Abstract
Background/Aim: The dynamic interplay between cancer cells and the microenvironment involves a wide range of intricate relationships that evolve during different stages of tumor progression. Recent attention has focused on high endothelial venules (HEVs), specialized endothelial cells in tumors with a unique cuboidal shape similar to those in lymph nodes. Previous animal studies have shown that normalization of tumor angiogenesis through anti-VEGFR2 therapy promotes HEV formation. However, few reports exist regarding the relationship between HEVs and preexisting blood vessels or interstitial fibers. In this study, we histologically examined whether tumor vascular structure correlates with HEV neogenesis. Patients and Methods: A total of 109 patients with pathological stage I lung adenocarcinoma who had undergone curative lung resection at our Institute between 2012 and 2016 were included. HEVs were identified by anti-peripheral node addressin (PNAd) staining. Immunostaining and Elastica-Masson-Goldner staining were performed on tumor sections and quantified. Results: PNAd-positive cells were identified in 102 (93.6%) patients. Nearly all PNAd-positive cells were located within or near immune cell clusters. We investigated the correlation between microvessel structures or interstitial fibers and the number/density of PNAd-positive vessels, but no significant correlation was found. Since PNAd-positive cells were concentrated in immune cell aggregates, we focused our analysis specifically on these regions. Immune cell aggregates with abundant PNAd-positive vessels had a greater microvessel density along with by rich collagen fiber production, and displayed a more mature morphological phenotype of HEVs. Conclusion: The generation of PNAd-positive cells in tumors is governed by an angiogenetic mechanism distinct from that of broader tumor microenvironment. Furthermore, the accumulation of immune cells is associated with increased HEV maturation.
Antitumor immunity is controlled by a cancer immunity cycle, which involves the capture of cancer-specific antigens, the presentation of tumor-specific antigens in secondary lymphatic organs, such as regional lymph nodes, the activation of tumor-specific lymphocytes and the infiltration of T cells into tumors (1). Activated immune cells are known to form follicular clusters to prepare for attacking tumors. These clusters are known as tertiary lymphoid structures (TLSs). The presence of TLSs is recognized as a critical element of the cancer immune cycle (2). TLSs contain a specialized vascular endothelium, that serves as a “gateway” for immune cell invasion of tumors; these TLSs are composed of high endothelial venules (HEVs), which are similar to vessels found in normal lymph nodes. HEVs are specialized vascular endothelial cells primarily found in secondary lymphoid organs such as lymph nodes (3). HEVs in lymph nodes have morphological characteristics such that the cells are cuboidal despite being vascular endothelium, have a thick basement membrane and are surrounded by a vascular sheath composed of reticular fibroblasts. HEVs promote the homing of T and B lymphocytes and the expression of sialomucin, a ligand for L-selectin. Anti-peripheral lymph node addressin (PNAd) antibodies are widely used to identify HEVs. HEV neogenesis has attracted increasing attention as a mechanism related to immune activation, but knowledge of the underlying molecular biological mechanism is limited. Two pathways are known: the i) lymphotoxin β receptor (LTβR)-mediated pathway and ii) the TNFα receptor-mediated pathway (4, 5). Research on the developmental process of lymph nodes shows that the former pathway is crucial for HEV neogenesis (4). LTβR activates the NF-
B pathway, and the release of proinflammatory cytokines induces HEV development and maintenance (6). Inflammatory cytokine production precedes HEV neogenesis. Fate mapping experiments using mouse tumor models have shown that HEVs originate from postcapillary venules (7). However, little is known about the relationship between HEVs and surrounding preexisting blood vessels in human tumors. Using mouse models of breast and pancreatic cancer, Allen et al. found that combining anti-PD-L1 and anti-VEGFR2 antibodies promoted HEV neogenesis and the immune response (8). The combination of antiangiogenic drugs and immune checkpoint therapy increases the antitumor immune response, and these findings have increased research interest in vascular normalization and HEV neogenesis (9).
Based on these results, we hypothesized that HEVs are abundant in tumors with vascular structures similar to those in normal tissue. In this study, we conducted a clinicopathological examination of the relationships between the morphological characteristics of blood vessels and interstitial structures and HEV neogenesis using resected human lung adenocarcinoma specimens.
Patients and Methods
Study design and patients. We performed a single-Institution retrospective study. A total 318 patients with non-small cell lung cancer who had undergone curative resection (segmentectomy or lobectomy) for lung cancer between April 2012 and March 2016 at our hospital. Among them, 126 stage I adenocarcinoma patients were included. Patients with a predominant lepidic growth pattern were excluded because they did not exhibit pathological angiogenesis, and quantitative evaluation of blood vessels was structurally difficult. Finally, we collected clinical information from the medical records of 109 patients with pathological stage I lung adenocarcinoma and conducted a clinicopathological review (Figure 1).
CONSORT diagram of the patient enrollment.
Histological sample preparation. Paraffin-embedded formalin-fixed tumor blocks on the largest cut surface of the tumor in the resected specimen were sectioned into 4-μm-thick sections. Hematoxylin-eosin staining and Elastica-Masson-Goldner (EMG) staining were performed.
To visualize endothelial cells, CD31 was stained immuno-histochemically. The anti-CD31 antibody (clone: JC70A, Dako, Carpinteria, CA, USA) was diluted at a 1:100 ratio, the anti-rabbit IgG goat polyclonal antibody (I-View DAB, UltraView DAB, Roche Diagnostics, Rotkreuz, Switzerland) was used as the secondary antibody, and BenchMark-VENTANA (Roche Diagnostics) was used for staining (heat activation for 64 min at 95°C).
For HEV staining, an anti-PNAd antibody (clone: MECA-79, Biolegend Japan, Tokyo, Japan) was diluted at a 1:100 ratio and used as the primary antibody, and an HRP-labeled anti-rabbit IgG antibody (Leica Microsystems, Wetzlar, Germany) was used as the secondary antibody. Immunohistochemical staining (ER2, antigen retrieval for 20 min) was performed using a Leica Biosystems-BONDIII system (Leica Microsystems).
All imaging and tiling processes were performed with a BZ-800 microscope (Keyence, Osaka, Japan). A Tiff or Jpeg format file was saved for quantification. The entire tissue section was photographed for EMG-stained sections using a low-magnification (4×) objective lens. Immunostained sections (PNAd and CD31) were photographed using a 20× objective lens, and “hot spots” with intense staining within the tumor tissue. Images (3×3 tiles) were taken of the area including the hot spot.
Morphological quantification. The saved image files were quantitatively analyzed using ImageJ software (NIH, USA). All stained vasculature was marked using ImageJ software and a liquid crystal pen tablet. We created an ImageJ macro that produces an approximate ellipse that covers the marked area and obtains measurements of its large diameter, small diameter, area, and angle (Figure 2A). The data obtained for each slide were compiled into a database using an Excel file and subjected to statistical analysis. To obtain robust data, we used stereological point counting (SPC) as an alternative method to quantify microvessel density (10). We designed a macro that consecutively superimposed a fixed number (13×10) of lattice structures on immunostained sections using ImageJ (Figure 2B). In this grid, the central 9×9 area was targeted for quantification. The leftmost and bottom grids of each object were excluded from the analysis. In addition, grids that did not overlap with the tissue structure were excluded from the calculation. Only vessels in the area where the grid overlapped with the vascular structure were counted and added together to determine microvessel density.
(A) An ImageJ macro was developed for the quantification of morphological vessel analysis. (B) A representative figure for grid counting of microvessels by SPC analysis. The area indicated by the yellow line was analyzed. The sky-blue circles show the grids that include the vessels. Pink circles indicate grids that do not include tissue, which were excluded from the analysis. (C) Collagen and elastic fiber quantification method using EMG-stained samples. First, the nontumor area was erased. Then, the image was stacked via the hue–saturation–brightness method. In the hue image, specific color pixels were quantified. SPC: Stereological point counting; EMG: Elastica-Masson-Goldner.
Relative quantification of the area containing interstitial fibers was performed using EMG-stained sections and ImageJ (Figure 2C). Surrounding nontumor areas were removed from the electronically obtained TIFF format file of the section image using ImageJ software. Next, only the hue information of the image was extracted using the hue–saturation–brightness method (HSB method). We selected pixels where the hue value indicated the presence of elastic fibers or elastic fibers and quantified the proportion of these pixels among all pixels in the tumor section.
For the measurement of the PNAd positive vessel count in the follicular lymphoid-like structures, 30 high-power view images with follicular lymphoid structures were randomly selected. The same area was selected for the evaluation of the collagen deposition quantification using the EMG-stained sections. Among the thirty high-power view images, sixty PNAd-positive endothelial cells were randomly selected. The height of the stained endothelial cells was measured using ImageJ.
Statistical analysis. Survival analysis was performed using the Kaplan-Meier method. Pearson’s product-moment correlation coefficient was used to compare two variables of the staining ratio, and one-way analysis of variance was used to compare variables among three groups. A p-value less than 0.05 was considered to indicate statistical significance. MedCalc ver 22.016 (MedCalc, Ostend, Belgium) was used for all the statistical analyses.
Results
Relationship between PNAd-positive vascular endothelial cells and prognosis of lung adenocarcinoma. Patient characteristics are summarized in Table I. Almost half of the patients were smokers. HEV neogenesis was assessed using PNAd staining. There were only 7 patients in which no PNAd-positive blood vessels were detected. PNAd-positive cells were confirmed in 102 patients (93.6%). Within the tumor, almost all PNAd-positive vascular endothelial cells were present in areas of lymphoid follicular immune cell clusters (Figure 3A). When we focused on the regions where PNAd-positive endothelial cells were observed, many were particularly connected to blood vessels close to immune cell clusters. Furthermore, even in the same blood vessel, multiple PNAd-positive vascular endothelial cells were observed only on the side where lymphoid follicles were present (arrows in Figure 3A). Typical HEVs in normal secondary lymphatic organs have a cuboidal cell shape (Figure 3B). However, many PNAd-positive endothelial cells in the tumor were flat (Figure 3C).
Patient characteristics.
Representative immunohistochemistry image of PNAd-positive vessels. (A) Most of the PNAd (+) vessels reside inside or adjacent to immune cell aggregates. (B) A representative image of control staining for PNAd. The image was taken from a tonsil. (C) Intratumoral PNAd (+) vessels often show a flat appearance. (D) Kaplan-Meier curves for recurrence-free survival for patients grouped according to PNAd positivity. Total number of the cases was 109. PNAd: Peripheral node addressin.
A total of 2,639 PNAd-positive vessels were quantified. The average number of PNAd-positive blood vessels per 200× field of view was 2.7±3.6. Patients were divided into two groups based on the density of PNAd-positive vascular endothelial cells, and the postoperative recurrence rate was assessed. The group with a high PNAd-positive vascular endothelial cell density tended to have a lower risk of recurrence than the group with a low density (5-year recurrence-free survival probability: 94.7% vs. 82.9%, p=0.18; Figure 3D). Recurrence only occurred in one patient recurrence, so the difference did not reach the threshold for statistical significance.
There was no correlation between the overall blood vessel density in tumor tissue and the density of PNAd-positive endothelial cells. To assess whether the appearance of PNAd-positive vascular endothelial cells was associated with the tumor vasculature, CD31-immunostained sections from the same patient were quantified. A total of 7,890 CD31-positive vessels were evaluated. When we performed an analysis of postoperative recurrence risk; when the patients were grouped according to microvessel density (high and low microvessel density groups), the group with high microvessel density had no recurrence events, and there was a trend toward a more favorable prognosis (5-year recurrence-free survival probability of 100% vs. 83.0%, p=0.12; Figure 4A), but the difference did not reach the threshold for statistical significance. In tissues with normalized microvessels, the average tumor vessel cross-sectional area was greater because the vessels are patent, so we performed a similar survival analysis for patients grouped according to the cross-sectional area per unit vessel. However, there was no significant difference in prognosis between the groups (5-year recurrence-free survival probability 93.0% vs. 82.0%, p=0.14; Figure 4B). As an alternative method for measuring tumor microvessel density, manual measurements were performed using the SPC method (10). The average vascular index according to the SPC method was 3.7±2.9. There was a robust positive correlation between the number of blood vessels measured using the SPC method and the automated quantification method according to ImageJ analysis (Pearson’s correlation coefficient=0.84, p<0.0001; Figure 4C). We performed a correlation analysis to determine whether there was a relationship between the tumor blood vessel density or vascular architecture features obtained above (i.e., average cross-sectional area of micro vessels, average length of microvessels, average short diameter of microvessels, total count of microvessels, and total length of microvessels) and the density of PNAd-positive vascular endothelial cells, but no factors showed a correlation (data not shown).
Kaplan-Meier curves for patients grouped according to microvessel density. (A) A high microvessel count tended to indicate favorable outcomes, but the difference was not statistically significant (N=109). (B) The average cross-sectional area was negatively correlated with survival, but this relationship was not statistically significant (N=109). (C) Two different methods for quantifying microvessel density were compared. The values from the digital pixel counting method and the stereological point counting method were strongly correlated (N=109).
Correlation analysis of the density of PNAd-positive vascular endothelial cells and stroma in the whole tumor. Elastic fibers and collagen fibers increase in the tumor microenvironment and affect prognosis. EMG staining was used to quantify the fibrous composition of the tumor stroma. The proportion of specific hue values in the EMG-stained sections was quantified. First, when comparing recurrence rates for patients grouped based on elastic fiber density, patients with poor elastic fiber production had a better prognosis (5-year recurrence-free survival probability: 100% vs. 81.1%, p=0.03; Figure 5A). Conversely, patients with high collagen fiber density tended to have a better prognosis than those with low collagen fiber density. However, the difference did not reach the threshold for statistical significance (5-year recurrence-free survival probability: 94% vs. 80%, p=0.06; Figure 5B).
