Correlation analysis of tertiary lymphoid structure parameters with the prognosis of patients with locally advanced rectal cancer after neoadjuvant chemotherapy: a retrospective study

Abstract

Background

The tertiary lymphoid structures (TLSs) are positively correlated with the prognosis of many solid tumors, including colorectal cancer. However, their prognostic significance in patients with locally advanced rectal cancer (LARC) after neoadjuvant chemotherapy remains unclear. This study aimed to explore the correlation between TLS parameters and the prognosis of LARC patients receiving neoadjuvant chemotherapy.

Methods

This retrospective study included patients with LARC treated at the Harbin Medical University Cancer Hospital from 2012 to 2021. The quantity, area, and density of TLSs in the tumor, normal, and total tissues from surgical specimens were determined. Overall survival (OS) was calculated from surgery to death from any cause. The correlation between TLS parameters and prognosis was assessed using Kaplan-Meier survival analysis and Cox regression analysis. Multiplex immunofluorescence (mIF) staining was used to analyze TLS maturity and immune composition.

Results

This study included 114 patients, of whom 46.5% were over 60 years old, and 70.2% were male. TLS parameters in tumor region were smaller than those in normal and total regions (P < 0.001). A larger TLS area and higher density in the total region (HR = 0.371, P = 0.023 for area; HR = 0.250, P = 0.005 for density) were significantly associated with better OS. Moreover, a higher total-region TLS density was correlated with low carcinoembryonic antigen (CEA) levels (P = 0.028), positive responses to neoadjuvant therapy (P < 0.001), and tumor regression (P < 0.001). Subgroup analysis revealed that combining total-region TLS density with clinicopathologic features such as sex, age, cTNM stage, CEA levels, and extramural vascular invasion further stratified prognosis. Additionally, mIF analysis showed that a high TLS density was associated with a higher TLS maturity (P = 0.014); mature TLSs exhibited greater infiltration of CD20⁺ B cells and CD21⁺ follicular dendritic cells compared to non-mature TLSs.

Conclusions

TLS parameters, particularly TLS density, are promising prognostic biomarkers for LARC patients undergoing neoadjuvant chemotherapy.

Trial registration

not applicable.

Introduction

Rectal cancer is the eighth most common cancer worldwide (estimated 732,210 new cases in 2020, or 3.8% of all cancers) and is the ninth cause of cancer-related death (estimated 339,022 deaths in 2020, or 3.4% of all cancer-related deaths) [1]. Despite the declining incidence and mortality rates of rectal cancer over the past few decades, the prognosis for patients diagnosed with locally advanced rectal cancer (LARC) remains poor, with a 5-year survival rate of 71% following optimal treatment, compared to 90% for those with localized disease [2]. Accurately predicting the prognosis of rectal cancer is crucial for effective treatment [3].

Tertiary lymphoid structures (TLSs) are histological features observed at sites of chronic inflammation, including cancer [4]. Increasing evidence suggests that TLSs can stimulate anti-tumor immune responses, and their presence is positively associated with the prognosis of solid tumors, independent of tumor, node, metastasis (TNM) staging and other prognostic factors [5,6,7]. This beneficial role of TLS has also been observed in colorectal cancer [8]. TLSs are associated with lower local recurrence rates, reduced distant metastasis, and improved survival in colorectal cancer patients [7, 9, 10].

However, previous studies have rarely distinguished between colon cancer and rectal cancer. Research has shown that colorectal cancers at different primary sites exhibit distinct tumor microenvironment (TME) immune phenotypes [11]. Proximal and distal colorectal cancers display various biological and clinical differences, which in turn affect disease prognosis [12]. Considering the heterogeneity of colorectal cancer, individually exploring TLS expression in LARC patients and its impact on prognosis can help to understand its biological role and clinical significance in different tumor sites, thereby providing more precise guidance for clinical treatment. In addition, the Preoperative Radiation or Selective Preoperative Radiation and Evaluation Before Chemotherapy and TME (PROSPECT) trial suggested that neoadjuvant chemotherapy may serve as an alternative to neoadjuvant chemoradiotherapy [13, 14]. However, previous research on LARC has primarily focused on the context of neoadjuvant chemoradiotherapy [15]. The prognostic value of TLSs in LARC patients treated with neoadjuvant chemotherapy alone warrants further investigation. Moreover, TLSs may interact with various antitumor agents in different ways. Nonetheless, no previous studies have examined the impact of different chemotherapy regimens on the prognostic value of TLSs in LARC patients [16].

