Predictive power of tertiary lymphoid structure for prognosis and neoadjuvant chemotherapy response in HER2-positive breast cancer

Abstract

This study evaluated the prognostic significance of tertiary lymphoid structures (TLS) in human epidermal growth factor receptor 2 (HER2)-positive breast cancer (BC), focusing on their associations with survival outcomes, response to neoadjuvant therapy, and potential as a biomarker for personalized treatment strategies. Data from patients with HER2-positive BC in the METABRIC and The Cancer Genome Atlas databases were analyzed. TLS expression scores were calculated using gene set variation analysis, and their associations with survival outcomes were assessed. Immune cell infiltration, immune checkpoint expression, tumor mutational burden, and pathway enrichment were also evaluated. Data from the I-SPY2 clinical trial and a clinicopathological cohort of 19 patients from Xiangya Hospital were used to assess the relationship between TLS expression and pathological complete response following neoadjuvant therapy. High TLS expression was associated with improved survival and increased infiltration of antitumor immune cells. TLS-high tumors were enriched in immune-related pathways, whereas TLS-low tumors showed activation of proliferation and metabolism pathways. Patients with high TLS expression had better responses to neoadjuvant therapy, while those with low TLS expression derived greater benefit from dual-targeted treatments. TLS represents a promising biomarker for predicting survival and response to neoadjuvant therapy in HER2-positive BC, with potential to support personalized treatment strategies.

1. Introduction

Breast cancer (BC) is the most common malignant tumor worldwide and a leading cause of cancer-related deaths among women, accounting for 15.5% of cancer-related mortality.^[1,2] Among patients with BC, 15% to 20% overexpress human epidermal growth factor receptor 2 (HER2), which is associated with high tumor aggressiveness and poor prognosis.^[3] The development of humanized monoclonal antibodies, such as trastuzumab and pertuzumab, has significantly improved the treatment outcomes of HER2-positive (HER2+) BC.^[4] Currently, pertuzumab is used in combination with trastuzumab and chemotherapy for neoadjuvant, adjuvant, and metastatic BC treatment.^[5]

Trastuzumab binds to subdomain IV of the extracellular domain of HER2, blocking intracellular signaling pathways (e.g., MAPK and PI3K/Akt) and inducing cell death. It also activates natural killer (NK) cells in the tumor immune microenvironment (TIME), indicating that its efficacy is partly mediated through antibody-dependent cell-mediated cytotoxicity (ADCC).^[6] Pertuzumab binds to subdomain II of the extracellular domain, preventing HER2 dimerization, inducing apoptosis, and inhibiting tumor growth.^[7] Similar to trastuzumab, pertuzumab contains an Fc region capable of binding to FcƴR-positive immune cells and initiating ADCC, thereby exhibiting comparable efficacy.^[5,8,9] Dual HER2 blockade with trastuzumab and pertuzumab can enhance ADCC, leading to stronger antitumor effects.^[6,10,11] By promoting NK cell–tumor cell interactions and generating an inflammatory environment via ADCC, anti-HER2 antibodies may modulate other immune cell populations and produce a “vaccine effect.”^[12,13] However, therapeutic responses vary considerably among patients,^[14] and most with metastatic BC eventually develop resistance. Long-term follow-up data show that only 11% of HER2 +  metastatic BC patients achieve durable complete remission.^[15] In cases of recurrence or progression, ADCC may be impaired by the immunosuppressive TIME and other factors.^[12] This variability in antitumor responses suggests that the differences in TIME modulation by anti-HER2 antibodies may contribute to variations in treatment outcomes in HER2 + BC.

Tertiary lymphoid structures (TLS), also known as ectopic lymphoid structures, are recognized components of the TIME.^[16] These structures are typically located within or around the invasive tumor stroma and serve as sites for lymphocyte recruitment and immune activation. TLS formation is often associated with increased inflammation, such as following autologous tumor vaccine administration.^[17] Numerous studies have confirmed the presence of TLS in various malignancies, including lung cancer,^[18–20] colorectal cancer,^[21,22] and malignant melanoma.^[23,24] In general, TLS presence is linked to better prognosis and favorable responses to immunotherapy,^[25,26] although exceptions have been reported.^[27] In HER2 + BC, TLS has been associated with active antitumor immune responses.^[28,29] However, the relationship between TLS and sensitivity to neoadjuvant therapy in HER2 + BC has not yet been established.

This study aimed to investigate the associations between TLS and prognosis, immune microenvironment, and tumor mutational burden (TMB) in HER2 + BC. It also evaluated the potential value of TLS in predicting the efficacy of neoadjuvant therapy. By assessing TLS, this study sought to enable early prediction of patient response to neoadjuvant treatment, thereby advancing the development of personalized treatment strategies for BC.

2. Materials and methods

2.1. HER2-positive BC cohort data acquisition

1. METABRIC cohort: clinical data and standardized gene expression profiles for BC cases were obtained from the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) via the cBioPortal for Cancer Genomics. HER2 + BC samples were identified using the PAM50 classification.

2. The Cancer Genome Atlas (TCGA) cohort: RNA sequencing gene expression data and clinical information were downloaded from TCGA database (https://portal.gdc.cancer.gov/). HER2 + samples were identified based on immunohistochemistry results.

3. I-SPY2 Cohort: Clinical information, gene expression data, and related pathway gene sets were downloaded from the I-SPY2 trial platform. HER2 + samples were selected based on immunohistochemistry results.

After obtaining the data from the above platforms, HER2 + BC samples were selected using clinical criteria, resulting in 3 independent HER2 + BC cohorts.

2.2. Xiangya Hospital clinical pathology cohort

Patients with BC from the Department of Breast Surgery at Xiangya Hospital, Central South University, who were initially diagnosed with HER2 + disease via biopsy and subsequently received neoadjuvant therapy followed by surgery, were included. The effectiveness of neoadjuvant therapy was determined by pathological evaluation after surgery. A total of 22 patients were enrolled, including 11 who achieved pathological complete response (pCR) and 11 who did not. Pretreatment biopsy specimens from these patients were collected for histological sectioning and analysis. All treatments followed standard clinical practice. The study was approved by the Ethics Committee of Xiangya Clinical Medical Research, Central South University (Ethics Approval No.: 2024111446), with informed consent waived. All procedures strictly followed the ethical guidelines of the committee, ensuring protection of patient privacy and compliance with applicable ethical standards.

2.3. Bioinformatics analysis

Nine pre-defined, highly expressed genes (CD79B, CD1D, CCR6, LAT, SKAP1, CETP, EIF1AY, RBP5, and PTGDS) were selected as the TLS gene set.^[30] Gene set variation analysis was used to calculate enrichment scores for each sample, with a Gaussian distribution applied as the Kernel density function to derive TLS values.

The potential association between TLS expression and overall survival (OS) and relapse-free survival (RFS) in HER2-positive (HER2+) BC was investigated. Kaplan–Meier survival curves were generated, and log-rank tests were performed using the survdiff function. Hazard ratios (HRs) with 95% confidence intervals (CI) were calculated based on logarithmic transformation and normal distribution assumptions. A Cox proportional hazards regression model was applied to evaluate the impact of predictive factors on OS.

Additional analyses included immune cell infiltration using the CIBERSORT algorithm,^[31] gene set enrichment analysis (GSEA), single-sample GSEA (ssGSEA), linear regression analysis, and univariate interaction analysis.

All statistical analyses and graphical outputs were performed in the R environment to ensure reproducibility and accuracy. Unless otherwise specified, a P-value < .05 was considered statistically significant.

2.4. Multiplex immunohistochemistry for TLS analysis

Multiplex immunohistochemistry was conducted on formalin-fixed, paraffin-embedded tissue sections to detect and quantify TLS. Tissue sections were deparaffinized in xylene and rehydrated through a graded ethanol series. Antigen retrieval was performed by heating the sections in citrate buffer (pH 6.0). Nonspecific binding was blocked by incubating the sections in 5% normal goat serum for 1 hour. Primary antibodies used for staining were rabbit-derived anti-human CD4 (Abcam, catalog no. ab133616, 1:500 dilution), CD8 (Abcam, catalog no. ab93278, 1:4000 dilution), and CD20 (Abcam, catalog no. ab78237, 1:50 dilution). Sections were incubated with primary antibodies overnight at 4°C, followed by washing with PBS and incubation with species-specific secondary antibodies conjugated to fluorophores for 1 hour at room temperature. Secondary antibodies were selected based on the host species of the primary antibodies and were labeled as follows: CD4 (Alexa Fluor 488), CD8 (Alexa Fluor 555), and CD20 (Alexa Fluor 647). Cell nuclei were counterstained with DAPI (1 μg/mL), and sections were mounted using anti-fade mounting medium. Multichannel fluorescence images were acquired under consistent settings using a fluorescence microscope. Five non-overlapping fields per sample were randomly captured at 10x magnification, and all images were calibrated with a scale bar.

TLS regions of interest (TLS_ROI) were defined by the colocalization of CD4, CD8, and CD20 signals.^[32] The average TLS area per sample was calculated, and TLS status (high/low) was determined using the median TLS area as the cutoff.

All immunohistochemistry image analyses were performed using ImageJ software, and data visualization was conducted in R.

2.5. Code availability

All data analyses and processing steps were performed using previously published software packages cited in Section 2. No new code or algorithms were developed for this study.

3. Results

3.1. Relationship between TLS signature and OS in HER2-positive BC patients

We first assessed the expression levels of the 9 TLS signature genes in TCGA HER2 + BC samples and corresponding normal tissues. Compared to normal tissues, CD79B, CCR6, CETP, and PTGDS expression was significantly downregulated, while SKAP1 expression was significantly upregulated in tumor tissues from HER2 + BC patients (Fig. 1A). Overall, TLS signature expression was significantly lower in tumor tissues than in nontumor tissues (Fig. 1B).

Figure 1.

Relationship between TLS expression and survival in HER2-positive breast cancer. (A) Expression levels of 9 genes in the TLS signature in HER2 + BRCA tumor tissues versus nontumor tissues from the TCGA dataset. (B) Overall TLS signature expression levels in HER2 + BC tumor tissues compared to nontumor tissues. (C) Kaplan–Meier survival analysis of overall survival (OS) in METABRIC HER2 + BC patients,stratified by TLS expression levels. (D) Kaplan–Meier survival analysis of OS in TCGA HER2 + BC patients, stratified by TLS expression levels. (E) Kaplan–Meier survival analysis of relapse-free survival (RFS) in METABRIC HER2 + BC patients, stratified by TLS expression levels. (F) Multivariate Cox proportional hazards analysis of OS in METABRIC HER2 + BC patients, including TLS expression and clinicopathological factors. BC = breast cancer, HER2 = human epidermal growth factor receptor 2, TCGA = The Cancer Genome Atlas, TLS = tertiary lymphoid structures.

We then evaluated the prognostic value of TLS expression in HER2 + BC patients from the METABRIC and TCGA cohorts. In the METABRIC dataset, 224 HER2 + BC patients were included. Using the res.cut function from the survminer R package, an optimal TLS expression cutoff value of 0.32 was determined (Figure S1, Supplemental Digital Content, https://links.lww.com/MD/P65), categorizing patients into high TLS (n = 53) and low TLS (n = 171) groups. Kaplan–Meier survival analysis (Fig. 1C) revealed that patients in the high TLS group had significantly better survival than those in the low TLS group. After a median follow-up of 60 months, the high TLS group had a survival probability of approximately 75%, compared to <60% in the low TLS group. The difference in survival was statistically significant (HR 1.644, 95% CI 1.082–2.500, P = .041). This survival advantage was validated in the TCGA cohort (Fig. 1D), which included 163 HER2 + BC patients. Among these, 34 were classified as high TLS. Compared to the low TLS group (n = 129), patients in the high TLS group had significantly longer OS (HR 6.088, 95% CI 2.163–17.132, P = .04).

In addition, clinical characteristics of patients with high and low TLS expression in the METABRIC cohort were compared (Table S1, Supplemental Digital Content, https://links.lww.com/MD/P66). No significant associations were found between TLS expression and clinical characteristics. We also evaluated the relationship between TLS expression and RFS (Fig. 1E). Although the difference in RFS between the 2 groups was not statistically significant (HR 1.311, 95% CI 0.848–2.025, P = .26), the high TLS group still showed a trend toward improved outcomes. Finally, we performed multivariate analysis using the Cox proportional hazards model to assess the effects of TLS expression and other clinicopathological variables on OS. As shown in Figure 1F, tumor stage and low TLS expression were identified as significant predictors of OS (pathological stage: HR 1.59, 95% CI 1.04–2.42, P = .032; low TLS expression: HR 2.02, 95% CI 0.67–0.92, P = .006).

3.2. Relationship between TLS signature and immune microenvironment

Immune cell infiltration in HER2-positive (HER2+) tumors was evaluated using data from the METABRIC and TCGA databases. Figure 2A and B shows differences in immune cell subsets between high and low TLS groups in both datasets. While the general infiltration trends were consistent across cohorts, some variations in group differences were noted. In the METABRIC HER2 + BC cohort, infiltration levels of B cells, CD8 + T cells, CD4 + T cells, regulatory T cells (Tregs), and M1 macrophages were significantly higher in the high TLS group compared to the low TLS group (P < .05). In the TCGA HER2 + BC cohort, although similar trends were observed, only CD8 + T cells, activated memory CD4 + T cells, and Tregs showed statistically significant differences (P < .01). Additionally, in the METABRIC cohort, M0 and M2 macrophages, as well as neutrophils, were significantly more abundant in the low TLS group (P < .0001). In the TCGA cohort, only M2 macrophages were significantly elevated in the low TLS group (P < .0001); other differences were not statistically significant.

Figure 2.

Immune cell infiltration, checkpoint gene expression, and pathway enrichment analysis in HER2-positive breast cancer. (A) Comparison of immune cell infiltration levels between high and low TLS groups in HER2 + BC from the METABRIC dataset. (B) Similar comparison in the TCGA dataset. (C, D) Differential expression of checkpoint genes between high and low TLS groups in the METABRIC and TCGA HER2 + BC cohorts, respectively. (E, F) Gene Set Enrichment Analysis (GSEA) of pathway enrichment in high and low TLS groups from the METABRIC (E) and TCGA (F) datasets. The high TLS group is shown in red, and the low TLS group in blue. Significance levels are indicated as: ****P < .0001, ***P < .001, **P < .01. BC = breast cancer, HER2 = human epidermal growth factor receptor 2, TCGA = The Cancer Genome Atlas, TLS = tertiary lymphoid structures.

The relationship between TLS signature and immune checkpoint gene expression was also examined. In both the METABRIC (Fig. 2C) and TCGA (Fig. 2D) HER2 + BC cohorts, most checkpoint genes were expressed at higher levels in the high TLS group, except CD276 (B7-H3), which was more highly expressed in the low TLS group. This difference reached statistical significance only in the TCGA cohort (P < .05). GSEA was then performed using gene sets from the Molecular Signatures Database (MSigDB).^[33] Enrichment criteria included NES > 1, P-value < .05, and adjusted P-value (p.adjust) < .25 (Fig. 2E and F). The high TLS group showed significant upregulation of immune and inflammation-related pathways, suggesting stronger immune responses. In contrast, cell proliferation and signal transduction pathways were enriched in the low TLS group, indicating increased tumor cell activity.

3.3. Relationship between TLS signature, driver gene mutations, and TMB

Figure 3A and B illustrates the mutation landscape of driver genes in the high and low TLS groups in both cohorts. In the METABRIC HER2 + BC cohort, 217 of 224 samples (96.88%) harbored mutations, while in the TCGA cohort, 132 of 163 samples (80.98%) exhibited mutations. However, mutation frequencies of individual genes were low, with only TP53 showing a mutation rate above 5%. The relationship between TLS signature and TMB was further explored. Boxplots in Figure 3C and D show no significant differences in TMB values between high and low TLS groups. Correlation analysis also revealed no significant association between TLS expression and TMB in either cohort (Fig. 3E: R = –0.09, P = .1662; Fig. 3F: R = –0.15, P = .7565).

Figure 3.

Mutation landscape and TMB in HER2-positive breast cancer. (A, B) Mutation landscape of driver genes in high and low TLS groups from the METABRIC (A) and TCGA (B) datasets. (C, D) Box plots of TMB values in high and low TLS groups from METABRIC (C) and TCGA (D) datasets. (E, F) Correlation between TLS signature and TMB in the METABRIC (E) and TCGA (F) datasets. . BC = breast cancer, HER2 = human epidermal growth factor receptor 2, TCGA = The Cancer Genome Atlas, TLS = tertiary lymphoid structures, TMB = tumor mutational burden.

3.4. Pathway enrichment and pCR rates in high and low TLS groups in I-SPY2 trial data

HER2 + samples were selected from the I-SPY2 trial, and the previously established cutoff (TLS = 0.32) was applied to classify samples into high and low TLS groups. Single-sample GSEA (ssGSEA) scores were visualized using a heatmap (Fig. 4A). The high TLS group showed strong enrichment in immune-related pathways (e.g., B cells, dendritic cells, macrophages) and inflammatory features (e.g., STAT1 signaling and chemokine activity). Conversely, proliferation-associated pathways (e.g., Module11_Prolif) and basal signatures were enriched in the low TLS group. These findings reflect notable biological differences in immune and proliferative activity between high and low TLS tumors, consistent with results from METABRIC and TCGA.

