Abstract
Background
Triple-negative breast cancer (TNBC) is a biologically heterogeneous, aggressive disease where immunohistochemistry (IHC)-based algorithms, defining luminal androgen receptor (LAR), immunomodulatory (IM), basal-like immunosuppressed (BLIS), mesenchymal (MES), and unclassifiable (UC) subtypes, offer a pragmatic alternative to transcriptomics. This study evaluated clinical trajectories and survival outcomes associated with IHC-based molecular subtypes and stromal tumor-infiltrating lymphocytes (sTILs) across distinct subgroups.
Methods
A retrospective analysis of 289 women with TNBC treated with chemotherapy-only regimens analyzed molecular subtypes across four clinical groups: non-recurrent, recurrent, progression during neoadjuvant chemotherapy, and de novo metastatic disease. IHC-based subclassification included androgen receptor (AR), CD8, FOXC1, and DCLK1 biomarkers. Overall survival (OS) and disease-free survival (DFS) were estimated using the Kaplan-Meier method, and multivariable Cox proportional hazards models.
Results
The cohort included non-recurrent (43.9%), recurrent (26.3%), neoadjuvant progressors (11.5%), and de novo stage IV (18.3%) cases. IM subtype (16.3%) was enriched in non-recurrent cases with the highest median sTILs levels (20.0%). BLIS (26.3%) and UC (39.4%) subtypes predominated in recurrent and metastatic disease, while MES subtype (9.0%) predominated in neoadjuvant chemotherapy progressors. In early-stage disease, sTILs ≥10% independently predicted superior OS (HR 0.63, 95% CI, 0.40-1.0; P = .049), showing no prognostic role in the de novo stage IV group. Molecular subtypes lost independent significance in multivariable models.
Conclusions
Clinicopathological factors and sTILs remain primary prognostic determinants in TNBC. Although IHC-based molecular subtypes were not independent survival predictors, their differential clinical distribution validates their biological relevance to identify therapeutic vulnerabilities, and to guide precision medicine and de-escalation strategies.
Keywords: triple-negative breast cancer, molecular subtypes, immunohistochemistry, stromal TILs, prognosis
Implications for Practice.
The integration of immunohistochemistry (IHC)-based molecular subtyping and stromal tumor-infiltrating lymphocytes (sTILs) into clinical practice provides a pragmatic, cost-effective framework for refining risk stratification in triple-negative breast cancer (TNBC). Clinicians should prioritize sTILs, as levels ≥10% predict superior overall survival. Patients with immunomodulator subtype and high sTILs have favorable prognosis and may benefit from treatment de-escalation trials to minimize toxicity and costs. Conversely, BLIS, UC, and MES subtypes in patients experiencing early recurrence or neoadjuvant progression identify high-risk populations that may require intensification or novel subtype-directed therapies. Using available IHC biomarkers (AR, CD8, FOXC1, DCLK1) supports precise, personalized TNBC management.
Introduction
Triple-negative breast cancer (TNBC) is an aggressive form of breast cancer (BC) associated with early recurrence, increased risk of visceral metastases, and variable survival outcomes, ranging from 5-year overall survival (OS) rates of 75-86% in early and locally advanced stages to a critical 11% in stage IV disease.1–4 Despite its generally unfavorable prognosis, TNBC is a biologically heterogeneous disease. A subset of patients with TNBC demonstrates marked sensitivity to systemic therapy, achieving high pathological complete response (pCR) rates and experiencing durable long-term survival.3,5 Thus, it is clear that not all cases of TNBC are equally aggressive, and determining the drivers behind these differences is critical for the future of escalation and de-escalation strategies.
Transcriptomic profiling studies classified TNBC into distinct molecular subtypes, highlighting heterogeneity in immune activation, androgen receptor (AR) signaling, basal-like features, and mesenchymal differentiation.6 Subsequent refinements have converged on clinically relevant categories such as luminal androgen receptor (LAR), immunomodulatory (IM), basal-like immunosuppressed (BLIS), mesenchymal (MES), and unclassifiable (UC) subtypes, each associated with distinct biological characteristics, prognoses, and potential therapeutic vulnerabilities.7,8
Although transcriptomic profiling remains the reference standard for molecular subtyping, its routine clinical implementation is limited by cost, availability, and infrastructure constraints. Consequently, immunohistochemistry (IHC)-based algorithms have been developed as pragmatic surrogates, enabling classification using widely available biomarkers such as AR, CD8, FOXC1, and DCLK1 to classify TNBC tumors into the aforementioned groups.8 Using this IHC-based approach, the IM subtype has been associated with a favorable prognosis, reflecting an immune-enriched tumor microenvironment, whereas BLIS tumors are characterized by immune suppression and have been linked to poorer outcomes. LAR tumors often display an intermediate prognosis and relative chemoresistance, while MES and UC subtypes show variable clinical behavior, underscoring the biological heterogeneity of TNBC.8,9
These IHC subtypes are closely intertwined with the tumor immune microenvironment. Tumor-infiltrating lymphocytes (TILs), particularly stromal TILs (sTILs), have emerged as robust predictors of chemotherapy sensitivity and survival.10 In the current treatment landscape, sTILs serve as a dynamic reflection of the tumor-host interface; higher levels are not only prognostic but are powerful indicators of pCR in the neoadjuvant setting, especially within “immune-hot” subtypes like IM. The success of pembrolizumab3 in early-stage TNBC has further cemented this synergy, as the clinical efficacy of immune checkpoint inhibition is fundamentally predicated on the preexisting immune environment that these biomarkers characterize.
Unfortunately, the distribution of IHC-based molecular subtypes and immune infiltration patterns across the spectrum of clinical presentations, from durable remission recurrence, primary resistance during neoadjuvant therapy, or de novo metastatic presentation, remains largely uncharacterized. Therefore, the objective of this study was to describe the clinical and histopathological characteristics of a cohort of women with TNBC treated with chemotherapy-only regimens and to evaluate the association between IHC-based molecular subtypes (as determined by AR, CD8, FOXC1, and DCLK1), sTILs, and clinical outcomes across different stages and patterns of disease presentation.
