Survival outcomes by molecular subtypes based on receptor status in medullary breast carcinoma: a SEER database analysis.

Introduction
Breast cancer persists as the most frequently diagnosed cancer in women globally, with approximately 310,000 new cases anticipated in the United States in 2024 (1). Molecular classification has categorized breast tumors into several subtypes—luminal A, luminal B, human epidermal growth factor receptor 2 (HER2)-enriched, and triple-negative/basal-like—each characterized by distinct gene expression patterns and receptor status, including estrogen receptor (ER), progesterone receptor (PR), and HER2. These subtypes display considerable differences in clinical behavior; for instance, luminal A tumors are associated with favorable survival, whereas triple-negative and HER2-enriched subtypes are linked to poorer prognosis and earlier recurrence (2-4).
Invasive breast carcinoma of no special type (IBC-NST), previously termed invasive ductal carcinoma, represents the most prevalent histological subtype of breast cancer, accounting for approximately 50% to 80% of all cases. However, Medullary breast carcinoma (MBC) is a rare histological variant, accounting for 3–6% of all invasive breast cancers (5). MBC typically presents as a high-grade carcinoma with a syncytial growth pattern, well-circumscribed pushing borders, and a prominent lymphoplasmacytic stromal infiltrate (6). Despite aggressive histopathological features—such as high nuclear grade, elevated Ki-67 indices, and a frequent triple-negative phenotype (ER−, PR−, HER2−)—some studies suggest that MBC is associated with more favorable survival outcomes compared to IBC-NST (7,8). In contrast, other investigations have reported no significant prognostic differences between the two entities (9).
Propensity score matching (PSM) is a statistical matching technique in which each case is matched to one or more controls based on their propensity scores, aiming to minimize the influence of confounding factors. To evaluate survival outcomes in MBC, we extracted a large patient cohort from the Surveillance, Epidemiology, and End Results (SEER) database and applied PSM to minimize confounding biases between MBC and IBC-NST. This approach enabled a systematic comparison of overall survival (OS) and breast cancer-specific survival (BCSS) between the two groups. The prognosis and treatment decisions for breast cancer are highly dependent on molecular subtypes (such as triple-negative, luminal and HER2-positive), and MBC itself also exhibits significant molecular subtype heterogeneity. Departing from previous research that often treats MBC as a uniform entity, this study aims to specifically investigate the prognostic differences among the different molecular subtypes based on receptor status of MBC itself. The findings aim to deliver evidence-based insights to support precise treatment and improve risk stratification for this distinct breast cancer subtype. We present this article in accordance with the STROBE reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2666/rc).

Methods

Patients
Data were obtained from the SEER Program, a population-based registry encompassing approximately 26.5% of the U.S. population and providing detailed information on cancer incidence, treatment, and survival. Using SEER Stat software (version 9.0.41), we identified patients diagnosed with primary breast cancer based on the following histology codes: 8510/3 (medullary carcinoma, not otherwise specified), 8512/3 (medullary carcinoma with lymphoid stroma), 8513/3 (atypical medullary carcinoma), and 8500/3 (invasive ductal carcinoma, not otherwise specified). Included patients were required to have definitive ER, PR, and HER2 status documentation, complete survival follow-up data, and a confirmed pathological diagnosis. Cases with American Joint Committee on Cancer (AJCC) stage IV disease or distant metastasis at diagnosis were excluded. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Statistical analysis
When comparing demographic and clinical characteristics between the two histological groups, continuous variables (e.g., age) were analyzed using Student’s t-test, while categorical variables were examined with Pearson’s Chi-squared test or Fisher’s exact test, as appropriate. Survival curves were constructed using the Kaplan-Meier method, and differences in BCSS and OS were evaluated with the log-rank test. Univariate and multivariate Cox proportional hazards regression models were used to identify independent prognostic factors. A two-sided P value <0.05 was considered statistically significant, and hazard ratios (HRs) are presented with 95% confidence intervals (CIs).
To address sample size imbalance and minimize selection bias, PSM was performed using a 1:5 nearest neighbor matching algorithm with a caliper width of 0.01. Matching covariates included race, age, tumor grade, tumor (T) stage and node (N) stage and ER/PR/HER2 status. This process ensured balanced baseline characteristics between the MBC and IBC-NST groups, providing a robust foundation for subsequent comparative analyses. All statistical analyses were conducted in R software (version 4.4.2), utilizing packages including “dplyr”, “stats”, “survival”, “survminer”, “MatchIt”, “ggplot2”, and “cobalt”.