Figure 4.

Pathway enrichment and pCR rates in HER2-positive breast cancer in the I-SPY2 trial. (A) Heatmap of ssGSEA scores for pathways in high and low TLS groups. (B) pCR rates in high and low TLS groups for patients receiving different treatments Arm. (C) Bar plots showing pCR rates in high and low TLS tumors according to treatment arms (Ctr or pertuzumab). HER2 = human epidermal growth factor receptor 2, pCR = pathological complete response, ssGSEA = single-sample gene set enrichment analysis, TLS = tertiary lymphoid structures.

TLS-based stratification of pCR rates across treatment arms is shown in Figure 4B, with no statistically significant differences observed between groups (Table S2, Supplemental Digital Content, https://links.lww.com/MD/P66). Further analysis of I-SPY2 data evaluated whether TLS could predict pCR in HER2 + BC patients undergoing targeted therapy and the added benefit of pertuzumab (Fig. 4C). In the control group (paclitaxel + trastuzumab), the pCR rate for high TLS tumors was 45% (4/9), compared to 18% (4/22) for low TLS tumors. In the pertuzumab group (control + pertuzumab), the pCR rate was 63% (5/8) for high TLS and 58% (21/36) for low TLS tumors. Notably, the addition of pertuzumab significantly increased the pCR rate in low TLS tumors by more than threefold (58% vs 18%, P = .0032), while no statistically significant difference was observed in high TLS tumors (63% vs 45%, P = .6372).

3.5. Clinical and pathological characteristics of TLS high and low groups and their relationship with neoadjuvant treatment efficacy

Pretreatment biopsy samples from 22 HER2 + early BC patients were initially included. Due to tissue damage or insufficient material, 3 samples were excluded, leaving 19 patients for analysis (Table 1). Among them, 52.6% (10/19) were classified as high TLS and 47.4% (9/19) as low TLS. No significant associations were found between TLS expression and tumor size or histological grade, consistent with METABRIC findings (Table S1, Supplemental Digital Content, https://links.lww.com/MD/P66).

Table 1.

Patient characteristics.

pCR
	TLS high	TLS low
N(%)	8 (80%)	2 (20%)
Mean age (years)	41.6	47.5
Range (years) tumour stage	23–57	44–51
T2	4 (50%)	2 (100%)
T3 histological stage	4 (50%)	0
2	8 (100%)	2 (100%)
3	0	0

non_pCR
N (%)	2 (22.2%)	7 (77.8%)
Mean age (years)	43	43.4
Range (years) tumour stage	33–53	33–52
T2	0	3 (42.9%)
T3 histological stage	2 (100%)	4 (57.1%)
2	1 (50%)	5 (71.4%)
3	1 (50%)	2 (28.6%)

Figure 5A shows representative multiplex immunofluorescence images from HER2 + BC biopsies in pCR and non-pCR cases. Figure 5B presents the relationship between average TLS area and neoadjuvant therapy response. The average TLS area was significantly larger in the pCR group than in the non-pCR group (P < .01). Specifically, the median TLS area in the pCR group was 52,351.34 µm², which was significantly greater than that in the non-pCR group. Figure 5C illustrates the pCR rates in high and low TLS groups. The high TLS group had a significantly higher pCR rate compared to the low TLS group (80% vs 22.2%, P = .023). These results suggest that increased TLS area is strongly associated with better neoadjuvant therapy response in HER2 + BC patients.

Figure 5.

TLS mean area and clinical features related to pCR. (A) Multiplex immunofluorescence images of HER2-positive breast cancer biopsies in pCR and non_pCR patients. (B) Comparison of TLS mean area between pCR and non-pCR groups. (C) pCR rates in high and low TLS groups. HER2 = human epidermal growth factor receptor 2, pCR = pathological complete response, TLS = tertiary lymphoid structures.