Methods and materials
Patient selection
Women aged 18 years or older with a diagnosis of TNBC were retrospectively identified from two tertiary care centers (Instituto Nacional de Cancerología in Mexico City and TecSalud in Monterrey, Mexico). Patients diagnosed between 2006 and 2021 were included, with clinical follow-up extending through 2023. TNBC was defined as the absence of estrogen receptor and progesterone receptor expression (<1%) and lack of HER2 overexpression or amplification according to contemporaneous ASCO/CAP guidelines. Exclusion criteria comprised synchronous or metachronous malignancies, bilateral BC, pregnancy-associated BC, incomplete clinical data, and insufficient, unavailable, or poor-quality formalin-fixed paraffin-embedded (FFPE) tissue for IHC analysis, as determined by a single senior pathologist specialized in breast pathology.
Clinical and histopathological data were extracted from electronic medical records and compiled into an institutional database under continuous supervision by board-certified oncologists. Variables collected at diagnosis included age, body mass index (BMI), menopausal status, tumor size, lymph node involvement, clinical stage, histological type, histologic grade, duration of follow-up, and clinical outcome.
Patients were categorized into four clinical groups according to their disease course: no evidence of recurrence, recurrent disease, progression during neoadjuvant chemotherapy, and de novo metastatic disease. Recurrent disease was defined as locoregional or distant relapse after completion of curative-intent therapy. Progression during neoadjuvant treatment was defined as radiologic or clinical disease progression according to treating physician assessment prior to surgery. De novo metastatic disease was defined as stage IV disease at diagnosis.
Crucially, defining the therapeutic context of this study, no patients in this cohort received neoadjuvant or adjuvant immune checkpoint inhibitors, including pembrolizumab, due to the historical treatment era spanned by the database from 2006 to 2021. While pembrolizumab established a new standard of care for early-stage high-risk TNBC following international trial data in 2020,11 eligible patients treated at our institutions during the 2020 and 2021 window received conventional chemotherapy alone. This restriction was dictated by contemporary institutional protocols, local regulatory approval timelines, and healthcare access limitations regarding immunotherapy agents during those final accrual years.
IHC and histopathological evaluation
Manual IHC was performed to determine molecular subtypes using pretreatment FFPE tumor tissue from cases with a confirmed diagnosis of TNBC and adequate tissue quality. Tissue sections (1 µm) were prepared by experienced histotechnologists and mounted on electrostatically charged slides.
The following primary antibodies were used: AR (Biocare, clone CM109A, 1:50 dilution), CD8 (Biocare, clone CRM311C, 1:25 dilution), FOXC1 (Abcam, clone EPR10685, 1:50 dilution), and DCLK1 (Abcam, clone EPR6085, 1:100 dilution). Appropriate external positive controls were included for each biomarker: prostate tissue for AR, appendix for CD8, fetal kidney for FOXC1, and fetal brain for DCLK1.
Slides were deparaffinized, rehydrated, and subjected to antigen retrieval using Target Retrieval Solution (Agilent Dako). Endogenous peroxidase activity was blocked prior to incubation with primary antibodies. After incubation with the primary antibodies, detection was performed using a MACH 4 detection system (Biocare), followed by visualization with diaminobenzidine chromogen. Slides were counterstained with hematoxylin, dehydrated, and coverslipped according to standard laboratory protocols.
Histopathological evaluation was performed using hematoxylin and eosin (H&E)-stained slides to confirm histological type and grade and to assess TILs, in accordance with the recommendations of the International TILs Working Group 2014.12 All slides were evaluated by a single senior pathologist specialized in breast pathology blinded to the patients’ baseline characteristics and clinical outcomes.
TNBC subclassification
TNBC cases were classified into five molecular subtypes using an IHC-based algorithm proposed by Zhao et al.8 Molecular biomarkers were assessed in a sequential order: AR (LAR) was evaluated first, followed by CD8 (IM), FOXC1 (BLIS), and DCLK1 (MES), with each subsequent biomarker assessed only if the preceding was negative (Figure 1). Positivity was defined as ≥10% positive tumor cells for AR, FOXC1, and DCLK1, and ≥20% for CD8. Tumor samples negative for all biomarkers were classified as UC.
Figure 1.
IHC-based diagnostic algorithm for TNBC molecular subtypes proposed by Zhao et al. Positivity was defined as ≥10% positive tumor cells for AR, FOXC1, and DCLK1, and ≥20% for CD8. Abbreviations: AR, androgen receptor; BLIS, basal-like immunosuppressed; IM, immunomodulatory; MES, mesenchymal; UC, unclassifiable.
Statistical analysis
Descriptive statistics were used to summarize demographic, clinical, and histopathological characteristics. Data distribution was assessed using the Kolmogorov-Smirnov test. Continuous variables were compared using the Student’s t test or the Mann-Whitney U test, as appropriate, and categorical variables were compared using the chi-square test or Fisher’s exact test.
OS in patients with early-stage (comprising non-recurrent, recurrent, and progression during neoadjuvant chemotherapy cohorts) and de novo stage IV disease, and disease-free survival (DFS) in patients with early-stage were estimated using the Kaplan-Meier method and compared between groups using the log-rank test. OS was defined as the time from diagnosis to death from any cause. DFS was calculated from the date of definitive surgery to the first evidence of locoregional recurrence, distant metastasis, a second primary invasive cancer, or death from any cause. Patients without an event were censored at the date of last clinical follow-up.
Univariate and multivariate Cox proportional hazards regression models were used to evaluate associations between clinicopathological variables, IHC-based molecular subtypes, sTILs, and survival outcomes. To evaluate the independent prognostic value of these factors within each distinct disease context, separate multivariate models were constructed. For patients with early-stage TNBC, the multivariate model for OS was adjusted for lymph node status, clinical stage, sTILs levels, molecular subtypes, and treatment era (up to 2014 vs 2015 onward). For the evaluation of DFS, the model included the same covariates. Conversely, for patients with de novo metastatic TNBC, the multivariate model for OS was adjusted for lymph node status, sTILs levels, molecular subtypes, and treatment era. To ensure model stability, variables with strong collinearity were excluded.
All statistical tests were two-sided, and statistical significance was defined as a P-value <.05. Statistical analyses were performed using R software version 4.5.2 (RStudio, version 2024.12.0 + 467).
Ethics statement
This study was conducted following the Declaration of Helsinki principles and approved by Research and Ethics in Research Institutional Boards (Project ID: P000550-SM-CMTN-CI/CEIC-CR001). All data were anonymized, and patients provided informed consent at the time of diagnosis allowing the use of their clinical data and tumor samples for research purposes.
Results
General characteristics
A total of 1247 patients with TNBC were screened. The comprehensive selection process with respective patient counts is illustrated in Figure 2.
Figure 2.
Flowchart of patient selection and clinical grouping.
A total of 289 women with TNBC were included in the final analysis. Baseline demographic, clinical, and histopathological characteristics are summarized in Table 1. According to disease course, 127 patients (43.9%) had no evidence of recurrence, 76 (26.3%) developed recurrent disease, 33 (11.5%) experienced progression during neoadjuvant chemotherapy, and 53 (18.3%) presented with de novo stage IV disease. In the non-recurrent group, the median follow-up reached 140 months, with a minimum observation period of 55 months, confirming the clinical robustness and long-term stability of this cohort’s disease-free status. Conversely, for the group of patients who developed recurrence, the median follow-up was 32 months (minimum observation of 10 months), and 98.7% of recurrence events occurred in the first 5 years of follow-up.
Table 1.
Demographic, clinical, and histopathological characteristics at baseline (n = 289).
| Variable | Global (n = 289) | No recurrence (n = 127) | Recurrence (n = 76) | Progression (n = 33) | De novo stage IV (n = 53) | P-value |
|---|---|---|---|---|---|---|
| Age, years [IQR] | 48.43 [40.77, 56.33] | 48.45 [41.31, 57.20] | 47.64 [40.57, 54.94] | 46.30 [40.28, 53.49] | 49.24 [40.49, 55.96] | .830 |
| BMI, kg/m2 [IQR] | 26.91 [24.38, 30.71] | 27.19 [24.55, 31.02] | 27.49 [23.94, 30.68] | 26.17 [23.90, 30.39] | 26.50 [24.44, 30.26] | .745 |
| sTILs, % [IQR] | 5.0 [3.0, 15.0] | 10.0 [3.0, 10.0] | 5.0 [3.0, 10.0] | 5.0 [2.0, 10.0] | 5.0 [3.0, 10.0] | .021 |
| BMI (categories) | ||||||
| Low weight | 3 (1.0) | 0 (0.0) | 2 (2.6) | 0 (0.0) | 1 (1.9) | .800 |
| Normal | 84 (29.1) | 35 (27.6) | 22 (28.9) | 12 (36.4) | 15 (28.3) | |
| Overweight | 116 (40.1) | 55 (43.3) | 29 (38.2) | 12 (36.4) | 20 (37.7) | |
| Obesity | 86 (29.8) | 37 (29.1) | 23 (30.3) | 9 (27.3) | 17 (32.1) | |
| Menopause status | ||||||
| Post-menopause | 149 (51.6) | 67 (52.8) | 38 (50.0) | 14 (42.4) | 30 (56.6) | .617 |
| Pre-menopause | 140 (48.4) | 60 (47.2) | 38 (50.0) | 19 (57.6) | 23 (43.4) | |
| Tumor size | ||||||
| T1 | 32 (11.1) | 20 (15.7) | 11 (14.5) | 1 (3.0) | 0 (0.0) | <.001 |
| T2 | 110 (38.1) | 67 (52.8) | 30 (39.5) | 4 (12.1) | 9 (17.0) | |
| T3 | 57 (19.7) | 28 (22.0) | 15 (19.7) | 8 (24.2) | 6 (11.3) | |
| T4 | 90 (31.1) | 12 (9.4) | 20 (26.3) | 20 (60.6) | 38 (71.7) | |
| Lymph nodes | ||||||
| N0 | 68 (23.5) | 53 (41.7) | 11 (14.5) | 1 (3.0) | 3 (5.7) | <.001 |
| N1 | 87 (30.1) | 48 (37.8) | 25 (32.9) | 8 (24.2) | 6 (11.3) | |
| N2 | 92 (31.8) | 23 (18.1) | 23 (30.3) | 19 (57.6) | 27 (50.9) | |
| N3 | 42 (14.5) | 3 (2.4) | 17 (22.4) | 5 (15.2) | 17 (32.1) | |
| Clinical stage | ||||||
| I | 12 (4.2) | 12 (9.4) | 0 (0.0) | 0 (0.0) | 0 (0.0) | <.001 |
| II | 94 (32.5) | 68 (53.5) | 25 (32.9) | 1 (3.0) | 0 (0.0) | |
| III | 130 (45.0) | 47 (37.0) | 51 (67.1) | 32 (97.0) | 0 (0.0) | |
| IV | 53 (18.3) | 0 (0.0) | 0 (0.0) | 0 (0.0) | 53 (100.0) | |
| Histologic type | ||||||
| Ductal | 265 (91.7) | 115 (90.6) | 71 (93.4) | 30 (90.9) | 49 (92.5) | .933 |
| Lobular | 12 (4.2) | 5 (3.9) | 4 (5.3) | 1 (3.0) | 2 (3.8) | |
| Mixed | 1 (0.3) | 1 (0.8) | 0 (0.0) | 0 (0.0) | 0 (0.0) | |
| Other | 11 (3.8) | 6 (4.7) | 1 (1.3) | 2 (6.1) | 2 (3.8) | |
| Histologic grade | ||||||
| Grade 1 | 4 (1.4) | 2 (1.6) | 1 (1.3) | 0 (0.0) | 1 (1.9) | .733 |
| Grade 2 | 42 (14.5) | 19 (15.0) | 11 (14.5) | 2 (6.1) | 10 (18.9) | |
| Grade 3 | 227 (78.5) | 100 (78.7) | 60 (78.9) | 30 (90.9) | 37 (69.8) | |
| ND | 16 (5.5) | 6 (4.7) | 4 (5.3) | 1 (3.0) | 5 (9.4) | |
| Molecular subtype | .038 | |||||
| IM | 47 (16.3) | 29 (22.8) | 11 (14.5) | 6 (18.2) | 1 (1.9) | |
| LAR | 26 (9.0) | 11 (8.7) | 7 (9.2) | 1 (3.0) | 7 (13.2) | |
| BLIS | 26 (9.0) | 32 (25.2) | 21 (27.6) | 8 (24.2) | 15 (28.3) | |
| MES | 76 (26.3) | 10 (7.9) | 6 (7.9) | 7 (21.2) | 3 (5.7) | |
| UC | 114 (39.4) | 45 (35.4) | 31 (40.8) | 11 (33.3) | 27 (50.9) | |
| Treatment regimens (no metastatic disease; n = 236) | .411 | |||||
| Anthracycline + Taxane | 198 (83.9) | 106 (83.5) | 66 (86.8) | 26 (78.8) | ||
| Anthracycline-based | 11 (4.7) | 6 (4.7) | 1 (1.3) | 4 (12.1) | ||
| Taxane-based | 20 (8.5) | 11 (8.7) | 7 (9.2) | 2 (6.1) | ||
| Other | 7 (2.9) | 4 (3.1) | 2 (2.6) | 1 (3.0) | ||
Characteristics by clinical group are also shown.
