Introduction

Tertiary lymphoid structures (TLSs) are immune cell clusters that resemble secondary lymphoid organs (SLOs). TLSs form postnatally in non-lymphoid tissues, often at sites of chronic antigen exposure, including cancer, organ transplants, and autoimmune conditions1,2. TLSs are characterized by distinct T and B cell zones, along with lymph node-like vasculature that expresses peripheral node addressin (PNAd), including mucosal addressin cell adhesion molecule 1 (MAdCAM1), vascular cell adhesion molecule 1 (VCAM1), and intercellular adhesion molecule 1 (ICAM1)3,4. They support local antigen presentation and the production of effector and memory T cells, promoting both humoral and cellular antitumor immune responses5. Growing evidence links TLS presence to improved prognosis and enhanced response to immune checkpoint inhibitors (ICIs) in various cancer types4,6,7. Additionally, the heterogeneity of TLSs, including their density, localization, and maturity, significantly impact their structural composition and immune function, shaping their role in the cancer immune landscape8.

Despite some studies on TLSs in melanoma, most of them have focused on TLSs in metastatic lymph nodes (LNs) or other organs, with limited analysis of primary lesions9,10,11.Additionally, many previous studies relied on tissue microarrays, which fail to display tissue structures as clearly as digital pathology slide scanning9,11. As a result, the role of TLSs, especially in primary melanoma lesions, in prognosis and immunotherapy response remains unclear.

Acral melanoma (AM) is a unique subtype of melanoma derived from melanocytes in sun-shielded areas, such as the palms, soles, and nails, and is the predominant subtype in Asians12. Compared with cutaneous melanoma (CM), AM has a lower tumor mutation burden (TMB) and rarely possess BRAF mutations. It is also associated with a markedly immunosuppressive environment, characterized by CD8+ T cell depletion, increased regulatory T (Treg) cells, and CD8+ T cell exhaustion13. Consequently, AM shows a poorer response to anti-PD1 monotherapy compared to CM14,15. Although TLSs have been extensively studied in melanoma, research on TLSs in AMs remains limited.

To address these gaps, we assembled a cohort of 46 AM patients and investigated TLS heterogeneity in primary lesions using hematoxylin and eosin (HE) staining and multiplex immunofluorescence (mIF). We analyzed the relationships among TLS characteristics, overall survival (OS), and immune cell infiltration within the tumor microenvironment (TME). Our findings suggest that the presence of intra-TLSs in primary lesions is associated with improved OS and enhanced immune infiltration in AMs.

Results

Baseline characteristics

A total of 46 patients with AM were included in the analysis (Table 1). The median age was 68 years (range: 43–86). The cohort included 22 males (47.8%) and 24 females (52.2%). According to the TNM staging system, 5 patients (10.8%) were stage I, 22 (47.8%) were stage II, and 19 (41.4%) were stage III. The median follow-up was 31.4 months (range: 7.7–124 months).

Table 1 Clinicopathological factors of included patients
Full size table

The number, localization, and maturation of TLSs within the TME were assessed by a pathologist using HE staining and mIF (Fig. 1A–D). TLSs were identified in 41 patients, both peritumoral and intratumoral. The number of each feature of the TLS is shown in Supplementary Table 1. No significant differences in intra-TLS or peri-TLS proportions were noted between different stages or sexes, however intra-TLS proportions were greater in stages I-II (Supplementary Fig. 1A, B). Quantification revealed that patients in stages I-II had significantly more intra-TLSs and eTLSs than did those in stage III (P = 0.053, P = 0.036, respectively) (Table 2). No significant differences in the number of different features of TLSs were observed between sexes.

Fig. 1: Representative structure of TLS in AM.
figure 1

A Representative H&E staining image of TLS (Indicated by the red arrow, Scale bars: 20×). B Representative mIF staining image of TLS (Markers: CD20, CD21, CD23. Scale bars: 100 μm). C Representative mIF sta ining image of TLS (Markers: CD4, CD8, CD68, CD10; Scale bars: 100 μm). D Representative mIF staining image of eTLS, pTLS, sTLS (left, middle, right column, respectively; Markers: CD20, CD21, CD23; Scale bars: 100 μm). TLS tertiary lymphoid structures, AM acral melanoma, mIF multiplex immunofluorescence, eTLS early tertiary lymphoid structure, pTLS primary follicle- like TLS, sTLS secondary follicle- like TLS.