In this study, we retrospectively collected and examined TLSs in the surgical specimens of LARC patients who underwent neoadjuvant chemotherapy. We hypothesized that TLS parameters (quantity, area, and density) are positively correlated with the prognosis of LARC patients.

Methods

Study design and patients

This single-center retrospective study included 114 patients with LARC who were treated at the Affiliated Cancer Hospital of Harbin Medical University between 2012 and 2021. The inclusion criteria were (1) a diagnosis of LARC confirmed by clinical, pathological, and radiological examinations; (2) preoperative neoadjuvant chemotherapy with capecitabine + oxaliplatin (XELOX) and 5-fluorouracil + folinic acid + oxaliplatin (FOLFOX6); (3) available surgical specimens; (4) available follow-up records; and (5) complete clinical information. Patients who had received preoperative radiotherapy were excluded from the study. The patient enrollment flowchart is shown in Figure. S1 in the Supplementary Materials.

Data collection

The clinical data were extracted from the patient charts. Demographic information included age and sex. Clinicopathological features included clinical staging (cTNM), mesorectal fascia (MRF) involvement, extramural vascular invasion (EMVI), lymph node involvement, response to neoadjuvant chemotherapy, and tumor regression grade. Additionally, preoperative inflammation markers, including CEA and carbohydrate antigen (CA) 199 levels, were collected. The primary outcome variable was overall survival (OS), which was calculated from the date of surgery to death from any cause. If no survival event occurred, the follow-up was censored at the date of last contact.

Response to neoadjuvant chemotherapy was assessed according to the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 [17] and categorized into two groups: response (complete response or partial response) and non-response (stable disease or progressive disease). Additionally, tumor regression grade was determined based on the American Joint Committee on Cancer (AJCC) / College of American Pathologists (CAP) tumor regression grading system [18] and classified into three groups: regression (grade 0), partial regression (grades 1–2), and no regression (grade 3).

Tertiary lymphoid structures

TLSs were defined as dense aggregates of B cells with an adjacent T-cell zone without a surrounding capsule [4]. Surgical specimens from patients with rectal cancer were collected. The presence of TLSs within these specimens was assessed using H&E staining. Sections were scanned and archived using the Hamamatsu digital slide scanner (#C12000-02; Hamamatsu Photonics K.K., Tokyo, Japan) and analyzed using the DNP.view2 software.

The identification and parameter calculation process for TLS is as follows: First, experienced pathologists delineate the tumor region (characterized by pronounced cellular atypia, disordered arrangement, enlarged nuclei, and increased pathological mitosis) and the normal region (areas outside the tumor) based on the cell morphology observed in HE staining. TLSs were then classified as tumor-region TLSs and normal-region TLSs based on the area in which they are distributed. Subsequently, pathologists label the TLS (determined by identifying dense mature lymphoid cell aggregates having curved and smooth outlines [19]) in the tumor and normal regions, respectively. The number of TLS is then automatically counted using StrataQuest (7.1.129), along with the areas of the different regions (as shown in Figure. S2). Finally, the TLS parameters are calculated according to the formula [20]:

