Abstract
Soft tissue sarcomas (STSs) are conventionally viewed as poorly immunogenic tumors; however, some human STSs have recently been reported to elicit an immune response, thus representing potential candidates for immunotherapy. Data regarding immune cell infiltrates in canine STSs are limited and reported without tumor-type stratification. The aim of this study was to retrospectively assess tumor-infiltrating lymphocytes (TILs) in canine STSs of 5 different histotypes. Eighty-seven canine STSs were collected: 22 perivascular wall tumors (PWTs), 19 liposarcomas, 17 fibrosarcomas, 16 myxosarcomas, and 13 leiomyosarcomas. The tumors were graded and immunolabeled for CD3, CD20, and FoxP3, and slides were scanned. T-cell, B-cell, Treg, and total TIL densities were quantified with QuPath software and expressed as cells/mm2. The B/T-cells ratio and Treg/T-cell proportions were calculated. Total TIL densities were higher in PWTs and myxosarcomas (median = 225 and 303, respectively). PWTs had higher T-cell density but lower Treg proportion (median = 152 and 7.6% respectively). Myxosarcomas had higher Treg densities and B/T-cell ratios (median = 24.4 and 1.57, respectively). No association with grade was found among STSs as a group. In myxosarcomas, higher grade was significantly associated with higher total TILs, and CD20+ and FoxP3+ cell densities (p < .05). The results suggest that PWTs and myxosarcomas may represent the most immunogenic STS types. Myxosarcomas elicit a B-cell and Treg-rich immune response; PWTs stimulate a T-cell-rich and Treg-poor reaction. The immune system response may contribute to the more aggressive behavior of myxosarcomas and the more indolent course of PWTs.
Soft tissue sarcomas (STSs) comprise a diverse group of neoplasms traditionally regarded as poorly immunogenic (“immunogenically cold”). 17 However, in recent years, there has been a surge of interest in immune therapies, with numerous studies in human medicine suggesting that STSs can indeed incite an immune response.8,9,12,16,20 This discovery carries significant implications, particularly as STSs often exhibit resistance to chemotherapy, and treatment options are scarce. Central to these investigations is the study of the tumor microenvironment (TME), focusing on tumor-infiltrating lymphocytes (TILs), tumor-associated macrophages (TAMs), and the expression of immune checkpoints, including the PD1-PDL1 (programmed cell death protein 1-programmed death-ligand 1) axis.8,9,12,16,20
In veterinary medicine, canine melanoma stands out as one of the most extensively studied malignancies regarding the immune TME, owing to its well-established immunogenicity and allowing for the development of effective immune therapies. 27 While TILs, TAMs, and PD1-PDL1 axis expression have been explored in various canine tumors,57,19 information specific to canine STSs remains limited. Some studies have reported on TILs and TAMs in canine STSs, correlating their presence with PD1-PDL1 axis expression and histologic grade, respectively.13,26 However, comprehensive data on TME differences among different canine STS histotypes are lacking. In contrast, in humans, complex karyotype STSs tend to provoke a stronger immune response compared with translocational STSs, likely due to their higher mutational burden. 12
While standardized guidelines exist for assessing TILs in other tumor types (e.g., mammary carcinomas) and can be applied to epithelial tumors,19,23 evaluating TILs in STSs is more complex and thus less standardized. This complexity stems from the relatively low number of TILs in STSs and the absence of clearly delineated supporting stroma in these tumors. Consequently, TIL assessment in STSs typically relies on immunolabeled sections rather than conventional hematoxylin and eosin staining.8,12,26
The aim of this study was to characterize the lymphocytic immune response in canine STSs by evaluating TILs in immunohistochemically labeled sections using image analysis. Additionally, the study aimed to evaluate the comparison of the immune response in relation to different STS histotypes, grades, and additional histologic parameters.