Abbreviations: IQR, interquartile range; BMI, body mass index; IM, immunomodulator; LAR, luminal androgen receptors; BLIS, basal-like immunosuppresed; MES, mesenchymal; UC, unclassifiable.
The median age at diagnosis for the entire cohort was 48.4 years (interquartile range [IQR] 40.8-56.3), and the median BMI was 26.9 kg/m2 (IQR 24.4-30.7), with overweight being the most frequent BMI category. A significant proportion of patients presented with locally advanced disease (stage III disease in 45.0%). Ductal carcinoma (91.7%) and histologic grade 3 tumors (78.5%) predominated. Clinicopathological characteristics differed significantly across clinical groups with respect to tumor size and lymph node status (Table 1).
Treatment regimens
Among patients without recurrence, the most frequent treatment strategy was adjuvant chemotherapy (n = 70, 55.6%), followed by neoadjuvant therapy (n = 50, 39.7%). Most patients in the adjuvant chemotherapy group had T1-T2 tumors (n = 59), while a minority had tumors that would otherwise have a stronger indication for neoadjuvant therapy (ie, T3-T4 tumors; n = 11). Conversely, in patients with recurrence, neoadjuvant therapy was the most common approach (n = 34, 44.7%), followed by adjuvant therapy (n = 27, 35.5%).
Sequential anthracycline- and taxane-based regimens were the predominant therapeutic strategy across all clinical subgroups, administered to 83.5% (n = 106) of patients without recurrence, 86.8% (n = 66) of those with recurrent disease, and 78.8% (n = 26) of patients exhibiting disease progression. Taxane monotherapy was utilized in a minority of cases, ranging from 6.1% in the progression group to 9.2% in the recurrence cohort. Anthracycline-only regimens were the least frequent, notably accounting for only 1.3% (n = 1) of patients with recurrence and 12.1% (n = 4) of those with primary disease progression. Regarding platinum-based therapy, among patients without recurrence, 62 (49.2%) received platinum salts. In the recurrence and the progression group, 37 (48.7%) and 20 (60.6%) patients received them, respectively.
In patients with metastatic disease, regimens included a heterogeneous variety of drug families. Taxane-based therapy was the most frequently used (n = 40, 75.5%), followed by anthracyclines (n = 38, 71.7%) and platinum salts (n = 32, 60.4%). Drugs such as capecitabine and gemcitabine were used less frequently (n = 18, 34.0%), as were alkylating agents including cyclophosphamide (n = 16, 30.2%).
Distribution of molecular subtypes and stromal TILs
In the overall cohort, 47 patients (16.3%) were classified as IM, 26 (9.0%) as LAR, 76 (26.3%) as BLIS, 26 (9.0%) as MES, and 114 (39.4%) as UC.
The distribution of molecular subtypes varied significantly according to clinical trajectory. IM tumors were more frequently observed among patients without recurrence, whereas BLIS and UC subtypes were enriched in patients with recurrent disease and de novo metastatic presentation. The MES subtype was most prevalent among patients who experienced progression during neoadjuvant chemotherapy. Evaluation of sTILs in patients with early disease demonstrated significantly higher median levels (%) in IM (20.0 [IQR 16.5-23.5]) tumors compared with other subtypes (LAR: 7.5 [IQR 5.9-9.1]; BLIS: 5.0 [IQR 4.1-5.9]; MES: 5.0 [IQR 2.8-7.2]; UC: 5.0 [IQR 3.9-6.0]) (Figure 3A). Conversely, in patients with metastatic disease, findings were strictly exploratory due to the limited sample size (Figure 3B).
Figure 3.
Violin plots for stromal TILs expression by molecular subtypes. (A) Expression of sTILs in early disease; IM showed statistically significant higher sTILs compared to other subtypes. (B) Expression of sTILs in metastatic disease; findings strictly exploratory.
Survival outcomes
The median follow-up for the entire cohort was 63 months (IQR 22-105). To address the heterogeneity of the study population, 3-, 5-, and 10-year OS rates were stratified separately. Detailed survival rates according to clinical stages and clinical groups are shown in Table S1. Stratification by molecular subtype in patients with early disease showed a trend toward improved OS in LAR and IM tumors and poorer outcomes in BLIS and UC tumors; however, these differences did not reach statistical significance in the overall cohort (P = .35; Figure 4A). Similarly, no significant differences in OS were found in the group of patients with de novo stage IV disease (P = .091; Figure 4B). Furthermore, DFS did not differ significantly across molecular subtypes (P = .68; Figure 4C).
Figure 4.
(A) Kaplan-Meier curves for OS by molecular subtype in early disease (n = 236); no statistically significant differences were found. (B) Kaplan-Meier curves for OS by molecular subtype in metastatic disease (n = 53); no statistically significant differences were found. (C) Kaplan-Meier curves for DFS by molecular subtype. Only the non-recurrent and recurrent clinical groups are included (n = 203). No statistically significant results were found.
Among patients with recurrent disease, molecular subtype was significantly associated with OS. IM, LAR, and MES subtypes demonstrated improved survival, whereas BLIS and UC tumors were associated with worse outcomes (P < .01; Figure S1). No statistically significant differences in OS among IHC subtypes were found in the groups of no recurrence, progression during neoadjuvant chemotherapy, and de novo stage IV disease.
Stratification by sTILs in patients with early disease revealed significantly improved OS in patients with sTILs ≥10% compared with those with sTILs <10% (5-year OS: 73.2% vs 51.8%, P < .01; Figure 5A). Conversely, no statistically significant differences were observed in patients with metastatic disease (P = .83; Figure 5B).
Figure 5.
(A) Kaplan-Meier curves for OS stratified by sTILs expression in early disease (<10% vs ≥10; n = 236; P < .01). (B) Kaplan-Meier for OS stratified by sTILs expression in metastatic disease (<10% vs ≥10%; n = 53; P = .83).