Correlation of extracellular matrix features with lung adenocarcinoma patient survival. (A) The abundant production of elastic fibers was related to tumor recurrence (N=109). (B) Collagen-rich tumors tended to have favorable outcomes, but the difference was not statistically significant (N=109).
There was no correlation between elastic fiber production and PNAd-positive vascular endothelial cell density (Pearson’s correlation coefficient=0.03, p=0.77). On the other hand, a weak correlation was observed between collagen fiber production and PNAd-positive vascular endothelial cell density (correlation coefficient=0.28, p=0.004). Furthermore, we analyzed the correlation between PNAd-positive blood vessel density and microvessel density by further stratifying patients according to collagen fiber abundance. However, even when collagen fiber abundance was considered, the percentage of PNAd-positive cells was not correlated with microvessel density (correlation coefficient=0.11, p=0.889).
When limited to lymphoid follicle-like structures, the density of PNAd-positive vascular endothelial cells correlated with the density of tumor blood vessels. As mentioned above, there was no correlation between microvessel density and PNAd-positive vascular endothelial cell density in whole tumor sections. On the other hand, since the localization of PNAd-positive vascular endothelial cells is concentrated in lymphoid follicle-like structures, quantitative analysis was performed focusing only on lymphoid follicle-like structures.
We separated the cohort into three groups according to the PNAd-positive vascular endothelial cell density, and randomly selected lymphoid follicle-like structures from each group were analyzed for their microenvironmental architecture. It was confirmed that lymphoid follicle microvessel density and HEV density were correlated (high: medium: low=23±15: 15±11: 10±6, p<0.001; Figure 6A). When comparing the interstitial composition within lymphoid follicle-like structures by quantifying EMG-stained sections, there was no difference in elastic fiber density, but there was a positive correlation between an increase in collagen fiber density and an increase in HEV density (high: medium: low=12.1±6.4%: 4.8±4.5%: 4.6±5.4%, p=0.003; Figure 6B). Next, the shapes of PNAd-positive cells within lymphoid follicles were compared. PNAd-positive cells were taller in the high group tumors (high: 5.8±2.1μm, medium: 3.1±1.2μm, low: 3.5±1.2μm, p<0.001; Figure 6C) and exhibited a mature shape similar to that of HEVs in noncancerous tissues. From this observation, it was inferred that PNAd-positive vascular endothelial cells in tumors undergo a maturation process leading to HEV formation.
Tumor microenvironment analysis of immune cell aggregates. (A) An abundance of PNAd-positive vessels was correlated with higher microvessel counts. Thirty high-power fields were measured from each group. (B) Collagen production was also related to PNAd positivity. Thirty high-power fields were measured from each group. (C) Endothelial height comparison among the PNAd-positive endothelial cells. More mature HEVs were found in PNAd-rich tumors. Sixty vessels were randomly measured from each group.
Discussion
This study involved analysis of the relationship between the neogenesis of PNAd-positive HEVs and the tumor microenvironment using lung cancer resection specimens. We hypothesized that the proportion of HEVs would be more significant in tumors with abundant vessel density and/or with open vasculature. Contrary to this hypothesis, there was no correlation between the overall tumor microvascular architecture and the density of PNAd-positive vascular endothelial cells. Since PNAd-positive vascular endothelial cells were abundant in lymphoid-like follicles, we randomly extracted these structures and analyzed them. There was a positive correlation between microvessel density and PNAd-positive cell density in these areas. Regarding the morphology of vascular endothelial cells, we found that the PNAd-positive endothelial cells in the PNAd-high group were cuboidal, exhibiting a mature shape similar to that of HEVs observed in typical lymphoid structures.
These findings suggest that the normality of tumor blood vessels and the maturity of PNAd-positive blood vessels with lymphoid follicle-like structures are not directly correlated. Different angiogenic mechanisms control the entire tumor’s vascular structure and follicle-like structures. To date, few analyses have investigated the microenvironment of lymphoid follicle-like structures, and to our knowledge, these results are novel.