Materials and Methods
Cases Selection
Formalin-fixed, paraffin-embedded samples of canine STSs with defined histotypes were retrospectively selected from the histologic archive of three institutions according to the following inclusion criteria: STS tumor types included perivascular wall tumor (PWT), fibrosarcoma, liposarcoma, myxosarcoma, and leiomyosarcoma and excluded visceral (e.g., splenic and intestinal) sarcomas; hematoxylin and eosin-stained slides were reviewed by two pathologists (GA and PR) to confirm the diagnosis; and the paraffin blocks checked for availability of tissue.
The histotype of the tumors was confirmed based on the following criteria:
PWTs: the presence of at least 1 of the 4 classical perivascular patterns as previously described, 22 and absence of patterns suggestive of nerve sheath tumors, such as nuclear palisading, Verocay bodies, and tactile-like structure.
Myxosarcoma: spindle to stellate neoplastic cells embedded in various amounts of mucinous matrix, which nevertheless represents the majority of the extracellular matrix of the tumor. Histologic patterns indicative of PWTs and patterns indicative of nerve sheath tumors are absent.
Liposarcoma: polygonal to spindle cells containing optically empty vacuole(s) with sharp margins, displacing and indenting the nucleus.
Leiomyosarcoma: intersecting bundles of spindle cells with abundant eosinophilic cytoplasm, scant to absent extracellular matrix, and immunohistochemical (IHC) expression of smooth muscle actin and/or desmin. Some of the leiomyosarcomas were included in a previously published study. 3
Fibrosarcoma: bundles and streams of spindle cells, perpendicularly arranged or forming herringbone pattern, with moderate amounts of cytoplasm, elongated nuclei, and embedded in various amounts of collagen, which represented the majority of the extracellular matrix of the tumor. These cases were differentiated from leiomyosarcomas by lack of smooth muscle actin expression and a vimentin-only phenotype.
The histologic grade was assigned according to the literature.4,22 The amount of necrosis was assessed at the microscope and classified as absent, ≤50% of the sarcoma, and >50% of the sarcoma.4,22 Mitotic count (MC) was assessed in 2.37 mm2 in the most cellular and proliferative areas of the tumor. 4
Immunohistochemistry
All cases underwent IHC assessment for CD3, CD20, and FoxP3. Three-micrometer-thick sections were dewaxed and rehydrated. Endogenous peroxidase was blocked by immersion in 3% H2O2 in methanol for 30 minutes. Source, dilution, and retrieval protocols for each antibody are reported in Table 1. Each slide was incubated with the primary antibody overnight at 4°C.
Antibody source and protocols.
The reaction was amplified by the avidin-biotin method (Vectastain® Elite ABC-HRP kit, Vector, Burlingame, CA, USA) and visualized with 0.04% 3,3′-diaminobenzidine (Code: 10-0048, Histoline Milano, IT) for 4 minutes. Sections were counterstained with hematoxylin, rinsed in tap water, and dehydrated before a coverslip was added. Sections of canine hyperplastic lymph nodes were used as positive controls. Negative controls comprised slides incubated with the omission of the primary antibody, and internal negative controls were constituted by the neoplastic cells of the same samples known to be nonreactive for the specific antibody.
IHC Evaluation and Image Analysis
The IHC slides were scanned with Grundium Ocus 20 (Tampere, Finland) at 20× magnification (0.25 μm/pixel) to obtain a whole-slide image (WSI). The digital images were analyzed with the open-source DIA software QuPath v0.5.0.3.
For each marker, positive cells were counted in WSIs. Automatic cell counting was performed using the Positive cell detection tool. As the intensity of labeling varied within the cohort, the IHC and hematoxylin stain estimates for each digitized slide were adjusted (estimate stain vectors tool in QuPath). Analysis was performed starting with manual selection of all the neoplastic tissue in each section and Positive cell detection tool (Analyze → Cell analysis → Positive cell detection → adjust parameters based on Optical Density Sum, Score compartment: Nucleus DAB OD Mean). Default parameters for cell detection were adjusted to best match the outline of lymphocytes, and the results for each case were visually checked by the operator for satisfactory quality of the analysis results (Fig. 1).