Results
Based on the predefined inclusion and exclusion criteria, a total of 555,988 patients were identified from the SEER database, comprising 555,025 cases of IBC-NST and 963 cases of MBC. The median follow-up time was 99 months for the MBC group and 55 months for the IBC-NST group.
As summarized in Table 1, patients with MBC were significantly younger than those with IBC-NST (mean age: 54.2 vs. 61.4 years, P<0.001) and were more frequently of Black race (24.1% vs. 10.8%, P<0.001). Tumors in the MBC group were characterized by markedly higher histologic grade, with approximately 80% classified as grade 3. A larger proportion of MBC cases presented at T2 stage (47.7% vs. 29.4%, P<0.001), while lymph node involvement was less common (N0 stage: 78.2% vs. 71.0%, P<0.001). Furthermore, MBC tumors showed significantly higher rates of ER negativity (65.2% vs. 17.9%, P<0.001), PR negativity (82.2% vs. 28.1%, P<0.001), and HER2 negativity (90.0% vs. 84.1%, P<0.001). Patients with MBC more frequently received chemotherapy (75.5% vs. 40.3%, P<0.001) and surgical intervention (97.9% vs. 94.3%, P<0.001). No significant difference was observed in gender distribution between the two groups. Radiotherapy data were not compared due to extensive missing values.
For BCSS analysis, 1,422 patients with unknown cause of death were excluded, as they constituted a relatively small proportion of the overall cohort. Survival analysis revealed that MBC patients had significantly better OS (10-year OS: 83.9% vs. 73.1%, P<0.001; Figure 1A) and BCSS (10-year BCSS: 93.4% vs. 88.1%, P<0.001; Figure 1B) compared to IBC-NST patients.
To minimize potential confounding, PSM was performed to balance key clinical variables -including age, T stage, N stage, and ER/PR/HER2 status—between the two groups. A total of 5,772 matched patients were included, consisting of 963 MBC and 4,810 IBC-NST cases (Table 1). After matching, the median follow-up was 99 months for the MBC group and 84 months for the IBC-NST group. Survival analyses continued to demonstrate significantly better OS (10-year OS: 84.0% vs. 73.3%, P<0.001; Figure 2A) and BCSS (10-year BCSS: 93.5% vs. 82.8%, P<0.001; Figure 2B) in the MBC group compared to the IBC-NST group. Furthermore, both univariate and multivariate regression analyses identified histological type as an independent prognostic factor in breast cancer, with MBC patients demonstrating superior BCSS and OS compared to those with IBC-NST (Table S1).
To further evaluate prognostic outcomes across molecular subtypes based on receptor status, MBC cases were classified into triple-negative breast cancer (TNBC; n=541, 55.8%), luminal (n=326, 33.9%), and HER2-positive (n=96, 10.0%) categories. The respective median follow-up durations were 99, 99.5, and 104 months. Log-rank tests showed no significant differences in OS (P=0.38; Figure 3A) or BCSS (P=0.65; Figure 3B) among the three molecular subtypes. Univariate and multivariate Cox regression analyses confirmed that molecular subtype was not an independent prognostic factor (Tables 2,3). Instead, age, T stage, N stage, and chemotherapy were identified as independent predictors of survival. Older age and advanced T or N stage were associated with worse outcomes, whereas chemotherapy was associated with significantly improved survival.