4. Discussion

We found that the overall expression of TLS in HER2-positive BC tissues was significantly lower than in normal, nontumor tissues (Fig. 1B). This reduction may be attributed to the aggressive and highly proliferative nature of HER2-positive tumors, which can induce local immunosuppression and hypoxia, thereby impairing TLS formation and maintenance. Among the 9 TLS-related genes analyzed, only SKAP1 (Src kinase-associated phosphoprotein 1) showed significantly elevated expression in HER2-positive tumors compared to normal tissues (Fig. 1A). This finding is consistent with the report by Lingqin Zhu et al in gastric cancer.^[34] SKAP1, an immune cell adaptor protein, plays a critical role in regulating T cell adhesion and proliferation, which is essential for T cell division and proper expression of cell cycle proteins.^[35] These functions are key to effective T cell clonal expansion following antigen activation. However, the full physiological role of SKAP1 in immune regulation remains under investigation. In BC, a SKAP1-COL14A1 gene fusion has been identified,^[36] which may partly explain the elevated SKAP1 expression observed in our analysis. Across 2 independent datasets, high TLS expression was significantly associated with better survival in HER2-positive BC patients (Fig. 1C and D), suggesting that TLS may serve as a prognostic marker for survival outcomes. Although RFS was not significantly different between high and low TLS groups (Fig. 1E), the high TLS group showed a trend toward improved outcomes, implying a protective role of TLS in disease progression. Multivariate analysis (Fig. 1F) identified low TLS expression and tumor size as independent predictors of poor prognosis, reinforcing the association between high TLS expression and favorable survival in HER2-positive BC.

The TIME comprises both pro-tumor and antitumor immune cells. Pro-tumor cells, such as neutrophils that promote metastasis and Tregs that support tumor growth, contrast with antitumor cells such as CD8 + T cells and NK cells, which mediate effective immune responses.^[37] This immune context is especially relevant in HER2-positive BC. Preclinical studies have demonstrated that immune activation, particularly of NK cells, is essential for trastuzumab efficacy.^[38,39] ADCC is a well-established mechanism of trastuzumab, and its effect is enhanced when combined with pertuzumab.^[6] Thus, the composition of the TIME can substantially influence clinical outcomes in HER2-positive metastatic BC. In our study, patients with high TLS signatures exhibited a more favorable immune microenvironment than those with low TLS (Fig. 2A and B). Specifically, immune cells linked to antitumor activity, such as B cells, CD8 + cytotoxic T cells, activated CD4 + T cells, NK cells, and M1 macrophages, were significantly more prevalent in the high TLS group. In contrast, immunosuppressive M2 macrophages and dendritic cells were more abundant in the low TLS group. These differences may help explain variations in treatment response and prognosis among patients with the same cancer subtype. The immune microenvironment in high TLS tumors may be more responsive to HER2-targeted antibody therapies. Interestingly, Tregs were also significantly more abundant in the high TLS group. A study by Tessa G et al reported that Treg infiltration in primary HER2-positive breast tumors was associated with poorer prognosis.^[40] This may reflect the structural similarity between tumor-associated TLS and secondary lymphoid organs, which are populated by diverse lymphocyte subtypes.^[41] Immunosuppressive cells are present within TLS, and their enrichment may contribute to the observed variability in outcomes among high TLS patients. This is supported by findings in melanoma, where TLS has been linked to both immune activation and the presence of suppressive cells.^[30] Similarly, Nikhil S et al reported Treg accumulation in TLS within tumor-bearing lungs in mice.^[42] These findings suggest that treatment outcomes in HER2-positive BC may depend on the balance between antitumor and pro-tumor immune cell populations in the tumor microenvironment.

In our immune checkpoint analysis (Fig. 2C and D), the expression of major checkpoint genes, except CD276, was significantly higher in the high TLS group. This may result from increased immune cell infiltration in high TLS tumors, as more active immune environments are known to upregulate immune checkpoint genes like PD-1, PD-L1, and CTLA-4 to evade immune destruction.^[43–45] Although checkpoint expression is elevated, the robust immune infiltration in high TLS tumors still promotes antitumor responses. While immune checkpoint blockade therapies are not yet standard in HER2-positive BC, our findings suggest that patients with high TLS expression may benefit from future immune checkpoint blockade-based strategies. CD276 (B7-H3) is a notable exception. It plays a critical role in immune evasion and regulation of the tumor microenvironment and is widely expressed in multiple tumor types, including BC, where it is associated with poor prognosis, large tumor size, advanced TNM stage, lymph node metastasis, and recurrence.^[46–48] CD276 suppresses T cell proliferation and promotes the production of IL-10 and TGF-β1, contributing to an immunosuppressive microenvironment.^[49–51] It also inhibits NK cell function and reduces cytokine secretion, such as IFN-γ and TNF-α,^[52–54] while promoting macrophage polarization toward the M2 phenotype.^[55] In our study, the low TLS group exhibited higher M2 macrophage infiltration and reduced NK and T cell presence, consistent with the immunosuppressive role of CD276. These observations may help explain the poor survival outcomes and reduced treatment responses in patients with low TLS. CD276 could therefore represent a potential therapeutic target for improving outcomes in this group.