Univariate and multivariate cox analyses
Univariate and multivariate Cox regression analyses for OS (in both early and metastatic disease) and DFS are presented in Tables 2, S2, and S3, respectively. In univariate analysis for OS in early disease, larger tumor size, increasing lymph node involvement, higher sTILs levels, and treatment era were significantly associated with survival. Compared with T1 tumors, T4 tumors (hazard ratio [HR] 3.83, 95% CI, 1.84-7.97; P < .001) demonstrated increased risk of death. Increasing nodal burden was associated with progressively worse outcomes. sTILs ≥10% were associated with a protective effect (HR 0.54, 95% CI, 0.36-0.81; P < .01).
Table 2.
Univariate and multivariate Cox proportional hazards model for overall survival in the population with early disease (n = 236).
| Variable | Univariate |
Multivariate |
||
|---|---|---|---|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value | |
| Lymph node status | ||||
| N0 | Ref. | Ref. | ||
| N1 | 3.07 (1.46-6.48) | <.01 | 2.67 (1.20-5.90) | .016 |
| N2 | 6.81 (3.31-14.06) | <.001 | 3.65 (1.52-8.80) | .004 |
| N3 | 11.47 (5.20-25.09) | <.001 | 6.13 (2.44-15.40) | <.001 |
| Clinical stage | ||||
| I-II | Ref. | Ref. | ||
| III | 4.45 (2.75-7.21) | <.001 | 2.31 (1.24-4.30) | .008 |
| Stromal TILs | ||||
| Low (<10%) | Ref. | Ref. | ||
| High (≥10%) | 0.54 (0.36-0.81) | <.01 | 0.63 (0.40-1.0) | .049 |
| Molecular subtype | ||||
| IM | Ref. | Ref. | ||
| LAR | 1.04 (0.42-2.57) | .934 | 0.56 (0.20-1.50) | .250 |
| BLIS | 1.69 (0.89-3.22) | .108 | 0.92 (0.45-1.90) | .808 |
| MES | 1.64 (0.73-3.69) | .234 | 1.34 (0.59-3.10) | .485 |
| UC | 1.70 (0.59-0.93) | .085 | 1.07 (0.55-2.10) | .835 |
| Treatment era | ||||
| Up to 2014 | Ref. | Ref. | ||
| From 2015 onward | 2.64 (1.47-4.77) | <.01 | 2.43 (1.28-4.60) | .007 |
| Tumor size | ||||
| T1 | Ref. | |||
| T2 | 1.15 (0.55-2.39) | .719 | ||
| T3 | 1.88 (0.87-4.06) | .109 | ||
| T4 | 3.83 (1.84-7.97) | <.001 | ||
| BMI | ||||
| Normal | Ref. | |||
| Underweight | 2.34 (0.56-9.78) | .246 | ||
| Overweight | 0.79 (0.49-1.26) | .325 | ||
| Obesity | 0.83 (0.49-1.38) | .471 | ||
| Menopause status | ||||
| Pre-menopause | Ref. | |||
| Post-menopause | 0.90 (0.61-1-33) | .603 | ||
| Histological type | ||||
| Ductal | Ref. | |||
| Ohers | 0.80 (0.37-1.73) | .579 | ||
| Histologic grade | ||||
| Grade 1 | Ref. | |||
| Grade 2 | 1.29 (0.17-9.97) | .809 | ||
| Grade 3 | 1.83 (0.25-13.16) | .548 | ||
Concordance index for multivariate analysis: 0.76.
Abbreviations: HR, hazard ratio; CI, confidence interval; BMI, body mass index; Ref., reference; TILs, tumor infiltrating lymphocytes; IM, immunomodulator; LAR, luminal androgen receptors; BLIS, basal-like immunosuppressed; MES, mesenchymal; UC, unclassifiable.
In multivariate analysis for OS, lymph node involvement, clinical stage, and treatment era remained independently associated with mortality. sTILs ≥10% retained a protective effect, reducing the risk of death by 37% (HR 0.63, 95% CI, 0.40-1.0; P = .049). Molecular subtypes were not independently associated with OS after adjustment.
In univariate analysis for OS in de novo stage IV disease, only T4 tumors were associated with mortality; however, in multivariate analysis, only N3 was associated with mortality.
For DFS, univariate analysis showed significant associations with lymph node status, clinical stage, and treatment era, while sTILs ≥10% demonstrated no association (HR 0.64, 95% CI, 0.41-1.02). In multivariate analysis, lymph node status and treatment era remained as independent predictors of recurrence.
Discussion
This study evaluated clinical, histopathological, and molecular characteristics of a cohort of women with TNBC, focusing on the distribution of IHC-based molecular subtypes and sTILs across distinct clinical disease trajectories and their association with survival. Our findings reinforce the biological heterogeneity of TNBC and underscore the prognostic relevance of the tumor immune microenvironment, while highlighting the complementary role of molecular subtyping in risk stratification, consistent with recent advances in neoadjuvant chemotherapy and immunotherapy for TNBC.13,14
The molecular subtype distribution in our cohort differed from that reported in prior studies and varied significantly across clinical trajectories.7–9,15,16 The IM subtype was enriched in non-recurrent cases and demonstrated a trend toward improved OS, consistent with prior reports linking immune-enriched tumors to favorable outcomes.17,18 Conversely, BLIS and UC subtypes predominated in recurrent and metastatic disease, supporting their association with poor prognosis and treatment resistance.6,19 The MES subtype was most prevalent in patients experiencing progression during neoadjuvant chemotherapy, highlighting a candidate group for treatment escalation.
Our analysis confirmed the prognostic importance of immune infiltration: patients with sTILs ≥10% experienced significantly improved OS.20 Higher sTILs levels predominated in the IM subtype, supporting its immune-activated phenotype21–23 and concordant with data linking high immune infiltration to superior treatment response and survival.
The clinical relevance of this immune activation is underscored in neoadjuvant immunotherapy trials. In KEYNOTE-522, adding pembrolizumab to neoadjuvant chemotherapy significantly increased pCR rates (64.8% vs 51.2%)11 and improved long-term outcomes: at 5 years, OS was 86.6% with pembrolizumab plus chemotherapy vs 81.7% with chemotherapy alone.3 In contrast, results from the NeoTRIP trial, which evaluated atezolizumab plus carboplatin/nab-paclitaxel in high-risk early TNBC, failed to demonstrate a statistically significant increase in pCR.24 Beyond differences in immune checkpoint inhibitors (ICIs), lower prevalence of the IM subtype or lower baseline immune infiltration in NeoTRIP may explain these divergent results. The same argument could be extended to the metastatic setting, where only pembrolizumab has consistently demonstrated an improvement in OS.25 It is plausible that inter-trial differences in the prevalence of molecular subtypes could explain the failure of other ICI agents in the same setting, such as atezolizumab in the IMpassion131 trial.