Discussion
MBC, also referred to as invasive breast cancer with medullary features, is a rare histological subtype. Due to its low incidence, previous studies on MBC have mostly been limited to small single-center series or analyses based on the SEER database. The SEER registry contains extensive breast cancer patient data, rendering it particularly valuable for investigating rare subtypes such as MBC. Accordingly, this study also leveraged the SEER database to assess the prognostic profile of MBC.
Our results demonstrate that MBC patients present with more aggressive clinicopathological features compared to those with IBC-NST. Specifically, MBC patients were younger at diagnosis, predominantly had grade 3 tumors, exhibited larger tumor size (more frequently classified as T2 stage), and showed a higher proportion of triple-negative receptor status. They also received chemotherapy more frequently, yet had lower lymph node involvement. These observations align with several prior studies, including those by Dai et al. (10) and Chen et al. (11). However, unlike our findings, Dai et al. reported lower rates of chemotherapy administration in MBC. It is worth noting that although both studies utilized the SEER database, the discrepancy may stem from differences in case selection criteria, with our study representing the largest MBC cohort extracted from SEER to date. Similarly, institutional studies by Budzik et al. (12) and Zangouri et al. (13) also reported younger age, higher tumor grade, predominant triple-negative phenotype, and less frequent lymph node metastasis in MBC. Inconsistent findings have been reported regarding tumor size: whereas Budzik et al. (12) noted a higher frequency of T2 tumors in MBC, Zangouri et al. found no significant difference in tumor size between MBC and IBC-NST. In contrast, Flucke et al. (14) reported that MBC tended to be smaller than IBC-NST tumors.
Our study demonstrated that in the overall cohort, patients with MBC had significantly better OS and BCSS than those with IBC-NST. This finding remained consistent after PSM was applied to minimize potential confounders. These results are in agreement with several previous reports (10-13). For example, Huober et al. (7) observed superior 14-year distant recurrence-free interval (DRFI) and OS in MBC patients compared to IBC-NST (DRFI: 76% vs. 64%, P=0.0005; OS: 66% vs. 57%, P=0.03). Similarly, a retrospective analysis by Cao et al. (8) reported significantly better 10-year recurrence-free survival (RFS) and OS in the MBC group (RFS: 74% vs. 64%, P=0.032; OS: 91% vs. 81%, P=0.042).
However, some studies have reached conflicting conclusions. For instance, a SEER-based analysis by Wang et al. (15) suggested that IBC-NST was associated with better BCSS than MBC, although no significant difference was observed in OS. This discrepancy may be partly explained by their relatively short follow-up duration. Moreover, Park et al. (9) found no significant differences in 5-year disease-free survival (DFS) (88.0% vs. 89.2%, P=0.920) or OS (93.4% vs. 94.4%, P=0.503) between MBC and IBC-NST. The divergence from other studies may be attributed to the limited sample size of the MBC cohort (n=52) in their analysis, which could reduce statistical power.
The molecular classification of breast cancer carries substantial prognostic and therapeutic relevance. Previous evidence indicates that TNBC is generally associated with poorer outcomes compared to other subtypes (2,3). To explore prognostic factors specific to MBC, this study categorized patients into three molecular subtypes based on receptor status: TNBC (ER−, PR−, HER2−), luminal (ER+ and/or PR+, HER2−), and HER2-overexpressing (HER2+), and subsequently compared their survival outcomes. Results showed no statistically significant differences in OS or BCSS among these molecular subtypes. Cox regression analysis further confirmed that molecular subtype and individual ER/PR/HER2 status were not significant prognostic factors in MBC. This key finding suggests that the characteristically favorable prognosis of MBC is a pervasive trait, likely rooted in its distinctive histopathological architecture and tumor immune microenvironment, rather than being differentially mediated by hormonal or HER2 signaling pathways. Instead, age at diagnosis, T stage, lymph node metastasis, and chemotherapy were identified as independent predictors of survival. Among these, older age, advanced T stage, and lymph node involvement were associated with worse outcomes. whereas receipt of chemotherapy was independently associated with improved BCSS and OS. It should be noted that this conclusion may be influenced by confounding by indication, for example, patients who received chemotherapy may have had a better performance status and younger age.
Multiple prior studies have similarly reported older age, higher T stage, and lymph node metastasis as adverse prognostic factors in MBC (11,15). However, the impact of chemotherapy on survival remains inconsistent across the literature. For example, Aihara et al. (16) demonstrated that chemotherapy reduced recurrence and metastasis risk and improved OS. In contrast, Dai et al. (10) observed no significant association between chemotherapy and prognosis in MBC. Meanwhile, Chen et al. (17) reported that chemotherapy improved OS in low-risk but not high-risk MBC patients. Similarly, conflicting evidence exists regarding the relationship between hormone receptor status and MBC outcomes. Dai et al. (10) and Qin et al. (18) found no link between ER/PR status and prognosis, whereas Martinez et al. (19) suggested that ER− and PR+ status were associated with better OS. Aksoy et al. (20) reported no prognostic impact of ER status, but identified PR− as a factor for improved RFS and OS. Due to the low incidence of HER2-positive MBC, few studies have evaluated its prognostic role. Only Chen et al. (11) previously indicated that HER2 status does not influence MBC survival, which aligns with our findings. To the best of our knowledge, no prior study has directly compared prognostic differences among molecular subtypes in MBC. This study is the first to demonstrate that molecular subtype does not independently influence prognosis in patients with MBC.
There are several limitations in this study. First, the SEER database lacks detailed information on endocrine therapy and anti-HER2 targeted therapy, precluding assessment of their influence on patient prognosis. Second, specific biomarker data such as Ki-67 expression levels and quantitative ER/PR values are not recorded in SEER, which limits further stratification between luminal A and luminal B subtypes. Third, approximately 40% of included patients had incomplete or missing radiotherapy records, hindering a robust evaluation of the impact of radiotherapy on survival outcomes. Fourth, due to the limitations of the SEER database, genomic/transcriptomic-level heterogeneity cannot be assessed. The molecular subtyping in this study was solely based on receptor expression status.

Conclusions
This study demonstrates that, compared to IBC-NST, MBC is characterized by younger age at diagnosis, higher tumor grade, larger tumor size (more frequently classified as T2 stage), and ER, PR, and HER2 negativity, yet a lower incidence of lymph node metastasis. Despite these aggressive clinicopathological features, patients with MBC exhibited significantly better OS and BCSS—an advantage that remained consistent after PSM. Key factors influencing prognosis in MBC included age at diagnosis, T stage, N stage, and chemotherapy. Older age and advanced T or N stage were associated with poorer survival outcomes, whereas chemotherapy was associated with better outcomes. In contrast, ER, PR, and HER2 status were not significantly correlated with prognosis in MBC, and no notable survival differences were observed across molecular subtypes.

Supplementary
The article’s supplementary files as