Full size image
Table 2 Association of clinical features and number of tertiary lymphoid structures in TME
Full size table

Features of TLSs correlate with OS

To examine the relationship between TLS features and OS in AM, we categorized patients on the basis of TLS number, localization, and maturation. Kaplan-Meier analysis revealed that patients with TLSs had significantly longer OS than did those without TLSs (P = 0.032) (Fig. 2A). However, grouping patients on the basis of the median number of TLSs in the TME (high vs. low) revealed no significant difference in OS (P = 0.29) (Fig. 2B). Compared to patients without intra-TLSs, patients with intra-TLSs had significantly longer OS (P = 0.037) (Fig. 2C), whereas whereas no significant difference in OS was noted between groups with and without peri-TLSs (P = 0.28) (Fig. 2D). Finally, TLS maturation did not correlate with OS, as no significant differences were observed among eTLS, pTLS, and sTLS (P = 0.26, P = 0.91, and P = 0.14, respectively) (Supplementary Fig. 2A–C).

Fig. 2: Kaplan-meier analysis of the relationship between TLS features and overall survival in AM Patients.
figure 2

A Kaplan-Meier survival analysis of the cohort grouped by the presence of TLS. B Kaplan-Meier survival analysis of the cohort grouped by the number of TLS. C Kaplan-Meier survival analysis of the cohort grouped by the presence of intratumoral TLS. D Kaplan-Meier survival analysis of the cohort grouped by the presence of peritumoral TLS. TLS tertiary lymphoid structures, AM acral melanoma.

Full size image

TLS features correlate with intratumoral immune infiltrates

Research has shown that TLSs are associated not only with OS but also with the response to ICB therapy. The response to ICB therapy is closely linked to the TME, which includes immune cells, tumor cells, and fibroblasts, all of which interact continuously16. Therefore, we analyzed the relationships between TLS features and immune cell proportions and distributions in the TME of AMs (Fig. 3A, Supplementary Table 2). Intra-TLSs presence was associated with increased T cell (CD4+ and CD8+) and CD8+ T cell infiltration in the TME (Fig. 3B, C). Conversely, peri-TLSs was correlated with a higher proportion of CD68+ macrophages (Supplementary Fig. 3B).

Fig. 3: Association between intratumoral TLS and immune infiltrates in AM.
figure 3

A Representative mIF image of immune infiltrates in AM. B Proportion of T cells (CD4+ and CD8+). C Proportion of CD8+ T cells. D Number of CD8+ T cells within 50μm of SOX10+ tumor cells. E Number of CD68+ Macrophages within 20μm of SOX10+ tumor cells. F Number of CD68+ Macrophages within 20 μm of CD8+ T cells. G Number of CD68+ Macrophages within 50μm of CD8+ T cells. H Nearest distance from CD8+T cells to CD68+ macrophages (unit: μm). TLS tertiary lymphoid structures; AM acral melanoma; mIF multiplex immunofluorescence.

Full size image

On the basis of the assumption that two cells are more likely to interact if the distance between their nuclei is less than 20 μm17 and the findings of Alberto et al. 18 linking the proximity of CD8+ T cells to cancer cells (within 50 μm) with the response to ICB therapy in urothelial cancer, we quantified immune cell numbers within 20 μm and 50 μm of tumor cells. We found that, compared with patients without intra-TLSs, those with intra-TLSs had significantly more CD8+ T cells within a 50 μm radius and fewer CD68+ macrophages within a 20 μm radius around tumor cells (Fig. 3D, E). Patients with sTLSs had more CD8+ T cells within 50 µm of tumor cells (Supplementary Fig. 3C) and within 20 µm of CD68+ macrophages (Supplementary Fig. 3D).

CD8+ T cells and CD68+ macrophages play key roles in the TME, and their interactions influence tumor immunity. Therefore, we analyzed the relationships between TLS features and the interactions between CD8+ T cells and CD68+ macrophages. Our study revealed that, compared with patients without intra-TLSs, those with intra-TLSs had more CD68+ macrophages within 20 µm and 50 µm of CD8+ T cells (Fig. 3F, G) and a shorter spatial distance between these two cell types (Fig. 3H). Additionally, TLS presence was associated with a greater proportion of CD4+ T cells (Supplementary Fig. 3A).

Role of the TLS signature gene score in AM

We analyzed transcriptomic sequencing data and clinical information from Liang WS et al.‘s study on AM19 to assess the impact of nine TLS-associated signature gene sets on OS and disease-free survival (DFS) (Fig. 4A, B). The results revealed that both the Th1/B cell signature gene set and the TLS Germinal Center Mature gene set were significantly associated with OS and DFS (Th1/B cell signature: OS: P = 0.029; DFS: P = 0.039; TLS Germinal Center Mature: OS: P = 0.012; DFS: P = 0.041). In contrast, the TLS gene set was significantly associated with OS (P = 0.027) but not with DFS (P = 0.12). The other gene had no significant impact on OS or DFS. When we evaluated the effects of these gene sets on OS using data from Farshidfar et al. 20, none were found to be associated with OS (Supplementary Fig. 4).

Fig. 4: Association of nine TLS-associated signature gene sets with prognosis and immunotherapy efficacy in AM patients.
figure 4

A Kaplan-Meier curve analysis of the relationship between TLS-associated signature gene sets and OS. B Kaplan-Meier curve analysis of the relationship between TLS-associated signature gene sets and DFS. C Differences in TLS-associated signature gene scores between the response and non-response groups to immunotherapy. TLS tertiary lymphoid structures, AM acral melanoma, OS overall survival, DFS disease-free survival.

Full size image

Given the influence of TLSs on immunotherapy, we further explored the relationship between nine gene sets—particularly the Th1/B cell signature and TLS Germinal Center Mature gene sets—and immunotherapy efficacy in AMs. Our analysis revealed that the Th1/B cell signature gene set was more strongly correlation with immunotherapy efficacy than the TLS Germinal Center Mature gene set (Fig. 4C).

Discussion

In this study, we observed TLSs in the TME of 41 out of 46 (89.13%) treatment-naïve patients with stage I-III AM. This prevalence is significantly higher than that reported in previous melanoma studies. For example, Cipponi A et al. 21 reported TLSs in 14 out of 29 (48%) melanoma patients. However, their study included various melanoma subtypes and focused primarily on lymph node metastases. Similarly, Lynch et al. 11 identified TLSs in 30 out of 64 (47%) melanoma patients used tissue microarrays to analyze metastatic samples. In contrast, our study focused on primary tumor samples from patients wiht stage I-III AM, employing mIF on whole pathological sections. Vanhersecke et al. reported that mIF, along with HE and double CD20/CD23 staining, was the most sensitive method for detecting TLSs22. They also reported a 2.0-fold greater likelihood of detecting TLSs in primary tumor samples compared to metastases. Among the 41 patients with TLSs in our study, 10 (24.4%) had sTLSs, which is consistent with the findings of Lynch et al. 11. Further analysis of the relationships between clinical characteristics and TLS numbers revealed that patients in stages I-II had greater numbers of intra-TLSs and eTLSs in the TME. A similar phenomenon was observed in non-small cell lung cancer, in which the number of TLSs significantly decreased in stage III patients as compared to stage II23. While various molecular components, such as cytokines, chemokines, and adhesion molecules, are known to play key roles in TLS formation, the underlying mechanisms remain unclear24. Notably, although each chemokine independently induces TLS, it produces structures with distinct characteristics. For example, CXCL13 induces B cell aggregates that lack FDC networks, whereas TNF and CXCL12 induce small lymphocytic infiltrates, predominantly B cells with few T cells, and, in the case of CXCL12, DCs25,26. Furthermore, the CCR7 ligands CCL19 and CCL21 induce similar aggregates, but CCL21 produces larger, more organized structures26.

Advances in mIF now allow for detailed evaluation of TLSs in the TME, including their size, density, localization, composition, and maturity. TLS presence has been linked to a better prognosis in cancers such as lung adenocarcinoma27, melanoma11, and dedifferentiated liposarcoma28. High-density TLSs are associated with a better prognosis in non-metastatic colorectal cancer29. In cancers such as intrahepatic cholangiocarcinoma, breast cancer, and hepatocellular carcinoma, intra-TLS is correlated with a better prognosis, whereas peri-TLS is linked to poorer outcomes30,31,32,33. In our study, only intra-TLSs in AM patients were associated with improved OS, while peri-TLSs showed no correlation. Additionally, neither the number nor the maturity of TLSs affected OS. These results are limited by the small sample size, highlighting the need for larger studies to further investigate the role of TLSs in patient outcomes.

TLSs play a crucial role in the antitumor immune response by facilitating antigen presentation to T cells via dendritic cells (DCs) and promoting the activation, proliferation, and differentiation of T and B cells4. T cells within TLSs exhibit increased activation markers compared with other tumor-resident T cells in melanoma34. In murine melanoma models, TLS-driven tumor-specific T cell responses led to immune cell infiltration and tumor regression35,36. Additionally, TLS-associated fibroblasts and other cell subsets secrete survival factors that support lymphocyte homeostasis, potentially contributing to the persistence of tumor-reactive T cells37,38. Our study found that patients with intra-TLSs had significantly greater proportions of T and CD8+ T cells in the TME compared to those without intra-TLSs. Similarly, a melanoma study revealed increased densities of CD8+ T cells, B cells, and plasma B cells in TLS+ cutaneous melanoma metastases (CMMs). They also observed that intratumoral CD8+ T cell density was correlated with a higher proportion of sTLSs, but not eTLSs or pTLSs11.

Although genetic and epigenetic alterations are commonly considered the root causes of cancer, their progression is also closely linked to crosstalk among tumor cells, immune cells, and the extracellular matrix (ECM)39. CD8+ T cells and tumor-associated macrophages (TAMs) are key immune cells in the TME. CD8+ T cells typically exert antitumor effects, while TAMs often promote tumor progression. The interactions between these cells and tumor cells influence cancer prognosis and affect the efficacy of therapeutic interventions40. Zhu et al. found that higher densities of CD8+ tumor-infiltrating lymphocytes within 20 μm of melanoma cells were associated with a good response to anti-PD-1 immunotherapy and improved progression-free survival (PFS)41. Wang et al. showed that PD-L1- TAMs, compared with PD-L1+ TAMs, interact more frequently with cancer cells and promote tumor progression17. Our study revealed that patients with sTLSs had more CD8+ T cells surrounding tumor cells, suggesting that TLS maturation may be linked to CD8+ T cell infiltration around tumors, despite the lack of an overall increase in CD8+ T cell proportion in the TME. Crosstalk between TAMs and CD8+ T cells is commonly observed in the TME. TAMs can modulate T cell responses by releasing CXCL9, CXCL10, IL-12, IL-15, and IL-1842,43, or by expressing T cell immunoglobulin and mucin ___domain-containing molecule 4 (TIM4) to constrain CD8+ T cells proliferation and activation44. FOLR2+ macrophages in perivascular tumor areas interact with CD8 + T cells and efficiently prime effector CD8+ T cells45. However, Kersten et al. reported that TAMs and CD8+ T cells engage in long-lasting, antigen-specific synaptic interactions that do not activate T cells but rather prime them for exhaustion46. In our study, patients with intra-TLSs had a significantly higher number of CD68+ macrophages surrounding CD8+ T cells in the TME, and a closer distance between the two cell types. However, the role of the interaction between CD68+ macrophages and CD8+ T cells in melanoma still requires further investigation.

Our study, along with previous research, demonstrated that TLSs serve as both prognostic markers and indicators of the tumor immune microenvironment (TIME), making them promising targets for cancer therapy. Understanding TLS formation and maturation is crucial, and recent advances in spatial transcriptomics and proteomics offer new insights. We compiled nine TLS signature gene sets, including the Th1/B cell and TLS Germinal Center Mature signatures, which predict patient prognosis and immunotherapy efficacy. The Th1/B cell signature shows strong potential for predicting treatment response, suggesting that Th1-B cell interactions play a key role in TLS formation. The immune infiltrate in tumors in which TLSs are present is often skewed toward a TH1 or cytotoxic effector state47. However, owing to the lack of spatial data in AM, studying these gene sets at the spatial level remains a challenge. As omics technologies advance, they will deepen our understanding of TLSs and accelerate progress in precision medicine.

In summary, this study is the first to comprehensively assess TLSs in AM patients and their clinicopathological associations. Our findings suggest that intra-TLS can serve as a prognostic marker and may contribute to an immune-responsive microenvironment. These insights open the door for therapeutic strategies aimed at enhancing TLS function, potentially improving clinical outcomes and increased responses to immunotherapy. However, due to the limited sample size, further studies with a larger cohort are needed to confirm and provide deeper insights into how TLSs influence immune cell infiltration in AMs.