Notably, although molecular subtypes showed prognostic associations in univariate and recurrent subgroup analyses, they lost independent significance in multivariate models adjusted for lymph node status and clinical stage. In contrast, sTILs remained independently associated with OS, suggesting that immune activation may represent a more direct and robust surrogate of prognosis than molecular subtyping alone. Rather than serving as independent prognostic markers, IHC-based molecular subtypes may provide complementary biological context that helps explain clinical behavior and therapeutic vulnerability.
The differential distribution of molecular subtypes across clinical disease trajectories observed in our study supports the biological relevance of TNBC subtyping. For instance, the FUTURE phase II umbrella trial showed that subtype-directed treatment strategies can improve outcomes in heavily pretreated metastatic TNBC.26 Furthermore, for the LAR subtype, clinical trials have evaluated AR inhibitors, such as bicalutamide and enzalutamide, both as monotherapy and in combination with CDK4/6 inhibitors. While these strategies have shown some clinical activity in improving progression-free survival, their efficacy as single agents remain limited.27 Conversely, treatment strategies for the BLIS and MES subtypes in heavily pretreated patients have focused on targeting the specific genomic drivers of each group. For BLIS, approaches have explored PARP inhibitors and anti-VEGF/VEGFR therapies, particularly in the context of BRCA1/2 mutations. For the MES subtype, VEGFR and mTOR inhibitors have been investigated to target characteristic PI3K/AKT pathway alterations.26 Thus, the classification of TNBC into molecular subtypes may not only provide prognostic information but also serve as a foundation for precision medicine. This perspective supports the rationale of our study: subtyping and tumor microenvironment profiling could help stratify patients, anticipate clinical behavior, and guide therapeutic decisions.
In the present study, patients who progressed during neoadjuvant chemotherapy were enriched in the MES subtype, suggesting that this group could benefit from neoadjuvant intensification, such as the KEYNOTE-522 regimen or potentially newer, more intensive protocols. A similar argument could be made for the BLIS and UC subtypes, which were more prevalent in recurrent and de novo stage IV cases; this indicates that novel intensified strategies are urgently required for these high-risk populations.
In contrast, the IM subtype presents a unique opportunity for treatment de-escalation due to its characteristic immune infiltration and favorable prognosis. For instance, the PLANeT trial demonstrated that combining chemotherapy with low dose pembrolizumab in stage II and III TNBC significantly increased pCR rates compared to chemotherapy alone.28 Such low-dose approaches are particularly relevant in resource-limited settings. Thus, it could be posited that patients with the IM subtype, whose high baseline immune infiltration serves as a robust biomarker for both pCR and superior survival, may not require the standard prolonged neoadjuvant and adjuvant ICI regimens. In these cases, a shortened course of neoadjuvant pembrolizumab might suffice to achieve pCR while minimizing long-term toxicity and healthcare costs. Beyond shortening the duration of immunotherapy, these immune-activated characteristics also provide an excellent rationale for exploring anthracycline-free neoadjuvant backbones to minimize severe systemic toxicities. This therapeutic evolution is exemplified by the ongoing randomized phase III SCARLET trial (SWOG S2212), which is actively evaluating whether an anthracycline-free regimen combining taxanes and platinum salts with pembrolizumab can deliver non-inferior event-free survival while substantially improving quality of life and cardiac safety for patients with early-stage disease. Fully integrating quantitative biomarker subtyping could help clinicians identify ideal candidates with high baseline immune scores who are most suitable for such anthracycline-omission strategies.
Several limitations of our study must be disclosed. Its retrospective design entails inherent limitations, including potential selection bias and constrained sample sizes within specific molecular subgroups. Additionally, a significant limitation is that the majority of patients in the non-recurrent subgroup (55.6%) received adjuvant chemotherapy only. Within this specific population, while the vast majority presented with early-stage disease (84% with T1-T2 tumors), a small subset presented with locally advanced T3-T4 tumors. The management of these larger tumors with upfront surgery rather than neoadjuvant systemic therapy reflects specific institutional referral patterns. Because our institutions serve as national oncology reference centers, a significant proportion of patients are referred to our medical oncology services for systemic treatment allocation only after having already undergone definitive primary surgery at outside community hospitals. Because these patients arrived postoperatively, the opportunity to offer guideline-recommended neoadjuvant chemotherapy was precluded. While this operational distribution is technically comparable to the neoadjuvant-free cohorts used to establish the Zhao et al. algorithm, it introduces a meaningful selection bias toward more favorable clinical outcomes within this specific subgroup, limiting the generalizability of our findings to contemporary clinical practice, where neoadjuvant chemoimmunotherapy has become the standard of care for early-stage high-risk disease.
We must also emphasize as a limitation that the immunohistochemical thresholds utilized to define biomarker positivity (≥10% for AR, FOXC1, and DCLK1, and ≥20% for CD8) were adopted directly from the single pioneering study by Zhao et al. (2020) and are not universally standardized across international pathology guidelines. While these specific cutoffs have begun to accumulate independent validation in recent geographic cohort analyses, such as the external replication study in Thai patients by Leeha et al.9 and parallel subtyping series published in 2024 that utilized identical threshold definitions,16 their descriptive application within our Mexican cohort has not been validated. The optimal diagnostic cutoffs for these markers may vary across different patient populations, and our reliance on these pre-specified thresholds represents a structural parameter that requires broader, prospective multi-institutional validation before true clinical implementation. Furthermore, the biological heterogeneity of the UC remains a challenge that warrants more investigation in future research. Prospective validation in larger cohorts will be essential to integrate molecular subtyping with quantitative immune profiling, PD-L1 expression, and novel biomarkers. Such integrative approaches are necessary to further refine the precision and robustness of prognostic and predictive models in TNBC.
Conclusions
Our findings reinforce the notion that classical clinicopathological factors, particularly tumor size, nodal status, and clinical stage, remain the primary determinants of prognosis in TNBC. Beyond these established parameters, the tumor immune contexture, most notably sTILs, provides incremental prognostic value. Specifically, sTILs levels ≥10% were associated with significantly improved OS, underscoring the pivotal role of the tumor immune microenvironment in defining TNBC outcomes. In the current era of neoadjuvant chemoimmunotherapy, immune-enriched tumors represent a subgroup with a particularly favorable prognosis and the highest potential for de-escalation strategies.