Variable	Calculation formula
Tumor (T) region
Quantity, No.	The total number of TLS in the T region
Area, mm2	The total area of TLS in the T region
Density, No./mm2	The total number of TLS in the T region / the total area of T region
Normal (N) region
Quantity, No.	The total number of TLS in the N region
Area, mm2	The total area of TLS in the N region
Density, No./mm2	The total number of TLS in the N region / the total area of N region
Total (T + N) region
Quantity, No.	The total number of TLS in the T + N region
Area, mm2	The total area of TLS in the T + N region
Density, No./mm2	The total number of TLS in the T + N region / the total area of T + N region

Multiplex immunofluorescence staining and evaluation

Multiplex immunofluorescence (mIF) was used to further assess the maturity and cellular composition of TLSs, including CD4⁺ T cells, CD8⁺ T cells, CD20⁺ B cells, and CD21⁺ follicular dendritic cells (FDCs). The experimental procedure was based on previous studies [21, 22], utilizing formalin-fixed paraffin-embedded tissue sections with a thickness of 4 μm for mIF. The antibody panel included CD4 (Abcam, ab196372), CD8 (eBioscience, 50-0008-82), CD20 (eBioscience, 50-0202-80), CD21 (Abcam, ab75985), and CD23 (Abcam, ab16702), with DAPI used for nuclear staining. The stained sections were imaged using the Polaris slide scanner (Akoya Biosciences), and five randomly selected regions at 200× magnification were analyzed using the HALO analysis software (Indica Labs).

TLS maturity was evaluated based on the criteria proposed by Vanhersecke et al. [23], which consider the presence of FDCs and germinal centers (GCs). Early-stage TLSs, which are enriched in B and T lymphocytes but lack FDCs and GCs, were classified as aggregates (Agg; CD4⁺, CD8⁺, CD20⁺, CD21⁻, and CD23⁻). TLSs containing FDCs but lacking GCs were categorized as primary follicle-like TLSs (FL-I; CD4⁺, CD8⁺, CD20⁺, CD21⁺, and CD23⁻), whereas those containing both FDCs and GCs were classified as secondary follicle-like TLSs (FL-II; CD4⁺, CD8⁺, CD20⁺, CD21⁺, and CD23⁺). For samples with multiple TLSs, the highest maturity level observed among the TLS was assigned as the sample’s TLS maturity status. FL-II was defined as mature TLSs, while the other subtypes were classified as non-mature TLSs.

Statistical analysis

All analyses were conducted using R software version 4.2.2 and GraphPad Prism version 9. Categorical data were expressed as frequencies (percentages). Continuous data were expressed as median (quartiles) and were compared using the Mann-Whitney U test or Kruskal-Wallis H test. Data distribution was visualized using violin plots. The surv_cutpoint function in the R package survminer was used to determine the optimal cutoff values for the TLS parameters based on patient survival data (including survival time and status), and to stratify the patients accordingly. The correlation between TLS parameters and prognosis were assessed using Kaplan-Meier survival analysis and Cox regression analysis. Differences in TLS maturity levels between groups with different TLS densities were analyzed using Fisher’s exact test. Comparisons of immune cell quantity were conducted using the Mann-Whitney U test. Two-sided P-values < 0.05 were considered statistically significant.

Ethics consideration

This study was approved by the Ethics Committee of the Harbin Medical University Cancer Hospital (Approval No.: KY2016-19). Due to the retrospective nature of the study, the requirement for informed consent was waived by the Ethics Committee of the Harbin Medical University Cancer Hospital. The study was performed in accordance with the Declaration of Helsinki.

Results

Characteristics of patients

Of the 114 LARC patients enrolled in the study, the majority (70%) were male, and 46% were over 60 years of age. Fourteen patients (12%) were classified as stage II, while 89 patients (78%) were classified as stage III according to the cTNM system. In terms of treatment outcomes, 59 patients (52%) responded to neoadjuvant chemotherapy, and 48 patients (42%) exhibited tumor regression. Additional characteristics are detailed in Table 1. In this study, the majority of patients (n = 99, 86.8%) received the XELOX regimen as preoperative neoadjuvant chemotherapy. Furthermore, 52.6% (n = 60) and 14.9% (n = 17) of patients received postoperative adjuvant chemotherapy and radiotherapy, respectively (Table 2).