Methods

Patients and clinical data collection

This study included a cohort of 46 patients diagnosed with acral melanoma (AM), all of whom were treated at the Nanjing Drum Tower Hospital, Affiliated Hospital of Medical School, Nanjing University. Patient enrollment spanned from January 2013 to June 2021, with a follow-up period extending until December 2023. All participants self-reported no prior history of cancer or any antitumor therapy before the surgical resection of their primary lesion. The clinicopathological features evaluated in this study included age, sex, and tumor stage (Table 1). The clinical stage of AM was determined in accordance with the 7th edition of the American Joint Committee on Cancer (AJCC) staging system. Overall survival (OS) was defined as the time interval from the date of surgical treatment to the date of patient death during the follow-up period. This study adheres to the Guidance of the Ministry of Science and Technology (MOST) for the Review and Approval of Human Genetic Resources. All patient materials were sampled and handled in compliance with the ethical principles of the Declaration of Helsinki. This included obtaining written informed consent from all donors and securing approval from the Institute Research Ethics Committees of the Affiliated Drum Tower Hospital of Nanjing University Medical School (Approval No. 2019-312-01).

Multiplex immunofluorescence staining and image analysis

For mIF analysis, we used a PANO 5-plex IHC kit based on tyramide signal amplification (TSA) following the manufacturer’s protocol (Panovue, China). Consecutive 4-μm thick sections from formalin-fixed paraffin-embedded (FFPE) tumor specimens were stained. Multispectral images of each stained slide were captured using the PanoVIEW VS200 System (Panovue, China). Acidic and alkaline antigen retrieval buffers (Panovue, China) were applied as necessary. The staining sequences and antibodies used are listed in Supplementary Table 3.

The multispectral images were spectrally unmixed using single-stain positive control images with the InForm software (Akoya Biosciences, Marlborough, Massachusetts, USA). The immune cell proportions in the tumor microenvironment (TME) were quantified as the percentage of positively stained cells among all nucleated cells. The normalized proportion of immune cells was calculated by dividing the immune cell proportion by the tumor cell proportion (e.g., the proportion of CD8+ T cells = CD8+/SOX10+ cells). Data for downstream analysis included the position of each cell (x- and y-coordinates of the nuclei) and its corresponding cell type. Spatial relationships between cells in the TME were examined using the first-nearest neighbor (1-NN) statistic, implemented in spatstats. mIF staining assay results were validated in at least three biological replicates.

Defining TLS and TLS features

First, lymphoid aggregates were observed on whole-slide images (WSIs) (Fig. 1A). TLSs feature a central CD20+ B cell zone encircled by CD3+ T cells7. Using mIF with CD20/CD21/CD23 markers, we first delineated the inner B cell zone (Fig. 1B). Subsequent mIF analysis of consecutive sections stained with CD4/CD8/CD68 confirmed the peripheral organization of CD4+ and CD8+ T cells around this core structure (Fig. 1C), thereby establishing TLS identification. The tumor invasive margins were identified using a combination of H&E staining and mIF (tumor cells: SOX10+ cells). TLS were classified into peritumoral TLSs (peri-TLSs) and intratumoral TLSs (intra-TLSs) on the basis of their ___location relative to the invasive margins of the tumor7. TLS maturation is a dynamic, multi-stage process starting from immature early TLSs (eTLSs), to primary follicle-like TLSs (pTLSs), to secondary follicle-like TLSs (sTLSs)48. As previously described11, the maturation state of TLSs can be categorized into eTLS, pTLS, and sTLS on the basis of the presence of a CD21+ follicular dendritic cell (FDC) network (pTLS), a CD23+ germinal center (sTLS), or the absence of these features (eTLS) (Fig. 1D). Based on the varying maturation states of peritumoral TLSs, they were further categorized into peritumoral eTLSs, pTLSs, and sTLSs. Similarly, intratumoral TLSs were classified into intratumoral eTLSs, pTLSs, and sTLSs.

Defining the signature score of the TLS signature gene

Single-sample gene set enrichment analysis (ssGSEA)49 was used to calculate the absolute enrichment score of the TLS signature genes, which were derived from relevant studies and are listed in Supplementary Table 450,51,52,53,54,55.

Statistical analysis

All the statistical analyses were conducted using R version 4.2.3, SPSS version 25, and GraphPad Prism version 9. The Student’s t-test and the Mann–Whitney U test were utilized to calculate P values. Overall survival was analyzed using Kaplan–Meier and log-rank survival analyses. All the statistical tests were two-sided, with P < 0.05 considered statistically significant.