed]