Breast cancer (BC) has been the most frequently diagnosed malignancy worldwide, and is the main cause of tumor-associated mortality in women (1, 2). Originating from mammary epithelial cells, BC as a kind of heterogeneous disease has divergent histological subtypes and biological characteristics, thus leading to distinct clinical behaviors and treatment sensitivity profiles (3). Although the recent success of immunotherapy has paved the way for various solid or hematological malignancies, most subtypes of BC exhibit little efficacy to immunotherapy with immune checkpoint inhibitors only approved in combination therapy for PD-L1-positive metastatic triple-negative breast cancer (TNBC) (4). Poor immunogenicity, lack of T-cell infiltration, and an immunosuppressive tumor microenvironment (TME) have been identified as major barriers to the success of immunotherapy in BC (4). The interaction between tumor cells and the immune TME is a complex, dynamic, and evolving process; thus, conventional tumor characteristics and biomarkers may not be adequate to predict immunotherapy effectiveness and prognostication. Data across large BC clinical trials supported that the high levels of tumor-infiltrating lymphocytes (TILs) are predictive biomarkers for favorable prognosis and of the response to immunotherapy, particularly in HER-2⁺ BC and TNBC (5). Besides TILs, recent evidence revealed that spatial organization plays a crucial role in determining prognosis and response to immunotherapy, with tertiary lymphoid structures (TLSs) attracting widespread attention (6, 7).

TLSs are ectopic cellular aggregates in nonlymphoid tissues under conditions of chronic inflammation including tumors, and share similar architectural and functional characteristics with secondary lymphoid organs (SLOs) (8). The architecture of mature TLSs is characterized by B-cell-enriched zones that consists of B-cell follicles surrounded by a network of follicular helper T cells and follicular dendritic cells, T-cell-enriched regions with dendritic cells (DCs), high endothelial venules (HEVs), as well as lymphatic vessels (6, 7). In addition to the relevant number of immune cells, TLSs emphasize the spatial proximity of specialized subsets of immune cells within TLSs. In contrast to SLOs, TLSs represent privileged sites for local lymphocyte differentiation and antigen presentation, which provide an important milieu for both cellular and humoral antitumor immunity (7). Accumulating research has indicated that TLS presence was deeply associated with positive immunoreactivity and favorable clinical outcomes in most types of solid tumors (6). However, some studies evaluated the prognostic value of TLSs limited to small study numbers and subsets of BC, with inconsistent and conflicting results. Although a previous meta-analysis by Zhang et al. suggested that TLSs were related to better prognosis, their result was based on a limited number of studies, with only two or three studies providing survival outcomes (9). Furthermore, all included studies in their meta-analysis showed that TLSs were beneficial for prognosis, but opposite conclusions have been reported in the recent study (10).

Hence, with the publication of new studies regarding this topic, further evaluation of the role of TLSs in BC is necessary. This study including more than 15 articles aimed to comprehensively assess clinicopathological and prognostic values of TLSs in BC, providing higher-level medical evidence for clinical practice. Simultaneously, given the unique difficulties in the detection and quantification of TLSs, the TLS-related gene signature based on the TCGA BC cohort was further used to validate and supplement our results.

The present study was performed in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) criteria (11). The protocol of this meta-analysis was registered in the PROSPERO (registration number: CRD42022302921).

As a complex network composed of a variety of immune subsets, the tumor immune microenvironment exerts a great impact on immunotherapeutic efficacy and prognosis (33). TLSs have attracted increasing attention as a unique structure of the TME. TLSs not only are prognostic biomarkers of improved clinical outcome among cancer patients but also shape a local and favorable site for generating antitumor humoral and cellular immune responses (6, 8). However, several studies exploring the impact of TLS on prognosis and tumor progression were limited to small study numbers and subsets of BCs, of which the results are conflicting and lack more comprehensive evaluations. To the best of our knowledge, this is the most comprehensive meta-analysis including 15 articles to assess the clinicopathological and prognostic value of TLSs in BC.

The prognosis of BC is well recognized to be influenced by host- and tumor-associated factors (age, tumor size, histological grade, lymph node, hormone and growth receptor status, etc.) (19). First, we synthesized eleven pieces of research to evaluate the correlation between TLSs and clinicopathological parameters in BC ( Figure 3 ). Our results suggested that the presence of TLSs was correlated with early TNM stage. Consistent with this, the density of TLSs was also found to be obviously increased in early TNM stage in oral squamous cell carcinoma and NSCLC (34, 35). A positive association was found between the presence of TLSs and TIL levels in our study, which might be associated with TLS function. Being nonencapsulated and close to tumor tissues compared to draining lymph nodes, TLSs facilitate rapid migration of APCs to TLSs and presentation of antigen peptides to T-cell APCs at the site of the tumor (36). Indeed, some studies also reported that TILs were the strongest independent factor predicting TLSs, but not all cases with high TILs showed TLS formation (37). We also found that TLSs were negatively related to ER and PR status, but were correlated with high expression of HER-2 and Ki-67. These results were in line with previous studies, which revealed that increased TILs are inversely related to the expression of ER or PR, and are positive with HER-2 status, the pathologic complete response rate, and improved survival outcomes (25, 38). In the current study, we did not find the relationship between TLSs and age, tumor size, LVI, or histologic grade.

We then systematically evaluated the prognostic impact of TLS presence on BC patients ( Figure 4 ). Our meta-analysis describes that HR = 0.68 for OS and HR = 0.54 for DFS, both of which were statistically significant. The study revealed that patients with TLS presence had better survival outcomes regarding DFS and OS. It was worth noting that that sensitivity analyses revealed that our results were reliable and robust, but moderate heterogeneity between included studies was observed in survival outcomes, which can be caused by different baseline characteristics of individual studies. Therefore, subgroup analyses were performed using median age, ethnicity, sample size, source of data, and detection method to explore the potential heterogeneity ( Table 2 ). The results revealed that ethnicity, source of data, and/or detection method may be a source of heterogeneity. Therefore, it is worth noting that TLSs are hardly accurately identified by H&E staining alone, and IHC with TLS markers is typically necessary to evaluate TLS characteristics. Moreover, the heterogeneity among studies might be due to the complex subtyping of BC. Recent studies have suggested that maturation degrees and distribution of TLSs are critical to determine the impact of TLSs on prognosis. However, due to a lack of data, subgroup analysis could not be conducted to assess the impact of different maturation degrees and distributions of TLSs on survival outcomes. A high proportion of mature TLSs containing GCs was associated with better prognostic outcome than total TLSs, and the prognostic value of TLSs was lost while GC formation was impaired (39, 40). TLSs could localize to the core of tumor tissues called intratumor TLSs and/or the invasive margin of tumor tissues, known as peritumor TLSs (41). Several studies have indicated that the density of peritumor TLSs is associated with improved prognoses, whereas there are a few opposite results. Sofopoulos et al. described that patients with invasive ductal carcinoma having peritumoral TLSs exhibited worse DFS and OS than patients lacking TLSs (10). High levels of tumor-infiltrating Treg cells observed at the peritumoral areas were demonstrated to be correlated with relapse and death in BC patients (42).