Image analysis for the quantification of CD3+ cells in a canine perivascular wall tumor. (a) Whole-slide image (WSI), CD3 immunohistochemistry. (b) Selection of the region of interest on the WSI, excluding nonneoplastic tissues, holes, cracks, and folds of the section. (c) Cell detection of CD3+ cells (red) and CD3-negative cells (blue).
CD3+, CD20+, and FoxP3+ cell densities were calculated as number of cells/mm2, as previously reported.8,12 The total TIL density was calculated as the total of CD3+ and CD20+ cells. The B-cell/T-cell ratio and the percentage of FoxP3+ T-cells were also calculated. WSIs were searched for the presence of tertiary lymphoid structures (TLSs) that were manually counted if present.
Statistical Analysis
A Kruskal-Wallis 1-way analysis of variance was used to assess the association of TIL variables with histotype, grade, MC, and amount of necrosis. MC and amount of necrosis were classified according to the STS grading scheme. 4 Post-hoc pairwise analysis was performed with the Dwass-Steel-Critchlow-Fligner test. MC was further analyzed as a continuous variable by calculating Spearman’s correlation coefficient with the TIL variables. Furthermore, for statistical purposes, grade 2 and grade 3 cases were grouped together, and necrosis was also categorized as present or absent. The Mann-Whitney U test was then used to assess the difference in TIL variables between these groups. Results were considered significant at a threshold of p ≤ .05. Statistical analyses were conducted using Jamovi (version 2.4.12.0).
Results
A total of 87 canine STSs were included: 22 PWTs, 19 liposarcomas, 17 fibrosarcomas, 16 myxosarcomas, and 13 leiomyosarcomas. Fifty-six cases were grade 1, 25 were grade 2, and 6 were grade 3. Necrosis was absent in 57 cases, scored ≤ 50% of the tumor in 29 cases, and > 50% of the tumor in 1 case. The median MC was 4 (range 0–63). TLSs were present in 4 cases, ranging from 1 to 3 per section. The median total TIL density was 139 cells/mm2 (range 0–3538). TILs were not detected in 4 cases: 2 liposarcomas and 2 fibrosarcomas. The median density of CD3+ cells was 59.2 cells/mm2 (range 0–3025), and CD3+ cells were absent in 2 fibrosarcomas that were infiltrated by CD20+ cells. The median density of FoxP3+ cells was 9.06 cells/mm2 (range 0–617); FoxP3+ cells were absent in 3 fibrosarcomas and 1 PWT that nevertheless contained CD3+ cells. The median percentage of T-cells that were FoxP3+ was 12.2% (range 0–91.8%). The median density of CD20+ cells was 34.4 cells/mm2 (range 0–686), and CD20+ cells were absent in 2 additional fibrosarcomas that were infiltrated by CD3+ cells. The median B-cell/T-cell ratio was 0.484 (range 0 to 114). TLSs were identified in 2 myxosarcomas and 2 PWTs. Results of the immunohistochemistry are depicted in Figure 2 and summarized in Figure 3. Descriptive statistics are reported in Supplemental Tables S1 and S2.
Tumor-infiltrating lymphocyte immunohistochemistry in canine soft tissue sarcomas. (a) Canine myxosarcoma characterized by numerous intratumoral CD20+ cells. CD20 immunohistochemistry (IHC). (b) Canine perivascular wall tumor (PWT) infiltrated by numerous CD3+ cells. CD3 IHC. (c) Canine PWT, same case as in (b). Numerous lymphocytes infiltrate the tumor and a small proportion express FoxP3. FoxP3 IHC. (d) Canine fibrosarcoma infiltrated by rare CD20+ cells. CD20 IHC. (e) Canine liposarcoma infiltrated by rare CD3+ cells. CD3 IHC. (f) Canine fibrosarcoma infiltrated by rare CD3+ cells. CD3 IHC.