Although IHC-based molecular subtypes did not emerge as independent predictors of survival in our multivariable analysis, their differential distribution across clinical disease trajectories validates their biological and clinical relevance. These results align with emerging evidence from precision medicine trials suggesting that molecular subtyping can identify therapeutic vulnerabilities in specific patient populations. Collectively, our study highlights the necessity of integrating nodal status and immune biomarkers into prognostic assessments, while future prospective research should continue to evaluate how molecular subtyping can be incorporated to further refine personalized treatment approaches in TNBC.
Supplementary Material
Acknowledgments
We would like to thank Víctor Hugo Olivera Rodríguez, Claudia Ali Montoya Alatriste, Eduardo Bautista Nava, Jessica Calleja Osnaya, Belén Betzaida del Castillo Hernández, and Alberto Rivera Ramírez for technical support in the initial processing of the samples.
Contributor Information
Jesus Edgardo Hernandez-Hernandez, Tecnologico de Monterrey, Escuela de Medicina y Ciencias de la Salud, Monterrey, Nuevo Leon, 64710, Mexico; Oncology Institute and Breast Cancer Center, Hospital Zambrano Hellion TecSalud, Tecnologico de Monterrey, San Pedro Garza Garcia, 66278, Mexico.
Alejandro Mohar, Unidad de Investigación Biomédica en Cáncer, Instituto Nacional de Cancerología e Instituto de Investigaciones Biomédicas, UNAM, Mexico City, 14080, Mexico.
César Octavio Lara-Torres, Subdirección de Patología, Instituto Nacional de Cancerología, Mexico City, 14080, Mexico.
Daniela Vazquez-Juarez, Oncology Institute and Breast Cancer Center, Hospital Zambrano Hellion TecSalud, Tecnologico de Monterrey, San Pedro Garza Garcia, 66278, Mexico.
Tamara Palacios, Unidad de Investigación Biomédica en Cáncer, Instituto Nacional de Cancerología e Instituto de Investigaciones Biomédicas, UNAM, Mexico City, 14080, Mexico.
Lourdes Peña-Torres, Subdirección de Patología, Instituto Nacional de Cancerología, Mexico City, 14080, Mexico.
Guadalupe Moncada-Claudio, Subdirección de Patología, Instituto Nacional de Cancerología, Mexico City, 14080, Mexico.
Areli Velazquez-Martinez, Division of Oncology and Hematology, Instituto Nacional de Cancerología, Mexico City, 14080, Mexico.
Paula Cabrera-Galeana, Division of Oncology and Hematology, Instituto Nacional de Cancerología, Mexico City, 14080, Mexico.
Gabriela Sofía Gómez-Macías, Oncology Institute and Breast Cancer Center, Hospital Zambrano Hellion TecSalud, Tecnologico de Monterrey, San Pedro Garza Garcia, 66278, Mexico; Facultad de Medicina y Hospital Universitario “Dr José Eleuterio González”, Universidad Autónoma de Nuevo León, Monterrey, Nuevo León, 64460, México.
Fany Iris Porras-Reyes, Subdirección de Patología, Instituto Nacional de Cancerología, Mexico City, 14080, Mexico.
Víctor Manuel Pérez-Sánchez, Subdirección de Patología, Instituto Nacional de Cancerología, Mexico City, 14080, Mexico.
Alejandro Aranda-Gutierrez, Oncology Institute and Breast Cancer Center, Hospital Zambrano Hellion TecSalud, Tecnologico de Monterrey, San Pedro Garza Garcia, 66278, Mexico.
Cynthia Villarreal-Garza, Tecnologico de Monterrey, Escuela de Medicina y Ciencias de la Salud, Monterrey, Nuevo Leon, 64710, Mexico; Oncology Institute and Breast Cancer Center, Hospital Zambrano Hellion TecSalud, Tecnologico de Monterrey, San Pedro Garza Garcia, 66278, Mexico.
Author contributions
Jesus Edgardo Hernandez-Hernandez (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing—original draft, Writing—review & editing), Alejandro Mohar (Conceptualization, Methodology, Project administration, Resources, Supervision, Writing—review & editing), César Octavio Lara-Torres (Investigation, Methodology, Resources, Writing—review & editing), Daniela Vazquez-Juarez (Data curation, Supervision), Tamara Palacios (Data curation, Investigation), Lourdes Peña-Torres (Investigation, Methodology), Guadalupe Moncada-Claudio (Investigation, Methodology), Areli Velazquez-Martinez (Data curation, Supervision), Paula Cabrera-Galeana (Resources, Supervision), Gabriela Sofía Gómez-Macías (Resources, Writing—review & editing), Fany Iris Porras-Reyes (Resources), Víctor Manuel Pérez-Sánchez (Resources), Alejandro Aranda-Gutierrez (Conceptualization, Data curation, Investigation, Methodology, Supervision, Writing—original draft, Writing—review & editing), and Cynthia Villarreal-Garza (Conceptualization, Funding acquisition, Methodology, Project administration, Resources, Supervision, Writing—review & editing)
Supplementary material
Supplementary material is available at The Oncologist online.
Funding
This project was funded with internal resources.
Conflicts of interest
The authors declare no conflicts of interest regarding this manuscript.
Data availability
The data underlying this article will be shared on reasonable request to the corresponding author.