Table 1 Clinical characteristics of the patients

Table 2 Characteristics of the patient’s treatment regimen

Characteristics of tertiary lymphoid structures

We evaluated the TLSs in surgical specimens from all patients, and images of representative sections are presented in Fig. 1A-C. As shown in Table 3, the median TLS quantity in the normal, tumor, and total regions was 2, 0, and 3, respectively. The median TLS area was 0.21, 0, and 0.25 mm², while the median density was 0.2, 0, and 0.04 TLSs/mm², respectively. Furthermore, the quantity, area, and density of TLSs in tumor region were significantly lower than those in normal region and total region (all P < 0.001, Fig. 1D-F).

Table 3 The quantity, area, and density of TLSs in normal, tumor, and total regions (n = 114)

Fig. 1

TLS expression in surrounding normal, tumor, and total regions. A-C: Representative HE staining images of TLSs (10x magnification). D-F: Comparison of the quantity (D), area (E), and density (F) of TLSs in different regions

Correlation between tertiary lymphoid structures and prognosis

The prognostic significance of TLSs was assessed using survival analyses. The results indicated that in both total-region and normal-region groups, patients with larger areas (Fig. 2D F) and higher densities (Fig. 2G and I) of TLSs had significantly better OS. Among these, the total-region TLS density was the most optimal indicator of prognostic efficacy (HR = 0.250, P = 0.005), and subsequent analyses primarily focused on this parameter. In addition, literature suggests that TLS area is a potential prognostic marker [24]. Consequently, the total-region TLS area was also included in the analysis for reference.

Fig. 2

Kaplan-Meier survival curves of the quantity (A-C), area (D-F), and density (G-I) of TLSs in different regions

Correlation between tertiary lymphoid structures and clinical characteristics

We compared the area and density of TLSs in the total region across different clinical characteristics using violin plots for visualization (Fig. 3). In terms of TLS density, patients who responded to neoadjuvant therapy or experienced tumor regression had significantly higher TLS densities than those who did not respond (P < 0.001) or had no tumor regression (vs. partial regression, P = 0.041; vs. no regression, P < 0.001). Moreover, compared to patients with high CEA levels, those with low CEA levels had higher TLS densities (P = 0.028). Similarly, larger TLS area was observed in patients with low CEA levels (vs. high CEA levels, P = 0.039), responses to neoadjuvant therapy (vs. non-response, P < 0.001), and tumor regression (vs. partial regression, P = 0.06; vs. no regression, P < 0.001).

Fig. 3

Violin plots of total-region TLS area (A-I) and density (J-R) in patients with different clinical features

Subgroup analysis of the correlation between tertiary lymphoid structures and prognosis

Given variations in total TLS area and density among patients with different clinical characteristics, we further performed a subgroup analysis to explore the correlation between TLSs and prognosis. The results revealed that high TLS density was significantly associated with better OS in non-elderly patients (P = 0.044), males (P = 0.002), stage-III cases (P < 0.001), patients with low CEA levels (P = 0.022), both high and low CA199 levels (P = 0.045 and P = 0.036, respectively), and MRF-negative cases (P = 0.023). Furthermore, significant associations were observed in patients with both positive (P = 0.033) and negative (P = 0.022) EMVI, non-lymph node metastatic cases (P = 0.015), neoadjuvant chemotherapy-responsive patients (P = 0.046), and patients without postoperative adjuvant chemotherapy (P = 0.012) or radiotherapy (P = 0.002). Detailed results are shown in Table 4 and Figure S3 in the Supplementary Materials. In contrast, the prognostic significance related to TLS area in these subgroups largely lack statistical significance (Table 4). Thus, TLS density appears to have a stronger correlation with prognosis compared to TLS area.