GSEA results further revealed significant differences between high and low TLS groups in immune and proliferative activity (Fig. 2E and F). These findings were validated by ssGSEA analysis of HER2-positive samples from the I-SPY2 trial (Fig. 4A). The high TLS group exhibited strong enrichment of immune and inflammatory pathways, suggesting an environment conducive to immune function and ADCC activity. In contrast, the low TLS group showed weaker immune responses and greater upregulation of proliferation pathways, indicating higher tumor growth and a less favorable microenvironment for antibody-based therapies such as trastuzumab.

We also investigated driver gene mutations and TMB in relation to TLS levels (Fig. 3). While TMB is a known predictor of immunotherapy response,^[56–58] we found no significant correlation between TLS expression and TMB in HER2-positive BC. This suggests that the benefits of high TLS expression may be driven by enhanced immune infiltration and immune pathway activity rather than by mutational burden.

In our analysis of pCR, patients with high TLS had higher pCR rates following standard treatment (Fig. 4B). However, the addition of pertuzumab significantly increased pCR rates in the low TLS group (Fig. 4C), indicating that patients with low TLS may respond better to dual-targeted therapy. High TLS tumors already exhibited strong immune activation under baseline treatment, and thus derived limited additional benefit from pertuzumab. These results have important clinical implications. TLS may serve as a predictive biomarker of neoadjuvant therapy response in HER2-positive BC. Patients with high TLS appear more responsive to standard therapy, while those with low TLS may benefit more from treatment intensification with pertuzumab. Further multiplex immunofluorescence analysis (Fig. 5) confirmed that TLS area was significantly larger in patients who achieved pCR than in those who did not. The marked difference in pCR rates between high and low TLS groups reinforces the association between TLS and treatment response. These findings suggest that TLS not only reflects an active immune microenvironment but also predicts better outcomes with neoadjuvant therapy. The substantial benefit of pertuzumab in low TLS tumors highlights the potential of using TLS expression to guide treatment selection and optimize outcomes.

The strengths of this study include validation across multiple datasets, comprehensive biological analysis, and clinical relevance. These factors support the utility of TLS as a predictive biomarker in HER2-positive BC. However, the study also has limitations. Despite the integration of multi-database data, sample sizes in each treatment group were relatively small, and further subdivision into TLS subgroups reduced statistical power. Additionally, the study focused solely on HER2-positive disease, limiting its applicability to other subtypes or cancers. The molecular mechanisms linking TLS to treatment response were not fully explored, indicating a need for further research. Overall, this study supports the strong association between TLS expression and prognosis and neoadjuvant therapy response in HER2-positive BC, while also highlighting the need for additional investigation.

5. Conclusion

This study demonstrates that tumor-associated TLS are strong predictors of therapeutic outcomes in HER2-positive BC. High TLS expression is linked to improved pCR rates, better OS, and enhanced immune cell infiltration, supporting its role as a treatment response biomarker. Notably, pertuzumab offers significant benefit to patients with low TLS expression, while high TLS patients respond well to standard therapy. Differences in immune checkpoint expression between TLS groups may influence immunotherapy responses, highlighting TLS as a potential guide for personalized treatment strategies.

Author contributions

Conceptualization: Kejing Zhang.

Data curation: Mengxi Li, Yueheng Wang, Ziru Zhao, Liqiang Ai.

Formal analysis: Mengxi Li, Jing Cao, Liqiang Ai.

Investigation: Mengxi Li, Jing Cao.

Methodology: Jing Cao, Kejing Zhang.

Project administration: Kejing Zhang.

Resources: Kejing Zhang.

Supervision: Kejing Zhang.

Validation: Jing Cao.

Writing – original draft: Mengxi Li.

Writing – review & editing: Mengxi Li, Jing Cao, Yueheng Wang, Ziru Zhao, Liqiang Ai, Kejing Zhang.