References
- 1. Cocco S, Piezzo M, Calabrese A, et al. Biomarkers in triple-negative breast cancer: state-of-the-art and future perspectives. Int J Mol Sci. 2020;21:4579. 10.3390/ijms21134579 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Hwang K-T, Kim J, Jung J, et al. Impact of breast cancer subtypes on prognosis of women with operable invasive breast cancer: a population-based study using SEER database. Clin Cancer Res. 2019;25:1970-1979. 10.1158/1078-0432.CCR-18-2782 [DOI] [PubMed] [Google Scholar]
- 3. Schmid P, Cortes J, Dent R, et al. Overall survival with pembrolizumab in early-stage triple-negative breast cancer. N Engl J Med. 2024;391:1981-1991. 10.1056/NEJMoa2409932 [DOI] [PubMed] [Google Scholar]
- 4. Kesireddy M, Elsayed L, Shostrom VK, et al. Overall survival and prognostic factors in metastatic Triple-Negative Breast Cancer: a National Cancer Database analysis. Cancers. 2024;16:1791. 10.3390/cancers16101791 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Brouckaert O, Wildiers H, Floris G, Neven P. Update on triple-negative breast cancer: prognosis and management strategies. Int J Womens Health. 2012;4:511-520. 10.2147/IJWH.S18541 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Lehmann BD, Bauer JA, Chen X, et al. Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies. J Clin Invest. 2011;121:2750-2767. 10.1172/JCI45014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Jiang Y-Z, Ma D, Suo C, et al. Genomic and transcriptomic landscape of triple-negative breast cancers: subtypes and treatment strategies. Cancer Cell. 2019;35:428-440.e5. 10.1016/j.ccell.2019.02.001 [DOI] [PubMed] [Google Scholar]
- 8. Zhao S, Ma D, Xiao Y, et al. Molecular subtyping of triple‐negative breast cancers by immunohistochemistry: molecular basis and clinical relevance. Oncologist. 2020;25:e1481-e1491. 10.1634/theoncologist.2019-0982 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Leeha M, Kanokwiroon K, Laohawiriyakamol S, Thongsuksai P. Immunohistochemistry-based molecular subtyping of triple-negative breast cancer and its prognostic significance. Pathol Oncol Res. 2023;29:1611162. 10.3389/pore.2023.1611162 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Leon-Ferre RA, Jonas SF, Salgado R, et al. International Immuno-Oncology Biomarker Working Group. Tumor-infiltrating lymphocytes in triple-negative breast cancer. JAMA. 2024;331:1135-1144. 10.1001/jama.2024.3056 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Schmid P, Cortes J, Pusztai L, et al. Pembrolizumab for early triple-negative breast cancer. N Engl J Med. 2020;382:810-821. 10.1056/nejmoa1910549 [DOI] [PubMed] [Google Scholar]
- 12. Salgado R, Denkert C, Demaria S, et al. International TILs Working Group 2014. The evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer: recommendations by an international TILs working group 2014. Ann Oncol. 2015;26:259-271. 10.1093/annonc/mdu450 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Asleh K, Riaz N, Nielsen TO. Heterogeneity of triple negative breast cancer: current advances in subtyping and treatment implications. J Exp Clin Cancer Res. 2022;41:265. 10.1186/s13046-022-02476-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Liu Y, Hu Y, Xue J, et al. Correction: advances in immunotherapy for triple-negative breast cancer. Mol Cancer. 2023;22:154. 10.1186/s12943-023-01858-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Liu Y-R, Jiang Y-Z, Xu X-E, et al. Comprehensive transcriptome analysis identifies novel molecular subtypes and subtype-specific RNAs of triple-negative breast cancer. Breast Cancer Res. 2016;18:33. 10.1186/s13058-016-0690-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Hu H, Tong K, Tsang JY, et al. Subtyping of triple-negative breast cancers: its prognostication and implications in diagnosis of breast origin. ESMO Open. 2024;9:102993. 10.1016/j.esmoop.2024.102993 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Serrano García L, Jávega B, Llombart Cussac A, et al. Patterns of immune evasion in triple-negative breast cancer and new potential therapeutic targets: a review. Front Immunol. 2024;15:1513421. 10.3389/fimmu.2024.1513421 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Syeda MF, Santos RA, Sousa LM, et al. Immune monitoring of neoadjuvant chemo-immunotherapy for triple-negative breast cancer. Front Immunol. 2025;16:1654748. 10.3389/fimmu.2025.1654748 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Burstein MD, Tsimelzon A, Poage GM, et al. Comprehensive genomic analysis identifies novel subtypes and targets of triple-negative breast cancer. Clinical Cancer Research. 2015;21:1688-1698. 10.1158/1078-0432.CCR-14-0432 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Denkert C, von Minckwitz G, Darb-Esfahani S, et al. Tumour-infiltrating lymphocytes and prognosis in different subtypes of breast cancer: a pooled analysis of 3771 patients treated with neoadjuvant therapy. Lancet Oncol. 2018;19:40-50. 10.1016/S1470-2045(17)30904-X [DOI] [PubMed] [Google Scholar]
- 21. Sabit H, Adel A, Abdelfattah MM, et al. The role of tumor microenvironment and immune cell crosstalk in triple-negative breast cancer (TNBC): emerging therapeutic opportunities. Cancer Lett. 2025;628:217865. 10.1016/j.canlet.2025.217865 [DOI] [PubMed] [Google Scholar]
- 22. Syrnioti A, Petousis S, Newman LA, et al. Triple negative breast cancer: molecular subtype-specific immune landscapes with therapeutic implications. Cancers (Basel). 2024;16:2094. 10.3390/cancers16112094 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Harano K, Wang Y, Lim B, et al. Rates of immune cell infiltration in patients with triple-negative breast cancer by molecular subtype. PLoS One. 2018;13:e0204513. 10.1371/journal.pone.0204513 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Garufi G, Carbognin L, Schettini F, et al. Updated neoadjuvant treatment landscape for early triple negative breast cancer: immunotherapy, potential predictive biomarkers, and novel agents. Cancers (Basel). 2022;14:4064. 10.3390/cancers14174064 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Cortes J, Rugo HS, Cescon DW, et al. Pembrolizumab plus chemotherapy in advanced triple-negative breast cancer. N Engl J Med. 2022;387:217-226. 10.1056/nejmoa2202809 [DOI] [PubMed] [Google Scholar]
- 26. Liu Y, Zhu X-Z, Xiao Y, et al. Subtyping-based platform guides precision medicine for heavily pretreated metastatic triple-negative breast cancer: the FUTURE phase II umbrella clinical trial. Cell Res. 2023;33:389-402. 10.1038/s41422-023-00795-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Xiang D l, Yue X. Research progress of molecular typing and targeted therapy for triple-negative breast cancer. Front Oncol. 2025;15:1666126. 10.3389/fonc.2025.1666126 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Arora A, Bhaskarane H, Tansir G, et al. A phase II, randomized, open-label study to evaluate low-dose pembrolizumab plus chemotherapy versus chemotherapy as neoadjuvant therapy for localized triple-negative breast cancer (TNBC) (PLANeT trial—pembrolizumab low dose in addition to NACT in TNBC). Ann Oncol. 2025;37:639-649. 10.1016/j.annonc.2025.12.015 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data underlying this article will be shared on reasonable request to the corresponding author.