Table 4 Cox regression analysis of the total-region TLS area and density in subgroups (n = 114)

Assessment of TLS maturity and immune composition via mIF

To further investigate the structure of TLS and their role in the TME, we selected 20 samples from the 114 patients for mIF analysis to evaluate the maturity, distribution, and immune composition of total-region TLSs. Representative staining images of TLSs at different maturity levels are presented in Fig. 4A. Among the selected samples, three (15%) were classified as Agg, nine (45%) as FL-I, and four (20%) as FL-II (Fig. 4B). Notably, all TLSs in the low-density TLS group were classified as non-mature (n = 12), whereas in the high-density group (n = 8), 50% were mature (Fig. 4C). Fisher’s exact test revealed a statistically significant difference in the proportion of mature TLS between the two groups (P = 0.014). Furthermore, we compared the immune composition of mature and non-mature TLSs. As shown in Figs. 4D–G, the numbers of CD20⁺ B cells and CD21⁺ FDCs were significantly higher in the mature TLS group than in the non-mature TLS group. In contrast, the numbers of CD4⁺ T cells and CD8⁺ T cells did not significantly differ, suggesting that TLS maturation may be primarily associated with the infiltration of B cells and FDCs.

Fig. 4

Maturity and immune composition of total-region TLS. A: Representative multiplex immunofluorescence images of no TLS (NA), aggregates (Agg), primary follicles (FL-I), and secondary follicles (FL-II) (200× magnification). Fluorescent labeling: CD4⁺ (red), CD8⁺ (green), CD20⁺ (blue), CD21⁺ (pink), and CD23⁺ (yellow). B: The number of patients with different TLS maturity levels. C: The percentage of different TLS maturity levels in the high- and low-TLS density groups. D-G: Comparisons of the numbers of CD4⁺ T cells (D), CD8⁺ T cells (E), CD20⁺ B cells (F), and CD21⁺ follicular dendritic cells (G) between mature and non-mature TLS groups

Discussion

TLSs have emerged as key players in the immune milieu of various cancers, but their specific prognostic value in patients with LARC undergoing neoadjuvant chemotherapy is not fully understood. Our study found that larger TLS areas and higher TLS densities in the total region were associated with better OS in LARC patients receiving neoadjuvant chemotherapy. Moreover, these TLS parameters correlated with lower CEA levels, positive response to neoadjuvant chemotherapy, and tumor regression. Subgroup analysis revealed that combining total-region TLS density with clinicopathologic features such as sex, age, cTNM stage, CEA levels, and extramural vascular invasion further stratified prognosis. Additionally, mIF analysis showed that a high TLS density was associated with a higher TLS maturity, and mature TLSs exhibited greater infiltration of CD20⁺ B cells and CD21⁺ FDCs compared to non-mature TLSs. These findings suggested that TLS parameters, particularly TLS density, may serve as positive prognostic indicators for LARC patients undergoing neoadjuvant chemotherapy.

The TME plays an important role in tumor progression and therapeutic response, especially in the immunotherapy era [25]. TLSs, as important components of the tumor immune environment, are the primary sites for the generation of antitumor immune responses [26, 27]. Consequently, it is necessary to assess their biological characteristics and clinical significance in cancer patients undergoing antitumor therapy. In this study, we observed that the quantity, area, and density of TLSs in the tumor tissues of LARC patients were significantly lower than TLSs in the surrounding normal tissues. This may be related to immune evasion caused by the enrichment of regulatory T cells [28]. Additionally, the disorganized vasculature generated within the tumor may also inhibit immune cell infiltration [29]. TLSs in the normal and total regions were associated with patient prognosis, whereas TLSs in the tumor region lacked this prognostic value. Consistent with this, previous studies indicated that LARC patients exhibited a higher percentage of regulatory T cells in intratumoral TLSs, which may contribute to the loss of their prognostic value [16, 20]. Notably, TLSs display high heterogeneity among patients with different cancer types [30]. For instance, in intrahepatic cholangiocarcinoma, intratumoral TLSs were significantly associated with more favorable OS compared to peritumoral TLSs [31]. This may be due to the different cellular composition of TLSs in various types of cancer, leading to differences in their anti-tumor and pro-tumor differences effects [30].