Moreover, given the unique difficulties in TLS detection and quantification, we validated and supplemented the results of our analysis by TLS-related gene signature in BC patients ( Figure 6 ). Accumulating evidence has confirmed that TLSs are highly associated with immune cell infiltration, which closely have an impact on the development, progression, and prognosis as well as the treatment of BC (43). Hence, the immune score, stromal score, and ESTIMATE score of BC samples were estimated via the ESTIMATE algorithm. Higher immune, stromal, or ESTIMATE scores and lower tumor purity were found in patients of the high-TLS signature group than those in the low-TLS signature group. Simultaneously, we observed that most of the 29 immune subsets, which represented immune cell types, functions, and pathways, in the high group were more abundant compared to the low group. Interestingly, the immunosuppressive subsets like Treg cells, which might lead to poor outcomes, were also higher in the high group. Indeed, immunosuppressive cells are also components of TLSs, and associations of TLSs with immunosuppressive cells have been reported in various solid tumors including BC, lung cancer, and melanoma (15, 42, 44). There was evidence that TLSs in combination with “immunoscore” defined by intratumoral immune cells might provide a comprehensive and most powerful prognosticator. Li et al. found that TLSs combined with CD8⁺ T cells and CD57⁺ NK cells provided a higher predictive prognostic accuracy (45). It was still noteworthy that all major checkpoint genes were obviously upregulated in the high-TLS signature group compared with the low-TLS signature group, suggesting that patients with high expression of TLS signature were more likely to benefit from immunotherapy. A study by Cabrita et al. observed that TLS-rich tumors in particular were related to significantly increased survival after CTLA4 inhibitor on the basis of the TLS signature (15). TLS-rich tumors were more infiltrated by CD8⁺ T cells, and these T cells might be depleted, explaining the correlation between immune checkpoint expression and TLSs and why checkpoint inhibitor might result in productive anti-tumor immunity in TLS-rich tumors (46). Intriguingly, checkpoint inhibitor therapy might also promote the formation of TLSs. Analysis of on-treatment tumor biopsies of urothelial carcinoma and melanoma has shown that tumors of responding patients showed a higher number of TLS-associated B cells relative to matched pretherapy samples after neoadjuvant immune checkpoint blockade (8). All these results demonstrated the significant correlations with TLS signature representing the major component of TLSs, which revealed that the 9-gene signature can efficiently reflect TLS enrichment in the TME. Our study also demonstrated that BC patients with a high TLS signature expression displayed improved survival, which showed that TLS signature could act as a favorable prognostic factor for BC patients. Based on the above results and discussion, multiple measures including chemotherapy, immunostimulants, vaccination, and TLS-associated cytokines and chemokines have been applied to explore the induction of TLS formation (40, 47). Considering some immunosuppressive factors such as regulatory T and B cells that impaired the antitumor of TLSs reported from recent studies, therapeutic strategies to induce TLS formation and maturation while inhibiting immunosuppressive factors might create bright prospects for enhancing tumor immunotherapeutic response (48).

This present study as the most comprehensive meta-analysis provides more substantial evidence for clinicopathological and prognostic significance of TLSs in BC. However, important considerations should be emphasized while interpreting the conclusions of this study. The cellular components, locations, and maturation degrees of TLSs might dictate treatment efficacy, tumor recurrence, and patient survival. The heterogeneity of the means used to quantify TLSs further confound their use in the clinic. Because the number of retrieved studies was not sufficient to be analyzed depending on the detection methods, no restriction was placed on the detection methods. Different scoring methods, scoring systems, and thresholds might lead to different results. Other limitations of our study were also noteworthy. First, partial survival data unavailable in the original article were extracted from Kaplan–Meier curves, which are less reliable than data directly acquired from research. Secondly, compared to multivariate analysis, data from univariate analysis may overestimate the effect sizes. Third, all the research data were derived from Asian and Caucasian patients. Accordingly, the global representation of data is insufficient and lacking. Finally, all studies included were retrospectively conducted and might have inherent structural biases. Therefore, prospective randomized trials are required to validate our results in the future.

In conclusion, TLS presence provides new insight into the clinicopathological features and prognosis of patients with BC. The presence of TLSs might have the potential to predict prognosis of BC patients, whereas factors discussed above limited the evidence quality of this study. We look forward to consistent methods to define and characterize TLSs, and more high-quality prospective clinical trials designed to validate the prognostic and predictive value of TLSs alone or in combination with other markers.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

Conception and design: BW, JL, and YJ. Wrote the manuscript: BW and JL. Acquired data: YH, YD, and J-ZL. Analyzed the data: BW. Discussed the results and implications of findings: JL, YD, and YH. Interpretation of data: BW and J-ZL. YJ performed critical evaluation, verification of the manuscript, and supported funding. All authors contributed to the article and approved the submitted version.

This work was funded by the National Natural Science Foundation of China (Grant Number 81903181) and the 1·3·5 Project for Disciplines of excellence–Clinical Research Incubation Project, West China Hospital (Grant Number 2021HXFH019).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.