Total tumor-infiltrating lymphocytes (TILs) and their subpopulations in canine soft tissue sarcomas separated by diagnosis. (a) Total TIL density, (b) CD20+ cell density, (c) CD3+ cell density, and (d) Treg proportion. The densities of total TILs, CD20+, and CD3+ cells are expressed as the number of lymphocytes/mm2, while Treg proportion is expressed as percentage of FoxP3+ cells/ CD3+ cells. PWT, perivascular wall tumor.
A statistically significant association was found between the diagnosis and the total TIL density; CD20+, CD3+, and FoxP3+ cell density; percentage FoxP3+ T-cells; and the B-cell/T-cell ratio (Table 2). Based on the Dwass-Steel-Critchlow-Fligner test, the total density of TILs was significantly higher in myxosarcomas and PWTs compared with fibrosarcomas; the CD20+ cell density was significantly higher in myxosarcomas compared with fibrosarcomas and liposarcomas, and in PWTs compared with fibrosarcomas; the CD3+ cell density was significantly higher in PWTs compared with fibrosarcomas; and the B-cell/T-cell ratio was higher in myxosarcomas compared with fibrosarcomas (Table 3).
Kruskall-Wallis analysis assessing the difference of tumor-infiltrating lymphocyte (TIL) variables in canine soft tissue sarcoma histotypes.
Dwass-Steel-Critchlow-Fligner pairwise comparison of tumor-infiltrating lymphocyte (TIL) variables among soft tissue sarcoma histotypes. Only statistically significant differences are reported.
Abbreviation: PWTs, perivascular wall tumors.
Considering all histotypes together, no statistically significant associations between grade, MC, amount of necrosis, and TIL variables were observed. Considering each single histotype separately (Fig. 4), a higher grade of myxosarcomas was significantly associated with higher total TILs (Chi square = 5.309, p = .021) and CD20+ (Chi square = 4.765, p = .029) and FoxP3+ (Chi square = 4.765, p = .029) cell densities; the amount of necrosis was significantly higher in cases with higher CD3+ cell densities (Chi square = 4.25, p = .039). No other statistically significant association was found. Analysis of the association of TIL variables with MC as a continuous variable, amount of necrosis (present vs absent), and grade (I vs II+III) did not find any associations (Supplemental Tables S3–S5, respectively).
Total tumor-infiltrating lymphocytes (TILs) and their subpopulations in canine soft tissue sarcomas separated by diagnosis and histological grade. (a) Total TIL density, (b) CD20+ cell density, (c) CD3+ cell density, and (d) Treg proportion. The densities of total TILs, CD20+, and CD3+ cells are expressed as the number of lymphocytes/mm2, while Treg proportion is expressed as percentage of FoxP3+ cells/ CD3+ cells. PWT, perivascular wall tumor.
Discussion
Interest in TME and TILs in canine cancers has increased in recent years, mainly because of their therapeutic implications. Among canine neoplasms, the immunogenicity of STSs has not been extensively studied because STSs are traditionally considered poorly immunogenic.8,9,12,21 Recently, in human medicine, the paradigm of “immune-cold” STS has been revised and several studies have reported that TIL infiltration can be significant, varies according to sarcoma type, and can be correlated with prognosis and/or response to therapies.8,9,12,21
Defined guidelines are available for breast cancer 23 and have recently been applied also to canine mammary tumors. 19 This protocol relies on a semiquantitative assessment of TILs in the stromal compartment of the tumor on hematoxylin and eosin-stained slides 23 and can be easily translated to other epithelial tumors; however, guidelines for the assessment of TILs in STSs are lacking. Furthermore, the lack of a clearly discernible supporting stroma in STSs and the smaller number of TILs does not allow the application of this approach; thus, some of the studies have assessed TILs by manual count on immunohistochemically labeled slides.15,26 The majority of studies report the amount of TILs as cell numbers in 1 mm2, but the criteria for the choice of the area in which to perform the count and its size vary among studies.12,21
In veterinary medicine, the few studies investigating TME in canine STSs have not specified tumor types and have not provided specific criteria for the quantification of intraneoplastic immune cells.13,26 Therefore, in order to provide the most reliable data, we quantified TILs with the aid of image analysis on WSIs, to express the density of TILs and TIL subpopulations as the number of cells/mm2 and to characterize TILs in 5 specific STS histotypes, allowing for a comparison among different tumor types.