This study also demonstrated that higher TLS density was associated with better OS in LARC patients compared to lower TLS density, which is consistent with the trends observed in most types of cancer, including lung cancer [32], pancreatic cancer [33], endometrial cancer [34], as well as colorectal cancer [7]. It is generally believed that the presence of TLSs enhances the adaptive immune response at their formation sites, which helps in clearing or rejecting tumor cells, thereby improving the prognosis of cancer patients [30]. In addition to OS, our study found that low TLS density was associated with factors known to negatively impact rectal cancer prognosis, including high CEA levels [35] and poor response to neoadjuvant therapy [36]. However, in this study, there were no significant differences in the area and density of TLSs between LARC patients treated with different chemotherapy regimens, suggesting that XELOX and FOLFOX6 regimens have a minimal impact on TLS expression.

In a recent study, Wang et al. found that TLS density predicted survival in LARC patients who did not receive neoadjuvant therapy [16]. Interestingly, their findings suggested that the prognostic value of TLS density and maturity was diminished after neoadjuvant chemoradiotherapy [16]. The discrepancy in results may be attributed to differences in demographic characteristics, clinical features, and treatment regimens between the two study populations. We further conducted subgroup analyses in this study. After stratifying by age, sex, cTNM stage, and other factors, TLS density demonstrated significant prognostic value in most subgroups. However, no significant survival differences were observed between the high and low TLS density groups among patients who were elderly, female, in stage II, MRF-positive, had lymph node metastases, or received postoperative adjuvant chemotherapy and radiotherapy. This may be due to the limited number of high TLS density samples in these subgroups. Additionally, the potential influence of certain clinicopathologic features on TLS cannot be ruled out. For instance, some studies have shown that TLS maturity is relatively lower in elderly patients, which may reduce their antitumor effect [37].

This study further revealed the relationship between TLS maturity and its immune composition through mIF analysis. We found that the TLSs in the high-density group had a higher maturity, with significant enrichment of CD20⁺ B cells and CD21⁺ FDCs in the mature TLSs. These results suggest that the prognostic value of TLS density may be closely associated with its structural integrity and the B cell/FDC-driven microenvironmental functions. CD4⁺ T cells, CD8⁺ T cells, CD20⁺ B cells, and CD21⁺ FDCs are important adaptive immune cells in the tumor immune microenvironment [38]. Among these, CD8⁺ T cells (cytotoxic T lymphocytes) are considered the primary effector cells in tumor control and elimination [39]. Previous studies have shown that FDCs recruit B cells through CXCL13 and maintain germinal center activity, while B cells enhance anti-tumor immunity and memory through antibody secretion and antigen presentation [40]. Therefore, we speculate that B cells in mature TLSs may activate T cells by presenting tumor antigens, enabling T cells to effectively target tumor cells and thereby improve patient prognosis [41]. Currently, the understanding of the formation and mechanisms of local anti-tumor immunity is still limited, and these results warrant further investigation, as TLSs may become promising biomarkers for predicting prognosis and therapeutic response, as well as potential targets for therapeutic induction in rectal cancer patients [42].

This study has several limitations. First, this was a single-center retrospective study, which may limit the generalizability of the findings to LARC patients in other regions or countries. Second, although the study utilized mIF to analyze the maturity and immune composition of TLSs, the small sample size and limited immune cell types analyzed constrain the findings. Future research should expand the sample size and include a broader range of immune cell types to more comprehensively elucidate the functions and mechanisms of TLS in the TME.

Conclusions

This study suggests that TLS parameters, particularly TLS density in the total region, are positively correlated with the prognosis of LARC patients undergoing neoadjuvant chemotherapy, potentially serving as promising prognostic biomarkers in this population. Since the combination of TLS density and clinical features can further differentiate prognosis, TLS detection in surgical specimens may be necessary.