The tumor and its microenvironment have diverse interactions depending on the stage of progression, altering biological traits. Immediately after cancer develops, the microenvironment becomes “hostile” to cancer (11). Cancer cells tend to be eliminated through cell competition. Epithelial defense against cancer (EDAC) is known as a cell competition mechanism by which cells with genetic abnormalities are eliminated (12). Normal cells and mutant cells compete for space to survive, with the winners surviving and the losers dropping off from the basement membrane and being eliminated. It is known that EDAC has two pathways: one in which transformed cells are eliminated via caspase-dependent apoptosis, and another in which transformed cells are pushed out by apical extrusion caused by activation of Ras, activated src, ErbB2, Yes-associated protein, etc., and then die by anoikis and are expelled from the body. However, over time, tumors adopt a stroma that is different from that of normal cells (a process referred to as “co-opting”) (13). Furthermore, the microenvironment undergoes remodeling to promote the survival and treatment resistance of cancer cells (14-16). Tumor blood vessels function as transport routes for oxygen and nutrients to cancer tissues, serve as cancer stem cell niches, and play essential roles as immune regulators (15). Due to the rapid growth process of tumors, the inner region of tumors lacks sufficient energy and nutrients, creating a particularly harsh environment characterized by hypoxia, malnutrition, low pH, and high solid stress (17). As a result, proangiogenic growth factors such as VEGF continue to be produced in excess inside the tumor, and this excess plays a part in suppressing the antitumor immune response (18).
As a treatment for pathological tumor angiogenesis, “normalization” of blood vessels by antagonism of excess angiogenic factors can be effective (vascular normalization hypothesis). VEGFR2 blockade by a monoclonal antibody is known to normalize the tumor microenvironment. Bergers et al. reported that HEV production occurs via activation of the LTβR pathway in mouse models (8). We hypothesized that the HEV density would be greater in tumors with a vascular structure similar to that in normal tissues. Nevertheless, we did not find a direct relationship between the vascular structure of the entire tumor tissue and HEV density.
PNAd-positive cells were present in 93.6% of the tumors. However, there was no difference in prognosis regardless of the number of PNAd-positive cells. This finding suggests that PNAd positivity alone may not be sufficient for the immune activation of HEVs. A group recently reported the heterogeneity of PNAd-positive cells within lymph nodes. It has been reported that the state of the HEV changes depending on the activation state of the lymph nodes (19). A comparison of the shapes of PNAd-positive endothelial cells in malignant melanoma revealed that intratumoral lymphocyte infiltration is promoted when PNAd-positive cells are cuboidal (20). In our study, although we confirmed that cells were deformed in PNAd-rich tumors, we could not evaluate the immunobiological activation status.
This research has certain imitations that are listed below. First, this was a single-center, retrospective-design study with a limited number of patients. Second, fractional analysis of immune cells was not performed. In particular, it is speculated that changes in immune cell components in lymphoid follicle-like structures are strongly related to the maturation of PNAd-positive cells, but we could not perform these analyses. Third, the biochemical EMG stain was used to stain the interstitium. Areas with a high density of elastic fibers were stained darkly, making accurate quantification difficult if fibers of other colors are present. Fourth, with this method, it is also impossible to obtain precise localization information, such as the presence of PNAd-positive cells and adjacent collagen fibers. Multiple immunostaining steps are needed to verify their colocalization. Last, it was not possible to analyze whether there are differences in the maturation level of PNAd-positive vascular endothelial cells in terms of their functions contributing to immune responses. These unresolved issues need to be pursued in the future.
Conclusion
The neogenesis of intratumoral PNAd-positive cells is governed by mechanisms distinct from those controlling the maturation of the entire tumor vasculature. Immune activation, accompanied by the expansion of lymphoid-like structures within tumors, is associated with neogenesis and maturation of HEVs.
Footnotes
Authors’ Contributions
Ato Sugiyama: Study design, data acquisition and analysis, drafting of the manuscript. Tai Hato: Conception, study design, data acquisition, analysis, interpretation, and drafting of the manuscript. Hiroaki Kashimada: Data acquisition. Masatoshi, Yamaguchi: Data acquisition. Yoshiaki Inoue: Data acquisition and drafting of the manuscript. Kohei Aoki: Data acquisition. Hiroki Fukuda: Data acquisition and drafting of the manuscript. Mitsuo Nakayama: Data acquisition, analysis, and drafting of the manuscript. Morihiro Higashi: Conception, study design, analysis, and drafting of the manuscript. Mitsutomo Kohno: Data acquisition analysis, interpretation, and drafting of the manuscript.
Funding
This study was carried out using academic research funds from Saitama Medical University.
Conflicts of Interest
All the Authors declare that there are no conflicts of interest regarding this research.
- Received July 20, 2024.
- Revision received August 19, 2024.
- Accepted September 5, 2024.
- Copyright © 2024 The Author(s). Published by the International Institute of Anticancer Research.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.