Overall, TILs were frequently observed in our caseload, with only a few cases being devoid of TILs, the so-called “immune-desert cases.” Nevertheless, high variability in total TIL density and type of lymphocytes were evident. This variability may be partially attributed to the striking differences in the immune response depending on the histotype, but not on the histologic grade, identified in this study. This finding is similar to what has been described in a large study of human STSs, where the grade did not correlate with TIL density and their subpopulation, with the exception of the number of regulatory T-cells (Tregs), which was positively associated with tumor grade. 12 The same study revealed that nontranslocation-associated sarcomas had a higher TIL density compared with translocation-associated sarcomas, likely because of their higher tumor mutational burden and probability of producing neo-antigens. 12
Canine myxosarcomas and PWTs had the highest total TIL densities, while fibrosarcomas had the lowest. Leiomyosarcomas and liposarcomas were generally TIL poor, even if the difference with the other tumor types was not statistically significant.
Interestingly, in human medicine, myxosarcoma is also the tumor type with the highest TIL density. 24 In canine myxosarcomas, TILs were characterized by a relatively high number of B-cells, with a B-cell/T-cell ratio of almost 1.6, and a median proportion of Tregs of 20% of T-cells. Increased B-cell infiltration has been found to be associated with prolonged survival and better prognosis in some human STSs. 21 Their role has not been fully elucidated, but it seems to be related to the presence of TLSs and their role in neo-antigen presentation, stimulating a cytotoxic T-cell response. 18 Hence, the finding of tertiary lymphoid structures is considered prognostically favorable. 21 On the contrary, other studies report that intratumoral TLSs and TAMs may sustain the immunosuppressive niche of STSs in tumor types where B-cells are concentrated in TLSs. 9 In our caseload, TLSs were present in only 2 myxosarcomas out of 16, and therefore, no conclusion can be drawn on the role of TLSs. Nevertheless, the overall TIL profile suggests the prevalence of a humoral immune response associated with an immune-suppressive Treg-rich TME. Further characterization of the myxosarcoma TME should be performed to better elucidate the role of B-cell infiltration. Notably, considering only the myxosarcoma group, a higher histologic grade was associated with higher TIL, B-cell, and Treg densities. This may suggest that the presence of B-cells and Tregs is associated with a poorer prognosis. Furthermore, it is interesting that myxosarcoma is the tumor type with the highest B-cell density, considering that it has been recently suggested to bear a more aggressive behavior and metastatic potential compared with the other canine STSs. 14 Unfortunately, no grade 3 myxosarcomas were present in our study population, and studies on a larger case series are needed to confirm this result.
In PWTs, TILs were characterized by a higher density of B- and T-cells and a lower Treg density compared with other STSs. Furthermore, the B-cell/T-cell ratio was significantly lower in PWTs compared with myxosarcomas, because of a higher proportion of T-cells. Thus, it seems that PWTs elicit a T-cell-rich/Treg-poor immune response that is likely to be cell mediated. Unfortunately, the lack of an anti-CD8 antibody working on formalin-fixed, paraffin-embedded canine tissues prevented further characterization of the T-cell component and did not allow confirmation of this hypothesis.
PWTs are the most common STS in the dog and are considered neoplasms with intermediate biological behavior, characterized by low metastatic and variable recurrence rates. 22 Recurrence of PWTs usually occurs in cases with tumor extension to the histologic tissue margins or histologic tumor-free margins smaller than 3 mm, but a variable proportion of tumors that extend to the tissue margins does not recur.2,10,11 While the recurrence of cases with clean margins has been explained with a false negative margin evaluation, that may be linked to the trimming protocol, the nonrecurrence of cases with infiltrated margins has been more difficult to explain. Based on our results, we hypothesize that the T-cell-rich/Treg-poor immune response may contribute to the control of the residual disease for these cases. This hypothesis is consistent with the finding, in human medicine, of an increased risk of local recurrence in STSs with high Treg density. 24 Furthermore, despite the lack of statistical significance, the Treg density and Treg proportion of T-cells were higher in grade 3 PWTs, similar to what has been described in human STSs. 12 Nevertheless, this finding needs further support considering the small number of grade 3 PWTs in this study.
Fibrosarcomas were the group of canine STSs with the lowest TIL density and were devoid of significant differences in TIL variables among the different grades. This is similar to what has been reported in a study of fibroblastic sarcoma in humans, where adult-type fibrosarcomas were the STS with the lowest CD4 and CD8-cell density, and myxosarcomas were the ones with the highest density. 8 The same study reported no differences in TIL density among different histologic grades, as occurred in our caseload. 8
Leiomyosarcomas had a similar TIL density to fibrosarcomas. This agrees with what has been described in human medicine since most soft tissue leiomyosarcomas are immune-low sarcomas. 21 In leiomyosarcomas, as in other human STSs, the presence of TLSs is a favorable prognostic feature. 21 Nevertheless, we did not find TLSs in our canine leiomyosarcoma series. Despite a low TIL density, the Treg proportion of T-cells in fibrosarcomas and leiomyosarcomas increased with the increase of histologic grade, as reported for human STSs. 12 On the contrary, liposarcomas showed a decreased Treg proportion of T-cells in cases with higher grades. This finding may deserve further investigation since liposarcoma is the only sarcoma type in our study with this trend. A possible reason for this discrepancy may be the peculiarity in the architecture of liposarcomas, which can be organized in lobules with supporting stroma as occurs in epithelial tumors. For this reason, the TIL evaluation protocol used for breast cancer,1,19 which assesses TIL density only in the supporting stroma, has also been successfully applied to human liposarcomas. 25 The specific role of stromal TILs should therefore be investigated to better elucidate liposarcoma TME.
Limits of the study include its retrospective nature, which does not allow for knowledge of possible presurgical treatments that could have affected the lymphocytic infiltrate in the tumors. Furthermore, the selection of cases diagnosed mainly based on histologic evaluation led to the exclusion of undifferentiated sarcomas, which could be an interesting topic for further investigations.
Summarizing, we characterized the TIL subpopulation density in canine STSs, finding differences among STS histotypes. Myxosarcomas and PWTs seem to be the most immunogenic canine STS types, which elicit very different immune responses: a B-cell/Treg-rich response in myxosarcomas and a T-cell-rich/Treg-poor response in PWTs. These two types of response may contribute to the higher propensity of myxosarcomas to metastasize 14 and the lack of recurrence in some PWTs with infiltrated margins.2,10,11
Further studies evaluating TAM density and the expression of the PD1-PDL1 axis may add useful information regarding the TME of canine STSs.
Supplemental Material
sj-pdf-1-vet-10.1177_03009858241300556 – Supplemental material for Tumor-infiltrating lymphocytes vary in different canine soft tissue sarcoma histological types
Supplemental material, sj-pdf-1-vet-10.1177_03009858241300556 for Tumor-infiltrating lymphocytes vary in different canine soft tissue sarcoma histological types by Giancarlo Avallone, Elena Brigandì, Chiara Tugnoli, Antonella Rigillo, Barbara Bacci and Paola Roccabianca in Veterinary Pathology
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
